diff --git a/contrib/other-builds/extract-mixed-syntax/.cproject b/contrib/other-builds/extract-mixed-syntax/.cproject new file mode 100644 index 000000000..b8f9cc79b --- /dev/null +++ b/contrib/other-builds/extract-mixed-syntax/.cproject @@ -0,0 +1,129 @@ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + diff --git a/contrib/other-builds/extract-mixed-syntax/.project b/contrib/other-builds/extract-mixed-syntax/.project new file mode 100644 index 000000000..b686d0cfd --- /dev/null +++ b/contrib/other-builds/extract-mixed-syntax/.project @@ -0,0 +1,219 @@ + + + extract-mixed-syntax + + + + + + org.eclipse.cdt.managedbuilder.core.genmakebuilder + clean,full,incremental, + + + + + org.eclipse.cdt.managedbuilder.core.ScannerConfigBuilder + full,incremental, + + + + + + org.eclipse.cdt.core.cnature + org.eclipse.cdt.core.ccnature + org.eclipse.cdt.managedbuilder.core.managedBuildNature + org.eclipse.cdt.managedbuilder.core.ScannerConfigNature + + + + AlignedSentence.cpp + 1 + PARENT-3-PROJECT_LOC/phrase-extract/extract-mixed-syntax/AlignedSentence.cpp + + + AlignedSentence.h + 1 + PARENT-3-PROJECT_LOC/phrase-extract/extract-mixed-syntax/AlignedSentence.h + + + AlignedSentenceSyntax.cpp + 1 + PARENT-3-PROJECT_LOC/phrase-extract/extract-mixed-syntax/AlignedSentenceSyntax.cpp + + + AlignedSentenceSyntax.h + 1 + PARENT-3-PROJECT_LOC/phrase-extract/extract-mixed-syntax/AlignedSentenceSyntax.h + + + ConsistentPhrase.cpp + 1 + PARENT-3-PROJECT_LOC/phrase-extract/extract-mixed-syntax/ConsistentPhrase.cpp + + + ConsistentPhrase.h + 1 + PARENT-3-PROJECT_LOC/phrase-extract/extract-mixed-syntax/ConsistentPhrase.h + + + ConsistentPhrases.cpp + 1 + PARENT-3-PROJECT_LOC/phrase-extract/extract-mixed-syntax/ConsistentPhrases.cpp + + + ConsistentPhrases.h + 1 + PARENT-3-PROJECT_LOC/phrase-extract/extract-mixed-syntax/ConsistentPhrases.h + + + InputFileStream.cpp + 1 + PARENT-3-PROJECT_LOC/phrase-extract/extract-mixed-syntax/InputFileStream.cpp + + + InputFileStream.h + 1 + PARENT-3-PROJECT_LOC/phrase-extract/extract-mixed-syntax/InputFileStream.h + + + Jamfile + 1 + PARENT-3-PROJECT_LOC/phrase-extract/extract-mixed-syntax/Jamfile + + + Main.cpp + 1 + PARENT-3-PROJECT_LOC/phrase-extract/extract-mixed-syntax/Main.cpp + + + Main.h + 1 + PARENT-3-PROJECT_LOC/phrase-extract/extract-mixed-syntax/Main.h + + + Makefile + 1 + PARENT-3-PROJECT_LOC/phrase-extract/extract-mixed-syntax/Makefile + + + NonTerm.cpp + 1 + PARENT-3-PROJECT_LOC/phrase-extract/extract-mixed-syntax/NonTerm.cpp + + + NonTerm.h + 1 + PARENT-3-PROJECT_LOC/phrase-extract/extract-mixed-syntax/NonTerm.h + + + OutputFileStream.cpp + 1 + PARENT-3-PROJECT_LOC/phrase-extract/extract-mixed-syntax/OutputFileStream.cpp + + + OutputFileStream.h + 1 + PARENT-3-PROJECT_LOC/phrase-extract/extract-mixed-syntax/OutputFileStream.h + + + Parameter.cpp + 1 + PARENT-3-PROJECT_LOC/phrase-extract/extract-mixed-syntax/Parameter.cpp + + + Parameter.h + 1 + PARENT-3-PROJECT_LOC/phrase-extract/extract-mixed-syntax/Parameter.h + + + Phrase.cpp + 1 + PARENT-3-PROJECT_LOC/phrase-extract/extract-mixed-syntax/Phrase.cpp + + + Phrase.h + 1 + PARENT-3-PROJECT_LOC/phrase-extract/extract-mixed-syntax/Phrase.h + + + Rule.cpp + 1 + PARENT-3-PROJECT_LOC/phrase-extract/extract-mixed-syntax/Rule.cpp + + + Rule.h + 1 + PARENT-3-PROJECT_LOC/phrase-extract/extract-mixed-syntax/Rule.h + + + RulePhrase.cpp + 1 + PARENT-3-PROJECT_LOC/phrase-extract/extract-mixed-syntax/RulePhrase.cpp + + + RulePhrase.h + 1 + PARENT-3-PROJECT_LOC/phrase-extract/extract-mixed-syntax/RulePhrase.h + + + RuleSymbol.cpp + 1 + PARENT-3-PROJECT_LOC/phrase-extract/extract-mixed-syntax/RuleSymbol.cpp + + + RuleSymbol.h + 1 + PARENT-3-PROJECT_LOC/phrase-extract/extract-mixed-syntax/RuleSymbol.h + + + Rules.cpp + 1 + PARENT-3-PROJECT_LOC/phrase-extract/extract-mixed-syntax/Rules.cpp + + + Rules.h + 1 + PARENT-3-PROJECT_LOC/phrase-extract/extract-mixed-syntax/Rules.h + + + SyntaxTree.cpp + 1 + PARENT-3-PROJECT_LOC/phrase-extract/extract-mixed-syntax/SyntaxTree.cpp + + + SyntaxTree.h + 1 + PARENT-3-PROJECT_LOC/phrase-extract/extract-mixed-syntax/SyntaxTree.h + + + Word.cpp + 1 + PARENT-3-PROJECT_LOC/phrase-extract/extract-mixed-syntax/Word.cpp + + + Word.h + 1 + PARENT-3-PROJECT_LOC/phrase-extract/extract-mixed-syntax/Word.h + + + gzfilebuf.h + 1 + PARENT-3-PROJECT_LOC/phrase-extract/extract-mixed-syntax/gzfilebuf.h + + + pugiconfig.hpp + 1 + PARENT-3-PROJECT_LOC/phrase-extract/extract-mixed-syntax/pugiconfig.hpp + + + pugixml.cpp + 1 + PARENT-3-PROJECT_LOC/phrase-extract/extract-mixed-syntax/pugixml.cpp + + + pugixml.hpp + 1 + PARENT-3-PROJECT_LOC/phrase-extract/extract-mixed-syntax/pugixml.hpp + + + diff --git a/phrase-extract/extract-mixed-syntax/AlignedSentence.cpp b/phrase-extract/extract-mixed-syntax/AlignedSentence.cpp deleted file mode 100644 index 082878c00..000000000 --- a/phrase-extract/extract-mixed-syntax/AlignedSentence.cpp +++ /dev/null @@ -1,194 +0,0 @@ -/* - * AlignedSentence.cpp - * - * Created on: 18 Feb 2014 - * Author: s0565741 - */ - -#include -#include "moses/Util.h" -#include "AlignedSentence.h" -#include "Parameter.h" - -using namespace std; - - -///////////////////////////////////////////////////////////////////////////////// -AlignedSentence::AlignedSentence(int lineNum, - const std::string &source, - const std::string &target, - const std::string &alignment) -:m_lineNum(lineNum) -{ - PopulateWordVec(m_source, source); - PopulateWordVec(m_target, target); - PopulateAlignment(alignment); -} - -AlignedSentence::~AlignedSentence() { - Moses::RemoveAllInColl(m_source); - Moses::RemoveAllInColl(m_target); -} - -void AlignedSentence::PopulateWordVec(Phrase &vec, const std::string &line) -{ - std::vector toks; - Moses::Tokenize(toks, line); - - vec.resize(toks.size()); - for (size_t i = 0; i < vec.size(); ++i) { - const string &tok = toks[i]; - Word *word = new Word(i, tok); - vec[i] = word; - } -} - -void AlignedSentence::PopulateAlignment(const std::string &line) -{ - vector alignStr; - Moses::Tokenize(alignStr, line); - - for (size_t i = 0; i < alignStr.size(); ++i) { - vector alignPair; - Moses::Tokenize(alignPair, alignStr[i], "-"); - assert(alignPair.size() == 2); - - int sourcePos = alignPair[0]; - int targetPos = alignPair[1]; - - if (sourcePos >= m_source.size()) { - cerr << "ERROR1:AlignedSentence=" << Debug() << endl; - cerr << "m_source=" << m_source.size() << endl; - abort(); - } - assert(sourcePos < m_source.size()); - assert(targetPos < m_target.size()); - Word *sourceWord = m_source[sourcePos]; - Word *targetWord = m_target[targetPos]; - - sourceWord->AddAlignment(targetWord); - targetWord->AddAlignment(sourceWord); - } -} - -std::string AlignedSentence::Debug() const -{ - stringstream out; - out << "m_lineNum:"; - out << m_lineNum; - out << endl; - - out << "m_source:"; - out << m_source.Debug(); - out << endl; - - out << "m_target:"; - out << m_target.Debug(); - out << endl; - - out << "consistent phrases:" << endl; - out << m_consistentPhrases.Debug(); - out << endl; - - return out.str(); -} - -std::vector AlignedSentence::GetSourceAlignmentCount() const -{ - vector ret(m_source.size()); - - for (size_t i = 0; i < m_source.size(); ++i) { - const Word &word = *m_source[i]; - ret[i] = word.GetAlignmentIndex().size(); - } - return ret; -} - -void AlignedSentence::Create(const Parameter ¶ms) -{ - CreateConsistentPhrases(params); - m_consistentPhrases.AddHieroNonTerms(params); -} - -void AlignedSentence::CreateConsistentPhrases(const Parameter ¶ms) -{ - int countT = m_target.size(); - int countS = m_source.size(); - - m_consistentPhrases.Initialize(countS); - - // check alignments for target phrase startT...endT - for(int lengthT=1; - lengthT <= params.maxSpan && lengthT <= countT; - lengthT++) { - for(int startT=0; startT < countT-(lengthT-1); startT++) { - - // that's nice to have - int endT = startT + lengthT - 1; - - // find find aligned source words - // first: find minimum and maximum source word - int minS = 9999; - int maxS = -1; - vector< int > usedS = GetSourceAlignmentCount(); - for(int ti=startT; ti<=endT; ti++) { - const Word &word = *m_target[ti]; - const std::set &alignment = word.GetAlignmentIndex(); - - std::set::const_iterator iterAlign; - for(iterAlign = alignment.begin(); iterAlign != alignment.end(); ++iterAlign) { - int si = *iterAlign; - if (simaxS) { - maxS = si; - } - usedS[ si ]--; - } - } - - // unaligned phrases are not allowed - if( maxS == -1 ) - continue; - - // source phrase has to be within limits - size_t width = maxS - minS + 1; - - if( width < params.minSpan ) - continue; - - if( width > params.maxSpan ) - continue; - - // check if source words are aligned to out of bound target words - bool out_of_bounds = false; - for(int si=minS; si<=maxS && !out_of_bounds; si++) - if (usedS[si]>0) { - out_of_bounds = true; - } - - // if out of bound, you gotta go - if (out_of_bounds) - continue; - - // done with all the checks, lets go over all consistent phrase pairs - // start point of source phrase may retreat over unaligned - for(int startS=minS; - (startS>=0 && - startS>maxS - params.maxSpan && // within length limit - (startS==minS || m_source[startS]->GetAlignment().size()==0)); // unaligned - startS--) { - // end point of source phrase may advance over unaligned - for(int endS=maxS; - (endSGetAlignment().size()==0)); // unaligned - endS++) { - - // take note that this is a valid phrase alignment - m_consistentPhrases.Add(startS, endS, startT, endT, params); - } - } - } - } -} diff --git a/phrase-extract/extract-mixed-syntax/AlignedSentence.h b/phrase-extract/extract-mixed-syntax/AlignedSentence.h deleted file mode 100644 index 915bdf90c..000000000 --- a/phrase-extract/extract-mixed-syntax/AlignedSentence.h +++ /dev/null @@ -1,51 +0,0 @@ -/* - * AlignedSentence.h - * - * Created on: 18 Feb 2014 - * Author: s0565741 - */ -#pragma once - -#include -#include -#include "ConsistentPhrases.h" -#include "Phrase.h" -#include "moses/TypeDef.h" - -class Parameter; - -class AlignedSentence { -public: - AlignedSentence(int lineNum) - :m_lineNum(lineNum) - {} - - AlignedSentence(int lineNum, - const std::string &source, - const std::string &target, - const std::string &alignment); - virtual ~AlignedSentence(); - virtual void Create(const Parameter ¶ms); - - const Phrase &GetPhrase(Moses::FactorDirection direction) const - { return (direction == Moses::Input) ? m_source : m_target; } - - const ConsistentPhrases &GetConsistentPhrases() const - { return m_consistentPhrases; } - - virtual std::string Debug() const; - - int m_lineNum; -protected: - Phrase m_source, m_target; - ConsistentPhrases m_consistentPhrases; - - void CreateConsistentPhrases(const Parameter ¶ms); - void PopulateWordVec(Phrase &vec, const std::string &line); - - // m_source and m_target MUST be populated before calling this - void PopulateAlignment(const std::string &line); - std::vector GetSourceAlignmentCount() const; -}; - - diff --git a/phrase-extract/extract-mixed-syntax/AlignedSentenceSyntax.cpp b/phrase-extract/extract-mixed-syntax/AlignedSentenceSyntax.cpp deleted file mode 100644 index 3d63ed044..000000000 --- a/phrase-extract/extract-mixed-syntax/AlignedSentenceSyntax.cpp +++ /dev/null @@ -1,183 +0,0 @@ -/* - * AlignedSentenceSyntax.cpp - * - * Created on: 26 Feb 2014 - * Author: hieu - */ - -#include "AlignedSentenceSyntax.h" -#include "Parameter.h" -#include "pugixml.hpp" -#include "moses/Util.h" - -using namespace std; - -AlignedSentenceSyntax::AlignedSentenceSyntax(int lineNum, - const std::string &source, - const std::string &target, - const std::string &alignment) -:AlignedSentence(lineNum) -,m_sourceStr(source) -,m_targetStr(target) -,m_alignmentStr(alignment) -{ -} - -AlignedSentenceSyntax::~AlignedSentenceSyntax() { - // TODO Auto-generated destructor stub -} - -void AlignedSentenceSyntax::Populate(bool isSyntax, int mixedSyntaxType, const Parameter ¶ms, - string line, Phrase &phrase, SyntaxTree &tree) -{ - // parse source and target string - if (isSyntax) { - line = "" + line + ""; - XMLParse(phrase, tree, line, params); - - if (mixedSyntaxType != 0) { - // mixed syntax. Always add [X] where there isn't 1 - tree.SetHieroLabel(params.hieroNonTerm); - if (mixedSyntaxType == 2) { - tree.AddToAll(params.hieroNonTerm); - } - } - } - else { - PopulateWordVec(phrase, line); - tree.SetHieroLabel(params.hieroNonTerm); - } - -} - -void AlignedSentenceSyntax::Create(const Parameter ¶ms) -{ - Populate(params.sourceSyntax, params.mixedSyntaxType, params, m_sourceStr, - m_source, m_sourceTree); - Populate(params.targetSyntax, params.mixedSyntaxType, params, m_targetStr, - m_target, m_targetTree); - - PopulateAlignment(m_alignmentStr); - CreateConsistentPhrases(params); - - // create labels - CreateNonTerms(); -} - -void Escape(string &text) -{ - text = Moses::Replace(text, "&", "&"); - text = Moses::Replace(text, "|", "|"); - text = Moses::Replace(text, "<", "<"); - text = Moses::Replace(text, ">", ">"); - text = Moses::Replace(text, "'", "'"); - text = Moses::Replace(text, "\"", """); - text = Moses::Replace(text, "[", "["); - text = Moses::Replace(text, "]", "]"); - -} - -void AlignedSentenceSyntax::XMLParse(Phrase &output, - SyntaxTree &tree, - const pugi::xml_node &parentNode, - const Parameter ¶ms) -{ - int childNum = 0; - for (pugi::xml_node childNode = parentNode.first_child(); childNode; childNode = childNode.next_sibling()) - { - string nodeName = childNode.name(); - - // span label - string label; - int startPos = output.size(); - - if (!nodeName.empty()) { - pugi::xml_attribute attribute = childNode.attribute("label"); - label = attribute.as_string(); - - // recursively call this function. For proper recursive trees - XMLParse(output, tree, childNode, params); - } - - - - // fill phrase vector - string text = childNode.value(); - Escape(text); - //cerr << childNum << " " << label << "=" << text << endl; - - std::vector toks; - Moses::Tokenize(toks, text); - - for (size_t i = 0; i < toks.size(); ++i) { - const string &tok = toks[i]; - Word *word = new Word(output.size(), tok); - output.push_back(word); - } - - // is it a labelled span? - int endPos = output.size() - 1; - - // fill syntax labels - if (!label.empty()) { - label = "[" + label + "]"; - tree.Add(startPos, endPos, label, params); - } - - ++childNum; - } - -} - -void AlignedSentenceSyntax::XMLParse(Phrase &output, - SyntaxTree &tree, - const std::string input, - const Parameter ¶ms) -{ - pugi::xml_document doc; - pugi::xml_parse_result result = doc.load(input.c_str(), - pugi::parse_default | pugi::parse_comments); - - pugi::xml_node topNode = doc.child("xml"); - XMLParse(output, tree, topNode, params); -} - -void AlignedSentenceSyntax::CreateNonTerms() -{ - for (int sourceStart = 0; sourceStart < m_source.size(); ++sourceStart) { - for (int sourceEnd = sourceStart; sourceEnd < m_source.size(); ++sourceEnd) { - ConsistentPhrases::Coll &coll = m_consistentPhrases.GetColl(sourceStart, sourceEnd); - const SyntaxTree::Labels &sourceLabels = m_sourceTree.Find(sourceStart, sourceEnd); - - ConsistentPhrases::Coll::iterator iter; - for (iter = coll.begin(); iter != coll.end(); ++iter) { - ConsistentPhrase &cp = **iter; - - int targetStart = cp.corners[2]; - int targetEnd = cp.corners[3]; - const SyntaxTree::Labels &targetLabels = m_targetTree.Find(targetStart, targetEnd); - - CreateNonTerms(cp, sourceLabels, targetLabels); - } - } - } - -} - -void AlignedSentenceSyntax::CreateNonTerms(ConsistentPhrase &cp, - const SyntaxTree::Labels &sourceLabels, - const SyntaxTree::Labels &targetLabels) -{ - SyntaxTree::Labels::const_iterator iterSource; - for (iterSource = sourceLabels.begin(); iterSource != sourceLabels.end(); ++iterSource) { - const string &sourceLabel = *iterSource; - - SyntaxTree::Labels::const_iterator iterTarget; - for (iterTarget = targetLabels.begin(); iterTarget != targetLabels.end(); ++iterTarget) { - const string &targetLabel = *iterTarget; - cp.AddNonTerms(sourceLabel, targetLabel); - } - } -} - - diff --git a/phrase-extract/extract-mixed-syntax/AlignedSentenceSyntax.h b/phrase-extract/extract-mixed-syntax/AlignedSentenceSyntax.h deleted file mode 100644 index 2e9431996..000000000 --- a/phrase-extract/extract-mixed-syntax/AlignedSentenceSyntax.h +++ /dev/null @@ -1,46 +0,0 @@ -/* - * AlignedSentenceSyntax.h - * - * Created on: 26 Feb 2014 - * Author: hieu - */ - -#pragma once - -#include "AlignedSentence.h" -#include "SyntaxTree.h" -#include "pugixml.hpp" - -class AlignedSentenceSyntax : public AlignedSentence -{ -public: - AlignedSentenceSyntax(int lineNum, - const std::string &source, - const std::string &target, - const std::string &alignment); - virtual ~AlignedSentenceSyntax(); - - void Create(const Parameter ¶ms); - - //virtual std::string Debug() const; -protected: - std::string m_sourceStr, m_targetStr, m_alignmentStr; - SyntaxTree m_sourceTree, m_targetTree; - - void XMLParse(Phrase &output, - SyntaxTree &tree, - const std::string input, - const Parameter ¶ms); - void XMLParse(Phrase &output, - SyntaxTree &tree, - const pugi::xml_node &parentNode, - const Parameter ¶ms); - void CreateNonTerms(); - void CreateNonTerms(ConsistentPhrase &cp, - const SyntaxTree::Labels &sourceLabels, - const SyntaxTree::Labels &targetLabels); - void Populate(bool isSyntax, int mixedSyntaxType, const Parameter ¶ms, - std::string line, Phrase &phrase, SyntaxTree &tree); - -}; - diff --git a/phrase-extract/extract-mixed-syntax/ConsistentPhrase.cpp b/phrase-extract/extract-mixed-syntax/ConsistentPhrase.cpp deleted file mode 100644 index bb913da5a..000000000 --- a/phrase-extract/extract-mixed-syntax/ConsistentPhrase.cpp +++ /dev/null @@ -1,66 +0,0 @@ -/* - * ConsistentPhrase.cpp - * - * Created on: 20 Feb 2014 - * Author: hieu - */ -#include -#include "ConsistentPhrase.h" -#include "Word.h" -#include "NonTerm.h" -#include "Parameter.h" - -using namespace std; - -ConsistentPhrase::ConsistentPhrase( - int sourceStart, int sourceEnd, - int targetStart, int targetEnd, - const Parameter ¶ms) -:corners(4) -,m_hieroNonTerm(*this, params.hieroNonTerm, params.hieroNonTerm) -{ - corners[0] = sourceStart; - corners[1] = sourceEnd; - corners[2] = targetStart; - corners[3] = targetEnd; -} - -ConsistentPhrase::~ConsistentPhrase() { - // TODO Auto-generated destructor stub -} - -bool ConsistentPhrase::operator<(const ConsistentPhrase &other) const -{ - return corners < other.corners; -} - -void ConsistentPhrase::AddNonTerms(const std::string &source, - const std::string &target) -{ - m_nonTerms.push_back(NonTerm(*this, source, target)); -} - -bool ConsistentPhrase::TargetOverlap(const ConsistentPhrase &other) const -{ - if ( other.corners[3] < corners[2] || other.corners[2] > corners[3]) - return false; - - return true; -} - -std::string ConsistentPhrase::Debug() const -{ - stringstream out; - out << "[" << corners[0] << "-" << corners[1] - << "][" << corners[2] << "-" << corners[3] << "]"; - - out << "NT:"; - for (size_t i = 0; i < m_nonTerms.size(); ++i) { - const NonTerm &nonTerm = m_nonTerms[i]; - out << nonTerm.GetLabel(Moses::Input) << ":" << nonTerm.GetLabel(Moses::Output); - } - - return out.str(); -} - - diff --git a/phrase-extract/extract-mixed-syntax/ConsistentPhrase.h b/phrase-extract/extract-mixed-syntax/ConsistentPhrase.h deleted file mode 100644 index 865b4386f..000000000 --- a/phrase-extract/extract-mixed-syntax/ConsistentPhrase.h +++ /dev/null @@ -1,51 +0,0 @@ -/* - * ConsistentPhrase.h - * - * Created on: 20 Feb 2014 - * Author: hieu - */ - -#pragma once - -#include -#include -#include -#include "moses/TypeDef.h" -#include "NonTerm.h" - -class ConsistentPhrase -{ -public: - typedef std::vector NonTerms; - - std::vector corners; - - ConsistentPhrase(const ConsistentPhrase ©); // do not implement - ConsistentPhrase(int sourceStart, int sourceEnd, - int targetStart, int targetEnd, - const Parameter ¶ms); - - virtual ~ConsistentPhrase(); - - int GetWidth(Moses::FactorDirection direction) const - { return (direction == Moses::Input) ? corners[1] - corners[0] + 1 : corners[3] - corners[2] + 1; } - - - void AddNonTerms(const std::string &source, - const std::string &target); - const NonTerms &GetNonTerms() const - { return m_nonTerms;} - const NonTerm &GetHieroNonTerm() const - { return m_hieroNonTerm;} - - bool TargetOverlap(const ConsistentPhrase &other) const; - - bool operator<(const ConsistentPhrase &other) const; - - std::string Debug() const; - -protected: - NonTerms m_nonTerms; - NonTerm m_hieroNonTerm; -}; - diff --git a/phrase-extract/extract-mixed-syntax/ConsistentPhrases.cpp b/phrase-extract/extract-mixed-syntax/ConsistentPhrases.cpp deleted file mode 100644 index 8978c88fa..000000000 --- a/phrase-extract/extract-mixed-syntax/ConsistentPhrases.cpp +++ /dev/null @@ -1,103 +0,0 @@ -/* - * ConsistentPhrases.cpp - * - * Created on: 20 Feb 2014 - * Author: hieu - */ -#include -#include -#include "ConsistentPhrases.h" -#include "NonTerm.h" -#include "Parameter.h" -#include "moses/Util.h" - -using namespace std; - -ConsistentPhrases::ConsistentPhrases() -{ -} - -ConsistentPhrases::~ConsistentPhrases() { - for (int start = 0; start < m_coll.size(); ++start) { - std::vector &allSourceStart = m_coll[start]; - - for (int size = 0; size < allSourceStart.size(); ++size) { - Coll &coll = allSourceStart[size]; - Moses::RemoveAllInColl(coll); - } - } -} - -void ConsistentPhrases::Initialize(size_t size) -{ - m_coll.resize(size); - - for (size_t sourceStart = 0; sourceStart < size; ++sourceStart) { - std::vector &allSourceStart = m_coll[sourceStart]; - allSourceStart.resize(size - sourceStart); - } -} - -void ConsistentPhrases::Add(int sourceStart, int sourceEnd, - int targetStart, int targetEnd, - const Parameter ¶ms) -{ - Coll &coll = m_coll[sourceStart][sourceEnd - sourceStart]; - ConsistentPhrase *cp = new ConsistentPhrase(sourceStart, sourceEnd, - targetStart, targetEnd, - params); - - pair inserted = coll.insert(cp); - assert(inserted.second); -} - -const ConsistentPhrases::Coll &ConsistentPhrases::GetColl(int sourceStart, int sourceEnd) const -{ - const std::vector &allSourceStart = m_coll[sourceStart]; - const Coll &ret = allSourceStart[sourceEnd - sourceStart]; - return ret; -} - -ConsistentPhrases::Coll &ConsistentPhrases::GetColl(int sourceStart, int sourceEnd) -{ - std::vector &allSourceStart = m_coll[sourceStart]; - Coll &ret = allSourceStart[sourceEnd - sourceStart]; - return ret; -} - -std::string ConsistentPhrases::Debug() const -{ - std::stringstream out; - for (int start = 0; start < m_coll.size(); ++start) { - const std::vector &allSourceStart = m_coll[start]; - - for (int size = 0; size < allSourceStart.size(); ++size) { - const Coll &coll = allSourceStart[size]; - - Coll::const_iterator iter; - for (iter = coll.begin(); iter != coll.end(); ++iter) { - const ConsistentPhrase &consistentPhrase = **iter; - out << consistentPhrase.Debug() << endl; - } - } - } - - return out.str(); -} - -void ConsistentPhrases::AddHieroNonTerms(const Parameter ¶ms) -{ - // add [X] labels everywhere - for (int i = 0; i < m_coll.size(); ++i) { - vector &inner = m_coll[i]; - for (int j = 0; j < inner.size(); ++j) { - ConsistentPhrases::Coll &coll = inner[j]; - ConsistentPhrases::Coll::iterator iter; - for (iter = coll.begin(); iter != coll.end(); ++iter) { - ConsistentPhrase &cp = **iter; - cp.AddNonTerms(params.hieroNonTerm, params.hieroNonTerm); - } - } - } -} - diff --git a/phrase-extract/extract-mixed-syntax/ConsistentPhrases.h b/phrase-extract/extract-mixed-syntax/ConsistentPhrases.h deleted file mode 100644 index 3daf6b7ff..000000000 --- a/phrase-extract/extract-mixed-syntax/ConsistentPhrases.h +++ /dev/null @@ -1,40 +0,0 @@ -/* - * ConsistentPhrases.h - * - * Created on: 20 Feb 2014 - * Author: hieu - */ -#pragma once - -#include -#include -#include -#include "ConsistentPhrase.h" - -class Word; -class Parameter; - -class ConsistentPhrases { -public: - typedef std::set Coll; - - ConsistentPhrases(); - virtual ~ConsistentPhrases(); - - void Initialize(size_t size); - - void Add(int sourceStart, int sourceEnd, - int targetStart, int targetEnd, - const Parameter ¶ms); - - void AddHieroNonTerms(const Parameter ¶ms); - - const Coll &GetColl(int sourceStart, int sourceEnd) const; - Coll &GetColl(int sourceStart, int sourceEnd); - - std::string Debug() const; - -protected: - std::vector< std::vector > m_coll; -}; - diff --git a/phrase-extract/extract-mixed-syntax/InputFileStream.cpp b/phrase-extract/extract-mixed-syntax/InputFileStream.cpp deleted file mode 100644 index b52d1f920..000000000 --- a/phrase-extract/extract-mixed-syntax/InputFileStream.cpp +++ /dev/null @@ -1,62 +0,0 @@ -// $Id: InputFileStream.cpp 2780 2010-01-29 17:11:17Z bojar $ - -/*********************************************************************** - Moses - factored phrase-based language decoder - Copyright (C) 2006 University of Edinburgh - - This library is free software; you can redistribute it and/or - modify it under the terms of the GNU Lesser General Public - License as published by the Free Software Foundation; either - version 2.1 of the License, or (at your option) any later version. - - This library is distributed in the hope that it will be useful, - but WITHOUT ANY WARRANTY; without even the implied warranty of - MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU - Lesser General Public License for more details. - - You should have received a copy of the GNU Lesser General Public - License along with this library; if not, write to the Free Software - Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA - ***********************************************************************/ - -#include "InputFileStream.h" -#include "gzfilebuf.h" -#include - -using namespace std; - -namespace Moses -{ - InputFileStream::InputFileStream(const std::string &filePath) - : std::istream(NULL) - , m_streambuf(NULL) - { - if (filePath.size() > 3 && - filePath.substr(filePath.size() - 3, 3) == ".gz") - { - m_streambuf = new gzfilebuf(filePath.c_str()); - } else { - std::filebuf* fb = new std::filebuf(); - fb = fb->open(filePath.c_str(), std::ios::in); - if (! fb) { - cerr << "Can't read " << filePath.c_str() << endl; - exit(1); - } - m_streambuf = fb; - } - this->init(m_streambuf); - } - - InputFileStream::~InputFileStream() - { - delete m_streambuf; - m_streambuf = NULL; - } - - void InputFileStream::Close() - { - } - - -} - diff --git a/phrase-extract/extract-mixed-syntax/InputFileStream.h b/phrase-extract/extract-mixed-syntax/InputFileStream.h deleted file mode 100644 index f10ec2164..000000000 --- a/phrase-extract/extract-mixed-syntax/InputFileStream.h +++ /dev/null @@ -1,48 +0,0 @@ -// $Id: InputFileStream.h 2939 2010-02-24 11:15:44Z jfouet $ - -/*********************************************************************** - Moses - factored phrase-based language decoder - Copyright (C) 2006 University of Edinburgh - - This library is free software; you can redistribute it and/or - modify it under the terms of the GNU Lesser General Public - License as published by the Free Software Foundation; either - version 2.1 of the License, or (at your option) any later version. - - This library is distributed in the hope that it will be useful, - but WITHOUT ANY WARRANTY; without even the implied warranty of - MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU - Lesser General Public License for more details. - - You should have received a copy of the GNU Lesser General Public - License along with this library; if not, write to the Free Software - Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA - ***********************************************************************/ - -#ifndef moses_InputFileStream_h -#define moses_InputFileStream_h - -#include -#include -#include - -namespace Moses -{ - - /** Used in place of std::istream, can read zipped files if it ends in .gz - */ - class InputFileStream : public std::istream - { - protected: - std::streambuf *m_streambuf; - public: - - InputFileStream(const std::string &filePath); - ~InputFileStream(); - - void Close(); - }; - -} - -#endif diff --git a/phrase-extract/extract-mixed-syntax/Jamfile b/phrase-extract/extract-mixed-syntax/Jamfile deleted file mode 100644 index 8072a54b0..000000000 --- a/phrase-extract/extract-mixed-syntax/Jamfile +++ /dev/null @@ -1,2 +0,0 @@ -exe extract-mixed-syntax : Main.cpp AlignedSentence.cpp AlignedSentenceSyntax.cpp ConsistentPhrase.cpp ConsistentPhrases.cpp NonTerm.cpp OutputFileStream.cpp Parameter.cpp Phrase.cpp pugixml.cpp Rule.cpp RulePhrase.cpp Rules.cpp RuleSymbol.cpp SyntaxTree.cpp Word.cpp ..//..//z ..//..//boost_iostreams ..//..//boost_program_options ..//..//moses ; - diff --git a/phrase-extract/extract-mixed-syntax/Main.cpp b/phrase-extract/extract-mixed-syntax/Main.cpp deleted file mode 100644 index 2fcb06a9b..000000000 --- a/phrase-extract/extract-mixed-syntax/Main.cpp +++ /dev/null @@ -1,203 +0,0 @@ -#include -#include -#include - -#include "Main.h" -#include "InputFileStream.h" -#include "OutputFileStream.h" -#include "AlignedSentence.h" -#include "AlignedSentenceSyntax.h" -#include "Parameter.h" -#include "Rules.h" - -using namespace std; - -bool g_debug = false; - -int main(int argc, char** argv) -{ - cerr << "Starting" << endl; - - Parameter params; - - namespace po = boost::program_options; - po::options_description desc("Options"); - desc.add_options() - ("help", "Print help messages") - ("MaxSpan", po::value()->default_value(params.maxSpan), "Max (source) span of a rule. ie. number of words in the source") - ("MinSpan", po::value()->default_value(params.minSpan), "Min (source) span of a rule.") - ("GlueGrammar", po::value()->default_value(params.gluePath), "Output glue grammar to here") - ("SentenceOffset", po::value()->default_value(params.sentenceOffset), "Starting sentence id. Not used") - ("GZOutput", "Compress extract files") - ("MaxNonTerm", po::value()->default_value(params.maxNonTerm), "Maximum number of non-terms allowed per rule") - ("MaxHieroNonTerm", po::value()->default_value(params.maxHieroNonTerm), "Maximum number of Hiero non-term. Usually, --MaxNonTerm is the normal constraint") - ("MinHoleSource", po::value()->default_value(params.minHoleSource), "Minimum source span for a non-term.") - - ("SourceSyntax", "Source sentence is a parse tree") - ("TargetSyntax", "Target sentence is a parse tree") - ("MixedSyntaxType", po::value()->default_value(params.mixedSyntaxType), "Hieu's Mixed syntax type. 0(default)=no mixed syntax, 1=add [X] only if no syntactic label. 2=add [X] everywhere") - ("MultiLabel", po::value()->default_value(params.multiLabel), "What to do with multiple labels on the same span. 0(default)=keep them all, 1=keep only top-most, 2=keep only bottom-most") - ("HieroSourceLHS", "Always use Hiero source LHS? Default = 0") - ("MaxSpanFreeNonTermSource", po::value()->default_value(params.maxSpanFreeNonTermSource), "Max number of words covered by beginning/end NT. Default = 0 (no limit)") - ("NoNieceTerminal", "Don't extract rule if 1 of the non-term covers the same word as 1 of the terminals") - ("MaxScope", po::value()->default_value(params.maxScope), "maximum scope (see Hopkins and Langmead (2010)). Default is HIGH") - ("MinScope", po::value()->default_value(params.minScope), "min scope.") - - ("SpanLength", "Property - span length of RHS each non-term") - - ("NonTermContext", "Property - (source) left and right, inside and outside words of each non-term ") - ("NonTermContextTarget", "Property - (target) left and right, inside and outside words of each non-term") - ("NonTermContextFactor", po::value()->default_value(params.nonTermContextFactor), "Factor to use for non-term context property.") - - ("NumSourceFactors", po::value()->default_value(params.numSourceFactors), "Number of source factors.") - ("NumTargetFactors", po::value()->default_value(params.numTargetFactors), "Number of target factors.") - - ("HieroNonTerm", po::value()->default_value(params.hieroNonTerm), "Hiero non-terminal label, including bracket") - ("ScopeSpan", po::value()->default_value(params.scopeSpanStr), "Min and max span for rules of each scope. Format is min,max:min,max...") - - ("NonTermConsecSource", "Allow consecutive non-terms on the source side"); - - - po::variables_map vm; - try - { - po::store(po::parse_command_line(argc, argv, desc), - vm); // can throw - - /** --help option - */ - if ( vm.count("help") || argc < 5 ) - { - std::cout << argv[0] << " target source alignment [options...]" << std::endl - << desc << std::endl; - return EXIT_SUCCESS; - } - - po::notify(vm); // throws on error, so do after help in case - // there are any problems - } - catch(po::error& e) - { - std::cerr << "ERROR: " << e.what() << std::endl << std::endl; - std::cerr << desc << std::endl; - return EXIT_FAILURE; - } - - if (vm.count("MaxSpan")) params.maxSpan = vm["MaxSpan"].as(); - if (vm.count("MinSpan")) params.minSpan = vm["MinSpan"].as(); - if (vm.count("GZOutput")) params.gzOutput = true; - if (vm.count("GlueGrammar")) params.gluePath = vm["GlueGrammar"].as(); - if (vm.count("SentenceOffset")) params.sentenceOffset = vm["SentenceOffset"].as(); - if (vm.count("MaxNonTerm")) params.maxNonTerm = vm["MaxNonTerm"].as(); - if (vm.count("MaxHieroNonTerm")) params.maxHieroNonTerm = vm["MaxHieroNonTerm"].as(); - if (vm.count("MinHoleSource")) params.minHoleSource = vm["MinHoleSource"].as(); - - if (vm.count("SourceSyntax")) params.sourceSyntax = true; - if (vm.count("TargetSyntax")) params.targetSyntax = true; - if (vm.count("MixedSyntaxType")) params.mixedSyntaxType = vm["MixedSyntaxType"].as(); - if (vm.count("MultiLabel")) params.multiLabel = vm["MultiLabel"].as(); - if (vm.count("HieroSourceLHS")) params.hieroSourceLHS = true; - if (vm.count("MaxSpanFreeNonTermSource")) params.maxSpanFreeNonTermSource = vm["MaxSpanFreeNonTermSource"].as(); - if (vm.count("NoNieceTerminal")) params.nieceTerminal = false; - if (vm.count("MaxScope")) params.maxScope = vm["MaxScope"].as(); - if (vm.count("MinScope")) params.minScope = vm["MinScope"].as(); - - // properties - if (vm.count("SpanLength")) params.spanLength = true; - if (vm.count("NonTermContext")) params.nonTermContext = true; - if (vm.count("NonTermContextTarget")) params.nonTermContextTarget = true; - if (vm.count("NonTermContextFactor")) params.nonTermContextFactor = vm["NonTermContextFactor"].as(); - - if (vm.count("NumSourceFactors")) params.numSourceFactors = vm["NumSourceFactors"].as(); - if (vm.count("NumTargetFactors")) params.numTargetFactors = vm["NumTargetFactors"].as(); - - if (vm.count("HieroNonTerm")) params.hieroNonTerm = vm["HieroNonTerm"].as(); - if (vm.count("ScopeSpan")) { - params.SetScopeSpan(vm["ScopeSpan"].as()); - } - - if (vm.count("NonTermConsecSource")) params.nonTermConsecSource = true; - - // input files; - string pathTarget = argv[1]; - string pathSource = argv[2]; - string pathAlignment = argv[3]; - - string pathExtract = argv[4]; - string pathExtractInv = pathExtract + ".inv"; - if (params.gzOutput) { - pathExtract += ".gz"; - pathExtractInv += ".gz"; - } - - Moses::InputFileStream strmTarget(pathTarget); - Moses::InputFileStream strmSource(pathSource); - Moses::InputFileStream strmAlignment(pathAlignment); - Moses::OutputFileStream extractFile(pathExtract); - Moses::OutputFileStream extractInvFile(pathExtractInv); - - - // MAIN LOOP - int lineNum = 1; - string lineTarget, lineSource, lineAlignment; - while (getline(strmTarget, lineTarget)) { - if (lineNum % 10000 == 0) { - cerr << lineNum << " "; - } - - bool success; - success = getline(strmSource, lineSource); - if (!success) { - throw "Couldn't read source"; - } - success = getline(strmAlignment, lineAlignment); - if (!success) { - throw "Couldn't read alignment"; - } - - /* - cerr << "lineTarget=" << lineTarget << endl; - cerr << "lineSource=" << lineSource << endl; - cerr << "lineAlignment=" << lineAlignment << endl; - */ - - AlignedSentence *alignedSentence; - - if (params.sourceSyntax || params.targetSyntax) { - alignedSentence = new AlignedSentenceSyntax(lineNum, lineSource, lineTarget, lineAlignment); - } - else { - alignedSentence = new AlignedSentence(lineNum, lineSource, lineTarget, lineAlignment); - } - - alignedSentence->Create(params); - //cerr << alignedSentence->Debug(); - - Rules rules(*alignedSentence); - rules.Extend(params); - rules.Consolidate(params); - //cerr << rules.Debug(); - - rules.Output(extractFile, true, params); - rules.Output(extractInvFile, false, params); - - delete alignedSentence; - - ++lineNum; - } - - if (!params.gluePath.empty()) { - Moses::OutputFileStream glueFile(params.gluePath); - CreateGlueGrammar(glueFile); - } - - cerr << "Finished" << endl; -} - -void CreateGlueGrammar(Moses::OutputFileStream &glueFile) -{ - glueFile << " [X] ||| [S] ||| 1 ||| ||| 0" << endl - << "[X][S] [X] ||| [X][S] [S] ||| 1 ||| 0-0 ||| 0" << endl - << "[X][S] [X][X] [X] ||| [X][S] [X][X] [S] ||| 2.718 ||| 0-0 1-1 ||| 0" << endl; - -} diff --git a/phrase-extract/extract-mixed-syntax/Main.h b/phrase-extract/extract-mixed-syntax/Main.h deleted file mode 100644 index 9744ba389..000000000 --- a/phrase-extract/extract-mixed-syntax/Main.h +++ /dev/null @@ -1,12 +0,0 @@ -/* - * Main.h - * - * Created on: 28 Feb 2014 - * Author: hieu - */ -#pragma once - -#include "OutputFileStream.h" - -void CreateGlueGrammar(Moses::OutputFileStream &glueFile); - diff --git a/phrase-extract/extract-mixed-syntax/Makefile b/phrase-extract/extract-mixed-syntax/Makefile deleted file mode 100644 index 6ecf15192..000000000 --- a/phrase-extract/extract-mixed-syntax/Makefile +++ /dev/null @@ -1,17 +0,0 @@ -all: extract-mixed-syntax - -clean: - rm -f *.o extract-mixed-syntax - -.cpp.o: - g++ -O4 -g -c -I../../boost/include -I../../ $< - -OBJECTS = AlignedSentence.o ConsistentPhrase.o ConsistentPhrases.o InputFileStream.o \ - Main.o OutputFileStream.o Parameter.o Phrase.o Rule.o Rules.o RuleSymbol.o \ - SyntaxTree.o Word.o NonTerm.o RulePhrase.o AlignedSentenceSyntax.o pugixml.o - -extract-mixed-syntax: $(OBJECTS) - - g++ $(OBJECTS) -L../../boost/lib64 -L../../lib -lz -lboost_iostreams-mt -lboost_program_options-mt -lmoses -o extract-mixed-syntax - - diff --git a/phrase-extract/extract-mixed-syntax/NonTerm.cpp b/phrase-extract/extract-mixed-syntax/NonTerm.cpp deleted file mode 100644 index 9e7d0dcaa..000000000 --- a/phrase-extract/extract-mixed-syntax/NonTerm.cpp +++ /dev/null @@ -1,65 +0,0 @@ -/* - * NonTerm.cpp - * - * Created on: 22 Feb 2014 - * Author: hieu - */ - -#include -#include "NonTerm.h" -#include "Word.h" -#include "ConsistentPhrase.h" -#include "Parameter.h" - -using namespace std; - -NonTerm::NonTerm(const ConsistentPhrase &consistentPhrase, - const std::string &source, - const std::string &target) -:m_consistentPhrase(&consistentPhrase) -,m_source(source) -,m_target(target) -{ - // TODO Auto-generated constructor stub - -} - -NonTerm::~NonTerm() { - // TODO Auto-generated destructor stub -} - -std::string NonTerm::Debug() const -{ - stringstream out; - out << m_source << m_target; - out << m_consistentPhrase->Debug(); - return out.str(); -} - -void NonTerm::Output(std::ostream &out) const -{ - out << m_source << m_target; -} - -void NonTerm::Output(std::ostream &out, Moses::FactorDirection direction) const -{ - out << GetLabel(direction); -} - -const std::string &NonTerm::GetLabel(Moses::FactorDirection direction) const -{ - return (direction == Moses::Input) ? m_source : m_target; -} - -bool NonTerm::IsHiero(Moses::FactorDirection direction, const Parameter ¶ms) const -{ - const std::string &label = NonTerm::GetLabel(direction); - return label == params.hieroNonTerm; -} - -bool NonTerm::IsHiero(const Parameter ¶ms) const -{ - return IsHiero(Moses::Input, params) && IsHiero(Moses::Output, params); -} -int NonTerm::GetWidth(Moses::FactorDirection direction) const -{ return GetConsistentPhrase().GetWidth(direction); } diff --git a/phrase-extract/extract-mixed-syntax/NonTerm.h b/phrase-extract/extract-mixed-syntax/NonTerm.h deleted file mode 100644 index 5b3bb9f04..000000000 --- a/phrase-extract/extract-mixed-syntax/NonTerm.h +++ /dev/null @@ -1,47 +0,0 @@ -/* - * NonTerm.h - * - * Created on: 22 Feb 2014 - * Author: hieu - */ -#pragma once -#include -#include "RuleSymbol.h" -#include "moses/TypeDef.h" - -class ConsistentPhrase; -class Parameter; - -class NonTerm : public RuleSymbol -{ -public: - - NonTerm(const ConsistentPhrase &consistentPhrase, - const std::string &source, - const std::string &target); - virtual ~NonTerm(); - - const ConsistentPhrase &GetConsistentPhrase() const - { return *m_consistentPhrase; } - - int GetWidth(Moses::FactorDirection direction) const; - - virtual bool IsNonTerm() const - { return true; } - - std::string GetString() const - { return m_source + m_target; } - - virtual std::string Debug() const; - virtual void Output(std::ostream &out) const; - void Output(std::ostream &out, Moses::FactorDirection direction) const; - - const std::string &GetLabel(Moses::FactorDirection direction) const; - bool IsHiero(Moses::FactorDirection direction, const Parameter ¶ms) const; - bool IsHiero(const Parameter ¶ms) const; - -protected: - const ConsistentPhrase *m_consistentPhrase; - std::string m_source, m_target; -}; - diff --git a/phrase-extract/extract-mixed-syntax/OutputFileStream.cpp b/phrase-extract/extract-mixed-syntax/OutputFileStream.cpp deleted file mode 100644 index a61ce1ab1..000000000 --- a/phrase-extract/extract-mixed-syntax/OutputFileStream.cpp +++ /dev/null @@ -1,79 +0,0 @@ -// $Id: OutputFileStream.cpp 2780 2010-01-29 17:11:17Z bojar $ - -/*********************************************************************** - Moses - factored phrase-based language decoder - Copyright (C) 2006 University of Edinburgh - - This library is free software; you can redistribute it and/or - modify it under the terms of the GNU Lesser General Public - License as published by the Free Software Foundation; either - version 2.1 of the License, or (at your option) any later version. - - This library is distributed in the hope that it will be useful, - but WITHOUT ANY WARRANTY; without even the implied warranty of - MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU - Lesser General Public License for more details. - - You should have received a copy of the GNU Lesser General Public - License along with this library; if not, write to the Free Software - Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA - ***********************************************************************/ - -#include -#include "OutputFileStream.h" -#include "gzfilebuf.h" - -using namespace std; - -namespace Moses -{ -OutputFileStream::OutputFileStream() - :boost::iostreams::filtering_ostream() - ,m_outFile(NULL) -{ -} - -OutputFileStream::OutputFileStream(const std::string &filePath) - : m_outFile(NULL) -{ - Open(filePath); -} - -OutputFileStream::~OutputFileStream() -{ - Close(); -} - -bool OutputFileStream::Open(const std::string &filePath) -{ - m_outFile = new ofstream(filePath.c_str(), ios_base::out | ios_base::binary); - if (m_outFile->fail()) { - return false; - } - - if (filePath.size() > 3 && filePath.substr(filePath.size() - 3, 3) == ".gz") { - this->push(boost::iostreams::gzip_compressor()); - } - this->push(*m_outFile); - - return true; -} - -void OutputFileStream::Close() -{ - if (m_outFile == NULL) { - return; - } - - this->flush(); - this->pop(); // file - - m_outFile->close(); - delete m_outFile; - m_outFile = NULL; - return; -} - - -} - diff --git a/phrase-extract/extract-mixed-syntax/OutputFileStream.h b/phrase-extract/extract-mixed-syntax/OutputFileStream.h deleted file mode 100644 index f52e36d76..000000000 --- a/phrase-extract/extract-mixed-syntax/OutputFileStream.h +++ /dev/null @@ -1,50 +0,0 @@ -// $Id: InputFileStream.h 2939 2010-02-24 11:15:44Z jfouet $ - -/*********************************************************************** - Moses - factored phrase-based language decoder - Copyright (C) 2006 University of Edinburgh - - This library is free software; you can redistribute it and/or - modify it under the terms of the GNU Lesser General Public - License as published by the Free Software Foundation; either - version 2.1 of the License, or (at your option) any later version. - - This library is distributed in the hope that it will be useful, - but WITHOUT ANY WARRANTY; without even the implied warranty of - MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU - Lesser General Public License for more details. - - You should have received a copy of the GNU Lesser General Public - License along with this library; if not, write to the Free Software - Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA - ***********************************************************************/ - -#pragma once - -#include -#include -#include -#include -#include - -namespace Moses -{ - -/** Used in place of std::istream, can read zipped files if it ends in .gz - */ -class OutputFileStream : public boost::iostreams::filtering_ostream -{ -protected: - std::ofstream *m_outFile; -public: - OutputFileStream(); - - OutputFileStream(const std::string &filePath); - virtual ~OutputFileStream(); - - bool Open(const std::string &filePath); - void Close(); -}; - -} - diff --git a/phrase-extract/extract-mixed-syntax/Parameter.cpp b/phrase-extract/extract-mixed-syntax/Parameter.cpp deleted file mode 100644 index 7f2238663..000000000 --- a/phrase-extract/extract-mixed-syntax/Parameter.cpp +++ /dev/null @@ -1,69 +0,0 @@ -/* - * Parameter.cpp - * - * Created on: 17 Feb 2014 - * Author: hieu - */ -#include "Parameter.h" -#include "moses/Util.h" -#include "util/exception.hh" - -using namespace std; - -Parameter::Parameter() -:maxSpan(10) -,minSpan(0) -,maxNonTerm(2) -,maxHieroNonTerm(999) -,maxSymbolsTarget(999) -,maxSymbolsSource(5) -,minHoleSource(2) -,sentenceOffset(0) -,nonTermConsecSource(false) -,requireAlignedWord(true) -,fractionalCounting(true) -,gzOutput(false) - -,hieroNonTerm("[X]") -,sourceSyntax(false) -,targetSyntax(false) - -,mixedSyntaxType(0) -,multiLabel(0) -,nonTermConsecSourceMixed(true) -,hieroSourceLHS(false) -,maxSpanFreeNonTermSource(0) -,nieceTerminal(true) -,maxScope(UNDEFINED) -,minScope(0) - -,spanLength(false) -,nonTermContext(false) -,nonTermContextTarget(false) -,nonTermContextFactor(0) - -,numSourceFactors(1) -,numTargetFactors(1) -{} - -Parameter::~Parameter() { - // TODO Auto-generated destructor stub -} - -void Parameter::SetScopeSpan(const std::string &str) -{ - scopeSpanStr = str; - vector toks1; - Moses::Tokenize(toks1, str, ":"); - - for (size_t i = 0; i < toks1.size(); ++i) { - const string &tok1 = toks1[i]; - - vector toks2; - Moses::Tokenize(toks2, tok1, ","); - UTIL_THROW_IF2(toks2.size() != 2, "Format is min,max:min,max... String is " << tok1); - - std::pair values(toks2[0], toks2[1]); - scopeSpan.push_back(values); - } -} diff --git a/phrase-extract/extract-mixed-syntax/Parameter.h b/phrase-extract/extract-mixed-syntax/Parameter.h deleted file mode 100644 index 26bad485d..000000000 --- a/phrase-extract/extract-mixed-syntax/Parameter.h +++ /dev/null @@ -1,62 +0,0 @@ -/* - * Parameter.h - * - * Created on: 17 Feb 2014 - * Author: hieu - */ -#pragma once - -#include -#include -#include - -#define UNDEFINED std::numeric_limits::max() - -class Parameter -{ -public: - Parameter(); - virtual ~Parameter(); - - int maxSpan; - int minSpan; - int maxNonTerm; - int maxHieroNonTerm; - int maxSymbolsTarget; - int maxSymbolsSource; - int minHoleSource; - - long sentenceOffset; - - bool nonTermConsecSource; - bool requireAlignedWord; - bool fractionalCounting; - bool gzOutput; - - std::string hieroNonTerm; - std::string gluePath; - - bool sourceSyntax, targetSyntax; - - int mixedSyntaxType, multiLabel; - bool nonTermConsecSourceMixed; - bool hieroSourceLHS; - int maxSpanFreeNonTermSource; - bool nieceTerminal; - int maxScope, minScope; - - // properties - bool spanLength; - bool nonTermContext; - bool nonTermContextTarget; - int nonTermContextFactor; - - int numSourceFactors, numTargetFactors; - - std::string scopeSpanStr; - std::vector > scopeSpan; - - void SetScopeSpan(const std::string &str); - -}; - diff --git a/phrase-extract/extract-mixed-syntax/Phrase.cpp b/phrase-extract/extract-mixed-syntax/Phrase.cpp deleted file mode 100644 index 535e10d6b..000000000 --- a/phrase-extract/extract-mixed-syntax/Phrase.cpp +++ /dev/null @@ -1,14 +0,0 @@ -#include -#include "Phrase.h" - -std::string Phrase::Debug() const -{ - std::stringstream out; - - for (size_t i = 0; i < size(); ++i) { - Word &word = *at(i); - out << word.Debug() << " "; - } - - return out.str(); -} diff --git a/phrase-extract/extract-mixed-syntax/Phrase.h b/phrase-extract/extract-mixed-syntax/Phrase.h deleted file mode 100644 index 13912cb95..000000000 --- a/phrase-extract/extract-mixed-syntax/Phrase.h +++ /dev/null @@ -1,19 +0,0 @@ -#pragma once - -#include -#include "Word.h" - -// a vector of terminals -class Phrase : public std::vector -{ -public: - Phrase() - {} - - Phrase(size_t size) - :std::vector(size) - {} - - std::string Debug() const; - -}; diff --git a/phrase-extract/extract-mixed-syntax/Rule.cpp b/phrase-extract/extract-mixed-syntax/Rule.cpp deleted file mode 100644 index 2f49bb873..000000000 --- a/phrase-extract/extract-mixed-syntax/Rule.cpp +++ /dev/null @@ -1,637 +0,0 @@ -/* - * Rule.cpp - * - * Created on: 20 Feb 2014 - * Author: hieu - */ - -#include -#include -#include "Rule.h" -#include "AlignedSentence.h" -#include "ConsistentPhrase.h" -#include "NonTerm.h" -#include "Parameter.h" - -using namespace std; - -Rule::Rule(const NonTerm &lhsNonTerm, const AlignedSentence &alignedSentence) -:m_lhs(lhsNonTerm) -,m_alignedSentence(alignedSentence) -,m_isValid(true) -,m_canRecurse(true) -{ - CreateSource(); -} - -Rule::Rule(const Rule ©, const NonTerm &nonTerm) -:m_lhs(copy.m_lhs) -,m_alignedSentence(copy.m_alignedSentence) -,m_isValid(true) -,m_canRecurse(true) -,m_nonterms(copy.m_nonterms) -{ - m_nonterms.push_back(&nonTerm); - CreateSource(); - -} - -Rule::~Rule() { - // TODO Auto-generated destructor stub -} - -const ConsistentPhrase &Rule::GetConsistentPhrase() const -{ return m_lhs.GetConsistentPhrase(); } - -void Rule::CreateSource() -{ - const NonTerm *cp = NULL; - size_t nonTermInd = 0; - if (nonTermInd < m_nonterms.size()) { - cp = m_nonterms[nonTermInd]; - } - - for (int sourcePos = m_lhs.GetConsistentPhrase().corners[0]; - sourcePos <= m_lhs.GetConsistentPhrase().corners[1]; - ++sourcePos) { - - const RuleSymbol *ruleSymbol; - if (cp && cp->GetConsistentPhrase().corners[0] <= sourcePos && sourcePos <= cp->GetConsistentPhrase().corners[1]) { - // replace words with non-term - ruleSymbol = cp; - sourcePos = cp->GetConsistentPhrase().corners[1]; - if (m_nonterms.size()) { - cp = m_nonterms[nonTermInd]; - } - - // move to next non-term - ++nonTermInd; - cp = (nonTermInd < m_nonterms.size()) ? m_nonterms[nonTermInd] : NULL; - } - else { - // terminal - ruleSymbol = m_alignedSentence.GetPhrase(Moses::Input)[sourcePos]; - } - - m_source.Add(ruleSymbol); - } -} - -int Rule::GetNextSourcePosForNonTerm() const -{ - if (m_nonterms.empty()) { - // no non-terms so far. Can start next non-term on left corner - return m_lhs.GetConsistentPhrase().corners[0]; - } - else { - // next non-term can start just left of previous - const ConsistentPhrase &cp = m_nonterms.back()->GetConsistentPhrase(); - int nextPos = cp.corners[1] + 1; - return nextPos; - } -} - -std::string Rule::Debug() const -{ - stringstream out; - - // source - for (size_t i = 0; i < m_source.GetSize(); ++i) { - const RuleSymbol &symbol = *m_source[i]; - out << symbol.Debug() << " "; - } - - // target - out << "||| "; - for (size_t i = 0; i < m_target.GetSize(); ++i) { - const RuleSymbol &symbol = *m_target[i]; - out << symbol.Debug() << " "; - } - - out << "||| "; - Alignments::const_iterator iterAlign; - for (iterAlign = m_alignments.begin(); iterAlign != m_alignments.end(); ++iterAlign) { - const std::pair &alignPair = *iterAlign; - out << alignPair.first << "-" << alignPair.second << " "; - } - - // overall range - out << "||| LHS=" << m_lhs.Debug(); - - return out.str(); -} - -void Rule::Output(std::ostream &out, bool forward, const Parameter ¶ms) const -{ - if (forward) { - // source - m_source.Output(out); - m_lhs.Output(out, Moses::Input); - - out << " ||| "; - - // target - m_target.Output(out); - m_lhs.Output(out, Moses::Output); - } - else { - // target - m_target.Output(out); - m_lhs.Output(out, Moses::Output); - - out << " ||| "; - - // source - m_source.Output(out); - m_lhs.Output(out, Moses::Input); - } - - out << " ||| "; - - // alignment - Alignments::const_iterator iterAlign; - for (iterAlign = m_alignments.begin(); iterAlign != m_alignments.end(); ++iterAlign) { - const std::pair &alignPair = *iterAlign; - - if (forward) { - out << alignPair.first << "-" << alignPair.second << " "; - } - else { - out << alignPair.second << "-" << alignPair.first << " "; - } - } - - out << "||| "; - - // count - out << m_count; - - out << " ||| "; - - // properties - - // span length - if (forward && params.spanLength && m_nonterms.size()) { - out << "{{SpanLength "; - - for (size_t i = 0; i < m_nonterms.size(); ++i) { - const NonTerm &nonTerm = *m_nonterms[i]; - const ConsistentPhrase &cp = nonTerm.GetConsistentPhrase(); - out << i << "," << cp.GetWidth(Moses::Input) << "," << cp.GetWidth(Moses::Output) << " "; - } - out << "}} "; - } - - // non-term context (source) - if (forward && params.nonTermContext && m_nonterms.size()) { - out << "{{NonTermContext "; - - int factor = params.nonTermContextFactor; - - for (size_t i = 0; i < m_nonterms.size(); ++i) { - const NonTerm &nonTerm = *m_nonterms[i]; - const ConsistentPhrase &cp = nonTerm.GetConsistentPhrase(); - NonTermContext(1, factor, i, cp, out); - } - out << "}} "; - } - - // non-term context (target) - if (forward && params.nonTermContextTarget && m_nonterms.size()) { - out << "{{NonTermContextTarget "; - - int factor = params.nonTermContextFactor; - - for (size_t i = 0; i < m_nonterms.size(); ++i) { - const NonTerm &nonTerm = *m_nonterms[i]; - const ConsistentPhrase &cp = nonTerm.GetConsistentPhrase(); - NonTermContext(2, factor, i, cp, out); - } - out << "}} "; - } - -} - -void Rule::NonTermContextFactor(int factor, const Word &word, std::ostream &out) const -{ - out << word.GetString(factor) << " "; -} - -void Rule::NonTermContext(int sourceTarget, int factor, size_t ntInd, const ConsistentPhrase &cp, std::ostream &out) const -{ - int startPos, endPos; - const Phrase *phrase; - - if (sourceTarget == 1) { - startPos = cp.corners[0]; - endPos = cp.corners[1]; - phrase = &m_alignedSentence.GetPhrase(Moses::Input); - } - else if (sourceTarget == 2) { - startPos = cp.corners[2]; - endPos = cp.corners[3]; - phrase = &m_alignedSentence.GetPhrase(Moses::Output); - } - else { - abort(); - } - - out << ntInd << " "; - - // left outside - if (startPos == 0) { - out << " "; - } - else { - NonTermContextFactor(factor, *phrase->at(startPos - 1), out); - } - - // left inside - NonTermContextFactor(factor, *phrase->at(startPos), out); - - // right inside - NonTermContextFactor(factor, *phrase->at(endPos), out); - - // right outside - if (endPos == phrase->size() - 1) { - out << " "; - } - else { - NonTermContextFactor(factor, *phrase->at(endPos + 1), out); - } - - -} - -void Rule::Prevalidate(const Parameter ¶ms) -{ - const ConsistentPhrase &cp = m_lhs.GetConsistentPhrase(); - - // check number of source symbols in rule - if (m_source.GetSize() > params.maxSymbolsSource) { - m_isValid = false; - } - - // check that last non-term added isn't too small - if (m_nonterms.size()) { - const NonTerm &lastNonTerm = *m_nonterms.back(); - const ConsistentPhrase &cp = lastNonTerm.GetConsistentPhrase(); - - int sourceWidth = cp.corners[1] - cp.corners[0] + 1; - if (sourceWidth < params.minHoleSource) { - m_isValid = false; - m_canRecurse = false; - return; - } - } - - // check number of non-terms - int numNonTerms = 0; - int numHieroNonTerms = 0; - for (size_t i = 0; i < m_source.GetSize(); ++i) { - const RuleSymbol *arc = m_source[i]; - if (arc->IsNonTerm()) { - ++numNonTerms; - const NonTerm &nonTerm = *static_cast(arc); - bool isHiero = nonTerm.IsHiero(params); - if (isHiero) { - ++numHieroNonTerms; - } - } - } - - if (numNonTerms >= params.maxNonTerm) { - m_canRecurse = false; - if (numNonTerms > params.maxNonTerm) { - m_isValid = false; - return; - } - } - - if (numHieroNonTerms >= params.maxHieroNonTerm) { - m_canRecurse = false; - if (numHieroNonTerms > params.maxHieroNonTerm) { - m_isValid = false; - return; - } - } - - // check if 2 consecutive non-terms in source - if (!params.nonTermConsecSource && m_nonterms.size() >= 2) { - const NonTerm &lastNonTerm = *m_nonterms.back(); - const NonTerm &secondLastNonTerm = *m_nonterms[m_nonterms.size() - 2]; - if (secondLastNonTerm.GetConsistentPhrase().corners[1] + 1 == - lastNonTerm.GetConsistentPhrase().corners[0]) { - if (params.mixedSyntaxType == 0) { - // ordinary hiero or syntax model - m_isValid = false; - m_canRecurse = false; - return; - } - else { - // Hieu's mixed syntax - if (lastNonTerm.IsHiero(Moses::Input, params) - && secondLastNonTerm.IsHiero(Moses::Input, params)) { - m_isValid = false; - m_canRecurse = false; - return; - } - } - - } - } - - //check to see if it overlaps with any other non-terms - if (m_nonterms.size() >= 2) { - const NonTerm &lastNonTerm = *m_nonterms.back(); - - for (size_t i = 0; i < m_nonterms.size() - 1; ++i) { - const NonTerm &otherNonTerm = *m_nonterms[i]; - bool overlap = lastNonTerm.GetConsistentPhrase().TargetOverlap(otherNonTerm.GetConsistentPhrase()); - - if (overlap) { - m_isValid = false; - m_canRecurse = false; - return; - } - } - } - - // check that at least 1 word is aligned - if (params.requireAlignedWord) { - bool ok = false; - for (size_t i = 0; i < m_source.GetSize(); ++i) { - const RuleSymbol &symbol = *m_source[i]; - if (!symbol.IsNonTerm()) { - const Word &word = static_cast(symbol); - if (word.GetAlignment().size()) { - ok = true; - break; - } - } - } - - if (!ok) { - m_isValid = false; - m_canRecurse = false; - return; - } - } - - if (params.maxSpanFreeNonTermSource) { - const NonTerm *front = dynamic_cast(m_source[0]); - if (front) { - int width = front->GetWidth(Moses::Input); - if (width > params.maxSpanFreeNonTermSource) { - m_isValid = false; - m_canRecurse = false; - return; - } - } - - const NonTerm *back = dynamic_cast(m_source.Back()); - if (back) { - int width = back->GetWidth(Moses::Input); - if (width > params.maxSpanFreeNonTermSource) { - m_isValid = false; - m_canRecurse = false; - return; - } - } - } - - if (!params.nieceTerminal) { - // collect terminal in a rule - std::set terms; - for (size_t i = 0; i < m_source.GetSize(); ++i) { - const Word *word = dynamic_cast(m_source[i]); - if (word) { - terms.insert(word); - } - } - - // look in non-terms - for (size_t i = 0; i < m_source.GetSize(); ++i) { - const NonTerm *nonTerm = dynamic_cast(m_source[i]); - if (nonTerm) { - const ConsistentPhrase &cp = nonTerm->GetConsistentPhrase(); - bool containTerm = ContainTerm(cp, terms); - - if (containTerm) { - //cerr << "ruleSource=" << *ruleSource << " "; - //cerr << "ntRange=" << ntRange << endl; - - // non-term contains 1 of the terms in the rule. - m_isValid = false; - m_canRecurse = false; - return; - } - } - } - } - - if (params.maxScope != UNDEFINED || params.minScope > 0) { - int scope = GetScope(params); - if (scope > params.maxScope) { - // scope of subsequent rules will be the same or increase - // therefore can NOT recurse - m_isValid = false; - m_canRecurse = false; - return; - } - - if (scope < params.minScope) { - // scope of subsequent rules may increase - // therefore can recurse - m_isValid = false; - } - } - - // min/max span per scope - if (params.scopeSpan.size()) { - int scope = GetScope(params); - if (scope >= params.scopeSpan.size()) { - // no constraint on it. It's ok - } - else { - const std::pair &constraint = params.scopeSpan[scope]; - int sourceWidth = m_lhs.GetWidth(Moses::Input); - if (sourceWidth < constraint.first || sourceWidth > constraint.second) { - m_isValid = false; - m_canRecurse = false; - return; - } - } - } -} - -int Rule::GetScope(const Parameter ¶ms) const -{ - size_t scope = 0; - bool previousIsAmbiguous = false; - - if (m_source[0]->IsNonTerm()) { - scope++; - previousIsAmbiguous = true; - } - - for (size_t i = 1; i < m_source.GetSize(); ++i) { - const RuleSymbol *symbol = m_source[i]; - bool isAmbiguous = symbol->IsNonTerm(); - if (isAmbiguous) { - // mixed syntax - const NonTerm *nt = static_cast(symbol); - isAmbiguous = nt->IsHiero(Moses::Input, params); - } - - if (isAmbiguous && previousIsAmbiguous) { - scope++; - } - previousIsAmbiguous = isAmbiguous; - } - - if (previousIsAmbiguous) { - scope++; - } - - return scope; - - /* - int scope = 0; - if (m_source.GetSize() > 1) { - const RuleSymbol &front = *m_source.Front(); - if (front.IsNonTerm()) { - ++scope; - } - - const RuleSymbol &back = *m_source.Back(); - if (back.IsNonTerm()) { - ++scope; - } - } - return scope; - */ -} - -template -bool Contains(const T *sought, const set &coll) -{ - std::set::const_iterator iter; - for (iter = coll.begin(); iter != coll.end(); ++iter) { - const Word *found = *iter; - if (sought->CompareString(*found) == 0) { - return true; - } - } - return false; -} - -bool Rule::ContainTerm(const ConsistentPhrase &cp, const std::set &terms) const -{ - const Phrase &sourceSentence = m_alignedSentence.GetPhrase(Moses::Input); - - for (int pos = cp.corners[0]; pos <= cp.corners[1]; ++pos) { - const Word *soughtWord = sourceSentence[pos]; - - // find same word in set - if (Contains(soughtWord, terms)) { - return true; - } - } - return false; -} - -bool CompareTargetNonTerms(const NonTerm *a, const NonTerm *b) -{ - // compare just start target pos - return a->GetConsistentPhrase().corners[2] < b->GetConsistentPhrase().corners[2]; -} - -void Rule::CreateTarget(const Parameter ¶ms) -{ - if (!m_isValid) { - return; - } - - vector targetNonTerm(m_nonterms); - std::sort(targetNonTerm.begin(), targetNonTerm.end(), CompareTargetNonTerms); - - const NonTerm *cp = NULL; - size_t nonTermInd = 0; - if (nonTermInd < targetNonTerm.size()) { - cp = targetNonTerm[nonTermInd]; - } - - for (int targetPos = m_lhs.GetConsistentPhrase().corners[2]; - targetPos <= m_lhs.GetConsistentPhrase().corners[3]; - ++targetPos) { - - const RuleSymbol *ruleSymbol; - if (cp && cp->GetConsistentPhrase().corners[2] <= targetPos && targetPos <= cp->GetConsistentPhrase().corners[3]) { - // replace words with non-term - ruleSymbol = cp; - targetPos = cp->GetConsistentPhrase().corners[3]; - if (targetNonTerm.size()) { - cp = targetNonTerm[nonTermInd]; - } - - // move to next non-term - ++nonTermInd; - cp = (nonTermInd < targetNonTerm.size()) ? targetNonTerm[nonTermInd] : NULL; - } - else { - // terminal - ruleSymbol = m_alignedSentence.GetPhrase(Moses::Output)[targetPos]; - } - - m_target.Add(ruleSymbol); - } - - CreateAlignments(); -} - - -void Rule::CreateAlignments() -{ - int sourceStart = GetConsistentPhrase().corners[0]; - int targetStart = GetConsistentPhrase().corners[2]; - - for (size_t sourcePos = 0; sourcePos < m_source.GetSize(); ++sourcePos) { - const RuleSymbol *symbol = m_source[sourcePos]; - if (!symbol->IsNonTerm()) { - // terminals - const Word &sourceWord = static_cast(*symbol); - const std::set &targetWords = sourceWord.GetAlignment(); - CreateAlignments(sourcePos, targetWords); - } - else { - // non-terms. same object in both source & target - CreateAlignments(sourcePos, symbol); - } - } -} - -void Rule::CreateAlignments(int sourcePos, const std::set &targetWords) -{ - std::set::const_iterator iterTarget; - for (iterTarget = targetWords.begin(); iterTarget != targetWords.end(); ++iterTarget) { - const Word *targetWord = *iterTarget; - CreateAlignments(sourcePos, targetWord); - } -} - -void Rule::CreateAlignments(int sourcePos, const RuleSymbol *targetSought) -{ - // should be in target phrase - for (size_t targetPos = 0; targetPos < m_target.GetSize(); ++targetPos) { - const RuleSymbol *foundSymbol = m_target[targetPos]; - if (targetSought == foundSymbol) { - pair alignPoint(sourcePos, targetPos); - m_alignments.insert(alignPoint); - return; - } - } - - throw "not found"; -} - diff --git a/phrase-extract/extract-mixed-syntax/Rule.h b/phrase-extract/extract-mixed-syntax/Rule.h deleted file mode 100644 index 15a142b97..000000000 --- a/phrase-extract/extract-mixed-syntax/Rule.h +++ /dev/null @@ -1,90 +0,0 @@ -/* - * Rule.h - * - * Created on: 20 Feb 2014 - * Author: hieu - */ -#pragma once -#include -#include "Phrase.h" -#include "RulePhrase.h" -#include "moses/TypeDef.h" - -class ConsistentPhrase; -class AlignedSentence; -class NonTerm; -class Parameter; - - -class Rule { -public: - typedef std::set > Alignments; - - Rule(const Rule ©); // do not implement - - // original rule with no non-term - Rule(const NonTerm &lhsNonTerm, const AlignedSentence &alignedSentence); - - // extend a rule, adding 1 new non-term - Rule(const Rule ©, const NonTerm &nonTerm); - - virtual ~Rule(); - - bool IsValid() const - { return m_isValid; } - - bool CanRecurse() const - { return m_canRecurse; } - - const NonTerm &GetLHS() const - { return m_lhs; } - - const ConsistentPhrase &GetConsistentPhrase() const; - - int GetNextSourcePosForNonTerm() const; - - void SetCount(float count) - { m_count = count; } - float GetCount() const - { return m_count; } - - const Alignments &GetAlignments() const - { return m_alignments; } - - std::string Debug() const; - void Output(std::ostream &out, bool forward, const Parameter ¶ms) const; - - void Prevalidate(const Parameter ¶ms); - void CreateTarget(const Parameter ¶ms); - - const RulePhrase &GetPhrase(Moses::FactorDirection direction) const - { return (direction == Moses::Input) ? m_source : m_target; } - -protected: - const NonTerm &m_lhs; - const AlignedSentence &m_alignedSentence; - RulePhrase m_source, m_target; - float m_count; - - Alignments m_alignments; - - // in source order - std::vector m_nonterms; - - bool m_isValid, m_canRecurse; - - void CreateSource(); - void CreateAlignments(); - void CreateAlignments(int sourcePos, const std::set &targetWords); - void CreateAlignments(int sourcePos, const RuleSymbol *targetSought); - - bool ContainTerm(const ConsistentPhrase &cp, const std::set &terms) const; - int GetScope(const Parameter ¶ms) const; - - void NonTermContext(int sourceTarget, int factors, size_t ntInd, const ConsistentPhrase &cp, std::ostream &out) const; - // sourceTarget: 1 = source, 2 = target - - void NonTermContextFactor(int factor, const Word &word, std::ostream &out) const; - -}; - diff --git a/phrase-extract/extract-mixed-syntax/RulePhrase.cpp b/phrase-extract/extract-mixed-syntax/RulePhrase.cpp deleted file mode 100644 index 5c629168b..000000000 --- a/phrase-extract/extract-mixed-syntax/RulePhrase.cpp +++ /dev/null @@ -1,50 +0,0 @@ -/* - * RulePhrase.cpp - * - * Created on: 26 Feb 2014 - * Author: hieu - */ - -#include -#include "RulePhrase.h" -#include "RuleSymbol.h" - -using namespace std; - -extern bool g_debug; - -int RulePhrase::Compare(const RulePhrase &other) const -{ - if (GetSize() != other.GetSize()) { - return GetSize() < other.GetSize() ? -1 : +1; - } - - for (size_t i = 0; i < m_coll.size(); ++i) { - const RuleSymbol &symbol = *m_coll[i]; - const RuleSymbol &otherSymbol = *other.m_coll[i]; - int compare = symbol.Compare(otherSymbol); - - if (compare) { - return compare; - } - } - - return 0; -} - -void RulePhrase::Output(std::ostream &out) const -{ - for (size_t i = 0; i < m_coll.size(); ++i) { - const RuleSymbol &symbol = *m_coll[i]; - symbol.Output(out); - out << " "; - } -} - -std::string RulePhrase::Debug() const -{ - std::stringstream out; - Output(out); - return out.str(); -} - diff --git a/phrase-extract/extract-mixed-syntax/RulePhrase.h b/phrase-extract/extract-mixed-syntax/RulePhrase.h deleted file mode 100644 index 412169b74..000000000 --- a/phrase-extract/extract-mixed-syntax/RulePhrase.h +++ /dev/null @@ -1,49 +0,0 @@ -/* - * RulePhrase.h - * - * Created on: 26 Feb 2014 - * Author: hieu - */ - -#ifndef RULEPHRASE_H_ -#define RULEPHRASE_H_ - -#include -#include -#include - -class RuleSymbol; - -// a phrase of terms and non-terms for 1 side of a rule -class RulePhrase -{ -public: - typedef std::vector Coll; - Coll m_coll; - - size_t GetSize() const - { return m_coll.size(); } - - void Add(const RuleSymbol *symbol) - { - m_coll.push_back(symbol); - } - - const RuleSymbol* operator[](size_t index) const { - return m_coll[index]; - } - - const RuleSymbol* Front() const { - return m_coll.front(); - } - const RuleSymbol* Back() const { - return m_coll.back(); - } - - int Compare(const RulePhrase &other) const; - - void Output(std::ostream &out) const; - std::string Debug() const; -}; - -#endif /* RULEPHRASE_H_ */ diff --git a/phrase-extract/extract-mixed-syntax/RuleSymbol.cpp b/phrase-extract/extract-mixed-syntax/RuleSymbol.cpp deleted file mode 100644 index 933ffc9c2..000000000 --- a/phrase-extract/extract-mixed-syntax/RuleSymbol.cpp +++ /dev/null @@ -1,36 +0,0 @@ -/* - * RuleSymbol.cpp - * - * Created on: 21 Feb 2014 - * Author: hieu - */ - -#include "RuleSymbol.h" - -using namespace std; - -RuleSymbol::RuleSymbol() { - // TODO Auto-generated constructor stub - -} - -RuleSymbol::~RuleSymbol() { - // TODO Auto-generated destructor stub -} - -int RuleSymbol::Compare(const RuleSymbol &other) const -{ - if (IsNonTerm() != other.IsNonTerm()) { - return IsNonTerm() ? -1 : +1; - } - - string str = GetString(); - string otherStr = other.GetString(); - - if (str == otherStr) { - return 0; - } - else { - return (str < otherStr) ? -1 : +1; - } -} diff --git a/phrase-extract/extract-mixed-syntax/RuleSymbol.h b/phrase-extract/extract-mixed-syntax/RuleSymbol.h deleted file mode 100644 index c292fcc0d..000000000 --- a/phrase-extract/extract-mixed-syntax/RuleSymbol.h +++ /dev/null @@ -1,31 +0,0 @@ -/* - * RuleSymbol.h - * - * Created on: 21 Feb 2014 - * Author: hieu - */ - -#ifndef RULESYMBOL_H_ -#define RULESYMBOL_H_ - -#include -#include - -// base class - terminal or non-term -class RuleSymbol { -public: - RuleSymbol(); - virtual ~RuleSymbol(); - - virtual bool IsNonTerm() const = 0; - - virtual std::string Debug() const = 0; - virtual void Output(std::ostream &out) const = 0; - - virtual std::string GetString() const = 0; - - int Compare(const RuleSymbol &other) const; - -}; - -#endif /* RULESYMBOL_H_ */ diff --git a/phrase-extract/extract-mixed-syntax/Rules.cpp b/phrase-extract/extract-mixed-syntax/Rules.cpp deleted file mode 100644 index 1b93430e2..000000000 --- a/phrase-extract/extract-mixed-syntax/Rules.cpp +++ /dev/null @@ -1,227 +0,0 @@ -/* - * Rules.cpp - * - * Created on: 20 Feb 2014 - * Author: hieu - */ - -#include -#include "Rules.h" -#include "ConsistentPhrase.h" -#include "ConsistentPhrases.h" -#include "AlignedSentence.h" -#include "Rule.h" -#include "Parameter.h" -#include "moses/Util.h" - -using namespace std; - -extern bool g_debug; - -Rules::Rules(const AlignedSentence &alignedSentence) -:m_alignedSentence(alignedSentence) -{ -} - -Rules::~Rules() { - Moses::RemoveAllInColl(m_keepRules); -} - -void Rules::CreateRules(const ConsistentPhrase &cp, - const Parameter ¶ms) -{ - if (params.hieroSourceLHS) { - const NonTerm &nonTerm = cp.GetHieroNonTerm(); - CreateRule(nonTerm, params); - } - else { - const ConsistentPhrase::NonTerms &nonTerms = cp.GetNonTerms(); - for (size_t i = 0; i < nonTerms.size(); ++i) { - const NonTerm &nonTerm = nonTerms[i]; - CreateRule(nonTerm, params); - } - } -} - -void Rules::CreateRule(const NonTerm &nonTerm, - const Parameter ¶ms) -{ - Rule *rule = new Rule(nonTerm, m_alignedSentence); - - rule->Prevalidate(params); - rule->CreateTarget(params); - - - if (rule->CanRecurse()) { - Extend(*rule, params); - } - - if (rule->IsValid()) { - m_keepRules.insert(rule); - } - else { - delete rule; - } - -} - -void Rules::Extend(const Parameter ¶ms) -{ - const ConsistentPhrases &allCPS = m_alignedSentence.GetConsistentPhrases(); - - size_t size = m_alignedSentence.GetPhrase(Moses::Input).size(); - for (size_t sourceStart = 0; sourceStart < size; ++sourceStart) { - for (size_t sourceEnd = sourceStart; sourceEnd < size; ++sourceEnd) { - const ConsistentPhrases::Coll &cps = allCPS.GetColl(sourceStart, sourceEnd); - - ConsistentPhrases::Coll::const_iterator iter; - for (iter = cps.begin(); iter != cps.end(); ++iter) { - const ConsistentPhrase &cp = **iter; - CreateRules(cp, params); - } - } - } -} - -void Rules::Extend(const Rule &rule, const Parameter ¶ms) -{ - const ConsistentPhrases &allCPS = m_alignedSentence.GetConsistentPhrases(); - int sourceMin = rule.GetNextSourcePosForNonTerm(); - - int ruleStart = rule.GetConsistentPhrase().corners[0]; - int ruleEnd = rule.GetConsistentPhrase().corners[1]; - - for (int sourceStart = sourceMin; sourceStart <= ruleEnd; ++sourceStart) { - for (int sourceEnd = sourceStart; sourceEnd <= ruleEnd; ++sourceEnd) { - if (sourceStart == ruleStart && sourceEnd == ruleEnd) { - // don't cover whole rule with 1 non-term - continue; - } - - const ConsistentPhrases::Coll &cps = allCPS.GetColl(sourceStart, sourceEnd); - Extend(rule, cps, params); - } - } -} - -void Rules::Extend(const Rule &rule, const ConsistentPhrases::Coll &cps, const Parameter ¶ms) -{ - ConsistentPhrases::Coll::const_iterator iter; - for (iter = cps.begin(); iter != cps.end(); ++iter) { - const ConsistentPhrase &cp = **iter; - Extend(rule, cp, params); - } -} - -void Rules::Extend(const Rule &rule, const ConsistentPhrase &cp, const Parameter ¶ms) -{ - const ConsistentPhrase::NonTerms &nonTerms = cp.GetNonTerms(); - for (size_t i = 0; i < nonTerms.size(); ++i) { - const NonTerm &nonTerm = nonTerms[i]; - - Rule *newRule = new Rule(rule, nonTerm); - newRule->Prevalidate(params); - newRule->CreateTarget(params); - - if (newRule->CanRecurse()) { - // recursively extend - Extend(*newRule, params); - } - - if (newRule->IsValid()) { - m_keepRules.insert(newRule); - } - else { - delete newRule; - } - } -} - -std::string Rules::Debug() const -{ - stringstream out; - - std::set::const_iterator iter; - out << "m_keepRules:" << endl; - for (iter = m_keepRules.begin(); iter != m_keepRules.end(); ++iter) { - const Rule &rule = **iter; - out << rule.Debug() << endl; - } - - return out.str(); -} - -void Rules::Output(std::ostream &out, bool forward, const Parameter ¶ms) const -{ - std::set::const_iterator iter; - for (iter = m_mergeRules.begin(); iter != m_mergeRules.end(); ++iter) { - const Rule &rule = **iter; - rule.Output(out, forward, params); - out << endl; - } -} - -void Rules::Consolidate(const Parameter ¶ms) -{ - if (params.fractionalCounting) { - CalcFractionalCount(); - } - else { - std::set::iterator iter; - for (iter = m_keepRules.begin(); iter != m_keepRules.end(); ++iter) { - Rule &rule = **iter; - rule.SetCount(1); - } - } - - MergeRules(params); -} - -void Rules::MergeRules(const Parameter ¶ms) -{ - typedef std::set MergeRules; - - std::set::const_iterator iterOrig; - for (iterOrig = m_keepRules.begin(); iterOrig != m_keepRules.end(); ++iterOrig) { - Rule *origRule = *iterOrig; - - pair inserted = m_mergeRules.insert(origRule); - if (!inserted.second) { - // already there, just add count - Rule &rule = **inserted.first; - float newCount = rule.GetCount() + origRule->GetCount(); - rule.SetCount(newCount); - } - } -} - -void Rules::CalcFractionalCount() -{ - typedef std::set RuleColl; - typedef std::map RuleByConsistentPhrase; - RuleByConsistentPhrase allRules; - - // sort by source AND target ranges - std::set::const_iterator iter; - for (iter = m_keepRules.begin(); iter != m_keepRules.end(); ++iter) { - Rule *rule = *iter; - const ConsistentPhrase &cp = rule->GetConsistentPhrase(); - RuleColl &ruleColl = allRules[&cp]; - ruleColl.insert(rule); - } - - // fractional count - RuleByConsistentPhrase::iterator iterOuter; - for (iterOuter = allRules.begin(); iterOuter != allRules.end(); ++iterOuter) { - RuleColl &rules = iterOuter->second; - - RuleColl::iterator iterInner; - for (iterInner = rules.begin(); iterInner != rules.end(); ++iterInner) { - Rule &rule = **iterInner; - rule.SetCount(1.0f / (float) rules.size()); - } - } - -} - - diff --git a/phrase-extract/extract-mixed-syntax/Rules.h b/phrase-extract/extract-mixed-syntax/Rules.h deleted file mode 100644 index 6d8cb122d..000000000 --- a/phrase-extract/extract-mixed-syntax/Rules.h +++ /dev/null @@ -1,72 +0,0 @@ -/* - * Rules.h - * - * Created on: 20 Feb 2014 - * Author: hieu - */ - -#pragma once - -#include -#include -#include "ConsistentPhrases.h" -#include "Rule.h" - -extern bool g_debug; - -class AlignedSentence; -class Parameter; - -struct CompareRules { - bool operator()(const Rule *a, const Rule *b) - { - int compare; - - compare = a->GetPhrase(Moses::Input).Compare(b->GetPhrase(Moses::Input)); - if (compare) return compare < 0; - - compare = a->GetPhrase(Moses::Output).Compare(b->GetPhrase(Moses::Output)); - if (compare) return compare < 0; - - if (a->GetAlignments() != b->GetAlignments()) { - return a->GetAlignments() < b->GetAlignments(); - } - - if (a->GetLHS().GetString() != b->GetLHS().GetString()) { - return a->GetLHS().GetString() < b->GetLHS().GetString(); - } - - return false; - } -}; - -class Rules { -public: - Rules(const AlignedSentence &alignedSentence); - virtual ~Rules(); - void Extend(const Parameter ¶ms); - void Consolidate(const Parameter ¶ms); - - std::string Debug() const; - void Output(std::ostream &out, bool forward, const Parameter ¶ms) const; - -protected: - const AlignedSentence &m_alignedSentence; - std::set m_keepRules; - std::set m_mergeRules; - - void Extend(const Rule &rule, const Parameter ¶ms); - void Extend(const Rule &rule, const ConsistentPhrases::Coll &cps, const Parameter ¶ms); - void Extend(const Rule &rule, const ConsistentPhrase &cp, const Parameter ¶ms); - - // create original rules - void CreateRules(const ConsistentPhrase &cp, - const Parameter ¶ms); - void CreateRule(const NonTerm &nonTerm, - const Parameter ¶ms); - - void MergeRules(const Parameter ¶ms); - void CalcFractionalCount(); - -}; - diff --git a/phrase-extract/extract-mixed-syntax/SyntaxTree.cpp b/phrase-extract/extract-mixed-syntax/SyntaxTree.cpp deleted file mode 100644 index 472444e7c..000000000 --- a/phrase-extract/extract-mixed-syntax/SyntaxTree.cpp +++ /dev/null @@ -1,47 +0,0 @@ -#include -#include -#include "SyntaxTree.h" -#include "Parameter.h" - -using namespace std; - -void SyntaxTree::Add(int startPos, int endPos, const std::string &label, const Parameter ¶ms) -{ - //cerr << "add " << label << " to " << "[" << startPos << "-" << endPos << "]" << endl; - - Range range(startPos, endPos); - Labels &labels = m_coll[range]; - - bool add = true; - if (labels.size()) { - if (params.multiLabel == 1) { - // delete the label in collection and add new - assert(labels.size() == 1); - labels.clear(); - } - else if (params.multiLabel == 2) { - // ignore this label - add = false; - } - } - - if (add) { - labels.push_back(label); - } -} - -void SyntaxTree::AddToAll(const std::string &label) -{ - Coll::iterator iter; - for (iter = m_coll.begin(); iter != m_coll.end(); ++iter) { - Labels &labels = iter->second; - labels.push_back(label); - } -} - -const SyntaxTree::Labels &SyntaxTree::Find(int startPos, int endPos) const -{ - Coll::const_iterator iter; - iter = m_coll.find(Range(startPos, endPos)); - return (iter == m_coll.end()) ? m_defaultLabels : iter->second; -} diff --git a/phrase-extract/extract-mixed-syntax/SyntaxTree.h b/phrase-extract/extract-mixed-syntax/SyntaxTree.h deleted file mode 100644 index 58f718151..000000000 --- a/phrase-extract/extract-mixed-syntax/SyntaxTree.h +++ /dev/null @@ -1,32 +0,0 @@ -#pragma once - -#include -#include -#include - -class Parameter; - -class SyntaxTree -{ -public: - typedef std::pair Range; - typedef std::vector Labels; - typedef std::map Coll; - - void Add(int startPos, int endPos, const std::string &label, const Parameter ¶ms); - void AddToAll(const std::string &label); - - const Labels &Find(int startPos, int endPos) const; - - void SetHieroLabel(const std::string &label) { - m_defaultLabels.push_back(label); - } - - -protected: - - Coll m_coll; - Labels m_defaultLabels; -}; - - diff --git a/phrase-extract/extract-mixed-syntax/Word.cpp b/phrase-extract/extract-mixed-syntax/Word.cpp deleted file mode 100644 index 8ce4f76c6..000000000 --- a/phrase-extract/extract-mixed-syntax/Word.cpp +++ /dev/null @@ -1,68 +0,0 @@ -/* - * Word.cpp - * - * Created on: 18 Feb 2014 - * Author: s0565741 - */ -#include -#include "Word.h" -#include "moses/Util.h" - -using namespace std; - -Word::Word(int pos, const std::string &str) -:m_pos(pos) -,m_str(str) -{ - // TODO Auto-generated constructor stub - -} - -Word::~Word() { - // TODO Auto-generated destructor stub -} - -void Word::AddAlignment(const Word *other) -{ - m_alignment.insert(other); -} - -std::set Word::GetAlignmentIndex() const -{ - std::set ret; - - std::set::const_iterator iter; - for (iter = m_alignment.begin(); iter != m_alignment.end(); ++iter) { - const Word &otherWord = **iter; - int otherPos = otherWord.GetPos(); - ret.insert(otherPos); - } - - return ret; -} - -void Word::Output(std::ostream &out) const -{ - out << m_str; -} - -std::string Word::Debug() const -{ - return m_str; -} - -int Word::CompareString(const Word &other) const -{ - return m_str.compare(other.m_str); -} - -std::string Word::GetString(int factor) const -{ - vector toks; - Moses::Tokenize(toks, m_str, "|"); - - assert(factor < toks.size()); - return toks[factor]; -} - - diff --git a/phrase-extract/extract-mixed-syntax/Word.h b/phrase-extract/extract-mixed-syntax/Word.h deleted file mode 100644 index 54419ceb0..000000000 --- a/phrase-extract/extract-mixed-syntax/Word.h +++ /dev/null @@ -1,49 +0,0 @@ -/* - * Word.h - * - * Created on: 18 Feb 2014 - * Author: s0565741 - */ -#pragma once - -#include -#include -#include "RuleSymbol.h" - -// a terminal -class Word : public RuleSymbol -{ -public: - Word(const Word&); // do not implement - Word(int pos, const std::string &str); - virtual ~Word(); - - virtual bool IsNonTerm() const - { return false; } - - std::string GetString() const - { return m_str; } - - std::string GetString(int factor) const; - - int GetPos() const - { return m_pos; } - - void AddAlignment(const Word *other); - - const std::set &GetAlignment() const - { return m_alignment; } - - std::set GetAlignmentIndex() const; - - void Output(std::ostream &out) const; - std::string Debug() const; - - int CompareString(const Word &other) const; - -protected: - int m_pos; // original position in sentence, NOT in lattice - std::string m_str; - std::set m_alignment; -}; - diff --git a/phrase-extract/extract-mixed-syntax/gzfilebuf.h b/phrase-extract/extract-mixed-syntax/gzfilebuf.h deleted file mode 100644 index 885c661f0..000000000 --- a/phrase-extract/extract-mixed-syntax/gzfilebuf.h +++ /dev/null @@ -1,81 +0,0 @@ -#ifndef moses_gzfile_buf_h -#define moses_gzfile_buf_h - -#include -#include -#include - -class gzfilebuf : public std::streambuf { -public: - gzfilebuf(const char *filename) - { _gzf = gzopen(filename, "rb"); - setg (_buff+sizeof(int), // beginning of putback area - _buff+sizeof(int), // read position - _buff+sizeof(int)); // end position - } - ~gzfilebuf() { gzclose(_gzf); } -protected: - virtual int_type overflow (int_type c) { - throw; - } - - // write multiple characters - virtual - std::streamsize xsputn (const char* s, - std::streamsize num) { - throw; - } - - virtual std::streampos seekpos ( std::streampos sp, std::ios_base::openmode which = std::ios_base::in | std::ios_base::out ){ throw; - } - - //read one character - virtual int_type underflow () { - // is read position before end of _buff? - if (gptr() < egptr()) { - return traits_type::to_int_type(*gptr()); - } - - /* process size of putback area - * - use number of characters read - * - but at most four - */ - unsigned int numPutback = gptr() - eback(); - if (numPutback > sizeof(int)) { - numPutback = sizeof(int); - } - - /* copy up to four characters previously read into - * the putback _buff (area of first four characters) - */ - std::memmove (_buff+(sizeof(int)-numPutback), gptr()-numPutback, - numPutback); - - // read new characters - int num = gzread(_gzf, _buff+sizeof(int), _buffsize-sizeof(int)); - if (num <= 0) { - // ERROR or EOF - return EOF; - } - - // reset _buff pointers - setg (_buff+(sizeof(int)-numPutback), // beginning of putback area - _buff+sizeof(int), // read position - _buff+sizeof(int)+num); // end of buffer - - // return next character - return traits_type::to_int_type(*gptr()); - } - - std::streamsize xsgetn (char* s, - std::streamsize num) { - return gzread(_gzf,s,num); - } - -private: - gzFile _gzf; - static const unsigned int _buffsize = 1024; - char _buff[_buffsize]; -}; - -#endif diff --git a/phrase-extract/extract-mixed-syntax/pugiconfig.hpp b/phrase-extract/extract-mixed-syntax/pugiconfig.hpp deleted file mode 100644 index c2196715c..000000000 --- a/phrase-extract/extract-mixed-syntax/pugiconfig.hpp +++ /dev/null @@ -1,69 +0,0 @@ -/** - * pugixml parser - version 1.2 - * -------------------------------------------------------- - * Copyright (C) 2006-2012, by Arseny Kapoulkine (arseny.kapoulkine@gmail.com) - * Report bugs and download new versions at http://pugixml.org/ - * - * This library is distributed under the MIT License. See notice at the end - * of this file. - * - * This work is based on the pugxml parser, which is: - * Copyright (C) 2003, by Kristen Wegner (kristen@tima.net) - */ - -#ifndef HEADER_PUGICONFIG_HPP -#define HEADER_PUGICONFIG_HPP - -// Uncomment this to enable wchar_t mode -// #define PUGIXML_WCHAR_MODE - -// Uncomment this to disable XPath -// #define PUGIXML_NO_XPATH - -// Uncomment this to disable STL -// #define PUGIXML_NO_STL - -// Uncomment this to disable exceptions -// #define PUGIXML_NO_EXCEPTIONS - -// Set this to control attributes for public classes/functions, i.e.: -// #define PUGIXML_API __declspec(dllexport) // to export all public symbols from DLL -// #define PUGIXML_CLASS __declspec(dllimport) // to import all classes from DLL -// #define PUGIXML_FUNCTION __fastcall // to set calling conventions to all public functions to fastcall -// In absence of PUGIXML_CLASS/PUGIXML_FUNCTION definitions PUGIXML_API is used instead - -// Uncomment this to switch to header-only version -// #define PUGIXML_HEADER_ONLY -// #include "pugixml.cpp" - -// Tune these constants to adjust memory-related behavior -// #define PUGIXML_MEMORY_PAGE_SIZE 32768 -// #define PUGIXML_MEMORY_OUTPUT_STACK 10240 -// #define PUGIXML_MEMORY_XPATH_PAGE_SIZE 4096 - -#endif - -/** - * Copyright (c) 2006-2012 Arseny Kapoulkine - * - * Permission is hereby granted, free of charge, to any person - * obtaining a copy of this software and associated documentation - * files (the "Software"), to deal in the Software without - * restriction, including without limitation the rights to use, - * copy, modify, merge, publish, distribute, sublicense, and/or sell - * copies of the Software, and to permit persons to whom the - * Software is furnished to do so, subject to the following - * conditions: - * - * The above copyright notice and this permission notice shall be - * included in all copies or substantial portions of the Software. - * - * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, - * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES - * OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND - * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT - * HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, - * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING - * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR - * OTHER DEALINGS IN THE SOFTWARE. - */ diff --git a/phrase-extract/extract-mixed-syntax/pugixml.cpp b/phrase-extract/extract-mixed-syntax/pugixml.cpp deleted file mode 100644 index 4035ab1cf..000000000 --- a/phrase-extract/extract-mixed-syntax/pugixml.cpp +++ /dev/null @@ -1,10250 +0,0 @@ -/** - * pugixml parser - version 1.2 - * -------------------------------------------------------- - * Copyright (C) 2006-2012, by Arseny Kapoulkine (arseny.kapoulkine@gmail.com) - * Report bugs and download new versions at http://pugixml.org/ - * - * This library is distributed under the MIT License. See notice at the end - * of this file. - * - * This work is based on the pugxml parser, which is: - * Copyright (C) 2003, by Kristen Wegner (kristen@tima.net) - */ - -#ifndef SOURCE_PUGIXML_CPP -#define SOURCE_PUGIXML_CPP - -#include "pugixml.hpp" - -#include -#include -#include -#include -#include - -#ifndef PUGIXML_NO_XPATH -# include -# include -# ifdef PUGIXML_NO_EXCEPTIONS -# include -# endif -#endif - -#ifndef PUGIXML_NO_STL -# include -# include -# include -#endif - -// For placement new -#include - -#ifdef _MSC_VER -# pragma warning(push) -# pragma warning(disable: 4127) // conditional expression is constant -# pragma warning(disable: 4324) // structure was padded due to __declspec(align()) -# pragma warning(disable: 4611) // interaction between '_setjmp' and C++ object destruction is non-portable -# pragma warning(disable: 4702) // unreachable code -# pragma warning(disable: 4996) // this function or variable may be unsafe -# pragma warning(disable: 4793) // function compiled as native: presence of '_setjmp' makes a function unmanaged -#endif - -#ifdef __INTEL_COMPILER -# pragma warning(disable: 177) // function was declared but never referenced -# pragma warning(disable: 279) // controlling expression is constant -# pragma warning(disable: 1478 1786) // function was declared "deprecated" -# pragma warning(disable: 1684) // conversion from pointer to same-sized integral type -#endif - -#if defined(__BORLANDC__) && defined(PUGIXML_HEADER_ONLY) -# pragma warn -8080 // symbol is declared but never used; disabling this inside push/pop bracket does not make the warning go away -#endif - -#ifdef __BORLANDC__ -# pragma option push -# pragma warn -8008 // condition is always false -# pragma warn -8066 // unreachable code -#endif - -#ifdef __SNC__ -// Using diag_push/diag_pop does not disable the warnings inside templates due to a compiler bug -# pragma diag_suppress=178 // function was declared but never referenced -# pragma diag_suppress=237 // controlling expression is constant -#endif - -// Inlining controls -#if defined(_MSC_VER) && _MSC_VER >= 1300 -# define PUGI__NO_INLINE __declspec(noinline) -#elif defined(__GNUC__) -# define PUGI__NO_INLINE __attribute__((noinline)) -#else -# define PUGI__NO_INLINE -#endif - -// Simple static assertion -#define PUGI__STATIC_ASSERT(cond) { static const char condition_failed[(cond) ? 1 : -1] = {0}; (void)condition_failed[0]; } - -// Digital Mars C++ bug workaround for passing char loaded from memory via stack -#ifdef __DMC__ -# define PUGI__DMC_VOLATILE volatile -#else -# define PUGI__DMC_VOLATILE -#endif - -// Borland C++ bug workaround for not defining ::memcpy depending on header include order (can't always use std::memcpy because some compilers don't have it at all) -#if defined(__BORLANDC__) && !defined(__MEM_H_USING_LIST) -using std::memcpy; -using std::memmove; -#endif - -// In some environments MSVC is a compiler but the CRT lacks certain MSVC-specific features -#if defined(_MSC_VER) && !defined(__S3E__) -# define PUGI__MSVC_CRT_VERSION _MSC_VER -#endif - -#ifdef PUGIXML_HEADER_ONLY -# define PUGI__NS_BEGIN namespace pugi { namespace impl { -# define PUGI__NS_END } } -# define PUGI__FN inline -# define PUGI__FN_NO_INLINE inline -#else -# if defined(_MSC_VER) && _MSC_VER < 1300 // MSVC6 seems to have an amusing bug with anonymous namespaces inside namespaces -# define PUGI__NS_BEGIN namespace pugi { namespace impl { -# define PUGI__NS_END } } -# else -# define PUGI__NS_BEGIN namespace pugi { namespace impl { namespace { -# define PUGI__NS_END } } } -# endif -# define PUGI__FN -# define PUGI__FN_NO_INLINE PUGI__NO_INLINE -#endif - -// uintptr_t -#if !defined(_MSC_VER) || _MSC_VER >= 1600 -# include -#else -# ifndef _UINTPTR_T_DEFINED -// No native uintptr_t in MSVC6 and in some WinCE versions -typedef size_t uintptr_t; -#define _UINTPTR_T_DEFINED -# endif -PUGI__NS_BEGIN - typedef unsigned __int8 uint8_t; - typedef unsigned __int16 uint16_t; - typedef unsigned __int32 uint32_t; -PUGI__NS_END -#endif - -// Memory allocation -PUGI__NS_BEGIN - PUGI__FN void* default_allocate(size_t size) - { - return malloc(size); - } - - PUGI__FN void default_deallocate(void* ptr) - { - free(ptr); - } - - template - struct xml_memory_management_function_storage - { - static allocation_function allocate; - static deallocation_function deallocate; - }; - - template allocation_function xml_memory_management_function_storage::allocate = default_allocate; - template deallocation_function xml_memory_management_function_storage::deallocate = default_deallocate; - - typedef xml_memory_management_function_storage xml_memory; -PUGI__NS_END - -// String utilities -PUGI__NS_BEGIN - // Get string length - PUGI__FN size_t strlength(const char_t* s) - { - assert(s); - - #ifdef PUGIXML_WCHAR_MODE - return wcslen(s); - #else - return strlen(s); - #endif - } - - // Compare two strings - PUGI__FN bool strequal(const char_t* src, const char_t* dst) - { - assert(src && dst); - - #ifdef PUGIXML_WCHAR_MODE - return wcscmp(src, dst) == 0; - #else - return strcmp(src, dst) == 0; - #endif - } - - // Compare lhs with [rhs_begin, rhs_end) - PUGI__FN bool strequalrange(const char_t* lhs, const char_t* rhs, size_t count) - { - for (size_t i = 0; i < count; ++i) - if (lhs[i] != rhs[i]) - return false; - - return lhs[count] == 0; - } - -#ifdef PUGIXML_WCHAR_MODE - // Convert string to wide string, assuming all symbols are ASCII - PUGI__FN void widen_ascii(wchar_t* dest, const char* source) - { - for (const char* i = source; *i; ++i) *dest++ = *i; - *dest = 0; - } -#endif -PUGI__NS_END - -#if !defined(PUGIXML_NO_STL) || !defined(PUGIXML_NO_XPATH) -// auto_ptr-like buffer holder for exception recovery -PUGI__NS_BEGIN - struct buffer_holder - { - void* data; - void (*deleter)(void*); - - buffer_holder(void* data_, void (*deleter_)(void*)): data(data_), deleter(deleter_) - { - } - - ~buffer_holder() - { - if (data) deleter(data); - } - - void* release() - { - void* result = data; - data = 0; - return result; - } - }; -PUGI__NS_END -#endif - -PUGI__NS_BEGIN - static const size_t xml_memory_page_size = - #ifdef PUGIXML_MEMORY_PAGE_SIZE - PUGIXML_MEMORY_PAGE_SIZE - #else - 32768 - #endif - ; - - static const uintptr_t xml_memory_page_alignment = 32; - static const uintptr_t xml_memory_page_pointer_mask = ~(xml_memory_page_alignment - 1); - static const uintptr_t xml_memory_page_name_allocated_mask = 16; - static const uintptr_t xml_memory_page_value_allocated_mask = 8; - static const uintptr_t xml_memory_page_type_mask = 7; - - struct xml_allocator; - - struct xml_memory_page - { - static xml_memory_page* construct(void* memory) - { - if (!memory) return 0; //$ redundant, left for performance - - xml_memory_page* result = static_cast(memory); - - result->allocator = 0; - result->memory = 0; - result->prev = 0; - result->next = 0; - result->busy_size = 0; - result->freed_size = 0; - - return result; - } - - xml_allocator* allocator; - - void* memory; - - xml_memory_page* prev; - xml_memory_page* next; - - size_t busy_size; - size_t freed_size; - - char data[1]; - }; - - struct xml_memory_string_header - { - uint16_t page_offset; // offset from page->data - uint16_t full_size; // 0 if string occupies whole page - }; - - struct xml_allocator - { - xml_allocator(xml_memory_page* root): _root(root), _busy_size(root->busy_size) - { - } - - xml_memory_page* allocate_page(size_t data_size) - { - size_t size = offsetof(xml_memory_page, data) + data_size; - - // allocate block with some alignment, leaving memory for worst-case padding - void* memory = xml_memory::allocate(size + xml_memory_page_alignment); - if (!memory) return 0; - - // align upwards to page boundary - void* page_memory = reinterpret_cast((reinterpret_cast(memory) + (xml_memory_page_alignment - 1)) & ~(xml_memory_page_alignment - 1)); - - // prepare page structure - xml_memory_page* page = xml_memory_page::construct(page_memory); - - page->memory = memory; - page->allocator = _root->allocator; - - return page; - } - - static void deallocate_page(xml_memory_page* page) - { - xml_memory::deallocate(page->memory); - } - - void* allocate_memory_oob(size_t size, xml_memory_page*& out_page); - - void* allocate_memory(size_t size, xml_memory_page*& out_page) - { - if (_busy_size + size > xml_memory_page_size) return allocate_memory_oob(size, out_page); - - void* buf = _root->data + _busy_size; - - _busy_size += size; - - out_page = _root; - - return buf; - } - - void deallocate_memory(void* ptr, size_t size, xml_memory_page* page) - { - if (page == _root) page->busy_size = _busy_size; - - assert(ptr >= page->data && ptr < page->data + page->busy_size); - (void)!ptr; - - page->freed_size += size; - assert(page->freed_size <= page->busy_size); - - if (page->freed_size == page->busy_size) - { - if (page->next == 0) - { - assert(_root == page); - - // top page freed, just reset sizes - page->busy_size = page->freed_size = 0; - _busy_size = 0; - } - else - { - assert(_root != page); - assert(page->prev); - - // remove from the list - page->prev->next = page->next; - page->next->prev = page->prev; - - // deallocate - deallocate_page(page); - } - } - } - - char_t* allocate_string(size_t length) - { - // allocate memory for string and header block - size_t size = sizeof(xml_memory_string_header) + length * sizeof(char_t); - - // round size up to pointer alignment boundary - size_t full_size = (size + (sizeof(void*) - 1)) & ~(sizeof(void*) - 1); - - xml_memory_page* page; - xml_memory_string_header* header = static_cast(allocate_memory(full_size, page)); - - if (!header) return 0; - - // setup header - ptrdiff_t page_offset = reinterpret_cast(header) - page->data; - - assert(page_offset >= 0 && page_offset < (1 << 16)); - header->page_offset = static_cast(page_offset); - - // full_size == 0 for large strings that occupy the whole page - assert(full_size < (1 << 16) || (page->busy_size == full_size && page_offset == 0)); - header->full_size = static_cast(full_size < (1 << 16) ? full_size : 0); - - // round-trip through void* to avoid 'cast increases required alignment of target type' warning - // header is guaranteed a pointer-sized alignment, which should be enough for char_t - return static_cast(static_cast(header + 1)); - } - - void deallocate_string(char_t* string) - { - // this function casts pointers through void* to avoid 'cast increases required alignment of target type' warnings - // we're guaranteed the proper (pointer-sized) alignment on the input string if it was allocated via allocate_string - - // get header - xml_memory_string_header* header = static_cast(static_cast(string)) - 1; - - // deallocate - size_t page_offset = offsetof(xml_memory_page, data) + header->page_offset; - xml_memory_page* page = reinterpret_cast(static_cast(reinterpret_cast(header) - page_offset)); - - // if full_size == 0 then this string occupies the whole page - size_t full_size = header->full_size == 0 ? page->busy_size : header->full_size; - - deallocate_memory(header, full_size, page); - } - - xml_memory_page* _root; - size_t _busy_size; - }; - - PUGI__FN_NO_INLINE void* xml_allocator::allocate_memory_oob(size_t size, xml_memory_page*& out_page) - { - const size_t large_allocation_threshold = xml_memory_page_size / 4; - - xml_memory_page* page = allocate_page(size <= large_allocation_threshold ? xml_memory_page_size : size); - out_page = page; - - if (!page) return 0; - - if (size <= large_allocation_threshold) - { - _root->busy_size = _busy_size; - - // insert page at the end of linked list - page->prev = _root; - _root->next = page; - _root = page; - - _busy_size = size; - } - else - { - // insert page before the end of linked list, so that it is deleted as soon as possible - // the last page is not deleted even if it's empty (see deallocate_memory) - assert(_root->prev); - - page->prev = _root->prev; - page->next = _root; - - _root->prev->next = page; - _root->prev = page; - } - - // allocate inside page - page->busy_size = size; - - return page->data; - } -PUGI__NS_END - -namespace pugi -{ - /// A 'name=value' XML attribute structure. - struct xml_attribute_struct - { - /// Default ctor - xml_attribute_struct(impl::xml_memory_page* page): header(reinterpret_cast(page)), name(0), value(0), prev_attribute_c(0), next_attribute(0) - { - } - - uintptr_t header; - - char_t* name; ///< Pointer to attribute name. - char_t* value; ///< Pointer to attribute value. - - xml_attribute_struct* prev_attribute_c; ///< Previous attribute (cyclic list) - xml_attribute_struct* next_attribute; ///< Next attribute - }; - - /// An XML document tree node. - struct xml_node_struct - { - /// Default ctor - /// \param type - node type - xml_node_struct(impl::xml_memory_page* page, xml_node_type type): header(reinterpret_cast(page) | (type - 1)), parent(0), name(0), value(0), first_child(0), prev_sibling_c(0), next_sibling(0), first_attribute(0) - { - } - - uintptr_t header; - - xml_node_struct* parent; ///< Pointer to parent - - char_t* name; ///< Pointer to element name. - char_t* value; ///< Pointer to any associated string data. - - xml_node_struct* first_child; ///< First child - - xml_node_struct* prev_sibling_c; ///< Left brother (cyclic list) - xml_node_struct* next_sibling; ///< Right brother - - xml_attribute_struct* first_attribute; ///< First attribute - }; -} - -PUGI__NS_BEGIN - struct xml_document_struct: public xml_node_struct, public xml_allocator - { - xml_document_struct(xml_memory_page* page): xml_node_struct(page, node_document), xml_allocator(page), buffer(0) - { - } - - const char_t* buffer; - }; - - inline xml_allocator& get_allocator(const xml_node_struct* node) - { - assert(node); - - return *reinterpret_cast(node->header & xml_memory_page_pointer_mask)->allocator; - } -PUGI__NS_END - -// Low-level DOM operations -PUGI__NS_BEGIN - inline xml_attribute_struct* allocate_attribute(xml_allocator& alloc) - { - xml_memory_page* page; - void* memory = alloc.allocate_memory(sizeof(xml_attribute_struct), page); - - return new (memory) xml_attribute_struct(page); - } - - inline xml_node_struct* allocate_node(xml_allocator& alloc, xml_node_type type) - { - xml_memory_page* page; - void* memory = alloc.allocate_memory(sizeof(xml_node_struct), page); - - return new (memory) xml_node_struct(page, type); - } - - inline void destroy_attribute(xml_attribute_struct* a, xml_allocator& alloc) - { - uintptr_t header = a->header; - - if (header & impl::xml_memory_page_name_allocated_mask) alloc.deallocate_string(a->name); - if (header & impl::xml_memory_page_value_allocated_mask) alloc.deallocate_string(a->value); - - alloc.deallocate_memory(a, sizeof(xml_attribute_struct), reinterpret_cast(header & xml_memory_page_pointer_mask)); - } - - inline void destroy_node(xml_node_struct* n, xml_allocator& alloc) - { - uintptr_t header = n->header; - - if (header & impl::xml_memory_page_name_allocated_mask) alloc.deallocate_string(n->name); - if (header & impl::xml_memory_page_value_allocated_mask) alloc.deallocate_string(n->value); - - for (xml_attribute_struct* attr = n->first_attribute; attr; ) - { - xml_attribute_struct* next = attr->next_attribute; - - destroy_attribute(attr, alloc); - - attr = next; - } - - for (xml_node_struct* child = n->first_child; child; ) - { - xml_node_struct* next = child->next_sibling; - - destroy_node(child, alloc); - - child = next; - } - - alloc.deallocate_memory(n, sizeof(xml_node_struct), reinterpret_cast(header & xml_memory_page_pointer_mask)); - } - - PUGI__FN_NO_INLINE xml_node_struct* append_node(xml_node_struct* node, xml_allocator& alloc, xml_node_type type = node_element) - { - xml_node_struct* child = allocate_node(alloc, type); - if (!child) return 0; - - child->parent = node; - - xml_node_struct* first_child = node->first_child; - - if (first_child) - { - xml_node_struct* last_child = first_child->prev_sibling_c; - - last_child->next_sibling = child; - child->prev_sibling_c = last_child; - first_child->prev_sibling_c = child; - } - else - { - node->first_child = child; - child->prev_sibling_c = child; - } - - return child; - } - - PUGI__FN_NO_INLINE xml_attribute_struct* append_attribute_ll(xml_node_struct* node, xml_allocator& alloc) - { - xml_attribute_struct* a = allocate_attribute(alloc); - if (!a) return 0; - - xml_attribute_struct* first_attribute = node->first_attribute; - - if (first_attribute) - { - xml_attribute_struct* last_attribute = first_attribute->prev_attribute_c; - - last_attribute->next_attribute = a; - a->prev_attribute_c = last_attribute; - first_attribute->prev_attribute_c = a; - } - else - { - node->first_attribute = a; - a->prev_attribute_c = a; - } - - return a; - } -PUGI__NS_END - -// Helper classes for code generation -PUGI__NS_BEGIN - struct opt_false - { - enum { value = 0 }; - }; - - struct opt_true - { - enum { value = 1 }; - }; -PUGI__NS_END - -// Unicode utilities -PUGI__NS_BEGIN - inline uint16_t endian_swap(uint16_t value) - { - return static_cast(((value & 0xff) << 8) | (value >> 8)); - } - - inline uint32_t endian_swap(uint32_t value) - { - return ((value & 0xff) << 24) | ((value & 0xff00) << 8) | ((value & 0xff0000) >> 8) | (value >> 24); - } - - struct utf8_counter - { - typedef size_t value_type; - - static value_type low(value_type result, uint32_t ch) - { - // U+0000..U+007F - if (ch < 0x80) return result + 1; - // U+0080..U+07FF - else if (ch < 0x800) return result + 2; - // U+0800..U+FFFF - else return result + 3; - } - - static value_type high(value_type result, uint32_t) - { - // U+10000..U+10FFFF - return result + 4; - } - }; - - struct utf8_writer - { - typedef uint8_t* value_type; - - static value_type low(value_type result, uint32_t ch) - { - // U+0000..U+007F - if (ch < 0x80) - { - *result = static_cast(ch); - return result + 1; - } - // U+0080..U+07FF - else if (ch < 0x800) - { - result[0] = static_cast(0xC0 | (ch >> 6)); - result[1] = static_cast(0x80 | (ch & 0x3F)); - return result + 2; - } - // U+0800..U+FFFF - else - { - result[0] = static_cast(0xE0 | (ch >> 12)); - result[1] = static_cast(0x80 | ((ch >> 6) & 0x3F)); - result[2] = static_cast(0x80 | (ch & 0x3F)); - return result + 3; - } - } - - static value_type high(value_type result, uint32_t ch) - { - // U+10000..U+10FFFF - result[0] = static_cast(0xF0 | (ch >> 18)); - result[1] = static_cast(0x80 | ((ch >> 12) & 0x3F)); - result[2] = static_cast(0x80 | ((ch >> 6) & 0x3F)); - result[3] = static_cast(0x80 | (ch & 0x3F)); - return result + 4; - } - - static value_type any(value_type result, uint32_t ch) - { - return (ch < 0x10000) ? low(result, ch) : high(result, ch); - } - }; - - struct utf16_counter - { - typedef size_t value_type; - - static value_type low(value_type result, uint32_t) - { - return result + 1; - } - - static value_type high(value_type result, uint32_t) - { - return result + 2; - } - }; - - struct utf16_writer - { - typedef uint16_t* value_type; - - static value_type low(value_type result, uint32_t ch) - { - *result = static_cast(ch); - - return result + 1; - } - - static value_type high(value_type result, uint32_t ch) - { - uint32_t msh = static_cast(ch - 0x10000) >> 10; - uint32_t lsh = static_cast(ch - 0x10000) & 0x3ff; - - result[0] = static_cast(0xD800 + msh); - result[1] = static_cast(0xDC00 + lsh); - - return result + 2; - } - - static value_type any(value_type result, uint32_t ch) - { - return (ch < 0x10000) ? low(result, ch) : high(result, ch); - } - }; - - struct utf32_counter - { - typedef size_t value_type; - - static value_type low(value_type result, uint32_t) - { - return result + 1; - } - - static value_type high(value_type result, uint32_t) - { - return result + 1; - } - }; - - struct utf32_writer - { - typedef uint32_t* value_type; - - static value_type low(value_type result, uint32_t ch) - { - *result = ch; - - return result + 1; - } - - static value_type high(value_type result, uint32_t ch) - { - *result = ch; - - return result + 1; - } - - static value_type any(value_type result, uint32_t ch) - { - *result = ch; - - return result + 1; - } - }; - - struct latin1_writer - { - typedef uint8_t* value_type; - - static value_type low(value_type result, uint32_t ch) - { - *result = static_cast(ch > 255 ? '?' : ch); - - return result + 1; - } - - static value_type high(value_type result, uint32_t ch) - { - (void)ch; - - *result = '?'; - - return result + 1; - } - }; - - template struct wchar_selector; - - template <> struct wchar_selector<2> - { - typedef uint16_t type; - typedef utf16_counter counter; - typedef utf16_writer writer; - }; - - template <> struct wchar_selector<4> - { - typedef uint32_t type; - typedef utf32_counter counter; - typedef utf32_writer writer; - }; - - typedef wchar_selector::counter wchar_counter; - typedef wchar_selector::writer wchar_writer; - - template struct utf_decoder - { - static inline typename Traits::value_type decode_utf8_block(const uint8_t* data, size_t size, typename Traits::value_type result) - { - const uint8_t utf8_byte_mask = 0x3f; - - while (size) - { - uint8_t lead = *data; - - // 0xxxxxxx -> U+0000..U+007F - if (lead < 0x80) - { - result = Traits::low(result, lead); - data += 1; - size -= 1; - - // process aligned single-byte (ascii) blocks - if ((reinterpret_cast(data) & 3) == 0) - { - // round-trip through void* to silence 'cast increases required alignment of target type' warnings - while (size >= 4 && (*static_cast(static_cast(data)) & 0x80808080) == 0) - { - result = Traits::low(result, data[0]); - result = Traits::low(result, data[1]); - result = Traits::low(result, data[2]); - result = Traits::low(result, data[3]); - data += 4; - size -= 4; - } - } - } - // 110xxxxx -> U+0080..U+07FF - else if (static_cast(lead - 0xC0) < 0x20 && size >= 2 && (data[1] & 0xc0) == 0x80) - { - result = Traits::low(result, ((lead & ~0xC0) << 6) | (data[1] & utf8_byte_mask)); - data += 2; - size -= 2; - } - // 1110xxxx -> U+0800-U+FFFF - else if (static_cast(lead - 0xE0) < 0x10 && size >= 3 && (data[1] & 0xc0) == 0x80 && (data[2] & 0xc0) == 0x80) - { - result = Traits::low(result, ((lead & ~0xE0) << 12) | ((data[1] & utf8_byte_mask) << 6) | (data[2] & utf8_byte_mask)); - data += 3; - size -= 3; - } - // 11110xxx -> U+10000..U+10FFFF - else if (static_cast(lead - 0xF0) < 0x08 && size >= 4 && (data[1] & 0xc0) == 0x80 && (data[2] & 0xc0) == 0x80 && (data[3] & 0xc0) == 0x80) - { - result = Traits::high(result, ((lead & ~0xF0) << 18) | ((data[1] & utf8_byte_mask) << 12) | ((data[2] & utf8_byte_mask) << 6) | (data[3] & utf8_byte_mask)); - data += 4; - size -= 4; - } - // 10xxxxxx or 11111xxx -> invalid - else - { - data += 1; - size -= 1; - } - } - - return result; - } - - static inline typename Traits::value_type decode_utf16_block(const uint16_t* data, size_t size, typename Traits::value_type result) - { - const uint16_t* end = data + size; - - while (data < end) - { - uint16_t lead = opt_swap::value ? endian_swap(*data) : *data; - - // U+0000..U+D7FF - if (lead < 0xD800) - { - result = Traits::low(result, lead); - data += 1; - } - // U+E000..U+FFFF - else if (static_cast(lead - 0xE000) < 0x2000) - { - result = Traits::low(result, lead); - data += 1; - } - // surrogate pair lead - else if (static_cast(lead - 0xD800) < 0x400 && data + 1 < end) - { - uint16_t next = opt_swap::value ? endian_swap(data[1]) : data[1]; - - if (static_cast(next - 0xDC00) < 0x400) - { - result = Traits::high(result, 0x10000 + ((lead & 0x3ff) << 10) + (next & 0x3ff)); - data += 2; - } - else - { - data += 1; - } - } - else - { - data += 1; - } - } - - return result; - } - - static inline typename Traits::value_type decode_utf32_block(const uint32_t* data, size_t size, typename Traits::value_type result) - { - const uint32_t* end = data + size; - - while (data < end) - { - uint32_t lead = opt_swap::value ? endian_swap(*data) : *data; - - // U+0000..U+FFFF - if (lead < 0x10000) - { - result = Traits::low(result, lead); - data += 1; - } - // U+10000..U+10FFFF - else - { - result = Traits::high(result, lead); - data += 1; - } - } - - return result; - } - - static inline typename Traits::value_type decode_latin1_block(const uint8_t* data, size_t size, typename Traits::value_type result) - { - for (size_t i = 0; i < size; ++i) - { - result = Traits::low(result, data[i]); - } - - return result; - } - - static inline typename Traits::value_type decode_wchar_block_impl(const uint16_t* data, size_t size, typename Traits::value_type result) - { - return decode_utf16_block(data, size, result); - } - - static inline typename Traits::value_type decode_wchar_block_impl(const uint32_t* data, size_t size, typename Traits::value_type result) - { - return decode_utf32_block(data, size, result); - } - - static inline typename Traits::value_type decode_wchar_block(const wchar_t* data, size_t size, typename Traits::value_type result) - { - return decode_wchar_block_impl(reinterpret_cast::type*>(data), size, result); - } - }; - - template PUGI__FN void convert_utf_endian_swap(T* result, const T* data, size_t length) - { - for (size_t i = 0; i < length; ++i) result[i] = endian_swap(data[i]); - } - -#ifdef PUGIXML_WCHAR_MODE - PUGI__FN void convert_wchar_endian_swap(wchar_t* result, const wchar_t* data, size_t length) - { - for (size_t i = 0; i < length; ++i) result[i] = static_cast(endian_swap(static_cast::type>(data[i]))); - } -#endif -PUGI__NS_END - -PUGI__NS_BEGIN - enum chartype_t - { - ct_parse_pcdata = 1, // \0, &, \r, < - ct_parse_attr = 2, // \0, &, \r, ', " - ct_parse_attr_ws = 4, // \0, &, \r, ', ", \n, tab - ct_space = 8, // \r, \n, space, tab - ct_parse_cdata = 16, // \0, ], >, \r - ct_parse_comment = 32, // \0, -, >, \r - ct_symbol = 64, // Any symbol > 127, a-z, A-Z, 0-9, _, :, -, . - ct_start_symbol = 128 // Any symbol > 127, a-z, A-Z, _, : - }; - - static const unsigned char chartype_table[256] = - { - 55, 0, 0, 0, 0, 0, 0, 0, 0, 12, 12, 0, 0, 63, 0, 0, // 0-15 - 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, // 16-31 - 8, 0, 6, 0, 0, 0, 7, 6, 0, 0, 0, 0, 0, 96, 64, 0, // 32-47 - 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 192, 0, 1, 0, 48, 0, // 48-63 - 0, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, // 64-79 - 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 0, 0, 16, 0, 192, // 80-95 - 0, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, // 96-111 - 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 0, 0, 0, 0, 0, // 112-127 - - 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, // 128+ - 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, - 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, - 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, - 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, - 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, - 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, - 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192 - }; - - enum chartypex_t - { - ctx_special_pcdata = 1, // Any symbol >= 0 and < 32 (except \t, \r, \n), &, <, > - ctx_special_attr = 2, // Any symbol >= 0 and < 32 (except \t), &, <, >, " - ctx_start_symbol = 4, // Any symbol > 127, a-z, A-Z, _ - ctx_digit = 8, // 0-9 - ctx_symbol = 16 // Any symbol > 127, a-z, A-Z, 0-9, _, -, . - }; - - static const unsigned char chartypex_table[256] = - { - 3, 3, 3, 3, 3, 3, 3, 3, 3, 0, 2, 3, 3, 2, 3, 3, // 0-15 - 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, // 16-31 - 0, 0, 2, 0, 0, 0, 3, 0, 0, 0, 0, 0, 0, 16, 16, 0, // 32-47 - 24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 0, 0, 3, 0, 3, 0, // 48-63 - - 0, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, // 64-79 - 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 0, 0, 0, 0, 20, // 80-95 - 0, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, // 96-111 - 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 0, 0, 0, 0, 0, // 112-127 - - 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, // 128+ - 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, - 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, - 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, - 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, - 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, - 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, - 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20 - }; - -#ifdef PUGIXML_WCHAR_MODE - #define PUGI__IS_CHARTYPE_IMPL(c, ct, table) ((static_cast(c) < 128 ? table[static_cast(c)] : table[128]) & (ct)) -#else - #define PUGI__IS_CHARTYPE_IMPL(c, ct, table) (table[static_cast(c)] & (ct)) -#endif - - #define PUGI__IS_CHARTYPE(c, ct) PUGI__IS_CHARTYPE_IMPL(c, ct, chartype_table) - #define PUGI__IS_CHARTYPEX(c, ct) PUGI__IS_CHARTYPE_IMPL(c, ct, chartypex_table) - - PUGI__FN bool is_little_endian() - { - unsigned int ui = 1; - - return *reinterpret_cast(&ui) == 1; - } - - PUGI__FN xml_encoding get_wchar_encoding() - { - PUGI__STATIC_ASSERT(sizeof(wchar_t) == 2 || sizeof(wchar_t) == 4); - - if (sizeof(wchar_t) == 2) - return is_little_endian() ? encoding_utf16_le : encoding_utf16_be; - else - return is_little_endian() ? encoding_utf32_le : encoding_utf32_be; - } - - PUGI__FN xml_encoding guess_buffer_encoding(uint8_t d0, uint8_t d1, uint8_t d2, uint8_t d3) - { - // look for BOM in first few bytes - if (d0 == 0 && d1 == 0 && d2 == 0xfe && d3 == 0xff) return encoding_utf32_be; - if (d0 == 0xff && d1 == 0xfe && d2 == 0 && d3 == 0) return encoding_utf32_le; - if (d0 == 0xfe && d1 == 0xff) return encoding_utf16_be; - if (d0 == 0xff && d1 == 0xfe) return encoding_utf16_le; - if (d0 == 0xef && d1 == 0xbb && d2 == 0xbf) return encoding_utf8; - - // look for <, (contents); - - PUGI__DMC_VOLATILE uint8_t d0 = data[0], d1 = data[1], d2 = data[2], d3 = data[3]; - - return guess_buffer_encoding(d0, d1, d2, d3); - } - - PUGI__FN bool get_mutable_buffer(char_t*& out_buffer, size_t& out_length, const void* contents, size_t size, bool is_mutable) - { - if (is_mutable) - { - out_buffer = static_cast(const_cast(contents)); - } - else - { - void* buffer = xml_memory::allocate(size > 0 ? size : 1); - if (!buffer) return false; - - memcpy(buffer, contents, size); - - out_buffer = static_cast(buffer); - } - - out_length = size / sizeof(char_t); - - return true; - } - -#ifdef PUGIXML_WCHAR_MODE - PUGI__FN bool need_endian_swap_utf(xml_encoding le, xml_encoding re) - { - return (le == encoding_utf16_be && re == encoding_utf16_le) || (le == encoding_utf16_le && re == encoding_utf16_be) || - (le == encoding_utf32_be && re == encoding_utf32_le) || (le == encoding_utf32_le && re == encoding_utf32_be); - } - - PUGI__FN bool convert_buffer_endian_swap(char_t*& out_buffer, size_t& out_length, const void* contents, size_t size, bool is_mutable) - { - const char_t* data = static_cast(contents); - - if (is_mutable) - { - out_buffer = const_cast(data); - } - else - { - out_buffer = static_cast(xml_memory::allocate(size > 0 ? size : 1)); - if (!out_buffer) return false; - } - - out_length = size / sizeof(char_t); - - convert_wchar_endian_swap(out_buffer, data, out_length); - - return true; - } - - PUGI__FN bool convert_buffer_utf8(char_t*& out_buffer, size_t& out_length, const void* contents, size_t size) - { - const uint8_t* data = static_cast(contents); - - // first pass: get length in wchar_t units - out_length = utf_decoder::decode_utf8_block(data, size, 0); - - // allocate buffer of suitable length - out_buffer = static_cast(xml_memory::allocate((out_length > 0 ? out_length : 1) * sizeof(char_t))); - if (!out_buffer) return false; - - // second pass: convert utf8 input to wchar_t - wchar_writer::value_type out_begin = reinterpret_cast(out_buffer); - wchar_writer::value_type out_end = utf_decoder::decode_utf8_block(data, size, out_begin); - - assert(out_end == out_begin + out_length); - (void)!out_end; - - return true; - } - - template PUGI__FN bool convert_buffer_utf16(char_t*& out_buffer, size_t& out_length, const void* contents, size_t size, opt_swap) - { - const uint16_t* data = static_cast(contents); - size_t length = size / sizeof(uint16_t); - - // first pass: get length in wchar_t units - out_length = utf_decoder::decode_utf16_block(data, length, 0); - - // allocate buffer of suitable length - out_buffer = static_cast(xml_memory::allocate((out_length > 0 ? out_length : 1) * sizeof(char_t))); - if (!out_buffer) return false; - - // second pass: convert utf16 input to wchar_t - wchar_writer::value_type out_begin = reinterpret_cast(out_buffer); - wchar_writer::value_type out_end = utf_decoder::decode_utf16_block(data, length, out_begin); - - assert(out_end == out_begin + out_length); - (void)!out_end; - - return true; - } - - template PUGI__FN bool convert_buffer_utf32(char_t*& out_buffer, size_t& out_length, const void* contents, size_t size, opt_swap) - { - const uint32_t* data = static_cast(contents); - size_t length = size / sizeof(uint32_t); - - // first pass: get length in wchar_t units - out_length = utf_decoder::decode_utf32_block(data, length, 0); - - // allocate buffer of suitable length - out_buffer = static_cast(xml_memory::allocate((out_length > 0 ? out_length : 1) * sizeof(char_t))); - if (!out_buffer) return false; - - // second pass: convert utf32 input to wchar_t - wchar_writer::value_type out_begin = reinterpret_cast(out_buffer); - wchar_writer::value_type out_end = utf_decoder::decode_utf32_block(data, length, out_begin); - - assert(out_end == out_begin + out_length); - (void)!out_end; - - return true; - } - - PUGI__FN bool convert_buffer_latin1(char_t*& out_buffer, size_t& out_length, const void* contents, size_t size) - { - const uint8_t* data = static_cast(contents); - - // get length in wchar_t units - out_length = size; - - // allocate buffer of suitable length - out_buffer = static_cast(xml_memory::allocate((out_length > 0 ? out_length : 1) * sizeof(char_t))); - if (!out_buffer) return false; - - // convert latin1 input to wchar_t - wchar_writer::value_type out_begin = reinterpret_cast(out_buffer); - wchar_writer::value_type out_end = utf_decoder::decode_latin1_block(data, size, out_begin); - - assert(out_end == out_begin + out_length); - (void)!out_end; - - return true; - } - - PUGI__FN bool convert_buffer(char_t*& out_buffer, size_t& out_length, xml_encoding encoding, const void* contents, size_t size, bool is_mutable) - { - // get native encoding - xml_encoding wchar_encoding = get_wchar_encoding(); - - // fast path: no conversion required - if (encoding == wchar_encoding) return get_mutable_buffer(out_buffer, out_length, contents, size, is_mutable); - - // only endian-swapping is required - if (need_endian_swap_utf(encoding, wchar_encoding)) return convert_buffer_endian_swap(out_buffer, out_length, contents, size, is_mutable); - - // source encoding is utf8 - if (encoding == encoding_utf8) return convert_buffer_utf8(out_buffer, out_length, contents, size); - - // source encoding is utf16 - if (encoding == encoding_utf16_be || encoding == encoding_utf16_le) - { - xml_encoding native_encoding = is_little_endian() ? encoding_utf16_le : encoding_utf16_be; - - return (native_encoding == encoding) ? - convert_buffer_utf16(out_buffer, out_length, contents, size, opt_false()) : - convert_buffer_utf16(out_buffer, out_length, contents, size, opt_true()); - } - - // source encoding is utf32 - if (encoding == encoding_utf32_be || encoding == encoding_utf32_le) - { - xml_encoding native_encoding = is_little_endian() ? encoding_utf32_le : encoding_utf32_be; - - return (native_encoding == encoding) ? - convert_buffer_utf32(out_buffer, out_length, contents, size, opt_false()) : - convert_buffer_utf32(out_buffer, out_length, contents, size, opt_true()); - } - - // source encoding is latin1 - if (encoding == encoding_latin1) return convert_buffer_latin1(out_buffer, out_length, contents, size); - - assert(!"Invalid encoding"); - return false; - } -#else - template PUGI__FN bool convert_buffer_utf16(char_t*& out_buffer, size_t& out_length, const void* contents, size_t size, opt_swap) - { - const uint16_t* data = static_cast(contents); - size_t length = size / sizeof(uint16_t); - - // first pass: get length in utf8 units - out_length = utf_decoder::decode_utf16_block(data, length, 0); - - // allocate buffer of suitable length - out_buffer = static_cast(xml_memory::allocate((out_length > 0 ? out_length : 1) * sizeof(char_t))); - if (!out_buffer) return false; - - // second pass: convert utf16 input to utf8 - uint8_t* out_begin = reinterpret_cast(out_buffer); - uint8_t* out_end = utf_decoder::decode_utf16_block(data, length, out_begin); - - assert(out_end == out_begin + out_length); - (void)!out_end; - - return true; - } - - template PUGI__FN bool convert_buffer_utf32(char_t*& out_buffer, size_t& out_length, const void* contents, size_t size, opt_swap) - { - const uint32_t* data = static_cast(contents); - size_t length = size / sizeof(uint32_t); - - // first pass: get length in utf8 units - out_length = utf_decoder::decode_utf32_block(data, length, 0); - - // allocate buffer of suitable length - out_buffer = static_cast(xml_memory::allocate((out_length > 0 ? out_length : 1) * sizeof(char_t))); - if (!out_buffer) return false; - - // second pass: convert utf32 input to utf8 - uint8_t* out_begin = reinterpret_cast(out_buffer); - uint8_t* out_end = utf_decoder::decode_utf32_block(data, length, out_begin); - - assert(out_end == out_begin + out_length); - (void)!out_end; - - return true; - } - - PUGI__FN size_t get_latin1_7bit_prefix_length(const uint8_t* data, size_t size) - { - for (size_t i = 0; i < size; ++i) - if (data[i] > 127) - return i; - - return size; - } - - PUGI__FN bool convert_buffer_latin1(char_t*& out_buffer, size_t& out_length, const void* contents, size_t size, bool is_mutable) - { - const uint8_t* data = static_cast(contents); - - // get size of prefix that does not need utf8 conversion - size_t prefix_length = get_latin1_7bit_prefix_length(data, size); - assert(prefix_length <= size); - - const uint8_t* postfix = data + prefix_length; - size_t postfix_length = size - prefix_length; - - // if no conversion is needed, just return the original buffer - if (postfix_length == 0) return get_mutable_buffer(out_buffer, out_length, contents, size, is_mutable); - - // first pass: get length in utf8 units - out_length = prefix_length + utf_decoder::decode_latin1_block(postfix, postfix_length, 0); - - // allocate buffer of suitable length - out_buffer = static_cast(xml_memory::allocate((out_length > 0 ? out_length : 1) * sizeof(char_t))); - if (!out_buffer) return false; - - // second pass: convert latin1 input to utf8 - memcpy(out_buffer, data, prefix_length); - - uint8_t* out_begin = reinterpret_cast(out_buffer); - uint8_t* out_end = utf_decoder::decode_latin1_block(postfix, postfix_length, out_begin + prefix_length); - - assert(out_end == out_begin + out_length); - (void)!out_end; - - return true; - } - - PUGI__FN bool convert_buffer(char_t*& out_buffer, size_t& out_length, xml_encoding encoding, const void* contents, size_t size, bool is_mutable) - { - // fast path: no conversion required - if (encoding == encoding_utf8) return get_mutable_buffer(out_buffer, out_length, contents, size, is_mutable); - - // source encoding is utf16 - if (encoding == encoding_utf16_be || encoding == encoding_utf16_le) - { - xml_encoding native_encoding = is_little_endian() ? encoding_utf16_le : encoding_utf16_be; - - return (native_encoding == encoding) ? - convert_buffer_utf16(out_buffer, out_length, contents, size, opt_false()) : - convert_buffer_utf16(out_buffer, out_length, contents, size, opt_true()); - } - - // source encoding is utf32 - if (encoding == encoding_utf32_be || encoding == encoding_utf32_le) - { - xml_encoding native_encoding = is_little_endian() ? encoding_utf32_le : encoding_utf32_be; - - return (native_encoding == encoding) ? - convert_buffer_utf32(out_buffer, out_length, contents, size, opt_false()) : - convert_buffer_utf32(out_buffer, out_length, contents, size, opt_true()); - } - - // source encoding is latin1 - if (encoding == encoding_latin1) return convert_buffer_latin1(out_buffer, out_length, contents, size, is_mutable); - - assert(!"Invalid encoding"); - return false; - } -#endif - - PUGI__FN size_t as_utf8_begin(const wchar_t* str, size_t length) - { - // get length in utf8 characters - return utf_decoder::decode_wchar_block(str, length, 0); - } - - PUGI__FN void as_utf8_end(char* buffer, size_t size, const wchar_t* str, size_t length) - { - // convert to utf8 - uint8_t* begin = reinterpret_cast(buffer); - uint8_t* end = utf_decoder::decode_wchar_block(str, length, begin); - - assert(begin + size == end); - (void)!end; - - // zero-terminate - buffer[size] = 0; - } - -#ifndef PUGIXML_NO_STL - PUGI__FN std::string as_utf8_impl(const wchar_t* str, size_t length) - { - // first pass: get length in utf8 characters - size_t size = as_utf8_begin(str, length); - - // allocate resulting string - std::string result; - result.resize(size); - - // second pass: convert to utf8 - if (size > 0) as_utf8_end(&result[0], size, str, length); - - return result; - } - - PUGI__FN std::basic_string as_wide_impl(const char* str, size_t size) - { - const uint8_t* data = reinterpret_cast(str); - - // first pass: get length in wchar_t units - size_t length = utf_decoder::decode_utf8_block(data, size, 0); - - // allocate resulting string - std::basic_string result; - result.resize(length); - - // second pass: convert to wchar_t - if (length > 0) - { - wchar_writer::value_type begin = reinterpret_cast(&result[0]); - wchar_writer::value_type end = utf_decoder::decode_utf8_block(data, size, begin); - - assert(begin + length == end); - (void)!end; - } - - return result; - } -#endif - - inline bool strcpy_insitu_allow(size_t length, uintptr_t allocated, char_t* target) - { - assert(target); - size_t target_length = strlength(target); - - // always reuse document buffer memory if possible - if (!allocated) return target_length >= length; - - // reuse heap memory if waste is not too great - const size_t reuse_threshold = 32; - - return target_length >= length && (target_length < reuse_threshold || target_length - length < target_length / 2); - } - - PUGI__FN bool strcpy_insitu(char_t*& dest, uintptr_t& header, uintptr_t header_mask, const char_t* source) - { - size_t source_length = strlength(source); - - if (source_length == 0) - { - // empty string and null pointer are equivalent, so just deallocate old memory - xml_allocator* alloc = reinterpret_cast(header & xml_memory_page_pointer_mask)->allocator; - - if (header & header_mask) alloc->deallocate_string(dest); - - // mark the string as not allocated - dest = 0; - header &= ~header_mask; - - return true; - } - else if (dest && strcpy_insitu_allow(source_length, header & header_mask, dest)) - { - // we can reuse old buffer, so just copy the new data (including zero terminator) - memcpy(dest, source, (source_length + 1) * sizeof(char_t)); - - return true; - } - else - { - xml_allocator* alloc = reinterpret_cast(header & xml_memory_page_pointer_mask)->allocator; - - // allocate new buffer - char_t* buf = alloc->allocate_string(source_length + 1); - if (!buf) return false; - - // copy the string (including zero terminator) - memcpy(buf, source, (source_length + 1) * sizeof(char_t)); - - // deallocate old buffer (*after* the above to protect against overlapping memory and/or allocation failures) - if (header & header_mask) alloc->deallocate_string(dest); - - // the string is now allocated, so set the flag - dest = buf; - header |= header_mask; - - return true; - } - } - - struct gap - { - char_t* end; - size_t size; - - gap(): end(0), size(0) - { - } - - // Push new gap, move s count bytes further (skipping the gap). - // Collapse previous gap. - void push(char_t*& s, size_t count) - { - if (end) // there was a gap already; collapse it - { - // Move [old_gap_end, new_gap_start) to [old_gap_start, ...) - assert(s >= end); - memmove(end - size, end, reinterpret_cast(s) - reinterpret_cast(end)); - } - - s += count; // end of current gap - - // "merge" two gaps - end = s; - size += count; - } - - // Collapse all gaps, return past-the-end pointer - char_t* flush(char_t* s) - { - if (end) - { - // Move [old_gap_end, current_pos) to [old_gap_start, ...) - assert(s >= end); - memmove(end - size, end, reinterpret_cast(s) - reinterpret_cast(end)); - - return s - size; - } - else return s; - } - }; - - PUGI__FN char_t* strconv_escape(char_t* s, gap& g) - { - char_t* stre = s + 1; - - switch (*stre) - { - case '#': // &#... - { - unsigned int ucsc = 0; - - if (stre[1] == 'x') // &#x... (hex code) - { - stre += 2; - - char_t ch = *stre; - - if (ch == ';') return stre; - - for (;;) - { - if (static_cast(ch - '0') <= 9) - ucsc = 16 * ucsc + (ch - '0'); - else if (static_cast((ch | ' ') - 'a') <= 5) - ucsc = 16 * ucsc + ((ch | ' ') - 'a' + 10); - else if (ch == ';') - break; - else // cancel - return stre; - - ch = *++stre; - } - - ++stre; - } - else // &#... (dec code) - { - char_t ch = *++stre; - - if (ch == ';') return stre; - - for (;;) - { - if (static_cast(ch - '0') <= 9) - ucsc = 10 * ucsc + (ch - '0'); - else if (ch == ';') - break; - else // cancel - return stre; - - ch = *++stre; - } - - ++stre; - } - - #ifdef PUGIXML_WCHAR_MODE - s = reinterpret_cast(wchar_writer::any(reinterpret_cast(s), ucsc)); - #else - s = reinterpret_cast(utf8_writer::any(reinterpret_cast(s), ucsc)); - #endif - - g.push(s, stre - s); - return stre; - } - - case 'a': // &a - { - ++stre; - - if (*stre == 'm') // &am - { - if (*++stre == 'p' && *++stre == ';') // & - { - *s++ = '&'; - ++stre; - - g.push(s, stre - s); - return stre; - } - } - else if (*stre == 'p') // &ap - { - if (*++stre == 'o' && *++stre == 's' && *++stre == ';') // ' - { - *s++ = '\''; - ++stre; - - g.push(s, stre - s); - return stre; - } - } - break; - } - - case 'g': // &g - { - if (*++stre == 't' && *++stre == ';') // > - { - *s++ = '>'; - ++stre; - - g.push(s, stre - s); - return stre; - } - break; - } - - case 'l': // &l - { - if (*++stre == 't' && *++stre == ';') // < - { - *s++ = '<'; - ++stre; - - g.push(s, stre - s); - return stre; - } - break; - } - - case 'q': // &q - { - if (*++stre == 'u' && *++stre == 'o' && *++stre == 't' && *++stre == ';') // " - { - *s++ = '"'; - ++stre; - - g.push(s, stre - s); - return stre; - } - break; - } - - default: - break; - } - - return stre; - } - - // Utility macro for last character handling - #define ENDSWITH(c, e) ((c) == (e) || ((c) == 0 && endch == (e))) - - PUGI__FN char_t* strconv_comment(char_t* s, char_t endch) - { - gap g; - - while (true) - { - while (!PUGI__IS_CHARTYPE(*s, ct_parse_comment)) ++s; - - if (*s == '\r') // Either a single 0x0d or 0x0d 0x0a pair - { - *s++ = '\n'; // replace first one with 0x0a - - if (*s == '\n') g.push(s, 1); - } - else if (s[0] == '-' && s[1] == '-' && ENDSWITH(s[2], '>')) // comment ends here - { - *g.flush(s) = 0; - - return s + (s[2] == '>' ? 3 : 2); - } - else if (*s == 0) - { - return 0; - } - else ++s; - } - } - - PUGI__FN char_t* strconv_cdata(char_t* s, char_t endch) - { - gap g; - - while (true) - { - while (!PUGI__IS_CHARTYPE(*s, ct_parse_cdata)) ++s; - - if (*s == '\r') // Either a single 0x0d or 0x0d 0x0a pair - { - *s++ = '\n'; // replace first one with 0x0a - - if (*s == '\n') g.push(s, 1); - } - else if (s[0] == ']' && s[1] == ']' && ENDSWITH(s[2], '>')) // CDATA ends here - { - *g.flush(s) = 0; - - return s + 1; - } - else if (*s == 0) - { - return 0; - } - else ++s; - } - } - - typedef char_t* (*strconv_pcdata_t)(char_t*); - - template struct strconv_pcdata_impl - { - static char_t* parse(char_t* s) - { - gap g; - - while (true) - { - while (!PUGI__IS_CHARTYPE(*s, ct_parse_pcdata)) ++s; - - if (*s == '<') // PCDATA ends here - { - *g.flush(s) = 0; - - return s + 1; - } - else if (opt_eol::value && *s == '\r') // Either a single 0x0d or 0x0d 0x0a pair - { - *s++ = '\n'; // replace first one with 0x0a - - if (*s == '\n') g.push(s, 1); - } - else if (opt_escape::value && *s == '&') - { - s = strconv_escape(s, g); - } - else if (*s == 0) - { - return s; - } - else ++s; - } - } - }; - - PUGI__FN strconv_pcdata_t get_strconv_pcdata(unsigned int optmask) - { - PUGI__STATIC_ASSERT(parse_escapes == 0x10 && parse_eol == 0x20); - - switch ((optmask >> 4) & 3) // get bitmask for flags (eol escapes) - { - case 0: return strconv_pcdata_impl::parse; - case 1: return strconv_pcdata_impl::parse; - case 2: return strconv_pcdata_impl::parse; - case 3: return strconv_pcdata_impl::parse; - default: return 0; // should not get here - } - } - - typedef char_t* (*strconv_attribute_t)(char_t*, char_t); - - template struct strconv_attribute_impl - { - static char_t* parse_wnorm(char_t* s, char_t end_quote) - { - gap g; - - // trim leading whitespaces - if (PUGI__IS_CHARTYPE(*s, ct_space)) - { - char_t* str = s; - - do ++str; - while (PUGI__IS_CHARTYPE(*str, ct_space)); - - g.push(s, str - s); - } - - while (true) - { - while (!PUGI__IS_CHARTYPE(*s, ct_parse_attr_ws | ct_space)) ++s; - - if (*s == end_quote) - { - char_t* str = g.flush(s); - - do *str-- = 0; - while (PUGI__IS_CHARTYPE(*str, ct_space)); - - return s + 1; - } - else if (PUGI__IS_CHARTYPE(*s, ct_space)) - { - *s++ = ' '; - - if (PUGI__IS_CHARTYPE(*s, ct_space)) - { - char_t* str = s + 1; - while (PUGI__IS_CHARTYPE(*str, ct_space)) ++str; - - g.push(s, str - s); - } - } - else if (opt_escape::value && *s == '&') - { - s = strconv_escape(s, g); - } - else if (!*s) - { - return 0; - } - else ++s; - } - } - - static char_t* parse_wconv(char_t* s, char_t end_quote) - { - gap g; - - while (true) - { - while (!PUGI__IS_CHARTYPE(*s, ct_parse_attr_ws)) ++s; - - if (*s == end_quote) - { - *g.flush(s) = 0; - - return s + 1; - } - else if (PUGI__IS_CHARTYPE(*s, ct_space)) - { - if (*s == '\r') - { - *s++ = ' '; - - if (*s == '\n') g.push(s, 1); - } - else *s++ = ' '; - } - else if (opt_escape::value && *s == '&') - { - s = strconv_escape(s, g); - } - else if (!*s) - { - return 0; - } - else ++s; - } - } - - static char_t* parse_eol(char_t* s, char_t end_quote) - { - gap g; - - while (true) - { - while (!PUGI__IS_CHARTYPE(*s, ct_parse_attr)) ++s; - - if (*s == end_quote) - { - *g.flush(s) = 0; - - return s + 1; - } - else if (*s == '\r') - { - *s++ = '\n'; - - if (*s == '\n') g.push(s, 1); - } - else if (opt_escape::value && *s == '&') - { - s = strconv_escape(s, g); - } - else if (!*s) - { - return 0; - } - else ++s; - } - } - - static char_t* parse_simple(char_t* s, char_t end_quote) - { - gap g; - - while (true) - { - while (!PUGI__IS_CHARTYPE(*s, ct_parse_attr)) ++s; - - if (*s == end_quote) - { - *g.flush(s) = 0; - - return s + 1; - } - else if (opt_escape::value && *s == '&') - { - s = strconv_escape(s, g); - } - else if (!*s) - { - return 0; - } - else ++s; - } - } - }; - - PUGI__FN strconv_attribute_t get_strconv_attribute(unsigned int optmask) - { - PUGI__STATIC_ASSERT(parse_escapes == 0x10 && parse_eol == 0x20 && parse_wconv_attribute == 0x40 && parse_wnorm_attribute == 0x80); - - switch ((optmask >> 4) & 15) // get bitmask for flags (wconv wnorm eol escapes) - { - case 0: return strconv_attribute_impl::parse_simple; - case 1: return strconv_attribute_impl::parse_simple; - case 2: return strconv_attribute_impl::parse_eol; - case 3: return strconv_attribute_impl::parse_eol; - case 4: return strconv_attribute_impl::parse_wconv; - case 5: return strconv_attribute_impl::parse_wconv; - case 6: return strconv_attribute_impl::parse_wconv; - case 7: return strconv_attribute_impl::parse_wconv; - case 8: return strconv_attribute_impl::parse_wnorm; - case 9: return strconv_attribute_impl::parse_wnorm; - case 10: return strconv_attribute_impl::parse_wnorm; - case 11: return strconv_attribute_impl::parse_wnorm; - case 12: return strconv_attribute_impl::parse_wnorm; - case 13: return strconv_attribute_impl::parse_wnorm; - case 14: return strconv_attribute_impl::parse_wnorm; - case 15: return strconv_attribute_impl::parse_wnorm; - default: return 0; // should not get here - } - } - - inline xml_parse_result make_parse_result(xml_parse_status status, ptrdiff_t offset = 0) - { - xml_parse_result result; - result.status = status; - result.offset = offset; - - return result; - } - - struct xml_parser - { - xml_allocator alloc; - char_t* error_offset; - xml_parse_status error_status; - - // Parser utilities. - #define PUGI__SKIPWS() { while (PUGI__IS_CHARTYPE(*s, ct_space)) ++s; } - #define PUGI__OPTSET(OPT) ( optmsk & (OPT) ) - #define PUGI__PUSHNODE(TYPE) { cursor = append_node(cursor, alloc, TYPE); if (!cursor) PUGI__THROW_ERROR(status_out_of_memory, s); } - #define PUGI__POPNODE() { cursor = cursor->parent; } - #define PUGI__SCANFOR(X) { while (*s != 0 && !(X)) ++s; } - #define PUGI__SCANWHILE(X) { while ((X)) ++s; } - #define PUGI__ENDSEG() { ch = *s; *s = 0; ++s; } - #define PUGI__THROW_ERROR(err, m) return error_offset = m, error_status = err, static_cast(0) - #define PUGI__CHECK_ERROR(err, m) { if (*s == 0) PUGI__THROW_ERROR(err, m); } - - xml_parser(const xml_allocator& alloc_): alloc(alloc_), error_offset(0), error_status(status_ok) - { - } - - // DOCTYPE consists of nested sections of the following possible types: - // , , "...", '...' - // - // - // First group can not contain nested groups - // Second group can contain nested groups of the same type - // Third group can contain all other groups - char_t* parse_doctype_primitive(char_t* s) - { - if (*s == '"' || *s == '\'') - { - // quoted string - char_t ch = *s++; - PUGI__SCANFOR(*s == ch); - if (!*s) PUGI__THROW_ERROR(status_bad_doctype, s); - - s++; - } - else if (s[0] == '<' && s[1] == '?') - { - // - s += 2; - PUGI__SCANFOR(s[0] == '?' && s[1] == '>'); // no need for ENDSWITH because ?> can't terminate proper doctype - if (!*s) PUGI__THROW_ERROR(status_bad_doctype, s); - - s += 2; - } - else if (s[0] == '<' && s[1] == '!' && s[2] == '-' && s[3] == '-') - { - s += 4; - PUGI__SCANFOR(s[0] == '-' && s[1] == '-' && s[2] == '>'); // no need for ENDSWITH because --> can't terminate proper doctype - if (!*s) PUGI__THROW_ERROR(status_bad_doctype, s); - - s += 4; - } - else PUGI__THROW_ERROR(status_bad_doctype, s); - - return s; - } - - char_t* parse_doctype_ignore(char_t* s) - { - assert(s[0] == '<' && s[1] == '!' && s[2] == '['); - s++; - - while (*s) - { - if (s[0] == '<' && s[1] == '!' && s[2] == '[') - { - // nested ignore section - s = parse_doctype_ignore(s); - if (!s) return s; - } - else if (s[0] == ']' && s[1] == ']' && s[2] == '>') - { - // ignore section end - s += 3; - - return s; - } - else s++; - } - - PUGI__THROW_ERROR(status_bad_doctype, s); - } - - char_t* parse_doctype_group(char_t* s, char_t endch, bool toplevel) - { - assert(s[0] == '<' && s[1] == '!'); - s++; - - while (*s) - { - if (s[0] == '<' && s[1] == '!' && s[2] != '-') - { - if (s[2] == '[') - { - // ignore - s = parse_doctype_ignore(s); - if (!s) return s; - } - else - { - // some control group - s = parse_doctype_group(s, endch, false); - if (!s) return s; - } - } - else if (s[0] == '<' || s[0] == '"' || s[0] == '\'') - { - // unknown tag (forbidden), or some primitive group - s = parse_doctype_primitive(s); - if (!s) return s; - } - else if (*s == '>') - { - s++; - - return s; - } - else s++; - } - - if (!toplevel || endch != '>') PUGI__THROW_ERROR(status_bad_doctype, s); - - return s; - } - - char_t* parse_exclamation(char_t* s, xml_node_struct* cursor, unsigned int optmsk, char_t endch) - { - // parse node contents, starting with exclamation mark - ++s; - - if (*s == '-') // 'value = s; // Save the offset. - } - - if (PUGI__OPTSET(parse_eol) && PUGI__OPTSET(parse_comments)) - { - s = strconv_comment(s, endch); - - if (!s) PUGI__THROW_ERROR(status_bad_comment, cursor->value); - } - else - { - // Scan for terminating '-->'. - PUGI__SCANFOR(s[0] == '-' && s[1] == '-' && ENDSWITH(s[2], '>')); - PUGI__CHECK_ERROR(status_bad_comment, s); - - if (PUGI__OPTSET(parse_comments)) - *s = 0; // Zero-terminate this segment at the first terminating '-'. - - s += (s[2] == '>' ? 3 : 2); // Step over the '\0->'. - } - } - else PUGI__THROW_ERROR(status_bad_comment, s); - } - else if (*s == '[') - { - // 'value = s; // Save the offset. - - if (PUGI__OPTSET(parse_eol)) - { - s = strconv_cdata(s, endch); - - if (!s) PUGI__THROW_ERROR(status_bad_cdata, cursor->value); - } - else - { - // Scan for terminating ']]>'. - PUGI__SCANFOR(s[0] == ']' && s[1] == ']' && ENDSWITH(s[2], '>')); - PUGI__CHECK_ERROR(status_bad_cdata, s); - - *s++ = 0; // Zero-terminate this segment. - } - } - else // Flagged for discard, but we still have to scan for the terminator. - { - // Scan for terminating ']]>'. - PUGI__SCANFOR(s[0] == ']' && s[1] == ']' && ENDSWITH(s[2], '>')); - PUGI__CHECK_ERROR(status_bad_cdata, s); - - ++s; - } - - s += (s[1] == '>' ? 2 : 1); // Step over the last ']>'. - } - else PUGI__THROW_ERROR(status_bad_cdata, s); - } - else if (s[0] == 'D' && s[1] == 'O' && s[2] == 'C' && s[3] == 'T' && s[4] == 'Y' && s[5] == 'P' && ENDSWITH(s[6], 'E')) - { - s -= 2; - - if (cursor->parent) PUGI__THROW_ERROR(status_bad_doctype, s); - - char_t* mark = s + 9; - - s = parse_doctype_group(s, endch, true); - if (!s) return s; - - if (PUGI__OPTSET(parse_doctype)) - { - while (PUGI__IS_CHARTYPE(*mark, ct_space)) ++mark; - - PUGI__PUSHNODE(node_doctype); - - cursor->value = mark; - - assert((s[0] == 0 && endch == '>') || s[-1] == '>'); - s[*s == 0 ? 0 : -1] = 0; - - PUGI__POPNODE(); - } - } - else if (*s == 0 && endch == '-') PUGI__THROW_ERROR(status_bad_comment, s); - else if (*s == 0 && endch == '[') PUGI__THROW_ERROR(status_bad_cdata, s); - else PUGI__THROW_ERROR(status_unrecognized_tag, s); - - return s; - } - - char_t* parse_question(char_t* s, xml_node_struct*& ref_cursor, unsigned int optmsk, char_t endch) - { - // load into registers - xml_node_struct* cursor = ref_cursor; - char_t ch = 0; - - // parse node contents, starting with question mark - ++s; - - // read PI target - char_t* target = s; - - if (!PUGI__IS_CHARTYPE(*s, ct_start_symbol)) PUGI__THROW_ERROR(status_bad_pi, s); - - PUGI__SCANWHILE(PUGI__IS_CHARTYPE(*s, ct_symbol)); - PUGI__CHECK_ERROR(status_bad_pi, s); - - // determine node type; stricmp / strcasecmp is not portable - bool declaration = (target[0] | ' ') == 'x' && (target[1] | ' ') == 'm' && (target[2] | ' ') == 'l' && target + 3 == s; - - if (declaration ? PUGI__OPTSET(parse_declaration) : PUGI__OPTSET(parse_pi)) - { - if (declaration) - { - // disallow non top-level declarations - if (cursor->parent) PUGI__THROW_ERROR(status_bad_pi, s); - - PUGI__PUSHNODE(node_declaration); - } - else - { - PUGI__PUSHNODE(node_pi); - } - - cursor->name = target; - - PUGI__ENDSEG(); - - // parse value/attributes - if (ch == '?') - { - // empty node - if (!ENDSWITH(*s, '>')) PUGI__THROW_ERROR(status_bad_pi, s); - s += (*s == '>'); - - PUGI__POPNODE(); - } - else if (PUGI__IS_CHARTYPE(ch, ct_space)) - { - PUGI__SKIPWS(); - - // scan for tag end - char_t* value = s; - - PUGI__SCANFOR(s[0] == '?' && ENDSWITH(s[1], '>')); - PUGI__CHECK_ERROR(status_bad_pi, s); - - if (declaration) - { - // replace ending ? with / so that 'element' terminates properly - *s = '/'; - - // we exit from this function with cursor at node_declaration, which is a signal to parse() to go to LOC_ATTRIBUTES - s = value; - } - else - { - // store value and step over > - cursor->value = value; - PUGI__POPNODE(); - - PUGI__ENDSEG(); - - s += (*s == '>'); - } - } - else PUGI__THROW_ERROR(status_bad_pi, s); - } - else - { - // scan for tag end - PUGI__SCANFOR(s[0] == '?' && ENDSWITH(s[1], '>')); - PUGI__CHECK_ERROR(status_bad_pi, s); - - s += (s[1] == '>' ? 2 : 1); - } - - // store from registers - ref_cursor = cursor; - - return s; - } - - char_t* parse(char_t* s, xml_node_struct* xmldoc, unsigned int optmsk, char_t endch) - { - strconv_attribute_t strconv_attribute = get_strconv_attribute(optmsk); - strconv_pcdata_t strconv_pcdata = get_strconv_pcdata(optmsk); - - char_t ch = 0; - xml_node_struct* cursor = xmldoc; - char_t* mark = s; - - while (*s != 0) - { - if (*s == '<') - { - ++s; - - LOC_TAG: - if (PUGI__IS_CHARTYPE(*s, ct_start_symbol)) // '<#...' - { - PUGI__PUSHNODE(node_element); // Append a new node to the tree. - - cursor->name = s; - - PUGI__SCANWHILE(PUGI__IS_CHARTYPE(*s, ct_symbol)); // Scan for a terminator. - PUGI__ENDSEG(); // Save char in 'ch', terminate & step over. - - if (ch == '>') - { - // end of tag - } - else if (PUGI__IS_CHARTYPE(ch, ct_space)) - { - LOC_ATTRIBUTES: - while (true) - { - PUGI__SKIPWS(); // Eat any whitespace. - - if (PUGI__IS_CHARTYPE(*s, ct_start_symbol)) // <... #... - { - xml_attribute_struct* a = append_attribute_ll(cursor, alloc); // Make space for this attribute. - if (!a) PUGI__THROW_ERROR(status_out_of_memory, s); - - a->name = s; // Save the offset. - - PUGI__SCANWHILE(PUGI__IS_CHARTYPE(*s, ct_symbol)); // Scan for a terminator. - PUGI__CHECK_ERROR(status_bad_attribute, s); //$ redundant, left for performance - - PUGI__ENDSEG(); // Save char in 'ch', terminate & step over. - PUGI__CHECK_ERROR(status_bad_attribute, s); //$ redundant, left for performance - - if (PUGI__IS_CHARTYPE(ch, ct_space)) - { - PUGI__SKIPWS(); // Eat any whitespace. - PUGI__CHECK_ERROR(status_bad_attribute, s); //$ redundant, left for performance - - ch = *s; - ++s; - } - - if (ch == '=') // '<... #=...' - { - PUGI__SKIPWS(); // Eat any whitespace. - - if (*s == '"' || *s == '\'') // '<... #="...' - { - ch = *s; // Save quote char to avoid breaking on "''" -or- '""'. - ++s; // Step over the quote. - a->value = s; // Save the offset. - - s = strconv_attribute(s, ch); - - if (!s) PUGI__THROW_ERROR(status_bad_attribute, a->value); - - // After this line the loop continues from the start; - // Whitespaces, / and > are ok, symbols and EOF are wrong, - // everything else will be detected - if (PUGI__IS_CHARTYPE(*s, ct_start_symbol)) PUGI__THROW_ERROR(status_bad_attribute, s); - } - else PUGI__THROW_ERROR(status_bad_attribute, s); - } - else PUGI__THROW_ERROR(status_bad_attribute, s); - } - else if (*s == '/') - { - ++s; - - if (*s == '>') - { - PUGI__POPNODE(); - s++; - break; - } - else if (*s == 0 && endch == '>') - { - PUGI__POPNODE(); - break; - } - else PUGI__THROW_ERROR(status_bad_start_element, s); - } - else if (*s == '>') - { - ++s; - - break; - } - else if (*s == 0 && endch == '>') - { - break; - } - else PUGI__THROW_ERROR(status_bad_start_element, s); - } - - // !!! - } - else if (ch == '/') // '<#.../' - { - if (!ENDSWITH(*s, '>')) PUGI__THROW_ERROR(status_bad_start_element, s); - - PUGI__POPNODE(); // Pop. - - s += (*s == '>'); - } - else if (ch == 0) - { - // we stepped over null terminator, backtrack & handle closing tag - --s; - - if (endch != '>') PUGI__THROW_ERROR(status_bad_start_element, s); - } - else PUGI__THROW_ERROR(status_bad_start_element, s); - } - else if (*s == '/') - { - ++s; - - char_t* name = cursor->name; - if (!name) PUGI__THROW_ERROR(status_end_element_mismatch, s); - - while (PUGI__IS_CHARTYPE(*s, ct_symbol)) - { - if (*s++ != *name++) PUGI__THROW_ERROR(status_end_element_mismatch, s); - } - - if (*name) - { - if (*s == 0 && name[0] == endch && name[1] == 0) PUGI__THROW_ERROR(status_bad_end_element, s); - else PUGI__THROW_ERROR(status_end_element_mismatch, s); - } - - PUGI__POPNODE(); // Pop. - - PUGI__SKIPWS(); - - if (*s == 0) - { - if (endch != '>') PUGI__THROW_ERROR(status_bad_end_element, s); - } - else - { - if (*s != '>') PUGI__THROW_ERROR(status_bad_end_element, s); - ++s; - } - } - else if (*s == '?') // 'header & xml_memory_page_type_mask) + 1 == node_declaration) goto LOC_ATTRIBUTES; - } - else if (*s == '!') // 'first_child) continue; - } - } - - s = mark; - - if (cursor->parent) - { - PUGI__PUSHNODE(node_pcdata); // Append a new node on the tree. - cursor->value = s; // Save the offset. - - s = strconv_pcdata(s); - - PUGI__POPNODE(); // Pop since this is a standalone. - - if (!*s) break; - } - else - { - PUGI__SCANFOR(*s == '<'); // '...<' - if (!*s) break; - - ++s; - } - - // We're after '<' - goto LOC_TAG; - } - } - - // check that last tag is closed - if (cursor != xmldoc) PUGI__THROW_ERROR(status_end_element_mismatch, s); - - return s; - } - - static xml_parse_result parse(char_t* buffer, size_t length, xml_node_struct* root, unsigned int optmsk) - { - xml_document_struct* xmldoc = static_cast(root); - - // store buffer for offset_debug - xmldoc->buffer = buffer; - - // early-out for empty documents - if (length == 0) return make_parse_result(status_ok); - - // create parser on stack - xml_parser parser(*xmldoc); - - // save last character and make buffer zero-terminated (speeds up parsing) - char_t endch = buffer[length - 1]; - buffer[length - 1] = 0; - - // perform actual parsing - parser.parse(buffer, xmldoc, optmsk, endch); - - xml_parse_result result = make_parse_result(parser.error_status, parser.error_offset ? parser.error_offset - buffer : 0); - assert(result.offset >= 0 && static_cast(result.offset) <= length); - - // update allocator state - *static_cast(xmldoc) = parser.alloc; - - // since we removed last character, we have to handle the only possible false positive - if (result && endch == '<') - { - // there's no possible well-formed document with < at the end - return make_parse_result(status_unrecognized_tag, length); - } - - return result; - } - }; - - // Output facilities - PUGI__FN xml_encoding get_write_native_encoding() - { - #ifdef PUGIXML_WCHAR_MODE - return get_wchar_encoding(); - #else - return encoding_utf8; - #endif - } - - PUGI__FN xml_encoding get_write_encoding(xml_encoding encoding) - { - // replace wchar encoding with utf implementation - if (encoding == encoding_wchar) return get_wchar_encoding(); - - // replace utf16 encoding with utf16 with specific endianness - if (encoding == encoding_utf16) return is_little_endian() ? encoding_utf16_le : encoding_utf16_be; - - // replace utf32 encoding with utf32 with specific endianness - if (encoding == encoding_utf32) return is_little_endian() ? encoding_utf32_le : encoding_utf32_be; - - // only do autodetection if no explicit encoding is requested - if (encoding != encoding_auto) return encoding; - - // assume utf8 encoding - return encoding_utf8; - } - -#ifdef PUGIXML_WCHAR_MODE - PUGI__FN size_t get_valid_length(const char_t* data, size_t length) - { - assert(length > 0); - - // discard last character if it's the lead of a surrogate pair - return (sizeof(wchar_t) == 2 && static_cast(static_cast(data[length - 1]) - 0xD800) < 0x400) ? length - 1 : length; - } - - PUGI__FN size_t convert_buffer(char_t* r_char, uint8_t* r_u8, uint16_t* r_u16, uint32_t* r_u32, const char_t* data, size_t length, xml_encoding encoding) - { - // only endian-swapping is required - if (need_endian_swap_utf(encoding, get_wchar_encoding())) - { - convert_wchar_endian_swap(r_char, data, length); - - return length * sizeof(char_t); - } - - // convert to utf8 - if (encoding == encoding_utf8) - { - uint8_t* dest = r_u8; - uint8_t* end = utf_decoder::decode_wchar_block(data, length, dest); - - return static_cast(end - dest); - } - - // convert to utf16 - if (encoding == encoding_utf16_be || encoding == encoding_utf16_le) - { - uint16_t* dest = r_u16; - - // convert to native utf16 - uint16_t* end = utf_decoder::decode_wchar_block(data, length, dest); - - // swap if necessary - xml_encoding native_encoding = is_little_endian() ? encoding_utf16_le : encoding_utf16_be; - - if (native_encoding != encoding) convert_utf_endian_swap(dest, dest, static_cast(end - dest)); - - return static_cast(end - dest) * sizeof(uint16_t); - } - - // convert to utf32 - if (encoding == encoding_utf32_be || encoding == encoding_utf32_le) - { - uint32_t* dest = r_u32; - - // convert to native utf32 - uint32_t* end = utf_decoder::decode_wchar_block(data, length, dest); - - // swap if necessary - xml_encoding native_encoding = is_little_endian() ? encoding_utf32_le : encoding_utf32_be; - - if (native_encoding != encoding) convert_utf_endian_swap(dest, dest, static_cast(end - dest)); - - return static_cast(end - dest) * sizeof(uint32_t); - } - - // convert to latin1 - if (encoding == encoding_latin1) - { - uint8_t* dest = r_u8; - uint8_t* end = utf_decoder::decode_wchar_block(data, length, dest); - - return static_cast(end - dest); - } - - assert(!"Invalid encoding"); - return 0; - } -#else - PUGI__FN size_t get_valid_length(const char_t* data, size_t length) - { - assert(length > 4); - - for (size_t i = 1; i <= 4; ++i) - { - uint8_t ch = static_cast(data[length - i]); - - // either a standalone character or a leading one - if ((ch & 0xc0) != 0x80) return length - i; - } - - // there are four non-leading characters at the end, sequence tail is broken so might as well process the whole chunk - return length; - } - - PUGI__FN size_t convert_buffer(char_t* /* r_char */, uint8_t* r_u8, uint16_t* r_u16, uint32_t* r_u32, const char_t* data, size_t length, xml_encoding encoding) - { - if (encoding == encoding_utf16_be || encoding == encoding_utf16_le) - { - uint16_t* dest = r_u16; - - // convert to native utf16 - uint16_t* end = utf_decoder::decode_utf8_block(reinterpret_cast(data), length, dest); - - // swap if necessary - xml_encoding native_encoding = is_little_endian() ? encoding_utf16_le : encoding_utf16_be; - - if (native_encoding != encoding) convert_utf_endian_swap(dest, dest, static_cast(end - dest)); - - return static_cast(end - dest) * sizeof(uint16_t); - } - - if (encoding == encoding_utf32_be || encoding == encoding_utf32_le) - { - uint32_t* dest = r_u32; - - // convert to native utf32 - uint32_t* end = utf_decoder::decode_utf8_block(reinterpret_cast(data), length, dest); - - // swap if necessary - xml_encoding native_encoding = is_little_endian() ? encoding_utf32_le : encoding_utf32_be; - - if (native_encoding != encoding) convert_utf_endian_swap(dest, dest, static_cast(end - dest)); - - return static_cast(end - dest) * sizeof(uint32_t); - } - - if (encoding == encoding_latin1) - { - uint8_t* dest = r_u8; - uint8_t* end = utf_decoder::decode_utf8_block(reinterpret_cast(data), length, dest); - - return static_cast(end - dest); - } - - assert(!"Invalid encoding"); - return 0; - } -#endif - - class xml_buffered_writer - { - xml_buffered_writer(const xml_buffered_writer&); - xml_buffered_writer& operator=(const xml_buffered_writer&); - - public: - xml_buffered_writer(xml_writer& writer_, xml_encoding user_encoding): writer(writer_), bufsize(0), encoding(get_write_encoding(user_encoding)) - { - PUGI__STATIC_ASSERT(bufcapacity >= 8); - } - - ~xml_buffered_writer() - { - flush(); - } - - void flush() - { - flush(buffer, bufsize); - bufsize = 0; - } - - void flush(const char_t* data, size_t size) - { - if (size == 0) return; - - // fast path, just write data - if (encoding == get_write_native_encoding()) - writer.write(data, size * sizeof(char_t)); - else - { - // convert chunk - size_t result = convert_buffer(scratch.data_char, scratch.data_u8, scratch.data_u16, scratch.data_u32, data, size, encoding); - assert(result <= sizeof(scratch)); - - // write data - writer.write(scratch.data_u8, result); - } - } - - void write(const char_t* data, size_t length) - { - if (bufsize + length > bufcapacity) - { - // flush the remaining buffer contents - flush(); - - // handle large chunks - if (length > bufcapacity) - { - if (encoding == get_write_native_encoding()) - { - // fast path, can just write data chunk - writer.write(data, length * sizeof(char_t)); - return; - } - - // need to convert in suitable chunks - while (length > bufcapacity) - { - // get chunk size by selecting such number of characters that are guaranteed to fit into scratch buffer - // and form a complete codepoint sequence (i.e. discard start of last codepoint if necessary) - size_t chunk_size = get_valid_length(data, bufcapacity); - - // convert chunk and write - flush(data, chunk_size); - - // iterate - data += chunk_size; - length -= chunk_size; - } - - // small tail is copied below - bufsize = 0; - } - } - - memcpy(buffer + bufsize, data, length * sizeof(char_t)); - bufsize += length; - } - - void write(const char_t* data) - { - write(data, strlength(data)); - } - - void write(char_t d0) - { - if (bufsize + 1 > bufcapacity) flush(); - - buffer[bufsize + 0] = d0; - bufsize += 1; - } - - void write(char_t d0, char_t d1) - { - if (bufsize + 2 > bufcapacity) flush(); - - buffer[bufsize + 0] = d0; - buffer[bufsize + 1] = d1; - bufsize += 2; - } - - void write(char_t d0, char_t d1, char_t d2) - { - if (bufsize + 3 > bufcapacity) flush(); - - buffer[bufsize + 0] = d0; - buffer[bufsize + 1] = d1; - buffer[bufsize + 2] = d2; - bufsize += 3; - } - - void write(char_t d0, char_t d1, char_t d2, char_t d3) - { - if (bufsize + 4 > bufcapacity) flush(); - - buffer[bufsize + 0] = d0; - buffer[bufsize + 1] = d1; - buffer[bufsize + 2] = d2; - buffer[bufsize + 3] = d3; - bufsize += 4; - } - - void write(char_t d0, char_t d1, char_t d2, char_t d3, char_t d4) - { - if (bufsize + 5 > bufcapacity) flush(); - - buffer[bufsize + 0] = d0; - buffer[bufsize + 1] = d1; - buffer[bufsize + 2] = d2; - buffer[bufsize + 3] = d3; - buffer[bufsize + 4] = d4; - bufsize += 5; - } - - void write(char_t d0, char_t d1, char_t d2, char_t d3, char_t d4, char_t d5) - { - if (bufsize + 6 > bufcapacity) flush(); - - buffer[bufsize + 0] = d0; - buffer[bufsize + 1] = d1; - buffer[bufsize + 2] = d2; - buffer[bufsize + 3] = d3; - buffer[bufsize + 4] = d4; - buffer[bufsize + 5] = d5; - bufsize += 6; - } - - // utf8 maximum expansion: x4 (-> utf32) - // utf16 maximum expansion: x2 (-> utf32) - // utf32 maximum expansion: x1 - enum - { - bufcapacitybytes = - #ifdef PUGIXML_MEMORY_OUTPUT_STACK - PUGIXML_MEMORY_OUTPUT_STACK - #else - 10240 - #endif - , - bufcapacity = bufcapacitybytes / (sizeof(char_t) + 4) - }; - - char_t buffer[bufcapacity]; - - union - { - uint8_t data_u8[4 * bufcapacity]; - uint16_t data_u16[2 * bufcapacity]; - uint32_t data_u32[bufcapacity]; - char_t data_char[bufcapacity]; - } scratch; - - xml_writer& writer; - size_t bufsize; - xml_encoding encoding; - }; - - PUGI__FN void text_output_escaped(xml_buffered_writer& writer, const char_t* s, chartypex_t type) - { - while (*s) - { - const char_t* prev = s; - - // While *s is a usual symbol - while (!PUGI__IS_CHARTYPEX(*s, type)) ++s; - - writer.write(prev, static_cast(s - prev)); - - switch (*s) - { - case 0: break; - case '&': - writer.write('&', 'a', 'm', 'p', ';'); - ++s; - break; - case '<': - writer.write('&', 'l', 't', ';'); - ++s; - break; - case '>': - writer.write('&', 'g', 't', ';'); - ++s; - break; - case '"': - writer.write('&', 'q', 'u', 'o', 't', ';'); - ++s; - break; - default: // s is not a usual symbol - { - unsigned int ch = static_cast(*s++); - assert(ch < 32); - - writer.write('&', '#', static_cast((ch / 10) + '0'), static_cast((ch % 10) + '0'), ';'); - } - } - } - } - - PUGI__FN void text_output(xml_buffered_writer& writer, const char_t* s, chartypex_t type, unsigned int flags) - { - if (flags & format_no_escapes) - writer.write(s); - else - text_output_escaped(writer, s, type); - } - - PUGI__FN void text_output_cdata(xml_buffered_writer& writer, const char_t* s) - { - do - { - writer.write('<', '!', '[', 'C', 'D'); - writer.write('A', 'T', 'A', '['); - - const char_t* prev = s; - - // look for ]]> sequence - we can't output it as is since it terminates CDATA - while (*s && !(s[0] == ']' && s[1] == ']' && s[2] == '>')) ++s; - - // skip ]] if we stopped at ]]>, > will go to the next CDATA section - if (*s) s += 2; - - writer.write(prev, static_cast(s - prev)); - - writer.write(']', ']', '>'); - } - while (*s); - } - - PUGI__FN void node_output_attributes(xml_buffered_writer& writer, const xml_node& node, unsigned int flags) - { - const char_t* default_name = PUGIXML_TEXT(":anonymous"); - - for (xml_attribute a = node.first_attribute(); a; a = a.next_attribute()) - { - writer.write(' '); - writer.write(a.name()[0] ? a.name() : default_name); - writer.write('=', '"'); - - text_output(writer, a.value(), ctx_special_attr, flags); - - writer.write('"'); - } - } - - PUGI__FN void node_output(xml_buffered_writer& writer, const xml_node& node, const char_t* indent, unsigned int flags, unsigned int depth) - { - const char_t* default_name = PUGIXML_TEXT(":anonymous"); - - if ((flags & format_indent) != 0 && (flags & format_raw) == 0) - for (unsigned int i = 0; i < depth; ++i) writer.write(indent); - - switch (node.type()) - { - case node_document: - { - for (xml_node n = node.first_child(); n; n = n.next_sibling()) - node_output(writer, n, indent, flags, depth); - break; - } - - case node_element: - { - const char_t* name = node.name()[0] ? node.name() : default_name; - - writer.write('<'); - writer.write(name); - - node_output_attributes(writer, node, flags); - - if (flags & format_raw) - { - if (!node.first_child()) - writer.write(' ', '/', '>'); - else - { - writer.write('>'); - - for (xml_node n = node.first_child(); n; n = n.next_sibling()) - node_output(writer, n, indent, flags, depth + 1); - - writer.write('<', '/'); - writer.write(name); - writer.write('>'); - } - } - else if (!node.first_child()) - writer.write(' ', '/', '>', '\n'); - else if (node.first_child() == node.last_child() && (node.first_child().type() == node_pcdata || node.first_child().type() == node_cdata)) - { - writer.write('>'); - - if (node.first_child().type() == node_pcdata) - text_output(writer, node.first_child().value(), ctx_special_pcdata, flags); - else - text_output_cdata(writer, node.first_child().value()); - - writer.write('<', '/'); - writer.write(name); - writer.write('>', '\n'); - } - else - { - writer.write('>', '\n'); - - for (xml_node n = node.first_child(); n; n = n.next_sibling()) - node_output(writer, n, indent, flags, depth + 1); - - if ((flags & format_indent) != 0 && (flags & format_raw) == 0) - for (unsigned int i = 0; i < depth; ++i) writer.write(indent); - - writer.write('<', '/'); - writer.write(name); - writer.write('>', '\n'); - } - - break; - } - - case node_pcdata: - text_output(writer, node.value(), ctx_special_pcdata, flags); - if ((flags & format_raw) == 0) writer.write('\n'); - break; - - case node_cdata: - text_output_cdata(writer, node.value()); - if ((flags & format_raw) == 0) writer.write('\n'); - break; - - case node_comment: - writer.write('<', '!', '-', '-'); - writer.write(node.value()); - writer.write('-', '-', '>'); - if ((flags & format_raw) == 0) writer.write('\n'); - break; - - case node_pi: - case node_declaration: - writer.write('<', '?'); - writer.write(node.name()[0] ? node.name() : default_name); - - if (node.type() == node_declaration) - { - node_output_attributes(writer, node, flags); - } - else if (node.value()[0]) - { - writer.write(' '); - writer.write(node.value()); - } - - writer.write('?', '>'); - if ((flags & format_raw) == 0) writer.write('\n'); - break; - - case node_doctype: - writer.write('<', '!', 'D', 'O', 'C'); - writer.write('T', 'Y', 'P', 'E'); - - if (node.value()[0]) - { - writer.write(' '); - writer.write(node.value()); - } - - writer.write('>'); - if ((flags & format_raw) == 0) writer.write('\n'); - break; - - default: - assert(!"Invalid node type"); - } - } - - inline bool has_declaration(const xml_node& node) - { - for (xml_node child = node.first_child(); child; child = child.next_sibling()) - { - xml_node_type type = child.type(); - - if (type == node_declaration) return true; - if (type == node_element) return false; - } - - return false; - } - - inline bool allow_insert_child(xml_node_type parent, xml_node_type child) - { - if (parent != node_document && parent != node_element) return false; - if (child == node_document || child == node_null) return false; - if (parent != node_document && (child == node_declaration || child == node_doctype)) return false; - - return true; - } - - PUGI__FN void recursive_copy_skip(xml_node& dest, const xml_node& source, const xml_node& skip) - { - assert(dest.type() == source.type()); - - switch (source.type()) - { - case node_element: - { - dest.set_name(source.name()); - - for (xml_attribute a = source.first_attribute(); a; a = a.next_attribute()) - dest.append_attribute(a.name()).set_value(a.value()); - - for (xml_node c = source.first_child(); c; c = c.next_sibling()) - { - if (c == skip) continue; - - xml_node cc = dest.append_child(c.type()); - assert(cc); - - recursive_copy_skip(cc, c, skip); - } - - break; - } - - case node_pcdata: - case node_cdata: - case node_comment: - case node_doctype: - dest.set_value(source.value()); - break; - - case node_pi: - dest.set_name(source.name()); - dest.set_value(source.value()); - break; - - case node_declaration: - { - dest.set_name(source.name()); - - for (xml_attribute a = source.first_attribute(); a; a = a.next_attribute()) - dest.append_attribute(a.name()).set_value(a.value()); - - break; - } - - default: - assert(!"Invalid node type"); - } - } - - inline bool is_text_node(xml_node_struct* node) - { - xml_node_type type = static_cast((node->header & impl::xml_memory_page_type_mask) + 1); - - return type == node_pcdata || type == node_cdata; - } - - // get value with conversion functions - PUGI__FN int get_value_int(const char_t* value, int def) - { - if (!value) return def; - - #ifdef PUGIXML_WCHAR_MODE - return static_cast(wcstol(value, 0, 10)); - #else - return static_cast(strtol(value, 0, 10)); - #endif - } - - PUGI__FN unsigned int get_value_uint(const char_t* value, unsigned int def) - { - if (!value) return def; - - #ifdef PUGIXML_WCHAR_MODE - return static_cast(wcstoul(value, 0, 10)); - #else - return static_cast(strtoul(value, 0, 10)); - #endif - } - - PUGI__FN double get_value_double(const char_t* value, double def) - { - if (!value) return def; - - #ifdef PUGIXML_WCHAR_MODE - return wcstod(value, 0); - #else - return strtod(value, 0); - #endif - } - - PUGI__FN float get_value_float(const char_t* value, float def) - { - if (!value) return def; - - #ifdef PUGIXML_WCHAR_MODE - return static_cast(wcstod(value, 0)); - #else - return static_cast(strtod(value, 0)); - #endif - } - - PUGI__FN bool get_value_bool(const char_t* value, bool def) - { - if (!value) return def; - - // only look at first char - char_t first = *value; - - // 1*, t* (true), T* (True), y* (yes), Y* (YES) - return (first == '1' || first == 't' || first == 'T' || first == 'y' || first == 'Y'); - } - - // set value with conversion functions - PUGI__FN bool set_value_buffer(char_t*& dest, uintptr_t& header, uintptr_t header_mask, char (&buf)[128]) - { - #ifdef PUGIXML_WCHAR_MODE - char_t wbuf[128]; - impl::widen_ascii(wbuf, buf); - - return strcpy_insitu(dest, header, header_mask, wbuf); - #else - return strcpy_insitu(dest, header, header_mask, buf); - #endif - } - - PUGI__FN bool set_value_convert(char_t*& dest, uintptr_t& header, uintptr_t header_mask, int value) - { - char buf[128]; - sprintf(buf, "%d", value); - - return set_value_buffer(dest, header, header_mask, buf); - } - - PUGI__FN bool set_value_convert(char_t*& dest, uintptr_t& header, uintptr_t header_mask, unsigned int value) - { - char buf[128]; - sprintf(buf, "%u", value); - - return set_value_buffer(dest, header, header_mask, buf); - } - - PUGI__FN bool set_value_convert(char_t*& dest, uintptr_t& header, uintptr_t header_mask, double value) - { - char buf[128]; - sprintf(buf, "%g", value); - - return set_value_buffer(dest, header, header_mask, buf); - } - - PUGI__FN bool set_value_convert(char_t*& dest, uintptr_t& header, uintptr_t header_mask, bool value) - { - return strcpy_insitu(dest, header, header_mask, value ? PUGIXML_TEXT("true") : PUGIXML_TEXT("false")); - } - - // we need to get length of entire file to load it in memory; the only (relatively) sane way to do it is via seek/tell trick - PUGI__FN xml_parse_status get_file_size(FILE* file, size_t& out_result) - { - #if defined(PUGI__MSVC_CRT_VERSION) && PUGI__MSVC_CRT_VERSION >= 1400 && !defined(_WIN32_WCE) - // there are 64-bit versions of fseek/ftell, let's use them - typedef __int64 length_type; - - _fseeki64(file, 0, SEEK_END); - length_type length = _ftelli64(file); - _fseeki64(file, 0, SEEK_SET); - #elif defined(__MINGW32__) && !defined(__NO_MINGW_LFS) && !defined(__STRICT_ANSI__) - // there are 64-bit versions of fseek/ftell, let's use them - typedef off64_t length_type; - - fseeko64(file, 0, SEEK_END); - length_type length = ftello64(file); - fseeko64(file, 0, SEEK_SET); - #else - // if this is a 32-bit OS, long is enough; if this is a unix system, long is 64-bit, which is enough; otherwise we can't do anything anyway. - typedef long length_type; - - fseek(file, 0, SEEK_END); - length_type length = ftell(file); - fseek(file, 0, SEEK_SET); - #endif - - // check for I/O errors - if (length < 0) return status_io_error; - - // check for overflow - size_t result = static_cast(length); - - if (static_cast(result) != length) return status_out_of_memory; - - // finalize - out_result = result; - - return status_ok; - } - - PUGI__FN xml_parse_result load_file_impl(xml_document& doc, FILE* file, unsigned int options, xml_encoding encoding) - { - if (!file) return make_parse_result(status_file_not_found); - - // get file size (can result in I/O errors) - size_t size = 0; - xml_parse_status size_status = get_file_size(file, size); - - if (size_status != status_ok) - { - fclose(file); - return make_parse_result(size_status); - } - - // allocate buffer for the whole file - char* contents = static_cast(xml_memory::allocate(size > 0 ? size : 1)); - - if (!contents) - { - fclose(file); - return make_parse_result(status_out_of_memory); - } - - // read file in memory - size_t read_size = fread(contents, 1, size, file); - fclose(file); - - if (read_size != size) - { - xml_memory::deallocate(contents); - return make_parse_result(status_io_error); - } - - return doc.load_buffer_inplace_own(contents, size, options, encoding); - } - -#ifndef PUGIXML_NO_STL - template struct xml_stream_chunk - { - static xml_stream_chunk* create() - { - void* memory = xml_memory::allocate(sizeof(xml_stream_chunk)); - - return new (memory) xml_stream_chunk(); - } - - static void destroy(void* ptr) - { - xml_stream_chunk* chunk = static_cast(ptr); - - // free chunk chain - while (chunk) - { - xml_stream_chunk* next = chunk->next; - xml_memory::deallocate(chunk); - chunk = next; - } - } - - xml_stream_chunk(): next(0), size(0) - { - } - - xml_stream_chunk* next; - size_t size; - - T data[xml_memory_page_size / sizeof(T)]; - }; - - template PUGI__FN xml_parse_status load_stream_data_noseek(std::basic_istream& stream, void** out_buffer, size_t* out_size) - { - buffer_holder chunks(0, xml_stream_chunk::destroy); - - // read file to a chunk list - size_t total = 0; - xml_stream_chunk* last = 0; - - while (!stream.eof()) - { - // allocate new chunk - xml_stream_chunk* chunk = xml_stream_chunk::create(); - if (!chunk) return status_out_of_memory; - - // append chunk to list - if (last) last = last->next = chunk; - else chunks.data = last = chunk; - - // read data to chunk - stream.read(chunk->data, static_cast(sizeof(chunk->data) / sizeof(T))); - chunk->size = static_cast(stream.gcount()) * sizeof(T); - - // read may set failbit | eofbit in case gcount() is less than read length, so check for other I/O errors - if (stream.bad() || (!stream.eof() && stream.fail())) return status_io_error; - - // guard against huge files (chunk size is small enough to make this overflow check work) - if (total + chunk->size < total) return status_out_of_memory; - total += chunk->size; - } - - // copy chunk list to a contiguous buffer - char* buffer = static_cast(xml_memory::allocate(total)); - if (!buffer) return status_out_of_memory; - - char* write = buffer; - - for (xml_stream_chunk* chunk = static_cast*>(chunks.data); chunk; chunk = chunk->next) - { - assert(write + chunk->size <= buffer + total); - memcpy(write, chunk->data, chunk->size); - write += chunk->size; - } - - assert(write == buffer + total); - - // return buffer - *out_buffer = buffer; - *out_size = total; - - return status_ok; - } - - template PUGI__FN xml_parse_status load_stream_data_seek(std::basic_istream& stream, void** out_buffer, size_t* out_size) - { - // get length of remaining data in stream - typename std::basic_istream::pos_type pos = stream.tellg(); - stream.seekg(0, std::ios::end); - std::streamoff length = stream.tellg() - pos; - stream.seekg(pos); - - if (stream.fail() || pos < 0) return status_io_error; - - // guard against huge files - size_t read_length = static_cast(length); - - if (static_cast(read_length) != length || length < 0) return status_out_of_memory; - - // read stream data into memory (guard against stream exceptions with buffer holder) - buffer_holder buffer(xml_memory::allocate((read_length > 0 ? read_length : 1) * sizeof(T)), xml_memory::deallocate); - if (!buffer.data) return status_out_of_memory; - - stream.read(static_cast(buffer.data), static_cast(read_length)); - - // read may set failbit | eofbit in case gcount() is less than read_length (i.e. line ending conversion), so check for other I/O errors - if (stream.bad() || (!stream.eof() && stream.fail())) return status_io_error; - - // return buffer - size_t actual_length = static_cast(stream.gcount()); - assert(actual_length <= read_length); - - *out_buffer = buffer.release(); - *out_size = actual_length * sizeof(T); - - return status_ok; - } - - template PUGI__FN xml_parse_result load_stream_impl(xml_document& doc, std::basic_istream& stream, unsigned int options, xml_encoding encoding) - { - void* buffer = 0; - size_t size = 0; - - // load stream to memory (using seek-based implementation if possible, since it's faster and takes less memory) - xml_parse_status status = (stream.tellg() < 0) ? load_stream_data_noseek(stream, &buffer, &size) : load_stream_data_seek(stream, &buffer, &size); - if (status != status_ok) return make_parse_result(status); - - return doc.load_buffer_inplace_own(buffer, size, options, encoding); - } -#endif - -#if defined(PUGI__MSVC_CRT_VERSION) || defined(__BORLANDC__) || (defined(__MINGW32__) && !defined(__STRICT_ANSI__)) - PUGI__FN FILE* open_file_wide(const wchar_t* path, const wchar_t* mode) - { - return _wfopen(path, mode); - } -#else - PUGI__FN char* convert_path_heap(const wchar_t* str) - { - assert(str); - - // first pass: get length in utf8 characters - size_t length = wcslen(str); - size_t size = as_utf8_begin(str, length); - - // allocate resulting string - char* result = static_cast(xml_memory::allocate(size + 1)); - if (!result) return 0; - - // second pass: convert to utf8 - as_utf8_end(result, size, str, length); - - return result; - } - - PUGI__FN FILE* open_file_wide(const wchar_t* path, const wchar_t* mode) - { - // there is no standard function to open wide paths, so our best bet is to try utf8 path - char* path_utf8 = convert_path_heap(path); - if (!path_utf8) return 0; - - // convert mode to ASCII (we mirror _wfopen interface) - char mode_ascii[4] = {0}; - for (size_t i = 0; mode[i]; ++i) mode_ascii[i] = static_cast(mode[i]); - - // try to open the utf8 path - FILE* result = fopen(path_utf8, mode_ascii); - - // free dummy buffer - xml_memory::deallocate(path_utf8); - - return result; - } -#endif - - PUGI__FN bool save_file_impl(const xml_document& doc, FILE* file, const char_t* indent, unsigned int flags, xml_encoding encoding) - { - if (!file) return false; - - xml_writer_file writer(file); - doc.save(writer, indent, flags, encoding); - - int result = ferror(file); - - fclose(file); - - return result == 0; - } -PUGI__NS_END - -namespace pugi -{ - PUGI__FN xml_writer_file::xml_writer_file(void* file_): file(file_) - { - } - - PUGI__FN void xml_writer_file::write(const void* data, size_t size) - { - size_t result = fwrite(data, 1, size, static_cast(file)); - (void)!result; // unfortunately we can't do proper error handling here - } - -#ifndef PUGIXML_NO_STL - PUGI__FN xml_writer_stream::xml_writer_stream(std::basic_ostream >& stream): narrow_stream(&stream), wide_stream(0) - { - } - - PUGI__FN xml_writer_stream::xml_writer_stream(std::basic_ostream >& stream): narrow_stream(0), wide_stream(&stream) - { - } - - PUGI__FN void xml_writer_stream::write(const void* data, size_t size) - { - if (narrow_stream) - { - assert(!wide_stream); - narrow_stream->write(reinterpret_cast(data), static_cast(size)); - } - else - { - assert(wide_stream); - assert(size % sizeof(wchar_t) == 0); - - wide_stream->write(reinterpret_cast(data), static_cast(size / sizeof(wchar_t))); - } - } -#endif - - PUGI__FN xml_tree_walker::xml_tree_walker(): _depth(0) - { - } - - PUGI__FN xml_tree_walker::~xml_tree_walker() - { - } - - PUGI__FN int xml_tree_walker::depth() const - { - return _depth; - } - - PUGI__FN bool xml_tree_walker::begin(xml_node&) - { - return true; - } - - PUGI__FN bool xml_tree_walker::end(xml_node&) - { - return true; - } - - PUGI__FN xml_attribute::xml_attribute(): _attr(0) - { - } - - PUGI__FN xml_attribute::xml_attribute(xml_attribute_struct* attr): _attr(attr) - { - } - - PUGI__FN static void unspecified_bool_xml_attribute(xml_attribute***) - { - } - - PUGI__FN xml_attribute::operator xml_attribute::unspecified_bool_type() const - { - return _attr ? unspecified_bool_xml_attribute : 0; - } - - PUGI__FN bool xml_attribute::operator!() const - { - return !_attr; - } - - PUGI__FN bool xml_attribute::operator==(const xml_attribute& r) const - { - return (_attr == r._attr); - } - - PUGI__FN bool xml_attribute::operator!=(const xml_attribute& r) const - { - return (_attr != r._attr); - } - - PUGI__FN bool xml_attribute::operator<(const xml_attribute& r) const - { - return (_attr < r._attr); - } - - PUGI__FN bool xml_attribute::operator>(const xml_attribute& r) const - { - return (_attr > r._attr); - } - - PUGI__FN bool xml_attribute::operator<=(const xml_attribute& r) const - { - return (_attr <= r._attr); - } - - PUGI__FN bool xml_attribute::operator>=(const xml_attribute& r) const - { - return (_attr >= r._attr); - } - - PUGI__FN xml_attribute xml_attribute::next_attribute() const - { - return _attr ? xml_attribute(_attr->next_attribute) : xml_attribute(); - } - - PUGI__FN xml_attribute xml_attribute::previous_attribute() const - { - return _attr && _attr->prev_attribute_c->next_attribute ? xml_attribute(_attr->prev_attribute_c) : xml_attribute(); - } - - PUGI__FN const char_t* xml_attribute::as_string(const char_t* def) const - { - return (_attr && _attr->value) ? _attr->value : def; - } - - PUGI__FN int xml_attribute::as_int(int def) const - { - return impl::get_value_int(_attr ? _attr->value : 0, def); - } - - PUGI__FN unsigned int xml_attribute::as_uint(unsigned int def) const - { - return impl::get_value_uint(_attr ? _attr->value : 0, def); - } - - PUGI__FN double xml_attribute::as_double(double def) const - { - return impl::get_value_double(_attr ? _attr->value : 0, def); - } - - PUGI__FN float xml_attribute::as_float(float def) const - { - return impl::get_value_float(_attr ? _attr->value : 0, def); - } - - PUGI__FN bool xml_attribute::as_bool(bool def) const - { - return impl::get_value_bool(_attr ? _attr->value : 0, def); - } - - PUGI__FN bool xml_attribute::empty() const - { - return !_attr; - } - - PUGI__FN const char_t* xml_attribute::name() const - { - return (_attr && _attr->name) ? _attr->name : PUGIXML_TEXT(""); - } - - PUGI__FN const char_t* xml_attribute::value() const - { - return (_attr && _attr->value) ? _attr->value : PUGIXML_TEXT(""); - } - - PUGI__FN size_t xml_attribute::hash_value() const - { - return static_cast(reinterpret_cast(_attr) / sizeof(xml_attribute_struct)); - } - - PUGI__FN xml_attribute_struct* xml_attribute::internal_object() const - { - return _attr; - } - - PUGI__FN xml_attribute& xml_attribute::operator=(const char_t* rhs) - { - set_value(rhs); - return *this; - } - - PUGI__FN xml_attribute& xml_attribute::operator=(int rhs) - { - set_value(rhs); - return *this; - } - - PUGI__FN xml_attribute& xml_attribute::operator=(unsigned int rhs) - { - set_value(rhs); - return *this; - } - - PUGI__FN xml_attribute& xml_attribute::operator=(double rhs) - { - set_value(rhs); - return *this; - } - - PUGI__FN xml_attribute& xml_attribute::operator=(bool rhs) - { - set_value(rhs); - return *this; - } - - PUGI__FN bool xml_attribute::set_name(const char_t* rhs) - { - if (!_attr) return false; - - return impl::strcpy_insitu(_attr->name, _attr->header, impl::xml_memory_page_name_allocated_mask, rhs); - } - - PUGI__FN bool xml_attribute::set_value(const char_t* rhs) - { - if (!_attr) return false; - - return impl::strcpy_insitu(_attr->value, _attr->header, impl::xml_memory_page_value_allocated_mask, rhs); - } - - PUGI__FN bool xml_attribute::set_value(int rhs) - { - if (!_attr) return false; - - return impl::set_value_convert(_attr->value, _attr->header, impl::xml_memory_page_value_allocated_mask, rhs); - } - - PUGI__FN bool xml_attribute::set_value(unsigned int rhs) - { - if (!_attr) return false; - - return impl::set_value_convert(_attr->value, _attr->header, impl::xml_memory_page_value_allocated_mask, rhs); - } - - PUGI__FN bool xml_attribute::set_value(double rhs) - { - if (!_attr) return false; - - return impl::set_value_convert(_attr->value, _attr->header, impl::xml_memory_page_value_allocated_mask, rhs); - } - - PUGI__FN bool xml_attribute::set_value(bool rhs) - { - if (!_attr) return false; - - return impl::set_value_convert(_attr->value, _attr->header, impl::xml_memory_page_value_allocated_mask, rhs); - } - -#ifdef __BORLANDC__ - PUGI__FN bool operator&&(const xml_attribute& lhs, bool rhs) - { - return (bool)lhs && rhs; - } - - PUGI__FN bool operator||(const xml_attribute& lhs, bool rhs) - { - return (bool)lhs || rhs; - } -#endif - - PUGI__FN xml_node::xml_node(): _root(0) - { - } - - PUGI__FN xml_node::xml_node(xml_node_struct* p): _root(p) - { - } - - PUGI__FN static void unspecified_bool_xml_node(xml_node***) - { - } - - PUGI__FN xml_node::operator xml_node::unspecified_bool_type() const - { - return _root ? unspecified_bool_xml_node : 0; - } - - PUGI__FN bool xml_node::operator!() const - { - return !_root; - } - - PUGI__FN xml_node::iterator xml_node::begin() const - { - return iterator(_root ? _root->first_child : 0, _root); - } - - PUGI__FN xml_node::iterator xml_node::end() const - { - return iterator(0, _root); - } - - PUGI__FN xml_node::attribute_iterator xml_node::attributes_begin() const - { - return attribute_iterator(_root ? _root->first_attribute : 0, _root); - } - - PUGI__FN xml_node::attribute_iterator xml_node::attributes_end() const - { - return attribute_iterator(0, _root); - } - - PUGI__FN xml_object_range xml_node::children() const - { - return xml_object_range(begin(), end()); - } - - PUGI__FN xml_object_range xml_node::children(const char_t* name_) const - { - return xml_object_range(xml_named_node_iterator(child(name_), name_), xml_named_node_iterator()); - } - - PUGI__FN xml_object_range xml_node::attributes() const - { - return xml_object_range(attributes_begin(), attributes_end()); - } - - PUGI__FN bool xml_node::operator==(const xml_node& r) const - { - return (_root == r._root); - } - - PUGI__FN bool xml_node::operator!=(const xml_node& r) const - { - return (_root != r._root); - } - - PUGI__FN bool xml_node::operator<(const xml_node& r) const - { - return (_root < r._root); - } - - PUGI__FN bool xml_node::operator>(const xml_node& r) const - { - return (_root > r._root); - } - - PUGI__FN bool xml_node::operator<=(const xml_node& r) const - { - return (_root <= r._root); - } - - PUGI__FN bool xml_node::operator>=(const xml_node& r) const - { - return (_root >= r._root); - } - - PUGI__FN bool xml_node::empty() const - { - return !_root; - } - - PUGI__FN const char_t* xml_node::name() const - { - return (_root && _root->name) ? _root->name : PUGIXML_TEXT(""); - } - - PUGI__FN xml_node_type xml_node::type() const - { - return _root ? static_cast((_root->header & impl::xml_memory_page_type_mask) + 1) : node_null; - } - - PUGI__FN const char_t* xml_node::value() const - { - return (_root && _root->value) ? _root->value : PUGIXML_TEXT(""); - } - - PUGI__FN xml_node xml_node::child(const char_t* name_) const - { - if (!_root) return xml_node(); - - for (xml_node_struct* i = _root->first_child; i; i = i->next_sibling) - if (i->name && impl::strequal(name_, i->name)) return xml_node(i); - - return xml_node(); - } - - PUGI__FN xml_attribute xml_node::attribute(const char_t* name_) const - { - if (!_root) return xml_attribute(); - - for (xml_attribute_struct* i = _root->first_attribute; i; i = i->next_attribute) - if (i->name && impl::strequal(name_, i->name)) - return xml_attribute(i); - - return xml_attribute(); - } - - PUGI__FN xml_node xml_node::next_sibling(const char_t* name_) const - { - if (!_root) return xml_node(); - - for (xml_node_struct* i = _root->next_sibling; i; i = i->next_sibling) - if (i->name && impl::strequal(name_, i->name)) return xml_node(i); - - return xml_node(); - } - - PUGI__FN xml_node xml_node::next_sibling() const - { - if (!_root) return xml_node(); - - if (_root->next_sibling) return xml_node(_root->next_sibling); - else return xml_node(); - } - - PUGI__FN xml_node xml_node::previous_sibling(const char_t* name_) const - { - if (!_root) return xml_node(); - - for (xml_node_struct* i = _root->prev_sibling_c; i->next_sibling; i = i->prev_sibling_c) - if (i->name && impl::strequal(name_, i->name)) return xml_node(i); - - return xml_node(); - } - - PUGI__FN xml_node xml_node::previous_sibling() const - { - if (!_root) return xml_node(); - - if (_root->prev_sibling_c->next_sibling) return xml_node(_root->prev_sibling_c); - else return xml_node(); - } - - PUGI__FN xml_node xml_node::parent() const - { - return _root ? xml_node(_root->parent) : xml_node(); - } - - PUGI__FN xml_node xml_node::root() const - { - if (!_root) return xml_node(); - - impl::xml_memory_page* page = reinterpret_cast(_root->header & impl::xml_memory_page_pointer_mask); - - return xml_node(static_cast(page->allocator)); - } - - PUGI__FN xml_text xml_node::text() const - { - return xml_text(_root); - } - - PUGI__FN const char_t* xml_node::child_value() const - { - if (!_root) return PUGIXML_TEXT(""); - - for (xml_node_struct* i = _root->first_child; i; i = i->next_sibling) - if (i->value && impl::is_text_node(i)) - return i->value; - - return PUGIXML_TEXT(""); - } - - PUGI__FN const char_t* xml_node::child_value(const char_t* name_) const - { - return child(name_).child_value(); - } - - PUGI__FN xml_attribute xml_node::first_attribute() const - { - return _root ? xml_attribute(_root->first_attribute) : xml_attribute(); - } - - PUGI__FN xml_attribute xml_node::last_attribute() const - { - return _root && _root->first_attribute ? xml_attribute(_root->first_attribute->prev_attribute_c) : xml_attribute(); - } - - PUGI__FN xml_node xml_node::first_child() const - { - return _root ? xml_node(_root->first_child) : xml_node(); - } - - PUGI__FN xml_node xml_node::last_child() const - { - return _root && _root->first_child ? xml_node(_root->first_child->prev_sibling_c) : xml_node(); - } - - PUGI__FN bool xml_node::set_name(const char_t* rhs) - { - switch (type()) - { - case node_pi: - case node_declaration: - case node_element: - return impl::strcpy_insitu(_root->name, _root->header, impl::xml_memory_page_name_allocated_mask, rhs); - - default: - return false; - } - } - - PUGI__FN bool xml_node::set_value(const char_t* rhs) - { - switch (type()) - { - case node_pi: - case node_cdata: - case node_pcdata: - case node_comment: - case node_doctype: - return impl::strcpy_insitu(_root->value, _root->header, impl::xml_memory_page_value_allocated_mask, rhs); - - default: - return false; - } - } - - PUGI__FN xml_attribute xml_node::append_attribute(const char_t* name_) - { - if (type() != node_element && type() != node_declaration) return xml_attribute(); - - xml_attribute a(impl::append_attribute_ll(_root, impl::get_allocator(_root))); - a.set_name(name_); - - return a; - } - - PUGI__FN xml_attribute xml_node::prepend_attribute(const char_t* name_) - { - if (type() != node_element && type() != node_declaration) return xml_attribute(); - - xml_attribute a(impl::allocate_attribute(impl::get_allocator(_root))); - if (!a) return xml_attribute(); - - a.set_name(name_); - - xml_attribute_struct* head = _root->first_attribute; - - if (head) - { - a._attr->prev_attribute_c = head->prev_attribute_c; - head->prev_attribute_c = a._attr; - } - else - a._attr->prev_attribute_c = a._attr; - - a._attr->next_attribute = head; - _root->first_attribute = a._attr; - - return a; - } - - PUGI__FN xml_attribute xml_node::insert_attribute_before(const char_t* name_, const xml_attribute& attr) - { - if ((type() != node_element && type() != node_declaration) || attr.empty()) return xml_attribute(); - - // check that attribute belongs to *this - xml_attribute_struct* cur = attr._attr; - - while (cur->prev_attribute_c->next_attribute) cur = cur->prev_attribute_c; - - if (cur != _root->first_attribute) return xml_attribute(); - - xml_attribute a(impl::allocate_attribute(impl::get_allocator(_root))); - if (!a) return xml_attribute(); - - a.set_name(name_); - - if (attr._attr->prev_attribute_c->next_attribute) - attr._attr->prev_attribute_c->next_attribute = a._attr; - else - _root->first_attribute = a._attr; - - a._attr->prev_attribute_c = attr._attr->prev_attribute_c; - a._attr->next_attribute = attr._attr; - attr._attr->prev_attribute_c = a._attr; - - return a; - } - - PUGI__FN xml_attribute xml_node::insert_attribute_after(const char_t* name_, const xml_attribute& attr) - { - if ((type() != node_element && type() != node_declaration) || attr.empty()) return xml_attribute(); - - // check that attribute belongs to *this - xml_attribute_struct* cur = attr._attr; - - while (cur->prev_attribute_c->next_attribute) cur = cur->prev_attribute_c; - - if (cur != _root->first_attribute) return xml_attribute(); - - xml_attribute a(impl::allocate_attribute(impl::get_allocator(_root))); - if (!a) return xml_attribute(); - - a.set_name(name_); - - if (attr._attr->next_attribute) - attr._attr->next_attribute->prev_attribute_c = a._attr; - else - _root->first_attribute->prev_attribute_c = a._attr; - - a._attr->next_attribute = attr._attr->next_attribute; - a._attr->prev_attribute_c = attr._attr; - attr._attr->next_attribute = a._attr; - - return a; - } - - PUGI__FN xml_attribute xml_node::append_copy(const xml_attribute& proto) - { - if (!proto) return xml_attribute(); - - xml_attribute result = append_attribute(proto.name()); - result.set_value(proto.value()); - - return result; - } - - PUGI__FN xml_attribute xml_node::prepend_copy(const xml_attribute& proto) - { - if (!proto) return xml_attribute(); - - xml_attribute result = prepend_attribute(proto.name()); - result.set_value(proto.value()); - - return result; - } - - PUGI__FN xml_attribute xml_node::insert_copy_after(const xml_attribute& proto, const xml_attribute& attr) - { - if (!proto) return xml_attribute(); - - xml_attribute result = insert_attribute_after(proto.name(), attr); - result.set_value(proto.value()); - - return result; - } - - PUGI__FN xml_attribute xml_node::insert_copy_before(const xml_attribute& proto, const xml_attribute& attr) - { - if (!proto) return xml_attribute(); - - xml_attribute result = insert_attribute_before(proto.name(), attr); - result.set_value(proto.value()); - - return result; - } - - PUGI__FN xml_node xml_node::append_child(xml_node_type type_) - { - if (!impl::allow_insert_child(this->type(), type_)) return xml_node(); - - xml_node n(impl::append_node(_root, impl::get_allocator(_root), type_)); - - if (type_ == node_declaration) n.set_name(PUGIXML_TEXT("xml")); - - return n; - } - - PUGI__FN xml_node xml_node::prepend_child(xml_node_type type_) - { - if (!impl::allow_insert_child(this->type(), type_)) return xml_node(); - - xml_node n(impl::allocate_node(impl::get_allocator(_root), type_)); - if (!n) return xml_node(); - - n._root->parent = _root; - - xml_node_struct* head = _root->first_child; - - if (head) - { - n._root->prev_sibling_c = head->prev_sibling_c; - head->prev_sibling_c = n._root; - } - else - n._root->prev_sibling_c = n._root; - - n._root->next_sibling = head; - _root->first_child = n._root; - - if (type_ == node_declaration) n.set_name(PUGIXML_TEXT("xml")); - - return n; - } - - PUGI__FN xml_node xml_node::insert_child_before(xml_node_type type_, const xml_node& node) - { - if (!impl::allow_insert_child(this->type(), type_)) return xml_node(); - if (!node._root || node._root->parent != _root) return xml_node(); - - xml_node n(impl::allocate_node(impl::get_allocator(_root), type_)); - if (!n) return xml_node(); - - n._root->parent = _root; - - if (node._root->prev_sibling_c->next_sibling) - node._root->prev_sibling_c->next_sibling = n._root; - else - _root->first_child = n._root; - - n._root->prev_sibling_c = node._root->prev_sibling_c; - n._root->next_sibling = node._root; - node._root->prev_sibling_c = n._root; - - if (type_ == node_declaration) n.set_name(PUGIXML_TEXT("xml")); - - return n; - } - - PUGI__FN xml_node xml_node::insert_child_after(xml_node_type type_, const xml_node& node) - { - if (!impl::allow_insert_child(this->type(), type_)) return xml_node(); - if (!node._root || node._root->parent != _root) return xml_node(); - - xml_node n(impl::allocate_node(impl::get_allocator(_root), type_)); - if (!n) return xml_node(); - - n._root->parent = _root; - - if (node._root->next_sibling) - node._root->next_sibling->prev_sibling_c = n._root; - else - _root->first_child->prev_sibling_c = n._root; - - n._root->next_sibling = node._root->next_sibling; - n._root->prev_sibling_c = node._root; - node._root->next_sibling = n._root; - - if (type_ == node_declaration) n.set_name(PUGIXML_TEXT("xml")); - - return n; - } - - PUGI__FN xml_node xml_node::append_child(const char_t* name_) - { - xml_node result = append_child(node_element); - - result.set_name(name_); - - return result; - } - - PUGI__FN xml_node xml_node::prepend_child(const char_t* name_) - { - xml_node result = prepend_child(node_element); - - result.set_name(name_); - - return result; - } - - PUGI__FN xml_node xml_node::insert_child_after(const char_t* name_, const xml_node& node) - { - xml_node result = insert_child_after(node_element, node); - - result.set_name(name_); - - return result; - } - - PUGI__FN xml_node xml_node::insert_child_before(const char_t* name_, const xml_node& node) - { - xml_node result = insert_child_before(node_element, node); - - result.set_name(name_); - - return result; - } - - PUGI__FN xml_node xml_node::append_copy(const xml_node& proto) - { - xml_node result = append_child(proto.type()); - - if (result) impl::recursive_copy_skip(result, proto, result); - - return result; - } - - PUGI__FN xml_node xml_node::prepend_copy(const xml_node& proto) - { - xml_node result = prepend_child(proto.type()); - - if (result) impl::recursive_copy_skip(result, proto, result); - - return result; - } - - PUGI__FN xml_node xml_node::insert_copy_after(const xml_node& proto, const xml_node& node) - { - xml_node result = insert_child_after(proto.type(), node); - - if (result) impl::recursive_copy_skip(result, proto, result); - - return result; - } - - PUGI__FN xml_node xml_node::insert_copy_before(const xml_node& proto, const xml_node& node) - { - xml_node result = insert_child_before(proto.type(), node); - - if (result) impl::recursive_copy_skip(result, proto, result); - - return result; - } - - PUGI__FN bool xml_node::remove_attribute(const char_t* name_) - { - return remove_attribute(attribute(name_)); - } - - PUGI__FN bool xml_node::remove_attribute(const xml_attribute& a) - { - if (!_root || !a._attr) return false; - - // check that attribute belongs to *this - xml_attribute_struct* attr = a._attr; - - while (attr->prev_attribute_c->next_attribute) attr = attr->prev_attribute_c; - - if (attr != _root->first_attribute) return false; - - if (a._attr->next_attribute) a._attr->next_attribute->prev_attribute_c = a._attr->prev_attribute_c; - else if (_root->first_attribute) _root->first_attribute->prev_attribute_c = a._attr->prev_attribute_c; - - if (a._attr->prev_attribute_c->next_attribute) a._attr->prev_attribute_c->next_attribute = a._attr->next_attribute; - else _root->first_attribute = a._attr->next_attribute; - - impl::destroy_attribute(a._attr, impl::get_allocator(_root)); - - return true; - } - - PUGI__FN bool xml_node::remove_child(const char_t* name_) - { - return remove_child(child(name_)); - } - - PUGI__FN bool xml_node::remove_child(const xml_node& n) - { - if (!_root || !n._root || n._root->parent != _root) return false; - - if (n._root->next_sibling) n._root->next_sibling->prev_sibling_c = n._root->prev_sibling_c; - else if (_root->first_child) _root->first_child->prev_sibling_c = n._root->prev_sibling_c; - - if (n._root->prev_sibling_c->next_sibling) n._root->prev_sibling_c->next_sibling = n._root->next_sibling; - else _root->first_child = n._root->next_sibling; - - impl::destroy_node(n._root, impl::get_allocator(_root)); - - return true; - } - - PUGI__FN xml_node xml_node::find_child_by_attribute(const char_t* name_, const char_t* attr_name, const char_t* attr_value) const - { - if (!_root) return xml_node(); - - for (xml_node_struct* i = _root->first_child; i; i = i->next_sibling) - if (i->name && impl::strequal(name_, i->name)) - { - for (xml_attribute_struct* a = i->first_attribute; a; a = a->next_attribute) - if (impl::strequal(attr_name, a->name) && impl::strequal(attr_value, a->value)) - return xml_node(i); - } - - return xml_node(); - } - - PUGI__FN xml_node xml_node::find_child_by_attribute(const char_t* attr_name, const char_t* attr_value) const - { - if (!_root) return xml_node(); - - for (xml_node_struct* i = _root->first_child; i; i = i->next_sibling) - for (xml_attribute_struct* a = i->first_attribute; a; a = a->next_attribute) - if (impl::strequal(attr_name, a->name) && impl::strequal(attr_value, a->value)) - return xml_node(i); - - return xml_node(); - } - -#ifndef PUGIXML_NO_STL - PUGI__FN string_t xml_node::path(char_t delimiter) const - { - xml_node cursor = *this; // Make a copy. - - string_t result = cursor.name(); - - while (cursor.parent()) - { - cursor = cursor.parent(); - - string_t temp = cursor.name(); - temp += delimiter; - temp += result; - result.swap(temp); - } - - return result; - } -#endif - - PUGI__FN xml_node xml_node::first_element_by_path(const char_t* path_, char_t delimiter) const - { - xml_node found = *this; // Current search context. - - if (!_root || !path_ || !path_[0]) return found; - - if (path_[0] == delimiter) - { - // Absolute path; e.g. '/foo/bar' - found = found.root(); - ++path_; - } - - const char_t* path_segment = path_; - - while (*path_segment == delimiter) ++path_segment; - - const char_t* path_segment_end = path_segment; - - while (*path_segment_end && *path_segment_end != delimiter) ++path_segment_end; - - if (path_segment == path_segment_end) return found; - - const char_t* next_segment = path_segment_end; - - while (*next_segment == delimiter) ++next_segment; - - if (*path_segment == '.' && path_segment + 1 == path_segment_end) - return found.first_element_by_path(next_segment, delimiter); - else if (*path_segment == '.' && *(path_segment+1) == '.' && path_segment + 2 == path_segment_end) - return found.parent().first_element_by_path(next_segment, delimiter); - else - { - for (xml_node_struct* j = found._root->first_child; j; j = j->next_sibling) - { - if (j->name && impl::strequalrange(j->name, path_segment, static_cast(path_segment_end - path_segment))) - { - xml_node subsearch = xml_node(j).first_element_by_path(next_segment, delimiter); - - if (subsearch) return subsearch; - } - } - - return xml_node(); - } - } - - PUGI__FN bool xml_node::traverse(xml_tree_walker& walker) - { - walker._depth = -1; - - xml_node arg_begin = *this; - if (!walker.begin(arg_begin)) return false; - - xml_node cur = first_child(); - - if (cur) - { - ++walker._depth; - - do - { - xml_node arg_for_each = cur; - if (!walker.for_each(arg_for_each)) - return false; - - if (cur.first_child()) - { - ++walker._depth; - cur = cur.first_child(); - } - else if (cur.next_sibling()) - cur = cur.next_sibling(); - else - { - // Borland C++ workaround - while (!cur.next_sibling() && cur != *this && !cur.parent().empty()) - { - --walker._depth; - cur = cur.parent(); - } - - if (cur != *this) - cur = cur.next_sibling(); - } - } - while (cur && cur != *this); - } - - assert(walker._depth == -1); - - xml_node arg_end = *this; - return walker.end(arg_end); - } - - PUGI__FN size_t xml_node::hash_value() const - { - return static_cast(reinterpret_cast(_root) / sizeof(xml_node_struct)); - } - - PUGI__FN xml_node_struct* xml_node::internal_object() const - { - return _root; - } - - PUGI__FN void xml_node::print(xml_writer& writer, const char_t* indent, unsigned int flags, xml_encoding encoding, unsigned int depth) const - { - if (!_root) return; - - impl::xml_buffered_writer buffered_writer(writer, encoding); - - impl::node_output(buffered_writer, *this, indent, flags, depth); - } - -#ifndef PUGIXML_NO_STL - PUGI__FN void xml_node::print(std::basic_ostream >& stream, const char_t* indent, unsigned int flags, xml_encoding encoding, unsigned int depth) const - { - xml_writer_stream writer(stream); - - print(writer, indent, flags, encoding, depth); - } - - PUGI__FN void xml_node::print(std::basic_ostream >& stream, const char_t* indent, unsigned int flags, unsigned int depth) const - { - xml_writer_stream writer(stream); - - print(writer, indent, flags, encoding_wchar, depth); - } -#endif - - PUGI__FN ptrdiff_t xml_node::offset_debug() const - { - xml_node_struct* r = root()._root; - - if (!r) return -1; - - const char_t* buffer = static_cast(r)->buffer; - - if (!buffer) return -1; - - switch (type()) - { - case node_document: - return 0; - - case node_element: - case node_declaration: - case node_pi: - return (_root->header & impl::xml_memory_page_name_allocated_mask) ? -1 : _root->name - buffer; - - case node_pcdata: - case node_cdata: - case node_comment: - case node_doctype: - return (_root->header & impl::xml_memory_page_value_allocated_mask) ? -1 : _root->value - buffer; - - default: - return -1; - } - } - -#ifdef __BORLANDC__ - PUGI__FN bool operator&&(const xml_node& lhs, bool rhs) - { - return (bool)lhs && rhs; - } - - PUGI__FN bool operator||(const xml_node& lhs, bool rhs) - { - return (bool)lhs || rhs; - } -#endif - - PUGI__FN xml_text::xml_text(xml_node_struct* root): _root(root) - { - } - - PUGI__FN xml_node_struct* xml_text::_data() const - { - if (!_root || impl::is_text_node(_root)) return _root; - - for (xml_node_struct* node = _root->first_child; node; node = node->next_sibling) - if (impl::is_text_node(node)) - return node; - - return 0; - } - - PUGI__FN xml_node_struct* xml_text::_data_new() - { - xml_node_struct* d = _data(); - if (d) return d; - - return xml_node(_root).append_child(node_pcdata).internal_object(); - } - - PUGI__FN xml_text::xml_text(): _root(0) - { - } - - PUGI__FN static void unspecified_bool_xml_text(xml_text***) - { - } - - PUGI__FN xml_text::operator xml_text::unspecified_bool_type() const - { - return _data() ? unspecified_bool_xml_text : 0; - } - - PUGI__FN bool xml_text::operator!() const - { - return !_data(); - } - - PUGI__FN bool xml_text::empty() const - { - return _data() == 0; - } - - PUGI__FN const char_t* xml_text::get() const - { - xml_node_struct* d = _data(); - - return (d && d->value) ? d->value : PUGIXML_TEXT(""); - } - - PUGI__FN const char_t* xml_text::as_string(const char_t* def) const - { - xml_node_struct* d = _data(); - - return (d && d->value) ? d->value : def; - } - - PUGI__FN int xml_text::as_int(int def) const - { - xml_node_struct* d = _data(); - - return impl::get_value_int(d ? d->value : 0, def); - } - - PUGI__FN unsigned int xml_text::as_uint(unsigned int def) const - { - xml_node_struct* d = _data(); - - return impl::get_value_uint(d ? d->value : 0, def); - } - - PUGI__FN double xml_text::as_double(double def) const - { - xml_node_struct* d = _data(); - - return impl::get_value_double(d ? d->value : 0, def); - } - - PUGI__FN float xml_text::as_float(float def) const - { - xml_node_struct* d = _data(); - - return impl::get_value_float(d ? d->value : 0, def); - } - - PUGI__FN bool xml_text::as_bool(bool def) const - { - xml_node_struct* d = _data(); - - return impl::get_value_bool(d ? d->value : 0, def); - } - - PUGI__FN bool xml_text::set(const char_t* rhs) - { - xml_node_struct* dn = _data_new(); - - return dn ? impl::strcpy_insitu(dn->value, dn->header, impl::xml_memory_page_value_allocated_mask, rhs) : false; - } - - PUGI__FN bool xml_text::set(int rhs) - { - xml_node_struct* dn = _data_new(); - - return dn ? impl::set_value_convert(dn->value, dn->header, impl::xml_memory_page_value_allocated_mask, rhs) : false; - } - - PUGI__FN bool xml_text::set(unsigned int rhs) - { - xml_node_struct* dn = _data_new(); - - return dn ? impl::set_value_convert(dn->value, dn->header, impl::xml_memory_page_value_allocated_mask, rhs) : false; - } - - PUGI__FN bool xml_text::set(double rhs) - { - xml_node_struct* dn = _data_new(); - - return dn ? impl::set_value_convert(dn->value, dn->header, impl::xml_memory_page_value_allocated_mask, rhs) : false; - } - - PUGI__FN bool xml_text::set(bool rhs) - { - xml_node_struct* dn = _data_new(); - - return dn ? impl::set_value_convert(dn->value, dn->header, impl::xml_memory_page_value_allocated_mask, rhs) : false; - } - - PUGI__FN xml_text& xml_text::operator=(const char_t* rhs) - { - set(rhs); - return *this; - } - - PUGI__FN xml_text& xml_text::operator=(int rhs) - { - set(rhs); - return *this; - } - - PUGI__FN xml_text& xml_text::operator=(unsigned int rhs) - { - set(rhs); - return *this; - } - - PUGI__FN xml_text& xml_text::operator=(double rhs) - { - set(rhs); - return *this; - } - - PUGI__FN xml_text& xml_text::operator=(bool rhs) - { - set(rhs); - return *this; - } - - PUGI__FN xml_node xml_text::data() const - { - return xml_node(_data()); - } - -#ifdef __BORLANDC__ - PUGI__FN bool operator&&(const xml_text& lhs, bool rhs) - { - return (bool)lhs && rhs; - } - - PUGI__FN bool operator||(const xml_text& lhs, bool rhs) - { - return (bool)lhs || rhs; - } -#endif - - PUGI__FN xml_node_iterator::xml_node_iterator() - { - } - - PUGI__FN xml_node_iterator::xml_node_iterator(const xml_node& node): _wrap(node), _parent(node.parent()) - { - } - - PUGI__FN xml_node_iterator::xml_node_iterator(xml_node_struct* ref, xml_node_struct* parent): _wrap(ref), _parent(parent) - { - } - - PUGI__FN bool xml_node_iterator::operator==(const xml_node_iterator& rhs) const - { - return _wrap._root == rhs._wrap._root && _parent._root == rhs._parent._root; - } - - PUGI__FN bool xml_node_iterator::operator!=(const xml_node_iterator& rhs) const - { - return _wrap._root != rhs._wrap._root || _parent._root != rhs._parent._root; - } - - PUGI__FN xml_node& xml_node_iterator::operator*() const - { - assert(_wrap._root); - return _wrap; - } - - PUGI__FN xml_node* xml_node_iterator::operator->() const - { - assert(_wrap._root); - return const_cast(&_wrap); // BCC32 workaround - } - - PUGI__FN const xml_node_iterator& xml_node_iterator::operator++() - { - assert(_wrap._root); - _wrap._root = _wrap._root->next_sibling; - return *this; - } - - PUGI__FN xml_node_iterator xml_node_iterator::operator++(int) - { - xml_node_iterator temp = *this; - ++*this; - return temp; - } - - PUGI__FN const xml_node_iterator& xml_node_iterator::operator--() - { - _wrap = _wrap._root ? _wrap.previous_sibling() : _parent.last_child(); - return *this; - } - - PUGI__FN xml_node_iterator xml_node_iterator::operator--(int) - { - xml_node_iterator temp = *this; - --*this; - return temp; - } - - PUGI__FN xml_attribute_iterator::xml_attribute_iterator() - { - } - - PUGI__FN xml_attribute_iterator::xml_attribute_iterator(const xml_attribute& attr, const xml_node& parent): _wrap(attr), _parent(parent) - { - } - - PUGI__FN xml_attribute_iterator::xml_attribute_iterator(xml_attribute_struct* ref, xml_node_struct* parent): _wrap(ref), _parent(parent) - { - } - - PUGI__FN bool xml_attribute_iterator::operator==(const xml_attribute_iterator& rhs) const - { - return _wrap._attr == rhs._wrap._attr && _parent._root == rhs._parent._root; - } - - PUGI__FN bool xml_attribute_iterator::operator!=(const xml_attribute_iterator& rhs) const - { - return _wrap._attr != rhs._wrap._attr || _parent._root != rhs._parent._root; - } - - PUGI__FN xml_attribute& xml_attribute_iterator::operator*() const - { - assert(_wrap._attr); - return _wrap; - } - - PUGI__FN xml_attribute* xml_attribute_iterator::operator->() const - { - assert(_wrap._attr); - return const_cast(&_wrap); // BCC32 workaround - } - - PUGI__FN const xml_attribute_iterator& xml_attribute_iterator::operator++() - { - assert(_wrap._attr); - _wrap._attr = _wrap._attr->next_attribute; - return *this; - } - - PUGI__FN xml_attribute_iterator xml_attribute_iterator::operator++(int) - { - xml_attribute_iterator temp = *this; - ++*this; - return temp; - } - - PUGI__FN const xml_attribute_iterator& xml_attribute_iterator::operator--() - { - _wrap = _wrap._attr ? _wrap.previous_attribute() : _parent.last_attribute(); - return *this; - } - - PUGI__FN xml_attribute_iterator xml_attribute_iterator::operator--(int) - { - xml_attribute_iterator temp = *this; - --*this; - return temp; - } - - PUGI__FN xml_named_node_iterator::xml_named_node_iterator(): _name(0) - { - } - - PUGI__FN xml_named_node_iterator::xml_named_node_iterator(const xml_node& node, const char_t* name): _node(node), _name(name) - { - } - - PUGI__FN bool xml_named_node_iterator::operator==(const xml_named_node_iterator& rhs) const - { - return _node == rhs._node; - } - - PUGI__FN bool xml_named_node_iterator::operator!=(const xml_named_node_iterator& rhs) const - { - return _node != rhs._node; - } - - PUGI__FN xml_node& xml_named_node_iterator::operator*() const - { - assert(_node._root); - return _node; - } - - PUGI__FN xml_node* xml_named_node_iterator::operator->() const - { - assert(_node._root); - return const_cast(&_node); // BCC32 workaround - } - - PUGI__FN const xml_named_node_iterator& xml_named_node_iterator::operator++() - { - assert(_node._root); - _node = _node.next_sibling(_name); - return *this; - } - - PUGI__FN xml_named_node_iterator xml_named_node_iterator::operator++(int) - { - xml_named_node_iterator temp = *this; - ++*this; - return temp; - } - - PUGI__FN xml_parse_result::xml_parse_result(): status(status_internal_error), offset(0), encoding(encoding_auto) - { - } - - PUGI__FN xml_parse_result::operator bool() const - { - return status == status_ok; - } - - PUGI__FN const char* xml_parse_result::description() const - { - switch (status) - { - case status_ok: return "No error"; - - case status_file_not_found: return "File was not found"; - case status_io_error: return "Error reading from file/stream"; - case status_out_of_memory: return "Could not allocate memory"; - case status_internal_error: return "Internal error occurred"; - - case status_unrecognized_tag: return "Could not determine tag type"; - - case status_bad_pi: return "Error parsing document declaration/processing instruction"; - case status_bad_comment: return "Error parsing comment"; - case status_bad_cdata: return "Error parsing CDATA section"; - case status_bad_doctype: return "Error parsing document type declaration"; - case status_bad_pcdata: return "Error parsing PCDATA section"; - case status_bad_start_element: return "Error parsing start element tag"; - case status_bad_attribute: return "Error parsing element attribute"; - case status_bad_end_element: return "Error parsing end element tag"; - case status_end_element_mismatch: return "Start-end tags mismatch"; - - default: return "Unknown error"; - } - } - - PUGI__FN xml_document::xml_document(): _buffer(0) - { - create(); - } - - PUGI__FN xml_document::~xml_document() - { - destroy(); - } - - PUGI__FN void xml_document::reset() - { - destroy(); - create(); - } - - PUGI__FN void xml_document::reset(const xml_document& proto) - { - reset(); - - for (xml_node cur = proto.first_child(); cur; cur = cur.next_sibling()) - append_copy(cur); - } - - PUGI__FN void xml_document::create() - { - // initialize sentinel page - PUGI__STATIC_ASSERT(offsetof(impl::xml_memory_page, data) + sizeof(impl::xml_document_struct) + impl::xml_memory_page_alignment <= sizeof(_memory)); - - // align upwards to page boundary - void* page_memory = reinterpret_cast((reinterpret_cast(_memory) + (impl::xml_memory_page_alignment - 1)) & ~(impl::xml_memory_page_alignment - 1)); - - // prepare page structure - impl::xml_memory_page* page = impl::xml_memory_page::construct(page_memory); - - page->busy_size = impl::xml_memory_page_size; - - // allocate new root - _root = new (page->data) impl::xml_document_struct(page); - _root->prev_sibling_c = _root; - - // setup sentinel page - page->allocator = static_cast(_root); - } - - PUGI__FN void xml_document::destroy() - { - // destroy static storage - if (_buffer) - { - impl::xml_memory::deallocate(_buffer); - _buffer = 0; - } - - // destroy dynamic storage, leave sentinel page (it's in static memory) - if (_root) - { - impl::xml_memory_page* root_page = reinterpret_cast(_root->header & impl::xml_memory_page_pointer_mask); - assert(root_page && !root_page->prev && !root_page->memory); - - // destroy all pages - for (impl::xml_memory_page* page = root_page->next; page; ) - { - impl::xml_memory_page* next = page->next; - - impl::xml_allocator::deallocate_page(page); - - page = next; - } - - // cleanup root page - root_page->allocator = 0; - root_page->next = 0; - root_page->busy_size = root_page->freed_size = 0; - - _root = 0; - } - } - -#ifndef PUGIXML_NO_STL - PUGI__FN xml_parse_result xml_document::load(std::basic_istream >& stream, unsigned int options, xml_encoding encoding) - { - reset(); - - return impl::load_stream_impl(*this, stream, options, encoding); - } - - PUGI__FN xml_parse_result xml_document::load(std::basic_istream >& stream, unsigned int options) - { - reset(); - - return impl::load_stream_impl(*this, stream, options, encoding_wchar); - } -#endif - - PUGI__FN xml_parse_result xml_document::load(const char_t* contents, unsigned int options) - { - // Force native encoding (skip autodetection) - #ifdef PUGIXML_WCHAR_MODE - xml_encoding encoding = encoding_wchar; - #else - xml_encoding encoding = encoding_utf8; - #endif - - return load_buffer(contents, impl::strlength(contents) * sizeof(char_t), options, encoding); - } - - PUGI__FN xml_parse_result xml_document::load_file(const char* path_, unsigned int options, xml_encoding encoding) - { - reset(); - - FILE* file = fopen(path_, "rb"); - - return impl::load_file_impl(*this, file, options, encoding); - } - - PUGI__FN xml_parse_result xml_document::load_file(const wchar_t* path_, unsigned int options, xml_encoding encoding) - { - reset(); - - FILE* file = impl::open_file_wide(path_, L"rb"); - - return impl::load_file_impl(*this, file, options, encoding); - } - - PUGI__FN xml_parse_result xml_document::load_buffer_impl(void* contents, size_t size, unsigned int options, xml_encoding encoding, bool is_mutable, bool own) - { - reset(); - - // check input buffer - assert(contents || size == 0); - - // get actual encoding - xml_encoding buffer_encoding = impl::get_buffer_encoding(encoding, contents, size); - - // get private buffer - char_t* buffer = 0; - size_t length = 0; - - if (!impl::convert_buffer(buffer, length, buffer_encoding, contents, size, is_mutable)) return impl::make_parse_result(status_out_of_memory); - - // delete original buffer if we performed a conversion - if (own && buffer != contents && contents) impl::xml_memory::deallocate(contents); - - // parse - xml_parse_result res = impl::xml_parser::parse(buffer, length, _root, options); - - // remember encoding - res.encoding = buffer_encoding; - - // grab onto buffer if it's our buffer, user is responsible for deallocating contens himself - if (own || buffer != contents) _buffer = buffer; - - return res; - } - - PUGI__FN xml_parse_result xml_document::load_buffer(const void* contents, size_t size, unsigned int options, xml_encoding encoding) - { - return load_buffer_impl(const_cast(contents), size, options, encoding, false, false); - } - - PUGI__FN xml_parse_result xml_document::load_buffer_inplace(void* contents, size_t size, unsigned int options, xml_encoding encoding) - { - return load_buffer_impl(contents, size, options, encoding, true, false); - } - - PUGI__FN xml_parse_result xml_document::load_buffer_inplace_own(void* contents, size_t size, unsigned int options, xml_encoding encoding) - { - return load_buffer_impl(contents, size, options, encoding, true, true); - } - - PUGI__FN void xml_document::save(xml_writer& writer, const char_t* indent, unsigned int flags, xml_encoding encoding) const - { - impl::xml_buffered_writer buffered_writer(writer, encoding); - - if ((flags & format_write_bom) && encoding != encoding_latin1) - { - // BOM always represents the codepoint U+FEFF, so just write it in native encoding - #ifdef PUGIXML_WCHAR_MODE - unsigned int bom = 0xfeff; - buffered_writer.write(static_cast(bom)); - #else - buffered_writer.write('\xef', '\xbb', '\xbf'); - #endif - } - - if (!(flags & format_no_declaration) && !impl::has_declaration(*this)) - { - buffered_writer.write(PUGIXML_TEXT("'); - if (!(flags & format_raw)) buffered_writer.write('\n'); - } - - impl::node_output(buffered_writer, *this, indent, flags, 0); - } - -#ifndef PUGIXML_NO_STL - PUGI__FN void xml_document::save(std::basic_ostream >& stream, const char_t* indent, unsigned int flags, xml_encoding encoding) const - { - xml_writer_stream writer(stream); - - save(writer, indent, flags, encoding); - } - - PUGI__FN void xml_document::save(std::basic_ostream >& stream, const char_t* indent, unsigned int flags) const - { - xml_writer_stream writer(stream); - - save(writer, indent, flags, encoding_wchar); - } -#endif - - PUGI__FN bool xml_document::save_file(const char* path_, const char_t* indent, unsigned int flags, xml_encoding encoding) const - { - FILE* file = fopen(path_, (flags & format_save_file_text) ? "w" : "wb"); - return impl::save_file_impl(*this, file, indent, flags, encoding); - } - - PUGI__FN bool xml_document::save_file(const wchar_t* path_, const char_t* indent, unsigned int flags, xml_encoding encoding) const - { - FILE* file = impl::open_file_wide(path_, (flags & format_save_file_text) ? L"w" : L"wb"); - return impl::save_file_impl(*this, file, indent, flags, encoding); - } - - PUGI__FN xml_node xml_document::document_element() const - { - for (xml_node_struct* i = _root->first_child; i; i = i->next_sibling) - if ((i->header & impl::xml_memory_page_type_mask) + 1 == node_element) - return xml_node(i); - - return xml_node(); - } - -#ifndef PUGIXML_NO_STL - PUGI__FN std::string PUGIXML_FUNCTION as_utf8(const wchar_t* str) - { - assert(str); - - return impl::as_utf8_impl(str, wcslen(str)); - } - - PUGI__FN std::string PUGIXML_FUNCTION as_utf8(const std::basic_string& str) - { - return impl::as_utf8_impl(str.c_str(), str.size()); - } - - PUGI__FN std::basic_string PUGIXML_FUNCTION as_wide(const char* str) - { - assert(str); - - return impl::as_wide_impl(str, strlen(str)); - } - - PUGI__FN std::basic_string PUGIXML_FUNCTION as_wide(const std::string& str) - { - return impl::as_wide_impl(str.c_str(), str.size()); - } -#endif - - PUGI__FN void PUGIXML_FUNCTION set_memory_management_functions(allocation_function allocate, deallocation_function deallocate) - { - impl::xml_memory::allocate = allocate; - impl::xml_memory::deallocate = deallocate; - } - - PUGI__FN allocation_function PUGIXML_FUNCTION get_memory_allocation_function() - { - return impl::xml_memory::allocate; - } - - PUGI__FN deallocation_function PUGIXML_FUNCTION get_memory_deallocation_function() - { - return impl::xml_memory::deallocate; - } -} - -#if !defined(PUGIXML_NO_STL) && (defined(_MSC_VER) || defined(__ICC)) -namespace std -{ - // Workarounds for (non-standard) iterator category detection for older versions (MSVC7/IC8 and earlier) - PUGI__FN std::bidirectional_iterator_tag _Iter_cat(const pugi::xml_node_iterator&) - { - return std::bidirectional_iterator_tag(); - } - - PUGI__FN std::bidirectional_iterator_tag _Iter_cat(const pugi::xml_attribute_iterator&) - { - return std::bidirectional_iterator_tag(); - } - - PUGI__FN std::forward_iterator_tag _Iter_cat(const pugi::xml_named_node_iterator&) - { - return std::forward_iterator_tag(); - } -} -#endif - -#if !defined(PUGIXML_NO_STL) && defined(__SUNPRO_CC) -namespace std -{ - // Workarounds for (non-standard) iterator category detection - PUGI__FN std::bidirectional_iterator_tag __iterator_category(const pugi::xml_node_iterator&) - { - return std::bidirectional_iterator_tag(); - } - - PUGI__FN std::bidirectional_iterator_tag __iterator_category(const pugi::xml_attribute_iterator&) - { - return std::bidirectional_iterator_tag(); - } - - PUGI__FN std::forward_iterator_tag __iterator_category(const pugi::xml_named_node_iterator&) - { - return std::forward_iterator_tag(); - } -} -#endif - -#ifndef PUGIXML_NO_XPATH - -// STL replacements -PUGI__NS_BEGIN - struct equal_to - { - template bool operator()(const T& lhs, const T& rhs) const - { - return lhs == rhs; - } - }; - - struct not_equal_to - { - template bool operator()(const T& lhs, const T& rhs) const - { - return lhs != rhs; - } - }; - - struct less - { - template bool operator()(const T& lhs, const T& rhs) const - { - return lhs < rhs; - } - }; - - struct less_equal - { - template bool operator()(const T& lhs, const T& rhs) const - { - return lhs <= rhs; - } - }; - - template void swap(T& lhs, T& rhs) - { - T temp = lhs; - lhs = rhs; - rhs = temp; - } - - template I min_element(I begin, I end, const Pred& pred) - { - I result = begin; - - for (I it = begin + 1; it != end; ++it) - if (pred(*it, *result)) - result = it; - - return result; - } - - template void reverse(I begin, I end) - { - while (begin + 1 < end) swap(*begin++, *--end); - } - - template I unique(I begin, I end) - { - // fast skip head - while (begin + 1 < end && *begin != *(begin + 1)) begin++; - - if (begin == end) return begin; - - // last written element - I write = begin++; - - // merge unique elements - while (begin != end) - { - if (*begin != *write) - *++write = *begin++; - else - begin++; - } - - // past-the-end (write points to live element) - return write + 1; - } - - template void copy_backwards(I begin, I end, I target) - { - while (begin != end) *--target = *--end; - } - - template void insertion_sort(I begin, I end, const Pred& pred, T*) - { - assert(begin != end); - - for (I it = begin + 1; it != end; ++it) - { - T val = *it; - - if (pred(val, *begin)) - { - // move to front - copy_backwards(begin, it, it + 1); - *begin = val; - } - else - { - I hole = it; - - // move hole backwards - while (pred(val, *(hole - 1))) - { - *hole = *(hole - 1); - hole--; - } - - // fill hole with element - *hole = val; - } - } - } - - // std variant for elements with == - template void partition(I begin, I middle, I end, const Pred& pred, I* out_eqbeg, I* out_eqend) - { - I eqbeg = middle, eqend = middle + 1; - - // expand equal range - while (eqbeg != begin && *(eqbeg - 1) == *eqbeg) --eqbeg; - while (eqend != end && *eqend == *eqbeg) ++eqend; - - // process outer elements - I ltend = eqbeg, gtbeg = eqend; - - for (;;) - { - // find the element from the right side that belongs to the left one - for (; gtbeg != end; ++gtbeg) - if (!pred(*eqbeg, *gtbeg)) - { - if (*gtbeg == *eqbeg) swap(*gtbeg, *eqend++); - else break; - } - - // find the element from the left side that belongs to the right one - for (; ltend != begin; --ltend) - if (!pred(*(ltend - 1), *eqbeg)) - { - if (*eqbeg == *(ltend - 1)) swap(*(ltend - 1), *--eqbeg); - else break; - } - - // scanned all elements - if (gtbeg == end && ltend == begin) - { - *out_eqbeg = eqbeg; - *out_eqend = eqend; - return; - } - - // make room for elements by moving equal area - if (gtbeg == end) - { - if (--ltend != --eqbeg) swap(*ltend, *eqbeg); - swap(*eqbeg, *--eqend); - } - else if (ltend == begin) - { - if (eqend != gtbeg) swap(*eqbeg, *eqend); - ++eqend; - swap(*gtbeg++, *eqbeg++); - } - else swap(*gtbeg++, *--ltend); - } - } - - template void median3(I first, I middle, I last, const Pred& pred) - { - if (pred(*middle, *first)) swap(*middle, *first); - if (pred(*last, *middle)) swap(*last, *middle); - if (pred(*middle, *first)) swap(*middle, *first); - } - - template void median(I first, I middle, I last, const Pred& pred) - { - if (last - first <= 40) - { - // median of three for small chunks - median3(first, middle, last, pred); - } - else - { - // median of nine - size_t step = (last - first + 1) / 8; - - median3(first, first + step, first + 2 * step, pred); - median3(middle - step, middle, middle + step, pred); - median3(last - 2 * step, last - step, last, pred); - median3(first + step, middle, last - step, pred); - } - } - - template void sort(I begin, I end, const Pred& pred) - { - // sort large chunks - while (end - begin > 32) - { - // find median element - I middle = begin + (end - begin) / 2; - median(begin, middle, end - 1, pred); - - // partition in three chunks (< = >) - I eqbeg, eqend; - partition(begin, middle, end, pred, &eqbeg, &eqend); - - // loop on larger half - if (eqbeg - begin > end - eqend) - { - sort(eqend, end, pred); - end = eqbeg; - } - else - { - sort(begin, eqbeg, pred); - begin = eqend; - } - } - - // insertion sort small chunk - if (begin != end) insertion_sort(begin, end, pred, &*begin); - } -PUGI__NS_END - -// Allocator used for AST and evaluation stacks -PUGI__NS_BEGIN - struct xpath_memory_block - { - xpath_memory_block* next; - - char data[ - #ifdef PUGIXML_MEMORY_XPATH_PAGE_SIZE - PUGIXML_MEMORY_XPATH_PAGE_SIZE - #else - 4096 - #endif - ]; - }; - - class xpath_allocator - { - xpath_memory_block* _root; - size_t _root_size; - - public: - #ifdef PUGIXML_NO_EXCEPTIONS - jmp_buf* error_handler; - #endif - - xpath_allocator(xpath_memory_block* root, size_t root_size = 0): _root(root), _root_size(root_size) - { - #ifdef PUGIXML_NO_EXCEPTIONS - error_handler = 0; - #endif - } - - void* allocate_nothrow(size_t size) - { - const size_t block_capacity = sizeof(_root->data); - - // align size so that we're able to store pointers in subsequent blocks - size = (size + sizeof(void*) - 1) & ~(sizeof(void*) - 1); - - if (_root_size + size <= block_capacity) - { - void* buf = _root->data + _root_size; - _root_size += size; - return buf; - } - else - { - size_t block_data_size = (size > block_capacity) ? size : block_capacity; - size_t block_size = block_data_size + offsetof(xpath_memory_block, data); - - xpath_memory_block* block = static_cast(xml_memory::allocate(block_size)); - if (!block) return 0; - - block->next = _root; - - _root = block; - _root_size = size; - - return block->data; - } - } - - void* allocate(size_t size) - { - void* result = allocate_nothrow(size); - - if (!result) - { - #ifdef PUGIXML_NO_EXCEPTIONS - assert(error_handler); - longjmp(*error_handler, 1); - #else - throw std::bad_alloc(); - #endif - } - - return result; - } - - void* reallocate(void* ptr, size_t old_size, size_t new_size) - { - // align size so that we're able to store pointers in subsequent blocks - old_size = (old_size + sizeof(void*) - 1) & ~(sizeof(void*) - 1); - new_size = (new_size + sizeof(void*) - 1) & ~(sizeof(void*) - 1); - - // we can only reallocate the last object - assert(ptr == 0 || static_cast(ptr) + old_size == _root->data + _root_size); - - // adjust root size so that we have not allocated the object at all - bool only_object = (_root_size == old_size); - - if (ptr) _root_size -= old_size; - - // allocate a new version (this will obviously reuse the memory if possible) - void* result = allocate(new_size); - assert(result); - - // we have a new block - if (result != ptr && ptr) - { - // copy old data - assert(new_size > old_size); - memcpy(result, ptr, old_size); - - // free the previous page if it had no other objects - if (only_object) - { - assert(_root->data == result); - assert(_root->next); - - xpath_memory_block* next = _root->next->next; - - if (next) - { - // deallocate the whole page, unless it was the first one - xml_memory::deallocate(_root->next); - _root->next = next; - } - } - } - - return result; - } - - void revert(const xpath_allocator& state) - { - // free all new pages - xpath_memory_block* cur = _root; - - while (cur != state._root) - { - xpath_memory_block* next = cur->next; - - xml_memory::deallocate(cur); - - cur = next; - } - - // restore state - _root = state._root; - _root_size = state._root_size; - } - - void release() - { - xpath_memory_block* cur = _root; - assert(cur); - - while (cur->next) - { - xpath_memory_block* next = cur->next; - - xml_memory::deallocate(cur); - - cur = next; - } - } - }; - - struct xpath_allocator_capture - { - xpath_allocator_capture(xpath_allocator* alloc): _target(alloc), _state(*alloc) - { - } - - ~xpath_allocator_capture() - { - _target->revert(_state); - } - - xpath_allocator* _target; - xpath_allocator _state; - }; - - struct xpath_stack - { - xpath_allocator* result; - xpath_allocator* temp; - }; - - struct xpath_stack_data - { - xpath_memory_block blocks[2]; - xpath_allocator result; - xpath_allocator temp; - xpath_stack stack; - - #ifdef PUGIXML_NO_EXCEPTIONS - jmp_buf error_handler; - #endif - - xpath_stack_data(): result(blocks + 0), temp(blocks + 1) - { - blocks[0].next = blocks[1].next = 0; - - stack.result = &result; - stack.temp = &temp; - - #ifdef PUGIXML_NO_EXCEPTIONS - result.error_handler = temp.error_handler = &error_handler; - #endif - } - - ~xpath_stack_data() - { - result.release(); - temp.release(); - } - }; -PUGI__NS_END - -// String class -PUGI__NS_BEGIN - class xpath_string - { - const char_t* _buffer; - bool _uses_heap; - - static char_t* duplicate_string(const char_t* string, size_t length, xpath_allocator* alloc) - { - char_t* result = static_cast(alloc->allocate((length + 1) * sizeof(char_t))); - assert(result); - - memcpy(result, string, length * sizeof(char_t)); - result[length] = 0; - - return result; - } - - static char_t* duplicate_string(const char_t* string, xpath_allocator* alloc) - { - return duplicate_string(string, strlength(string), alloc); - } - - public: - xpath_string(): _buffer(PUGIXML_TEXT("")), _uses_heap(false) - { - } - - explicit xpath_string(const char_t* str, xpath_allocator* alloc) - { - bool empty_ = (*str == 0); - - _buffer = empty_ ? PUGIXML_TEXT("") : duplicate_string(str, alloc); - _uses_heap = !empty_; - } - - explicit xpath_string(const char_t* str, bool use_heap): _buffer(str), _uses_heap(use_heap) - { - } - - xpath_string(const char_t* begin, const char_t* end, xpath_allocator* alloc) - { - assert(begin <= end); - - bool empty_ = (begin == end); - - _buffer = empty_ ? PUGIXML_TEXT("") : duplicate_string(begin, static_cast(end - begin), alloc); - _uses_heap = !empty_; - } - - void append(const xpath_string& o, xpath_allocator* alloc) - { - // skip empty sources - if (!*o._buffer) return; - - // fast append for constant empty target and constant source - if (!*_buffer && !_uses_heap && !o._uses_heap) - { - _buffer = o._buffer; - } - else - { - // need to make heap copy - size_t target_length = strlength(_buffer); - size_t source_length = strlength(o._buffer); - size_t result_length = target_length + source_length; - - // allocate new buffer - char_t* result = static_cast(alloc->reallocate(_uses_heap ? const_cast(_buffer) : 0, (target_length + 1) * sizeof(char_t), (result_length + 1) * sizeof(char_t))); - assert(result); - - // append first string to the new buffer in case there was no reallocation - if (!_uses_heap) memcpy(result, _buffer, target_length * sizeof(char_t)); - - // append second string to the new buffer - memcpy(result + target_length, o._buffer, source_length * sizeof(char_t)); - result[result_length] = 0; - - // finalize - _buffer = result; - _uses_heap = true; - } - } - - const char_t* c_str() const - { - return _buffer; - } - - size_t length() const - { - return strlength(_buffer); - } - - char_t* data(xpath_allocator* alloc) - { - // make private heap copy - if (!_uses_heap) - { - _buffer = duplicate_string(_buffer, alloc); - _uses_heap = true; - } - - return const_cast(_buffer); - } - - bool empty() const - { - return *_buffer == 0; - } - - bool operator==(const xpath_string& o) const - { - return strequal(_buffer, o._buffer); - } - - bool operator!=(const xpath_string& o) const - { - return !strequal(_buffer, o._buffer); - } - - bool uses_heap() const - { - return _uses_heap; - } - }; - - PUGI__FN xpath_string xpath_string_const(const char_t* str) - { - return xpath_string(str, false); - } -PUGI__NS_END - -PUGI__NS_BEGIN - PUGI__FN bool starts_with(const char_t* string, const char_t* pattern) - { - while (*pattern && *string == *pattern) - { - string++; - pattern++; - } - - return *pattern == 0; - } - - PUGI__FN const char_t* find_char(const char_t* s, char_t c) - { - #ifdef PUGIXML_WCHAR_MODE - return wcschr(s, c); - #else - return strchr(s, c); - #endif - } - - PUGI__FN const char_t* find_substring(const char_t* s, const char_t* p) - { - #ifdef PUGIXML_WCHAR_MODE - // MSVC6 wcsstr bug workaround (if s is empty it always returns 0) - return (*p == 0) ? s : wcsstr(s, p); - #else - return strstr(s, p); - #endif - } - - // Converts symbol to lower case, if it is an ASCII one - PUGI__FN char_t tolower_ascii(char_t ch) - { - return static_cast(ch - 'A') < 26 ? static_cast(ch | ' ') : ch; - } - - PUGI__FN xpath_string string_value(const xpath_node& na, xpath_allocator* alloc) - { - if (na.attribute()) - return xpath_string_const(na.attribute().value()); - else - { - const xml_node& n = na.node(); - - switch (n.type()) - { - case node_pcdata: - case node_cdata: - case node_comment: - case node_pi: - return xpath_string_const(n.value()); - - case node_document: - case node_element: - { - xpath_string result; - - xml_node cur = n.first_child(); - - while (cur && cur != n) - { - if (cur.type() == node_pcdata || cur.type() == node_cdata) - result.append(xpath_string_const(cur.value()), alloc); - - if (cur.first_child()) - cur = cur.first_child(); - else if (cur.next_sibling()) - cur = cur.next_sibling(); - else - { - while (!cur.next_sibling() && cur != n) - cur = cur.parent(); - - if (cur != n) cur = cur.next_sibling(); - } - } - - return result; - } - - default: - return xpath_string(); - } - } - } - - PUGI__FN unsigned int node_height(xml_node n) - { - unsigned int result = 0; - - while (n) - { - ++result; - n = n.parent(); - } - - return result; - } - - PUGI__FN bool node_is_before(xml_node ln, unsigned int lh, xml_node rn, unsigned int rh) - { - // normalize heights - for (unsigned int i = rh; i < lh; i++) ln = ln.parent(); - for (unsigned int j = lh; j < rh; j++) rn = rn.parent(); - - // one node is the ancestor of the other - if (ln == rn) return lh < rh; - - // find common ancestor - while (ln.parent() != rn.parent()) - { - ln = ln.parent(); - rn = rn.parent(); - } - - // there is no common ancestor (the shared parent is null), nodes are from different documents - if (!ln.parent()) return ln < rn; - - // determine sibling order - for (; ln; ln = ln.next_sibling()) - if (ln == rn) - return true; - - return false; - } - - PUGI__FN bool node_is_ancestor(xml_node parent, xml_node node) - { - while (node && node != parent) node = node.parent(); - - return parent && node == parent; - } - - PUGI__FN const void* document_order(const xpath_node& xnode) - { - xml_node_struct* node = xnode.node().internal_object(); - - if (node) - { - if (node->name && (node->header & xml_memory_page_name_allocated_mask) == 0) return node->name; - if (node->value && (node->header & xml_memory_page_value_allocated_mask) == 0) return node->value; - return 0; - } - - xml_attribute_struct* attr = xnode.attribute().internal_object(); - - if (attr) - { - if ((attr->header & xml_memory_page_name_allocated_mask) == 0) return attr->name; - if ((attr->header & xml_memory_page_value_allocated_mask) == 0) return attr->value; - return 0; - } - - return 0; - } - - struct document_order_comparator - { - bool operator()(const xpath_node& lhs, const xpath_node& rhs) const - { - // optimized document order based check - const void* lo = document_order(lhs); - const void* ro = document_order(rhs); - - if (lo && ro) return lo < ro; - - // slow comparison - xml_node ln = lhs.node(), rn = rhs.node(); - - // compare attributes - if (lhs.attribute() && rhs.attribute()) - { - // shared parent - if (lhs.parent() == rhs.parent()) - { - // determine sibling order - for (xml_attribute a = lhs.attribute(); a; a = a.next_attribute()) - if (a == rhs.attribute()) - return true; - - return false; - } - - // compare attribute parents - ln = lhs.parent(); - rn = rhs.parent(); - } - else if (lhs.attribute()) - { - // attributes go after the parent element - if (lhs.parent() == rhs.node()) return false; - - ln = lhs.parent(); - } - else if (rhs.attribute()) - { - // attributes go after the parent element - if (rhs.parent() == lhs.node()) return true; - - rn = rhs.parent(); - } - - if (ln == rn) return false; - - unsigned int lh = node_height(ln); - unsigned int rh = node_height(rn); - - return node_is_before(ln, lh, rn, rh); - } - }; - - struct duplicate_comparator - { - bool operator()(const xpath_node& lhs, const xpath_node& rhs) const - { - if (lhs.attribute()) return rhs.attribute() ? lhs.attribute() < rhs.attribute() : true; - else return rhs.attribute() ? false : lhs.node() < rhs.node(); - } - }; - - PUGI__FN double gen_nan() - { - #if defined(__STDC_IEC_559__) || ((FLT_RADIX - 0 == 2) && (FLT_MAX_EXP - 0 == 128) && (FLT_MANT_DIG - 0 == 24)) - union { float f; uint32_t i; } u[sizeof(float) == sizeof(uint32_t) ? 1 : -1]; - u[0].i = 0x7fc00000; - return u[0].f; - #else - // fallback - const volatile double zero = 0.0; - return zero / zero; - #endif - } - - PUGI__FN bool is_nan(double value) - { - #if defined(PUGI__MSVC_CRT_VERSION) || defined(__BORLANDC__) - return !!_isnan(value); - #elif defined(fpclassify) && defined(FP_NAN) - return fpclassify(value) == FP_NAN; - #else - // fallback - const volatile double v = value; - return v != v; - #endif - } - - PUGI__FN const char_t* convert_number_to_string_special(double value) - { - #if defined(PUGI__MSVC_CRT_VERSION) || defined(__BORLANDC__) - if (_finite(value)) return (value == 0) ? PUGIXML_TEXT("0") : 0; - if (_isnan(value)) return PUGIXML_TEXT("NaN"); - return value > 0 ? PUGIXML_TEXT("Infinity") : PUGIXML_TEXT("-Infinity"); - #elif defined(fpclassify) && defined(FP_NAN) && defined(FP_INFINITE) && defined(FP_ZERO) - switch (fpclassify(value)) - { - case FP_NAN: - return PUGIXML_TEXT("NaN"); - - case FP_INFINITE: - return value > 0 ? PUGIXML_TEXT("Infinity") : PUGIXML_TEXT("-Infinity"); - - case FP_ZERO: - return PUGIXML_TEXT("0"); - - default: - return 0; - } - #else - // fallback - const volatile double v = value; - - if (v == 0) return PUGIXML_TEXT("0"); - if (v != v) return PUGIXML_TEXT("NaN"); - if (v * 2 == v) return value > 0 ? PUGIXML_TEXT("Infinity") : PUGIXML_TEXT("-Infinity"); - return 0; - #endif - } - - PUGI__FN bool convert_number_to_boolean(double value) - { - return (value != 0 && !is_nan(value)); - } - - PUGI__FN void truncate_zeros(char* begin, char* end) - { - while (begin != end && end[-1] == '0') end--; - - *end = 0; - } - - // gets mantissa digits in the form of 0.xxxxx with 0. implied and the exponent -#if defined(PUGI__MSVC_CRT_VERSION) && PUGI__MSVC_CRT_VERSION >= 1400 && !defined(_WIN32_WCE) - PUGI__FN void convert_number_to_mantissa_exponent(double value, char* buffer, size_t buffer_size, char** out_mantissa, int* out_exponent) - { - // get base values - int sign, exponent; - _ecvt_s(buffer, buffer_size, value, DBL_DIG + 1, &exponent, &sign); - - // truncate redundant zeros - truncate_zeros(buffer, buffer + strlen(buffer)); - - // fill results - *out_mantissa = buffer; - *out_exponent = exponent; - } -#else - PUGI__FN void convert_number_to_mantissa_exponent(double value, char* buffer, size_t buffer_size, char** out_mantissa, int* out_exponent) - { - // get a scientific notation value with IEEE DBL_DIG decimals - sprintf(buffer, "%.*e", DBL_DIG, value); - assert(strlen(buffer) < buffer_size); - (void)!buffer_size; - - // get the exponent (possibly negative) - char* exponent_string = strchr(buffer, 'e'); - assert(exponent_string); - - int exponent = atoi(exponent_string + 1); - - // extract mantissa string: skip sign - char* mantissa = buffer[0] == '-' ? buffer + 1 : buffer; - assert(mantissa[0] != '0' && mantissa[1] == '.'); - - // divide mantissa by 10 to eliminate integer part - mantissa[1] = mantissa[0]; - mantissa++; - exponent++; - - // remove extra mantissa digits and zero-terminate mantissa - truncate_zeros(mantissa, exponent_string); - - // fill results - *out_mantissa = mantissa; - *out_exponent = exponent; - } -#endif - - PUGI__FN xpath_string convert_number_to_string(double value, xpath_allocator* alloc) - { - // try special number conversion - const char_t* special = convert_number_to_string_special(value); - if (special) return xpath_string_const(special); - - // get mantissa + exponent form - char mantissa_buffer[64]; - - char* mantissa; - int exponent; - convert_number_to_mantissa_exponent(value, mantissa_buffer, sizeof(mantissa_buffer), &mantissa, &exponent); - - // make the number! - char_t result[512]; - char_t* s = result; - - // sign - if (value < 0) *s++ = '-'; - - // integer part - if (exponent <= 0) - { - *s++ = '0'; - } - else - { - while (exponent > 0) - { - assert(*mantissa == 0 || static_cast(*mantissa - '0') <= 9); - *s++ = *mantissa ? *mantissa++ : '0'; - exponent--; - } - } - - // fractional part - if (*mantissa) - { - // decimal point - *s++ = '.'; - - // extra zeroes from negative exponent - while (exponent < 0) - { - *s++ = '0'; - exponent++; - } - - // extra mantissa digits - while (*mantissa) - { - assert(static_cast(*mantissa - '0') <= 9); - *s++ = *mantissa++; - } - } - - // zero-terminate - assert(s < result + sizeof(result) / sizeof(result[0])); - *s = 0; - - return xpath_string(result, alloc); - } - - PUGI__FN bool check_string_to_number_format(const char_t* string) - { - // parse leading whitespace - while (PUGI__IS_CHARTYPE(*string, ct_space)) ++string; - - // parse sign - if (*string == '-') ++string; - - if (!*string) return false; - - // if there is no integer part, there should be a decimal part with at least one digit - if (!PUGI__IS_CHARTYPEX(string[0], ctx_digit) && (string[0] != '.' || !PUGI__IS_CHARTYPEX(string[1], ctx_digit))) return false; - - // parse integer part - while (PUGI__IS_CHARTYPEX(*string, ctx_digit)) ++string; - - // parse decimal part - if (*string == '.') - { - ++string; - - while (PUGI__IS_CHARTYPEX(*string, ctx_digit)) ++string; - } - - // parse trailing whitespace - while (PUGI__IS_CHARTYPE(*string, ct_space)) ++string; - - return *string == 0; - } - - PUGI__FN double convert_string_to_number(const char_t* string) - { - // check string format - if (!check_string_to_number_format(string)) return gen_nan(); - - // parse string - #ifdef PUGIXML_WCHAR_MODE - return wcstod(string, 0); - #else - return atof(string); - #endif - } - - PUGI__FN bool convert_string_to_number(const char_t* begin, const char_t* end, double* out_result) - { - char_t buffer[32]; - - size_t length = static_cast(end - begin); - char_t* scratch = buffer; - - if (length >= sizeof(buffer) / sizeof(buffer[0])) - { - // need to make dummy on-heap copy - scratch = static_cast(xml_memory::allocate((length + 1) * sizeof(char_t))); - if (!scratch) return false; - } - - // copy string to zero-terminated buffer and perform conversion - memcpy(scratch, begin, length * sizeof(char_t)); - scratch[length] = 0; - - *out_result = convert_string_to_number(scratch); - - // free dummy buffer - if (scratch != buffer) xml_memory::deallocate(scratch); - - return true; - } - - PUGI__FN double round_nearest(double value) - { - return floor(value + 0.5); - } - - PUGI__FN double round_nearest_nzero(double value) - { - // same as round_nearest, but returns -0 for [-0.5, -0] - // ceil is used to differentiate between +0 and -0 (we return -0 for [-0.5, -0] and +0 for +0) - return (value >= -0.5 && value <= 0) ? ceil(value) : floor(value + 0.5); - } - - PUGI__FN const char_t* qualified_name(const xpath_node& node) - { - return node.attribute() ? node.attribute().name() : node.node().name(); - } - - PUGI__FN const char_t* local_name(const xpath_node& node) - { - const char_t* name = qualified_name(node); - const char_t* p = find_char(name, ':'); - - return p ? p + 1 : name; - } - - struct namespace_uri_predicate - { - const char_t* prefix; - size_t prefix_length; - - namespace_uri_predicate(const char_t* name) - { - const char_t* pos = find_char(name, ':'); - - prefix = pos ? name : 0; - prefix_length = pos ? static_cast(pos - name) : 0; - } - - bool operator()(const xml_attribute& a) const - { - const char_t* name = a.name(); - - if (!starts_with(name, PUGIXML_TEXT("xmlns"))) return false; - - return prefix ? name[5] == ':' && strequalrange(name + 6, prefix, prefix_length) : name[5] == 0; - } - }; - - PUGI__FN const char_t* namespace_uri(const xml_node& node) - { - namespace_uri_predicate pred = node.name(); - - xml_node p = node; - - while (p) - { - xml_attribute a = p.find_attribute(pred); - - if (a) return a.value(); - - p = p.parent(); - } - - return PUGIXML_TEXT(""); - } - - PUGI__FN const char_t* namespace_uri(const xml_attribute& attr, const xml_node& parent) - { - namespace_uri_predicate pred = attr.name(); - - // Default namespace does not apply to attributes - if (!pred.prefix) return PUGIXML_TEXT(""); - - xml_node p = parent; - - while (p) - { - xml_attribute a = p.find_attribute(pred); - - if (a) return a.value(); - - p = p.parent(); - } - - return PUGIXML_TEXT(""); - } - - PUGI__FN const char_t* namespace_uri(const xpath_node& node) - { - return node.attribute() ? namespace_uri(node.attribute(), node.parent()) : namespace_uri(node.node()); - } - - PUGI__FN void normalize_space(char_t* buffer) - { - char_t* write = buffer; - - for (char_t* it = buffer; *it; ) - { - char_t ch = *it++; - - if (PUGI__IS_CHARTYPE(ch, ct_space)) - { - // replace whitespace sequence with single space - while (PUGI__IS_CHARTYPE(*it, ct_space)) it++; - - // avoid leading spaces - if (write != buffer) *write++ = ' '; - } - else *write++ = ch; - } - - // remove trailing space - if (write != buffer && PUGI__IS_CHARTYPE(write[-1], ct_space)) write--; - - // zero-terminate - *write = 0; - } - - PUGI__FN void translate(char_t* buffer, const char_t* from, const char_t* to) - { - size_t to_length = strlength(to); - - char_t* write = buffer; - - while (*buffer) - { - PUGI__DMC_VOLATILE char_t ch = *buffer++; - - const char_t* pos = find_char(from, ch); - - if (!pos) - *write++ = ch; // do not process - else if (static_cast(pos - from) < to_length) - *write++ = to[pos - from]; // replace - } - - // zero-terminate - *write = 0; - } - - struct xpath_variable_boolean: xpath_variable - { - xpath_variable_boolean(): value(false) - { - } - - bool value; - char_t name[1]; - }; - - struct xpath_variable_number: xpath_variable - { - xpath_variable_number(): value(0) - { - } - - double value; - char_t name[1]; - }; - - struct xpath_variable_string: xpath_variable - { - xpath_variable_string(): value(0) - { - } - - ~xpath_variable_string() - { - if (value) xml_memory::deallocate(value); - } - - char_t* value; - char_t name[1]; - }; - - struct xpath_variable_node_set: xpath_variable - { - xpath_node_set value; - char_t name[1]; - }; - - static const xpath_node_set dummy_node_set; - - PUGI__FN unsigned int hash_string(const char_t* str) - { - // Jenkins one-at-a-time hash (http://en.wikipedia.org/wiki/Jenkins_hash_function#one-at-a-time) - unsigned int result = 0; - - while (*str) - { - result += static_cast(*str++); - result += result << 10; - result ^= result >> 6; - } - - result += result << 3; - result ^= result >> 11; - result += result << 15; - - return result; - } - - template PUGI__FN T* new_xpath_variable(const char_t* name) - { - size_t length = strlength(name); - if (length == 0) return 0; // empty variable names are invalid - - // $$ we can't use offsetof(T, name) because T is non-POD, so we just allocate additional length characters - void* memory = xml_memory::allocate(sizeof(T) + length * sizeof(char_t)); - if (!memory) return 0; - - T* result = new (memory) T(); - - memcpy(result->name, name, (length + 1) * sizeof(char_t)); - - return result; - } - - PUGI__FN xpath_variable* new_xpath_variable(xpath_value_type type, const char_t* name) - { - switch (type) - { - case xpath_type_node_set: - return new_xpath_variable(name); - - case xpath_type_number: - return new_xpath_variable(name); - - case xpath_type_string: - return new_xpath_variable(name); - - case xpath_type_boolean: - return new_xpath_variable(name); - - default: - return 0; - } - } - - template PUGI__FN void delete_xpath_variable(T* var) - { - var->~T(); - xml_memory::deallocate(var); - } - - PUGI__FN void delete_xpath_variable(xpath_value_type type, xpath_variable* var) - { - switch (type) - { - case xpath_type_node_set: - delete_xpath_variable(static_cast(var)); - break; - - case xpath_type_number: - delete_xpath_variable(static_cast(var)); - break; - - case xpath_type_string: - delete_xpath_variable(static_cast(var)); - break; - - case xpath_type_boolean: - delete_xpath_variable(static_cast(var)); - break; - - default: - assert(!"Invalid variable type"); - } - } - - PUGI__FN xpath_variable* get_variable(xpath_variable_set* set, const char_t* begin, const char_t* end) - { - char_t buffer[32]; - - size_t length = static_cast(end - begin); - char_t* scratch = buffer; - - if (length >= sizeof(buffer) / sizeof(buffer[0])) - { - // need to make dummy on-heap copy - scratch = static_cast(xml_memory::allocate((length + 1) * sizeof(char_t))); - if (!scratch) return 0; - } - - // copy string to zero-terminated buffer and perform lookup - memcpy(scratch, begin, length * sizeof(char_t)); - scratch[length] = 0; - - xpath_variable* result = set->get(scratch); - - // free dummy buffer - if (scratch != buffer) xml_memory::deallocate(scratch); - - return result; - } -PUGI__NS_END - -// Internal node set class -PUGI__NS_BEGIN - PUGI__FN xpath_node_set::type_t xpath_sort(xpath_node* begin, xpath_node* end, xpath_node_set::type_t type, bool rev) - { - xpath_node_set::type_t order = rev ? xpath_node_set::type_sorted_reverse : xpath_node_set::type_sorted; - - if (type == xpath_node_set::type_unsorted) - { - sort(begin, end, document_order_comparator()); - - type = xpath_node_set::type_sorted; - } - - if (type != order) reverse(begin, end); - - return order; - } - - PUGI__FN xpath_node xpath_first(const xpath_node* begin, const xpath_node* end, xpath_node_set::type_t type) - { - if (begin == end) return xpath_node(); - - switch (type) - { - case xpath_node_set::type_sorted: - return *begin; - - case xpath_node_set::type_sorted_reverse: - return *(end - 1); - - case xpath_node_set::type_unsorted: - return *min_element(begin, end, document_order_comparator()); - - default: - assert(!"Invalid node set type"); - return xpath_node(); - } - } - - class xpath_node_set_raw - { - xpath_node_set::type_t _type; - - xpath_node* _begin; - xpath_node* _end; - xpath_node* _eos; - - public: - xpath_node_set_raw(): _type(xpath_node_set::type_unsorted), _begin(0), _end(0), _eos(0) - { - } - - xpath_node* begin() const - { - return _begin; - } - - xpath_node* end() const - { - return _end; - } - - bool empty() const - { - return _begin == _end; - } - - size_t size() const - { - return static_cast(_end - _begin); - } - - xpath_node first() const - { - return xpath_first(_begin, _end, _type); - } - - void push_back(const xpath_node& node, xpath_allocator* alloc) - { - if (_end == _eos) - { - size_t capacity = static_cast(_eos - _begin); - - // get new capacity (1.5x rule) - size_t new_capacity = capacity + capacity / 2 + 1; - - // reallocate the old array or allocate a new one - xpath_node* data = static_cast(alloc->reallocate(_begin, capacity * sizeof(xpath_node), new_capacity * sizeof(xpath_node))); - assert(data); - - // finalize - _begin = data; - _end = data + capacity; - _eos = data + new_capacity; - } - - *_end++ = node; - } - - void append(const xpath_node* begin_, const xpath_node* end_, xpath_allocator* alloc) - { - size_t size_ = static_cast(_end - _begin); - size_t capacity = static_cast(_eos - _begin); - size_t count = static_cast(end_ - begin_); - - if (size_ + count > capacity) - { - // reallocate the old array or allocate a new one - xpath_node* data = static_cast(alloc->reallocate(_begin, capacity * sizeof(xpath_node), (size_ + count) * sizeof(xpath_node))); - assert(data); - - // finalize - _begin = data; - _end = data + size_; - _eos = data + size_ + count; - } - - memcpy(_end, begin_, count * sizeof(xpath_node)); - _end += count; - } - - void sort_do() - { - _type = xpath_sort(_begin, _end, _type, false); - } - - void truncate(xpath_node* pos) - { - assert(_begin <= pos && pos <= _end); - - _end = pos; - } - - void remove_duplicates() - { - if (_type == xpath_node_set::type_unsorted) - sort(_begin, _end, duplicate_comparator()); - - _end = unique(_begin, _end); - } - - xpath_node_set::type_t type() const - { - return _type; - } - - void set_type(xpath_node_set::type_t value) - { - _type = value; - } - }; -PUGI__NS_END - -PUGI__NS_BEGIN - struct xpath_context - { - xpath_node n; - size_t position, size; - - xpath_context(const xpath_node& n_, size_t position_, size_t size_): n(n_), position(position_), size(size_) - { - } - }; - - enum lexeme_t - { - lex_none = 0, - lex_equal, - lex_not_equal, - lex_less, - lex_greater, - lex_less_or_equal, - lex_greater_or_equal, - lex_plus, - lex_minus, - lex_multiply, - lex_union, - lex_var_ref, - lex_open_brace, - lex_close_brace, - lex_quoted_string, - lex_number, - lex_slash, - lex_double_slash, - lex_open_square_brace, - lex_close_square_brace, - lex_string, - lex_comma, - lex_axis_attribute, - lex_dot, - lex_double_dot, - lex_double_colon, - lex_eof - }; - - struct xpath_lexer_string - { - const char_t* begin; - const char_t* end; - - xpath_lexer_string(): begin(0), end(0) - { - } - - bool operator==(const char_t* other) const - { - size_t length = static_cast(end - begin); - - return strequalrange(other, begin, length); - } - }; - - class xpath_lexer - { - const char_t* _cur; - const char_t* _cur_lexeme_pos; - xpath_lexer_string _cur_lexeme_contents; - - lexeme_t _cur_lexeme; - - public: - explicit xpath_lexer(const char_t* query): _cur(query) - { - next(); - } - - const char_t* state() const - { - return _cur; - } - - void next() - { - const char_t* cur = _cur; - - while (PUGI__IS_CHARTYPE(*cur, ct_space)) ++cur; - - // save lexeme position for error reporting - _cur_lexeme_pos = cur; - - switch (*cur) - { - case 0: - _cur_lexeme = lex_eof; - break; - - case '>': - if (*(cur+1) == '=') - { - cur += 2; - _cur_lexeme = lex_greater_or_equal; - } - else - { - cur += 1; - _cur_lexeme = lex_greater; - } - break; - - case '<': - if (*(cur+1) == '=') - { - cur += 2; - _cur_lexeme = lex_less_or_equal; - } - else - { - cur += 1; - _cur_lexeme = lex_less; - } - break; - - case '!': - if (*(cur+1) == '=') - { - cur += 2; - _cur_lexeme = lex_not_equal; - } - else - { - _cur_lexeme = lex_none; - } - break; - - case '=': - cur += 1; - _cur_lexeme = lex_equal; - - break; - - case '+': - cur += 1; - _cur_lexeme = lex_plus; - - break; - - case '-': - cur += 1; - _cur_lexeme = lex_minus; - - break; - - case '*': - cur += 1; - _cur_lexeme = lex_multiply; - - break; - - case '|': - cur += 1; - _cur_lexeme = lex_union; - - break; - - case '$': - cur += 1; - - if (PUGI__IS_CHARTYPEX(*cur, ctx_start_symbol)) - { - _cur_lexeme_contents.begin = cur; - - while (PUGI__IS_CHARTYPEX(*cur, ctx_symbol)) cur++; - - if (cur[0] == ':' && PUGI__IS_CHARTYPEX(cur[1], ctx_symbol)) // qname - { - cur++; // : - - while (PUGI__IS_CHARTYPEX(*cur, ctx_symbol)) cur++; - } - - _cur_lexeme_contents.end = cur; - - _cur_lexeme = lex_var_ref; - } - else - { - _cur_lexeme = lex_none; - } - - break; - - case '(': - cur += 1; - _cur_lexeme = lex_open_brace; - - break; - - case ')': - cur += 1; - _cur_lexeme = lex_close_brace; - - break; - - case '[': - cur += 1; - _cur_lexeme = lex_open_square_brace; - - break; - - case ']': - cur += 1; - _cur_lexeme = lex_close_square_brace; - - break; - - case ',': - cur += 1; - _cur_lexeme = lex_comma; - - break; - - case '/': - if (*(cur+1) == '/') - { - cur += 2; - _cur_lexeme = lex_double_slash; - } - else - { - cur += 1; - _cur_lexeme = lex_slash; - } - break; - - case '.': - if (*(cur+1) == '.') - { - cur += 2; - _cur_lexeme = lex_double_dot; - } - else if (PUGI__IS_CHARTYPEX(*(cur+1), ctx_digit)) - { - _cur_lexeme_contents.begin = cur; // . - - ++cur; - - while (PUGI__IS_CHARTYPEX(*cur, ctx_digit)) cur++; - - _cur_lexeme_contents.end = cur; - - _cur_lexeme = lex_number; - } - else - { - cur += 1; - _cur_lexeme = lex_dot; - } - break; - - case '@': - cur += 1; - _cur_lexeme = lex_axis_attribute; - - break; - - case '"': - case '\'': - { - char_t terminator = *cur; - - ++cur; - - _cur_lexeme_contents.begin = cur; - while (*cur && *cur != terminator) cur++; - _cur_lexeme_contents.end = cur; - - if (!*cur) - _cur_lexeme = lex_none; - else - { - cur += 1; - _cur_lexeme = lex_quoted_string; - } - - break; - } - - case ':': - if (*(cur+1) == ':') - { - cur += 2; - _cur_lexeme = lex_double_colon; - } - else - { - _cur_lexeme = lex_none; - } - break; - - default: - if (PUGI__IS_CHARTYPEX(*cur, ctx_digit)) - { - _cur_lexeme_contents.begin = cur; - - while (PUGI__IS_CHARTYPEX(*cur, ctx_digit)) cur++; - - if (*cur == '.') - { - cur++; - - while (PUGI__IS_CHARTYPEX(*cur, ctx_digit)) cur++; - } - - _cur_lexeme_contents.end = cur; - - _cur_lexeme = lex_number; - } - else if (PUGI__IS_CHARTYPEX(*cur, ctx_start_symbol)) - { - _cur_lexeme_contents.begin = cur; - - while (PUGI__IS_CHARTYPEX(*cur, ctx_symbol)) cur++; - - if (cur[0] == ':') - { - if (cur[1] == '*') // namespace test ncname:* - { - cur += 2; // :* - } - else if (PUGI__IS_CHARTYPEX(cur[1], ctx_symbol)) // namespace test qname - { - cur++; // : - - while (PUGI__IS_CHARTYPEX(*cur, ctx_symbol)) cur++; - } - } - - _cur_lexeme_contents.end = cur; - - _cur_lexeme = lex_string; - } - else - { - _cur_lexeme = lex_none; - } - } - - _cur = cur; - } - - lexeme_t current() const - { - return _cur_lexeme; - } - - const char_t* current_pos() const - { - return _cur_lexeme_pos; - } - - const xpath_lexer_string& contents() const - { - assert(_cur_lexeme == lex_var_ref || _cur_lexeme == lex_number || _cur_lexeme == lex_string || _cur_lexeme == lex_quoted_string); - - return _cur_lexeme_contents; - } - }; - - enum ast_type_t - { - ast_op_or, // left or right - ast_op_and, // left and right - ast_op_equal, // left = right - ast_op_not_equal, // left != right - ast_op_less, // left < right - ast_op_greater, // left > right - ast_op_less_or_equal, // left <= right - ast_op_greater_or_equal, // left >= right - ast_op_add, // left + right - ast_op_subtract, // left - right - ast_op_multiply, // left * right - ast_op_divide, // left / right - ast_op_mod, // left % right - ast_op_negate, // left - right - ast_op_union, // left | right - ast_predicate, // apply predicate to set; next points to next predicate - ast_filter, // select * from left where right - ast_filter_posinv, // select * from left where right; proximity position invariant - ast_string_constant, // string constant - ast_number_constant, // number constant - ast_variable, // variable - ast_func_last, // last() - ast_func_position, // position() - ast_func_count, // count(left) - ast_func_id, // id(left) - ast_func_local_name_0, // local-name() - ast_func_local_name_1, // local-name(left) - ast_func_namespace_uri_0, // namespace-uri() - ast_func_namespace_uri_1, // namespace-uri(left) - ast_func_name_0, // name() - ast_func_name_1, // name(left) - ast_func_string_0, // string() - ast_func_string_1, // string(left) - ast_func_concat, // concat(left, right, siblings) - ast_func_starts_with, // starts_with(left, right) - ast_func_contains, // contains(left, right) - ast_func_substring_before, // substring-before(left, right) - ast_func_substring_after, // substring-after(left, right) - ast_func_substring_2, // substring(left, right) - ast_func_substring_3, // substring(left, right, third) - ast_func_string_length_0, // string-length() - ast_func_string_length_1, // string-length(left) - ast_func_normalize_space_0, // normalize-space() - ast_func_normalize_space_1, // normalize-space(left) - ast_func_translate, // translate(left, right, third) - ast_func_boolean, // boolean(left) - ast_func_not, // not(left) - ast_func_true, // true() - ast_func_false, // false() - ast_func_lang, // lang(left) - ast_func_number_0, // number() - ast_func_number_1, // number(left) - ast_func_sum, // sum(left) - ast_func_floor, // floor(left) - ast_func_ceiling, // ceiling(left) - ast_func_round, // round(left) - ast_step, // process set left with step - ast_step_root // select root node - }; - - enum axis_t - { - axis_ancestor, - axis_ancestor_or_self, - axis_attribute, - axis_child, - axis_descendant, - axis_descendant_or_self, - axis_following, - axis_following_sibling, - axis_namespace, - axis_parent, - axis_preceding, - axis_preceding_sibling, - axis_self - }; - - enum nodetest_t - { - nodetest_none, - nodetest_name, - nodetest_type_node, - nodetest_type_comment, - nodetest_type_pi, - nodetest_type_text, - nodetest_pi, - nodetest_all, - nodetest_all_in_namespace - }; - - template struct axis_to_type - { - static const axis_t axis; - }; - - template const axis_t axis_to_type::axis = N; - - class xpath_ast_node - { - private: - // node type - char _type; - char _rettype; - - // for ast_step / ast_predicate - char _axis; - char _test; - - // tree node structure - xpath_ast_node* _left; - xpath_ast_node* _right; - xpath_ast_node* _next; - - union - { - // value for ast_string_constant - const char_t* string; - // value for ast_number_constant - double number; - // variable for ast_variable - xpath_variable* variable; - // node test for ast_step (node name/namespace/node type/pi target) - const char_t* nodetest; - } _data; - - xpath_ast_node(const xpath_ast_node&); - xpath_ast_node& operator=(const xpath_ast_node&); - - template static bool compare_eq(xpath_ast_node* lhs, xpath_ast_node* rhs, const xpath_context& c, const xpath_stack& stack, const Comp& comp) - { - xpath_value_type lt = lhs->rettype(), rt = rhs->rettype(); - - if (lt != xpath_type_node_set && rt != xpath_type_node_set) - { - if (lt == xpath_type_boolean || rt == xpath_type_boolean) - return comp(lhs->eval_boolean(c, stack), rhs->eval_boolean(c, stack)); - else if (lt == xpath_type_number || rt == xpath_type_number) - return comp(lhs->eval_number(c, stack), rhs->eval_number(c, stack)); - else if (lt == xpath_type_string || rt == xpath_type_string) - { - xpath_allocator_capture cr(stack.result); - - xpath_string ls = lhs->eval_string(c, stack); - xpath_string rs = rhs->eval_string(c, stack); - - return comp(ls, rs); - } - } - else if (lt == xpath_type_node_set && rt == xpath_type_node_set) - { - xpath_allocator_capture cr(stack.result); - - xpath_node_set_raw ls = lhs->eval_node_set(c, stack); - xpath_node_set_raw rs = rhs->eval_node_set(c, stack); - - for (const xpath_node* li = ls.begin(); li != ls.end(); ++li) - for (const xpath_node* ri = rs.begin(); ri != rs.end(); ++ri) - { - xpath_allocator_capture cri(stack.result); - - if (comp(string_value(*li, stack.result), string_value(*ri, stack.result))) - return true; - } - - return false; - } - else - { - if (lt == xpath_type_node_set) - { - swap(lhs, rhs); - swap(lt, rt); - } - - if (lt == xpath_type_boolean) - return comp(lhs->eval_boolean(c, stack), rhs->eval_boolean(c, stack)); - else if (lt == xpath_type_number) - { - xpath_allocator_capture cr(stack.result); - - double l = lhs->eval_number(c, stack); - xpath_node_set_raw rs = rhs->eval_node_set(c, stack); - - for (const xpath_node* ri = rs.begin(); ri != rs.end(); ++ri) - { - xpath_allocator_capture cri(stack.result); - - if (comp(l, convert_string_to_number(string_value(*ri, stack.result).c_str()))) - return true; - } - - return false; - } - else if (lt == xpath_type_string) - { - xpath_allocator_capture cr(stack.result); - - xpath_string l = lhs->eval_string(c, stack); - xpath_node_set_raw rs = rhs->eval_node_set(c, stack); - - for (const xpath_node* ri = rs.begin(); ri != rs.end(); ++ri) - { - xpath_allocator_capture cri(stack.result); - - if (comp(l, string_value(*ri, stack.result))) - return true; - } - - return false; - } - } - - assert(!"Wrong types"); - return false; - } - - template static bool compare_rel(xpath_ast_node* lhs, xpath_ast_node* rhs, const xpath_context& c, const xpath_stack& stack, const Comp& comp) - { - xpath_value_type lt = lhs->rettype(), rt = rhs->rettype(); - - if (lt != xpath_type_node_set && rt != xpath_type_node_set) - return comp(lhs->eval_number(c, stack), rhs->eval_number(c, stack)); - else if (lt == xpath_type_node_set && rt == xpath_type_node_set) - { - xpath_allocator_capture cr(stack.result); - - xpath_node_set_raw ls = lhs->eval_node_set(c, stack); - xpath_node_set_raw rs = rhs->eval_node_set(c, stack); - - for (const xpath_node* li = ls.begin(); li != ls.end(); ++li) - { - xpath_allocator_capture cri(stack.result); - - double l = convert_string_to_number(string_value(*li, stack.result).c_str()); - - for (const xpath_node* ri = rs.begin(); ri != rs.end(); ++ri) - { - xpath_allocator_capture crii(stack.result); - - if (comp(l, convert_string_to_number(string_value(*ri, stack.result).c_str()))) - return true; - } - } - - return false; - } - else if (lt != xpath_type_node_set && rt == xpath_type_node_set) - { - xpath_allocator_capture cr(stack.result); - - double l = lhs->eval_number(c, stack); - xpath_node_set_raw rs = rhs->eval_node_set(c, stack); - - for (const xpath_node* ri = rs.begin(); ri != rs.end(); ++ri) - { - xpath_allocator_capture cri(stack.result); - - if (comp(l, convert_string_to_number(string_value(*ri, stack.result).c_str()))) - return true; - } - - return false; - } - else if (lt == xpath_type_node_set && rt != xpath_type_node_set) - { - xpath_allocator_capture cr(stack.result); - - xpath_node_set_raw ls = lhs->eval_node_set(c, stack); - double r = rhs->eval_number(c, stack); - - for (const xpath_node* li = ls.begin(); li != ls.end(); ++li) - { - xpath_allocator_capture cri(stack.result); - - if (comp(convert_string_to_number(string_value(*li, stack.result).c_str()), r)) - return true; - } - - return false; - } - else - { - assert(!"Wrong types"); - return false; - } - } - - void apply_predicate(xpath_node_set_raw& ns, size_t first, xpath_ast_node* expr, const xpath_stack& stack) - { - assert(ns.size() >= first); - - size_t i = 1; - size_t size = ns.size() - first; - - xpath_node* last = ns.begin() + first; - - // remove_if... or well, sort of - for (xpath_node* it = last; it != ns.end(); ++it, ++i) - { - xpath_context c(*it, i, size); - - if (expr->rettype() == xpath_type_number) - { - if (expr->eval_number(c, stack) == i) - *last++ = *it; - } - else if (expr->eval_boolean(c, stack)) - *last++ = *it; - } - - ns.truncate(last); - } - - void apply_predicates(xpath_node_set_raw& ns, size_t first, const xpath_stack& stack) - { - if (ns.size() == first) return; - - for (xpath_ast_node* pred = _right; pred; pred = pred->_next) - { - apply_predicate(ns, first, pred->_left, stack); - } - } - - void step_push(xpath_node_set_raw& ns, const xml_attribute& a, const xml_node& parent, xpath_allocator* alloc) - { - if (!a) return; - - const char_t* name = a.name(); - - // There are no attribute nodes corresponding to attributes that declare namespaces - // That is, "xmlns:..." or "xmlns" - if (starts_with(name, PUGIXML_TEXT("xmlns")) && (name[5] == 0 || name[5] == ':')) return; - - switch (_test) - { - case nodetest_name: - if (strequal(name, _data.nodetest)) ns.push_back(xpath_node(a, parent), alloc); - break; - - case nodetest_type_node: - case nodetest_all: - ns.push_back(xpath_node(a, parent), alloc); - break; - - case nodetest_all_in_namespace: - if (starts_with(name, _data.nodetest)) - ns.push_back(xpath_node(a, parent), alloc); - break; - - default: - ; - } - } - - void step_push(xpath_node_set_raw& ns, const xml_node& n, xpath_allocator* alloc) - { - if (!n) return; - - switch (_test) - { - case nodetest_name: - if (n.type() == node_element && strequal(n.name(), _data.nodetest)) ns.push_back(n, alloc); - break; - - case nodetest_type_node: - ns.push_back(n, alloc); - break; - - case nodetest_type_comment: - if (n.type() == node_comment) - ns.push_back(n, alloc); - break; - - case nodetest_type_text: - if (n.type() == node_pcdata || n.type() == node_cdata) - ns.push_back(n, alloc); - break; - - case nodetest_type_pi: - if (n.type() == node_pi) - ns.push_back(n, alloc); - break; - - case nodetest_pi: - if (n.type() == node_pi && strequal(n.name(), _data.nodetest)) - ns.push_back(n, alloc); - break; - - case nodetest_all: - if (n.type() == node_element) - ns.push_back(n, alloc); - break; - - case nodetest_all_in_namespace: - if (n.type() == node_element && starts_with(n.name(), _data.nodetest)) - ns.push_back(n, alloc); - break; - - default: - assert(!"Unknown axis"); - } - } - - template void step_fill(xpath_node_set_raw& ns, const xml_node& n, xpath_allocator* alloc, T) - { - const axis_t axis = T::axis; - - switch (axis) - { - case axis_attribute: - { - for (xml_attribute a = n.first_attribute(); a; a = a.next_attribute()) - step_push(ns, a, n, alloc); - - break; - } - - case axis_child: - { - for (xml_node c = n.first_child(); c; c = c.next_sibling()) - step_push(ns, c, alloc); - - break; - } - - case axis_descendant: - case axis_descendant_or_self: - { - if (axis == axis_descendant_or_self) - step_push(ns, n, alloc); - - xml_node cur = n.first_child(); - - while (cur && cur != n) - { - step_push(ns, cur, alloc); - - if (cur.first_child()) - cur = cur.first_child(); - else if (cur.next_sibling()) - cur = cur.next_sibling(); - else - { - while (!cur.next_sibling() && cur != n) - cur = cur.parent(); - - if (cur != n) cur = cur.next_sibling(); - } - } - - break; - } - - case axis_following_sibling: - { - for (xml_node c = n.next_sibling(); c; c = c.next_sibling()) - step_push(ns, c, alloc); - - break; - } - - case axis_preceding_sibling: - { - for (xml_node c = n.previous_sibling(); c; c = c.previous_sibling()) - step_push(ns, c, alloc); - - break; - } - - case axis_following: - { - xml_node cur = n; - - // exit from this node so that we don't include descendants - while (cur && !cur.next_sibling()) cur = cur.parent(); - cur = cur.next_sibling(); - - for (;;) - { - step_push(ns, cur, alloc); - - if (cur.first_child()) - cur = cur.first_child(); - else if (cur.next_sibling()) - cur = cur.next_sibling(); - else - { - while (cur && !cur.next_sibling()) cur = cur.parent(); - cur = cur.next_sibling(); - - if (!cur) break; - } - } - - break; - } - - case axis_preceding: - { - xml_node cur = n; - - while (cur && !cur.previous_sibling()) cur = cur.parent(); - cur = cur.previous_sibling(); - - for (;;) - { - if (cur.last_child()) - cur = cur.last_child(); - else - { - // leaf node, can't be ancestor - step_push(ns, cur, alloc); - - if (cur.previous_sibling()) - cur = cur.previous_sibling(); - else - { - do - { - cur = cur.parent(); - if (!cur) break; - - if (!node_is_ancestor(cur, n)) step_push(ns, cur, alloc); - } - while (!cur.previous_sibling()); - - cur = cur.previous_sibling(); - - if (!cur) break; - } - } - } - - break; - } - - case axis_ancestor: - case axis_ancestor_or_self: - { - if (axis == axis_ancestor_or_self) - step_push(ns, n, alloc); - - xml_node cur = n.parent(); - - while (cur) - { - step_push(ns, cur, alloc); - - cur = cur.parent(); - } - - break; - } - - case axis_self: - { - step_push(ns, n, alloc); - - break; - } - - case axis_parent: - { - if (n.parent()) step_push(ns, n.parent(), alloc); - - break; - } - - default: - assert(!"Unimplemented axis"); - } - } - - template void step_fill(xpath_node_set_raw& ns, const xml_attribute& a, const xml_node& p, xpath_allocator* alloc, T v) - { - const axis_t axis = T::axis; - - switch (axis) - { - case axis_ancestor: - case axis_ancestor_or_self: - { - if (axis == axis_ancestor_or_self && _test == nodetest_type_node) // reject attributes based on principal node type test - step_push(ns, a, p, alloc); - - xml_node cur = p; - - while (cur) - { - step_push(ns, cur, alloc); - - cur = cur.parent(); - } - - break; - } - - case axis_descendant_or_self: - case axis_self: - { - if (_test == nodetest_type_node) // reject attributes based on principal node type test - step_push(ns, a, p, alloc); - - break; - } - - case axis_following: - { - xml_node cur = p; - - for (;;) - { - if (cur.first_child()) - cur = cur.first_child(); - else if (cur.next_sibling()) - cur = cur.next_sibling(); - else - { - while (cur && !cur.next_sibling()) cur = cur.parent(); - cur = cur.next_sibling(); - - if (!cur) break; - } - - step_push(ns, cur, alloc); - } - - break; - } - - case axis_parent: - { - step_push(ns, p, alloc); - - break; - } - - case axis_preceding: - { - // preceding:: axis does not include attribute nodes and attribute ancestors (they are the same as parent's ancestors), so we can reuse node preceding - step_fill(ns, p, alloc, v); - break; - } - - default: - assert(!"Unimplemented axis"); - } - } - - template xpath_node_set_raw step_do(const xpath_context& c, const xpath_stack& stack, T v) - { - const axis_t axis = T::axis; - bool attributes = (axis == axis_ancestor || axis == axis_ancestor_or_self || axis == axis_descendant_or_self || axis == axis_following || axis == axis_parent || axis == axis_preceding || axis == axis_self); - - xpath_node_set_raw ns; - ns.set_type((axis == axis_ancestor || axis == axis_ancestor_or_self || axis == axis_preceding || axis == axis_preceding_sibling) ? xpath_node_set::type_sorted_reverse : xpath_node_set::type_sorted); - - if (_left) - { - xpath_node_set_raw s = _left->eval_node_set(c, stack); - - // self axis preserves the original order - if (axis == axis_self) ns.set_type(s.type()); - - for (const xpath_node* it = s.begin(); it != s.end(); ++it) - { - size_t size = ns.size(); - - // in general, all axes generate elements in a particular order, but there is no order guarantee if axis is applied to two nodes - if (axis != axis_self && size != 0) ns.set_type(xpath_node_set::type_unsorted); - - if (it->node()) - step_fill(ns, it->node(), stack.result, v); - else if (attributes) - step_fill(ns, it->attribute(), it->parent(), stack.result, v); - - apply_predicates(ns, size, stack); - } - } - else - { - if (c.n.node()) - step_fill(ns, c.n.node(), stack.result, v); - else if (attributes) - step_fill(ns, c.n.attribute(), c.n.parent(), stack.result, v); - - apply_predicates(ns, 0, stack); - } - - // child, attribute and self axes always generate unique set of nodes - // for other axis, if the set stayed sorted, it stayed unique because the traversal algorithms do not visit the same node twice - if (axis != axis_child && axis != axis_attribute && axis != axis_self && ns.type() == xpath_node_set::type_unsorted) - ns.remove_duplicates(); - - return ns; - } - - public: - xpath_ast_node(ast_type_t type, xpath_value_type rettype_, const char_t* value): - _type(static_cast(type)), _rettype(static_cast(rettype_)), _axis(0), _test(0), _left(0), _right(0), _next(0) - { - assert(type == ast_string_constant); - _data.string = value; - } - - xpath_ast_node(ast_type_t type, xpath_value_type rettype_, double value): - _type(static_cast(type)), _rettype(static_cast(rettype_)), _axis(0), _test(0), _left(0), _right(0), _next(0) - { - assert(type == ast_number_constant); - _data.number = value; - } - - xpath_ast_node(ast_type_t type, xpath_value_type rettype_, xpath_variable* value): - _type(static_cast(type)), _rettype(static_cast(rettype_)), _axis(0), _test(0), _left(0), _right(0), _next(0) - { - assert(type == ast_variable); - _data.variable = value; - } - - xpath_ast_node(ast_type_t type, xpath_value_type rettype_, xpath_ast_node* left = 0, xpath_ast_node* right = 0): - _type(static_cast(type)), _rettype(static_cast(rettype_)), _axis(0), _test(0), _left(left), _right(right), _next(0) - { - } - - xpath_ast_node(ast_type_t type, xpath_ast_node* left, axis_t axis, nodetest_t test, const char_t* contents): - _type(static_cast(type)), _rettype(xpath_type_node_set), _axis(static_cast(axis)), _test(static_cast(test)), _left(left), _right(0), _next(0) - { - _data.nodetest = contents; - } - - void set_next(xpath_ast_node* value) - { - _next = value; - } - - void set_right(xpath_ast_node* value) - { - _right = value; - } - - bool eval_boolean(const xpath_context& c, const xpath_stack& stack) - { - switch (_type) - { - case ast_op_or: - return _left->eval_boolean(c, stack) || _right->eval_boolean(c, stack); - - case ast_op_and: - return _left->eval_boolean(c, stack) && _right->eval_boolean(c, stack); - - case ast_op_equal: - return compare_eq(_left, _right, c, stack, equal_to()); - - case ast_op_not_equal: - return compare_eq(_left, _right, c, stack, not_equal_to()); - - case ast_op_less: - return compare_rel(_left, _right, c, stack, less()); - - case ast_op_greater: - return compare_rel(_right, _left, c, stack, less()); - - case ast_op_less_or_equal: - return compare_rel(_left, _right, c, stack, less_equal()); - - case ast_op_greater_or_equal: - return compare_rel(_right, _left, c, stack, less_equal()); - - case ast_func_starts_with: - { - xpath_allocator_capture cr(stack.result); - - xpath_string lr = _left->eval_string(c, stack); - xpath_string rr = _right->eval_string(c, stack); - - return starts_with(lr.c_str(), rr.c_str()); - } - - case ast_func_contains: - { - xpath_allocator_capture cr(stack.result); - - xpath_string lr = _left->eval_string(c, stack); - xpath_string rr = _right->eval_string(c, stack); - - return find_substring(lr.c_str(), rr.c_str()) != 0; - } - - case ast_func_boolean: - return _left->eval_boolean(c, stack); - - case ast_func_not: - return !_left->eval_boolean(c, stack); - - case ast_func_true: - return true; - - case ast_func_false: - return false; - - case ast_func_lang: - { - if (c.n.attribute()) return false; - - xpath_allocator_capture cr(stack.result); - - xpath_string lang = _left->eval_string(c, stack); - - for (xml_node n = c.n.node(); n; n = n.parent()) - { - xml_attribute a = n.attribute(PUGIXML_TEXT("xml:lang")); - - if (a) - { - const char_t* value = a.value(); - - // strnicmp / strncasecmp is not portable - for (const char_t* lit = lang.c_str(); *lit; ++lit) - { - if (tolower_ascii(*lit) != tolower_ascii(*value)) return false; - ++value; - } - - return *value == 0 || *value == '-'; - } - } - - return false; - } - - case ast_variable: - { - assert(_rettype == _data.variable->type()); - - if (_rettype == xpath_type_boolean) - return _data.variable->get_boolean(); - - // fallthrough to type conversion - } - - default: - { - switch (_rettype) - { - case xpath_type_number: - return convert_number_to_boolean(eval_number(c, stack)); - - case xpath_type_string: - { - xpath_allocator_capture cr(stack.result); - - return !eval_string(c, stack).empty(); - } - - case xpath_type_node_set: - { - xpath_allocator_capture cr(stack.result); - - return !eval_node_set(c, stack).empty(); - } - - default: - assert(!"Wrong expression for return type boolean"); - return false; - } - } - } - } - - double eval_number(const xpath_context& c, const xpath_stack& stack) - { - switch (_type) - { - case ast_op_add: - return _left->eval_number(c, stack) + _right->eval_number(c, stack); - - case ast_op_subtract: - return _left->eval_number(c, stack) - _right->eval_number(c, stack); - - case ast_op_multiply: - return _left->eval_number(c, stack) * _right->eval_number(c, stack); - - case ast_op_divide: - return _left->eval_number(c, stack) / _right->eval_number(c, stack); - - case ast_op_mod: - return fmod(_left->eval_number(c, stack), _right->eval_number(c, stack)); - - case ast_op_negate: - return -_left->eval_number(c, stack); - - case ast_number_constant: - return _data.number; - - case ast_func_last: - return static_cast(c.size); - - case ast_func_position: - return static_cast(c.position); - - case ast_func_count: - { - xpath_allocator_capture cr(stack.result); - - return static_cast(_left->eval_node_set(c, stack).size()); - } - - case ast_func_string_length_0: - { - xpath_allocator_capture cr(stack.result); - - return static_cast(string_value(c.n, stack.result).length()); - } - - case ast_func_string_length_1: - { - xpath_allocator_capture cr(stack.result); - - return static_cast(_left->eval_string(c, stack).length()); - } - - case ast_func_number_0: - { - xpath_allocator_capture cr(stack.result); - - return convert_string_to_number(string_value(c.n, stack.result).c_str()); - } - - case ast_func_number_1: - return _left->eval_number(c, stack); - - case ast_func_sum: - { - xpath_allocator_capture cr(stack.result); - - double r = 0; - - xpath_node_set_raw ns = _left->eval_node_set(c, stack); - - for (const xpath_node* it = ns.begin(); it != ns.end(); ++it) - { - xpath_allocator_capture cri(stack.result); - - r += convert_string_to_number(string_value(*it, stack.result).c_str()); - } - - return r; - } - - case ast_func_floor: - { - double r = _left->eval_number(c, stack); - - return r == r ? floor(r) : r; - } - - case ast_func_ceiling: - { - double r = _left->eval_number(c, stack); - - return r == r ? ceil(r) : r; - } - - case ast_func_round: - return round_nearest_nzero(_left->eval_number(c, stack)); - - case ast_variable: - { - assert(_rettype == _data.variable->type()); - - if (_rettype == xpath_type_number) - return _data.variable->get_number(); - - // fallthrough to type conversion - } - - default: - { - switch (_rettype) - { - case xpath_type_boolean: - return eval_boolean(c, stack) ? 1 : 0; - - case xpath_type_string: - { - xpath_allocator_capture cr(stack.result); - - return convert_string_to_number(eval_string(c, stack).c_str()); - } - - case xpath_type_node_set: - { - xpath_allocator_capture cr(stack.result); - - return convert_string_to_number(eval_string(c, stack).c_str()); - } - - default: - assert(!"Wrong expression for return type number"); - return 0; - } - - } - } - } - - xpath_string eval_string_concat(const xpath_context& c, const xpath_stack& stack) - { - assert(_type == ast_func_concat); - - xpath_allocator_capture ct(stack.temp); - - // count the string number - size_t count = 1; - for (xpath_ast_node* nc = _right; nc; nc = nc->_next) count++; - - // gather all strings - xpath_string static_buffer[4]; - xpath_string* buffer = static_buffer; - - // allocate on-heap for large concats - if (count > sizeof(static_buffer) / sizeof(static_buffer[0])) - { - buffer = static_cast(stack.temp->allocate(count * sizeof(xpath_string))); - assert(buffer); - } - - // evaluate all strings to temporary stack - xpath_stack swapped_stack = {stack.temp, stack.result}; - - buffer[0] = _left->eval_string(c, swapped_stack); - - size_t pos = 1; - for (xpath_ast_node* n = _right; n; n = n->_next, ++pos) buffer[pos] = n->eval_string(c, swapped_stack); - assert(pos == count); - - // get total length - size_t length = 0; - for (size_t i = 0; i < count; ++i) length += buffer[i].length(); - - // create final string - char_t* result = static_cast(stack.result->allocate((length + 1) * sizeof(char_t))); - assert(result); - - char_t* ri = result; - - for (size_t j = 0; j < count; ++j) - for (const char_t* bi = buffer[j].c_str(); *bi; ++bi) - *ri++ = *bi; - - *ri = 0; - - return xpath_string(result, true); - } - - xpath_string eval_string(const xpath_context& c, const xpath_stack& stack) - { - switch (_type) - { - case ast_string_constant: - return xpath_string_const(_data.string); - - case ast_func_local_name_0: - { - xpath_node na = c.n; - - return xpath_string_const(local_name(na)); - } - - case ast_func_local_name_1: - { - xpath_allocator_capture cr(stack.result); - - xpath_node_set_raw ns = _left->eval_node_set(c, stack); - xpath_node na = ns.first(); - - return xpath_string_const(local_name(na)); - } - - case ast_func_name_0: - { - xpath_node na = c.n; - - return xpath_string_const(qualified_name(na)); - } - - case ast_func_name_1: - { - xpath_allocator_capture cr(stack.result); - - xpath_node_set_raw ns = _left->eval_node_set(c, stack); - xpath_node na = ns.first(); - - return xpath_string_const(qualified_name(na)); - } - - case ast_func_namespace_uri_0: - { - xpath_node na = c.n; - - return xpath_string_const(namespace_uri(na)); - } - - case ast_func_namespace_uri_1: - { - xpath_allocator_capture cr(stack.result); - - xpath_node_set_raw ns = _left->eval_node_set(c, stack); - xpath_node na = ns.first(); - - return xpath_string_const(namespace_uri(na)); - } - - case ast_func_string_0: - return string_value(c.n, stack.result); - - case ast_func_string_1: - return _left->eval_string(c, stack); - - case ast_func_concat: - return eval_string_concat(c, stack); - - case ast_func_substring_before: - { - xpath_allocator_capture cr(stack.temp); - - xpath_stack swapped_stack = {stack.temp, stack.result}; - - xpath_string s = _left->eval_string(c, swapped_stack); - xpath_string p = _right->eval_string(c, swapped_stack); - - const char_t* pos = find_substring(s.c_str(), p.c_str()); - - return pos ? xpath_string(s.c_str(), pos, stack.result) : xpath_string(); - } - - case ast_func_substring_after: - { - xpath_allocator_capture cr(stack.temp); - - xpath_stack swapped_stack = {stack.temp, stack.result}; - - xpath_string s = _left->eval_string(c, swapped_stack); - xpath_string p = _right->eval_string(c, swapped_stack); - - const char_t* pos = find_substring(s.c_str(), p.c_str()); - if (!pos) return xpath_string(); - - const char_t* result = pos + p.length(); - - return s.uses_heap() ? xpath_string(result, stack.result) : xpath_string_const(result); - } - - case ast_func_substring_2: - { - xpath_allocator_capture cr(stack.temp); - - xpath_stack swapped_stack = {stack.temp, stack.result}; - - xpath_string s = _left->eval_string(c, swapped_stack); - size_t s_length = s.length(); - - double first = round_nearest(_right->eval_number(c, stack)); - - if (is_nan(first)) return xpath_string(); // NaN - else if (first >= s_length + 1) return xpath_string(); - - size_t pos = first < 1 ? 1 : static_cast(first); - assert(1 <= pos && pos <= s_length + 1); - - const char_t* rbegin = s.c_str() + (pos - 1); - - return s.uses_heap() ? xpath_string(rbegin, stack.result) : xpath_string_const(rbegin); - } - - case ast_func_substring_3: - { - xpath_allocator_capture cr(stack.temp); - - xpath_stack swapped_stack = {stack.temp, stack.result}; - - xpath_string s = _left->eval_string(c, swapped_stack); - size_t s_length = s.length(); - - double first = round_nearest(_right->eval_number(c, stack)); - double last = first + round_nearest(_right->_next->eval_number(c, stack)); - - if (is_nan(first) || is_nan(last)) return xpath_string(); - else if (first >= s_length + 1) return xpath_string(); - else if (first >= last) return xpath_string(); - else if (last < 1) return xpath_string(); - - size_t pos = first < 1 ? 1 : static_cast(first); - size_t end = last >= s_length + 1 ? s_length + 1 : static_cast(last); - - assert(1 <= pos && pos <= end && end <= s_length + 1); - const char_t* rbegin = s.c_str() + (pos - 1); - const char_t* rend = s.c_str() + (end - 1); - - return (end == s_length + 1 && !s.uses_heap()) ? xpath_string_const(rbegin) : xpath_string(rbegin, rend, stack.result); - } - - case ast_func_normalize_space_0: - { - xpath_string s = string_value(c.n, stack.result); - - normalize_space(s.data(stack.result)); - - return s; - } - - case ast_func_normalize_space_1: - { - xpath_string s = _left->eval_string(c, stack); - - normalize_space(s.data(stack.result)); - - return s; - } - - case ast_func_translate: - { - xpath_allocator_capture cr(stack.temp); - - xpath_stack swapped_stack = {stack.temp, stack.result}; - - xpath_string s = _left->eval_string(c, stack); - xpath_string from = _right->eval_string(c, swapped_stack); - xpath_string to = _right->_next->eval_string(c, swapped_stack); - - translate(s.data(stack.result), from.c_str(), to.c_str()); - - return s; - } - - case ast_variable: - { - assert(_rettype == _data.variable->type()); - - if (_rettype == xpath_type_string) - return xpath_string_const(_data.variable->get_string()); - - // fallthrough to type conversion - } - - default: - { - switch (_rettype) - { - case xpath_type_boolean: - return xpath_string_const(eval_boolean(c, stack) ? PUGIXML_TEXT("true") : PUGIXML_TEXT("false")); - - case xpath_type_number: - return convert_number_to_string(eval_number(c, stack), stack.result); - - case xpath_type_node_set: - { - xpath_allocator_capture cr(stack.temp); - - xpath_stack swapped_stack = {stack.temp, stack.result}; - - xpath_node_set_raw ns = eval_node_set(c, swapped_stack); - return ns.empty() ? xpath_string() : string_value(ns.first(), stack.result); - } - - default: - assert(!"Wrong expression for return type string"); - return xpath_string(); - } - } - } - } - - xpath_node_set_raw eval_node_set(const xpath_context& c, const xpath_stack& stack) - { - switch (_type) - { - case ast_op_union: - { - xpath_allocator_capture cr(stack.temp); - - xpath_stack swapped_stack = {stack.temp, stack.result}; - - xpath_node_set_raw ls = _left->eval_node_set(c, swapped_stack); - xpath_node_set_raw rs = _right->eval_node_set(c, stack); - - // we can optimize merging two sorted sets, but this is a very rare operation, so don't bother - rs.set_type(xpath_node_set::type_unsorted); - - rs.append(ls.begin(), ls.end(), stack.result); - rs.remove_duplicates(); - - return rs; - } - - case ast_filter: - case ast_filter_posinv: - { - xpath_node_set_raw set = _left->eval_node_set(c, stack); - - // either expression is a number or it contains position() call; sort by document order - if (_type == ast_filter) set.sort_do(); - - apply_predicate(set, 0, _right, stack); - - return set; - } - - case ast_func_id: - return xpath_node_set_raw(); - - case ast_step: - { - switch (_axis) - { - case axis_ancestor: - return step_do(c, stack, axis_to_type()); - - case axis_ancestor_or_self: - return step_do(c, stack, axis_to_type()); - - case axis_attribute: - return step_do(c, stack, axis_to_type()); - - case axis_child: - return step_do(c, stack, axis_to_type()); - - case axis_descendant: - return step_do(c, stack, axis_to_type()); - - case axis_descendant_or_self: - return step_do(c, stack, axis_to_type()); - - case axis_following: - return step_do(c, stack, axis_to_type()); - - case axis_following_sibling: - return step_do(c, stack, axis_to_type()); - - case axis_namespace: - // namespaced axis is not supported - return xpath_node_set_raw(); - - case axis_parent: - return step_do(c, stack, axis_to_type()); - - case axis_preceding: - return step_do(c, stack, axis_to_type()); - - case axis_preceding_sibling: - return step_do(c, stack, axis_to_type()); - - case axis_self: - return step_do(c, stack, axis_to_type()); - - default: - assert(!"Unknown axis"); - return xpath_node_set_raw(); - } - } - - case ast_step_root: - { - assert(!_right); // root step can't have any predicates - - xpath_node_set_raw ns; - - ns.set_type(xpath_node_set::type_sorted); - - if (c.n.node()) ns.push_back(c.n.node().root(), stack.result); - else if (c.n.attribute()) ns.push_back(c.n.parent().root(), stack.result); - - return ns; - } - - case ast_variable: - { - assert(_rettype == _data.variable->type()); - - if (_rettype == xpath_type_node_set) - { - const xpath_node_set& s = _data.variable->get_node_set(); - - xpath_node_set_raw ns; - - ns.set_type(s.type()); - ns.append(s.begin(), s.end(), stack.result); - - return ns; - } - - // fallthrough to type conversion - } - - default: - assert(!"Wrong expression for return type node set"); - return xpath_node_set_raw(); - } - } - - bool is_posinv() - { - switch (_type) - { - case ast_func_position: - return false; - - case ast_string_constant: - case ast_number_constant: - case ast_variable: - return true; - - case ast_step: - case ast_step_root: - return true; - - case ast_predicate: - case ast_filter: - case ast_filter_posinv: - return true; - - default: - if (_left && !_left->is_posinv()) return false; - - for (xpath_ast_node* n = _right; n; n = n->_next) - if (!n->is_posinv()) return false; - - return true; - } - } - - xpath_value_type rettype() const - { - return static_cast(_rettype); - } - }; - - struct xpath_parser - { - xpath_allocator* _alloc; - xpath_lexer _lexer; - - const char_t* _query; - xpath_variable_set* _variables; - - xpath_parse_result* _result; - - #ifdef PUGIXML_NO_EXCEPTIONS - jmp_buf _error_handler; - #endif - - void throw_error(const char* message) - { - _result->error = message; - _result->offset = _lexer.current_pos() - _query; - - #ifdef PUGIXML_NO_EXCEPTIONS - longjmp(_error_handler, 1); - #else - throw xpath_exception(*_result); - #endif - } - - void throw_error_oom() - { - #ifdef PUGIXML_NO_EXCEPTIONS - throw_error("Out of memory"); - #else - throw std::bad_alloc(); - #endif - } - - void* alloc_node() - { - void* result = _alloc->allocate_nothrow(sizeof(xpath_ast_node)); - - if (!result) throw_error_oom(); - - return result; - } - - const char_t* alloc_string(const xpath_lexer_string& value) - { - if (value.begin) - { - size_t length = static_cast(value.end - value.begin); - - char_t* c = static_cast(_alloc->allocate_nothrow((length + 1) * sizeof(char_t))); - if (!c) throw_error_oom(); - - memcpy(c, value.begin, length * sizeof(char_t)); - c[length] = 0; - - return c; - } - else return 0; - } - - xpath_ast_node* parse_function_helper(ast_type_t type0, ast_type_t type1, size_t argc, xpath_ast_node* args[2]) - { - assert(argc <= 1); - - if (argc == 1 && args[0]->rettype() != xpath_type_node_set) throw_error("Function has to be applied to node set"); - - return new (alloc_node()) xpath_ast_node(argc == 0 ? type0 : type1, xpath_type_string, args[0]); - } - - xpath_ast_node* parse_function(const xpath_lexer_string& name, size_t argc, xpath_ast_node* args[2]) - { - switch (name.begin[0]) - { - case 'b': - if (name == PUGIXML_TEXT("boolean") && argc == 1) - return new (alloc_node()) xpath_ast_node(ast_func_boolean, xpath_type_boolean, args[0]); - - break; - - case 'c': - if (name == PUGIXML_TEXT("count") && argc == 1) - { - if (args[0]->rettype() != xpath_type_node_set) throw_error("Function has to be applied to node set"); - return new (alloc_node()) xpath_ast_node(ast_func_count, xpath_type_number, args[0]); - } - else if (name == PUGIXML_TEXT("contains") && argc == 2) - return new (alloc_node()) xpath_ast_node(ast_func_contains, xpath_type_string, args[0], args[1]); - else if (name == PUGIXML_TEXT("concat") && argc >= 2) - return new (alloc_node()) xpath_ast_node(ast_func_concat, xpath_type_string, args[0], args[1]); - else if (name == PUGIXML_TEXT("ceiling") && argc == 1) - return new (alloc_node()) xpath_ast_node(ast_func_ceiling, xpath_type_number, args[0]); - - break; - - case 'f': - if (name == PUGIXML_TEXT("false") && argc == 0) - return new (alloc_node()) xpath_ast_node(ast_func_false, xpath_type_boolean); - else if (name == PUGIXML_TEXT("floor") && argc == 1) - return new (alloc_node()) xpath_ast_node(ast_func_floor, xpath_type_number, args[0]); - - break; - - case 'i': - if (name == PUGIXML_TEXT("id") && argc == 1) - return new (alloc_node()) xpath_ast_node(ast_func_id, xpath_type_node_set, args[0]); - - break; - - case 'l': - if (name == PUGIXML_TEXT("last") && argc == 0) - return new (alloc_node()) xpath_ast_node(ast_func_last, xpath_type_number); - else if (name == PUGIXML_TEXT("lang") && argc == 1) - return new (alloc_node()) xpath_ast_node(ast_func_lang, xpath_type_boolean, args[0]); - else if (name == PUGIXML_TEXT("local-name") && argc <= 1) - return parse_function_helper(ast_func_local_name_0, ast_func_local_name_1, argc, args); - - break; - - case 'n': - if (name == PUGIXML_TEXT("name") && argc <= 1) - return parse_function_helper(ast_func_name_0, ast_func_name_1, argc, args); - else if (name == PUGIXML_TEXT("namespace-uri") && argc <= 1) - return parse_function_helper(ast_func_namespace_uri_0, ast_func_namespace_uri_1, argc, args); - else if (name == PUGIXML_TEXT("normalize-space") && argc <= 1) - return new (alloc_node()) xpath_ast_node(argc == 0 ? ast_func_normalize_space_0 : ast_func_normalize_space_1, xpath_type_string, args[0], args[1]); - else if (name == PUGIXML_TEXT("not") && argc == 1) - return new (alloc_node()) xpath_ast_node(ast_func_not, xpath_type_boolean, args[0]); - else if (name == PUGIXML_TEXT("number") && argc <= 1) - return new (alloc_node()) xpath_ast_node(argc == 0 ? ast_func_number_0 : ast_func_number_1, xpath_type_number, args[0]); - - break; - - case 'p': - if (name == PUGIXML_TEXT("position") && argc == 0) - return new (alloc_node()) xpath_ast_node(ast_func_position, xpath_type_number); - - break; - - case 'r': - if (name == PUGIXML_TEXT("round") && argc == 1) - return new (alloc_node()) xpath_ast_node(ast_func_round, xpath_type_number, args[0]); - - break; - - case 's': - if (name == PUGIXML_TEXT("string") && argc <= 1) - return new (alloc_node()) xpath_ast_node(argc == 0 ? ast_func_string_0 : ast_func_string_1, xpath_type_string, args[0]); - else if (name == PUGIXML_TEXT("string-length") && argc <= 1) - return new (alloc_node()) xpath_ast_node(argc == 0 ? ast_func_string_length_0 : ast_func_string_length_1, xpath_type_string, args[0]); - else if (name == PUGIXML_TEXT("starts-with") && argc == 2) - return new (alloc_node()) xpath_ast_node(ast_func_starts_with, xpath_type_boolean, args[0], args[1]); - else if (name == PUGIXML_TEXT("substring-before") && argc == 2) - return new (alloc_node()) xpath_ast_node(ast_func_substring_before, xpath_type_string, args[0], args[1]); - else if (name == PUGIXML_TEXT("substring-after") && argc == 2) - return new (alloc_node()) xpath_ast_node(ast_func_substring_after, xpath_type_string, args[0], args[1]); - else if (name == PUGIXML_TEXT("substring") && (argc == 2 || argc == 3)) - return new (alloc_node()) xpath_ast_node(argc == 2 ? ast_func_substring_2 : ast_func_substring_3, xpath_type_string, args[0], args[1]); - else if (name == PUGIXML_TEXT("sum") && argc == 1) - { - if (args[0]->rettype() != xpath_type_node_set) throw_error("Function has to be applied to node set"); - return new (alloc_node()) xpath_ast_node(ast_func_sum, xpath_type_number, args[0]); - } - - break; - - case 't': - if (name == PUGIXML_TEXT("translate") && argc == 3) - return new (alloc_node()) xpath_ast_node(ast_func_translate, xpath_type_string, args[0], args[1]); - else if (name == PUGIXML_TEXT("true") && argc == 0) - return new (alloc_node()) xpath_ast_node(ast_func_true, xpath_type_boolean); - - break; - - default: - break; - } - - throw_error("Unrecognized function or wrong parameter count"); - - return 0; - } - - axis_t parse_axis_name(const xpath_lexer_string& name, bool& specified) - { - specified = true; - - switch (name.begin[0]) - { - case 'a': - if (name == PUGIXML_TEXT("ancestor")) - return axis_ancestor; - else if (name == PUGIXML_TEXT("ancestor-or-self")) - return axis_ancestor_or_self; - else if (name == PUGIXML_TEXT("attribute")) - return axis_attribute; - - break; - - case 'c': - if (name == PUGIXML_TEXT("child")) - return axis_child; - - break; - - case 'd': - if (name == PUGIXML_TEXT("descendant")) - return axis_descendant; - else if (name == PUGIXML_TEXT("descendant-or-self")) - return axis_descendant_or_self; - - break; - - case 'f': - if (name == PUGIXML_TEXT("following")) - return axis_following; - else if (name == PUGIXML_TEXT("following-sibling")) - return axis_following_sibling; - - break; - - case 'n': - if (name == PUGIXML_TEXT("namespace")) - return axis_namespace; - - break; - - case 'p': - if (name == PUGIXML_TEXT("parent")) - return axis_parent; - else if (name == PUGIXML_TEXT("preceding")) - return axis_preceding; - else if (name == PUGIXML_TEXT("preceding-sibling")) - return axis_preceding_sibling; - - break; - - case 's': - if (name == PUGIXML_TEXT("self")) - return axis_self; - - break; - - default: - break; - } - - specified = false; - return axis_child; - } - - nodetest_t parse_node_test_type(const xpath_lexer_string& name) - { - switch (name.begin[0]) - { - case 'c': - if (name == PUGIXML_TEXT("comment")) - return nodetest_type_comment; - - break; - - case 'n': - if (name == PUGIXML_TEXT("node")) - return nodetest_type_node; - - break; - - case 'p': - if (name == PUGIXML_TEXT("processing-instruction")) - return nodetest_type_pi; - - break; - - case 't': - if (name == PUGIXML_TEXT("text")) - return nodetest_type_text; - - break; - - default: - break; - } - - return nodetest_none; - } - - // PrimaryExpr ::= VariableReference | '(' Expr ')' | Literal | Number | FunctionCall - xpath_ast_node* parse_primary_expression() - { - switch (_lexer.current()) - { - case lex_var_ref: - { - xpath_lexer_string name = _lexer.contents(); - - if (!_variables) - throw_error("Unknown variable: variable set is not provided"); - - xpath_variable* var = get_variable(_variables, name.begin, name.end); - - if (!var) - throw_error("Unknown variable: variable set does not contain the given name"); - - _lexer.next(); - - return new (alloc_node()) xpath_ast_node(ast_variable, var->type(), var); - } - - case lex_open_brace: - { - _lexer.next(); - - xpath_ast_node* n = parse_expression(); - - if (_lexer.current() != lex_close_brace) - throw_error("Unmatched braces"); - - _lexer.next(); - - return n; - } - - case lex_quoted_string: - { - const char_t* value = alloc_string(_lexer.contents()); - - xpath_ast_node* n = new (alloc_node()) xpath_ast_node(ast_string_constant, xpath_type_string, value); - _lexer.next(); - - return n; - } - - case lex_number: - { - double value = 0; - - if (!convert_string_to_number(_lexer.contents().begin, _lexer.contents().end, &value)) - throw_error_oom(); - - xpath_ast_node* n = new (alloc_node()) xpath_ast_node(ast_number_constant, xpath_type_number, value); - _lexer.next(); - - return n; - } - - case lex_string: - { - xpath_ast_node* args[2] = {0}; - size_t argc = 0; - - xpath_lexer_string function = _lexer.contents(); - _lexer.next(); - - xpath_ast_node* last_arg = 0; - - if (_lexer.current() != lex_open_brace) - throw_error("Unrecognized function call"); - _lexer.next(); - - if (_lexer.current() != lex_close_brace) - args[argc++] = parse_expression(); - - while (_lexer.current() != lex_close_brace) - { - if (_lexer.current() != lex_comma) - throw_error("No comma between function arguments"); - _lexer.next(); - - xpath_ast_node* n = parse_expression(); - - if (argc < 2) args[argc] = n; - else last_arg->set_next(n); - - argc++; - last_arg = n; - } - - _lexer.next(); - - return parse_function(function, argc, args); - } - - default: - throw_error("Unrecognizable primary expression"); - - return 0; - } - } - - // FilterExpr ::= PrimaryExpr | FilterExpr Predicate - // Predicate ::= '[' PredicateExpr ']' - // PredicateExpr ::= Expr - xpath_ast_node* parse_filter_expression() - { - xpath_ast_node* n = parse_primary_expression(); - - while (_lexer.current() == lex_open_square_brace) - { - _lexer.next(); - - xpath_ast_node* expr = parse_expression(); - - if (n->rettype() != xpath_type_node_set) throw_error("Predicate has to be applied to node set"); - - bool posinv = expr->rettype() != xpath_type_number && expr->is_posinv(); - - n = new (alloc_node()) xpath_ast_node(posinv ? ast_filter_posinv : ast_filter, xpath_type_node_set, n, expr); - - if (_lexer.current() != lex_close_square_brace) - throw_error("Unmatched square brace"); - - _lexer.next(); - } - - return n; - } - - // Step ::= AxisSpecifier NodeTest Predicate* | AbbreviatedStep - // AxisSpecifier ::= AxisName '::' | '@'? - // NodeTest ::= NameTest | NodeType '(' ')' | 'processing-instruction' '(' Literal ')' - // NameTest ::= '*' | NCName ':' '*' | QName - // AbbreviatedStep ::= '.' | '..' - xpath_ast_node* parse_step(xpath_ast_node* set) - { - if (set && set->rettype() != xpath_type_node_set) - throw_error("Step has to be applied to node set"); - - bool axis_specified = false; - axis_t axis = axis_child; // implied child axis - - if (_lexer.current() == lex_axis_attribute) - { - axis = axis_attribute; - axis_specified = true; - - _lexer.next(); - } - else if (_lexer.current() == lex_dot) - { - _lexer.next(); - - return new (alloc_node()) xpath_ast_node(ast_step, set, axis_self, nodetest_type_node, 0); - } - else if (_lexer.current() == lex_double_dot) - { - _lexer.next(); - - return new (alloc_node()) xpath_ast_node(ast_step, set, axis_parent, nodetest_type_node, 0); - } - - nodetest_t nt_type = nodetest_none; - xpath_lexer_string nt_name; - - if (_lexer.current() == lex_string) - { - // node name test - nt_name = _lexer.contents(); - _lexer.next(); - - // was it an axis name? - if (_lexer.current() == lex_double_colon) - { - // parse axis name - if (axis_specified) throw_error("Two axis specifiers in one step"); - - axis = parse_axis_name(nt_name, axis_specified); - - if (!axis_specified) throw_error("Unknown axis"); - - // read actual node test - _lexer.next(); - - if (_lexer.current() == lex_multiply) - { - nt_type = nodetest_all; - nt_name = xpath_lexer_string(); - _lexer.next(); - } - else if (_lexer.current() == lex_string) - { - nt_name = _lexer.contents(); - _lexer.next(); - } - else throw_error("Unrecognized node test"); - } - - if (nt_type == nodetest_none) - { - // node type test or processing-instruction - if (_lexer.current() == lex_open_brace) - { - _lexer.next(); - - if (_lexer.current() == lex_close_brace) - { - _lexer.next(); - - nt_type = parse_node_test_type(nt_name); - - if (nt_type == nodetest_none) throw_error("Unrecognized node type"); - - nt_name = xpath_lexer_string(); - } - else if (nt_name == PUGIXML_TEXT("processing-instruction")) - { - if (_lexer.current() != lex_quoted_string) - throw_error("Only literals are allowed as arguments to processing-instruction()"); - - nt_type = nodetest_pi; - nt_name = _lexer.contents(); - _lexer.next(); - - if (_lexer.current() != lex_close_brace) - throw_error("Unmatched brace near processing-instruction()"); - _lexer.next(); - } - else - throw_error("Unmatched brace near node type test"); - - } - // QName or NCName:* - else - { - if (nt_name.end - nt_name.begin > 2 && nt_name.end[-2] == ':' && nt_name.end[-1] == '*') // NCName:* - { - nt_name.end--; // erase * - - nt_type = nodetest_all_in_namespace; - } - else nt_type = nodetest_name; - } - } - } - else if (_lexer.current() == lex_multiply) - { - nt_type = nodetest_all; - _lexer.next(); - } - else throw_error("Unrecognized node test"); - - xpath_ast_node* n = new (alloc_node()) xpath_ast_node(ast_step, set, axis, nt_type, alloc_string(nt_name)); - - xpath_ast_node* last = 0; - - while (_lexer.current() == lex_open_square_brace) - { - _lexer.next(); - - xpath_ast_node* expr = parse_expression(); - - xpath_ast_node* pred = new (alloc_node()) xpath_ast_node(ast_predicate, xpath_type_node_set, expr); - - if (_lexer.current() != lex_close_square_brace) - throw_error("Unmatched square brace"); - _lexer.next(); - - if (last) last->set_next(pred); - else n->set_right(pred); - - last = pred; - } - - return n; - } - - // RelativeLocationPath ::= Step | RelativeLocationPath '/' Step | RelativeLocationPath '//' Step - xpath_ast_node* parse_relative_location_path(xpath_ast_node* set) - { - xpath_ast_node* n = parse_step(set); - - while (_lexer.current() == lex_slash || _lexer.current() == lex_double_slash) - { - lexeme_t l = _lexer.current(); - _lexer.next(); - - if (l == lex_double_slash) - n = new (alloc_node()) xpath_ast_node(ast_step, n, axis_descendant_or_self, nodetest_type_node, 0); - - n = parse_step(n); - } - - return n; - } - - // LocationPath ::= RelativeLocationPath | AbsoluteLocationPath - // AbsoluteLocationPath ::= '/' RelativeLocationPath? | '//' RelativeLocationPath - xpath_ast_node* parse_location_path() - { - if (_lexer.current() == lex_slash) - { - _lexer.next(); - - xpath_ast_node* n = new (alloc_node()) xpath_ast_node(ast_step_root, xpath_type_node_set); - - // relative location path can start from axis_attribute, dot, double_dot, multiply and string lexemes; any other lexeme means standalone root path - lexeme_t l = _lexer.current(); - - if (l == lex_string || l == lex_axis_attribute || l == lex_dot || l == lex_double_dot || l == lex_multiply) - return parse_relative_location_path(n); - else - return n; - } - else if (_lexer.current() == lex_double_slash) - { - _lexer.next(); - - xpath_ast_node* n = new (alloc_node()) xpath_ast_node(ast_step_root, xpath_type_node_set); - n = new (alloc_node()) xpath_ast_node(ast_step, n, axis_descendant_or_self, nodetest_type_node, 0); - - return parse_relative_location_path(n); - } - - // else clause moved outside of if because of bogus warning 'control may reach end of non-void function being inlined' in gcc 4.0.1 - return parse_relative_location_path(0); - } - - // PathExpr ::= LocationPath - // | FilterExpr - // | FilterExpr '/' RelativeLocationPath - // | FilterExpr '//' RelativeLocationPath - xpath_ast_node* parse_path_expression() - { - // Clarification. - // PathExpr begins with either LocationPath or FilterExpr. - // FilterExpr begins with PrimaryExpr - // PrimaryExpr begins with '$' in case of it being a variable reference, - // '(' in case of it being an expression, string literal, number constant or - // function call. - - if (_lexer.current() == lex_var_ref || _lexer.current() == lex_open_brace || - _lexer.current() == lex_quoted_string || _lexer.current() == lex_number || - _lexer.current() == lex_string) - { - if (_lexer.current() == lex_string) - { - // This is either a function call, or not - if not, we shall proceed with location path - const char_t* state = _lexer.state(); - - while (PUGI__IS_CHARTYPE(*state, ct_space)) ++state; - - if (*state != '(') return parse_location_path(); - - // This looks like a function call; however this still can be a node-test. Check it. - if (parse_node_test_type(_lexer.contents()) != nodetest_none) return parse_location_path(); - } - - xpath_ast_node* n = parse_filter_expression(); - - if (_lexer.current() == lex_slash || _lexer.current() == lex_double_slash) - { - lexeme_t l = _lexer.current(); - _lexer.next(); - - if (l == lex_double_slash) - { - if (n->rettype() != xpath_type_node_set) throw_error("Step has to be applied to node set"); - - n = new (alloc_node()) xpath_ast_node(ast_step, n, axis_descendant_or_self, nodetest_type_node, 0); - } - - // select from location path - return parse_relative_location_path(n); - } - - return n; - } - else return parse_location_path(); - } - - // UnionExpr ::= PathExpr | UnionExpr '|' PathExpr - xpath_ast_node* parse_union_expression() - { - xpath_ast_node* n = parse_path_expression(); - - while (_lexer.current() == lex_union) - { - _lexer.next(); - - xpath_ast_node* expr = parse_union_expression(); - - if (n->rettype() != xpath_type_node_set || expr->rettype() != xpath_type_node_set) - throw_error("Union operator has to be applied to node sets"); - - n = new (alloc_node()) xpath_ast_node(ast_op_union, xpath_type_node_set, n, expr); - } - - return n; - } - - // UnaryExpr ::= UnionExpr | '-' UnaryExpr - xpath_ast_node* parse_unary_expression() - { - if (_lexer.current() == lex_minus) - { - _lexer.next(); - - xpath_ast_node* expr = parse_unary_expression(); - - return new (alloc_node()) xpath_ast_node(ast_op_negate, xpath_type_number, expr); - } - else return parse_union_expression(); - } - - // MultiplicativeExpr ::= UnaryExpr - // | MultiplicativeExpr '*' UnaryExpr - // | MultiplicativeExpr 'div' UnaryExpr - // | MultiplicativeExpr 'mod' UnaryExpr - xpath_ast_node* parse_multiplicative_expression() - { - xpath_ast_node* n = parse_unary_expression(); - - while (_lexer.current() == lex_multiply || (_lexer.current() == lex_string && - (_lexer.contents() == PUGIXML_TEXT("mod") || _lexer.contents() == PUGIXML_TEXT("div")))) - { - ast_type_t op = _lexer.current() == lex_multiply ? ast_op_multiply : - _lexer.contents().begin[0] == 'd' ? ast_op_divide : ast_op_mod; - _lexer.next(); - - xpath_ast_node* expr = parse_unary_expression(); - - n = new (alloc_node()) xpath_ast_node(op, xpath_type_number, n, expr); - } - - return n; - } - - // AdditiveExpr ::= MultiplicativeExpr - // | AdditiveExpr '+' MultiplicativeExpr - // | AdditiveExpr '-' MultiplicativeExpr - xpath_ast_node* parse_additive_expression() - { - xpath_ast_node* n = parse_multiplicative_expression(); - - while (_lexer.current() == lex_plus || _lexer.current() == lex_minus) - { - lexeme_t l = _lexer.current(); - - _lexer.next(); - - xpath_ast_node* expr = parse_multiplicative_expression(); - - n = new (alloc_node()) xpath_ast_node(l == lex_plus ? ast_op_add : ast_op_subtract, xpath_type_number, n, expr); - } - - return n; - } - - // RelationalExpr ::= AdditiveExpr - // | RelationalExpr '<' AdditiveExpr - // | RelationalExpr '>' AdditiveExpr - // | RelationalExpr '<=' AdditiveExpr - // | RelationalExpr '>=' AdditiveExpr - xpath_ast_node* parse_relational_expression() - { - xpath_ast_node* n = parse_additive_expression(); - - while (_lexer.current() == lex_less || _lexer.current() == lex_less_or_equal || - _lexer.current() == lex_greater || _lexer.current() == lex_greater_or_equal) - { - lexeme_t l = _lexer.current(); - _lexer.next(); - - xpath_ast_node* expr = parse_additive_expression(); - - n = new (alloc_node()) xpath_ast_node(l == lex_less ? ast_op_less : l == lex_greater ? ast_op_greater : - l == lex_less_or_equal ? ast_op_less_or_equal : ast_op_greater_or_equal, xpath_type_boolean, n, expr); - } - - return n; - } - - // EqualityExpr ::= RelationalExpr - // | EqualityExpr '=' RelationalExpr - // | EqualityExpr '!=' RelationalExpr - xpath_ast_node* parse_equality_expression() - { - xpath_ast_node* n = parse_relational_expression(); - - while (_lexer.current() == lex_equal || _lexer.current() == lex_not_equal) - { - lexeme_t l = _lexer.current(); - - _lexer.next(); - - xpath_ast_node* expr = parse_relational_expression(); - - n = new (alloc_node()) xpath_ast_node(l == lex_equal ? ast_op_equal : ast_op_not_equal, xpath_type_boolean, n, expr); - } - - return n; - } - - // AndExpr ::= EqualityExpr | AndExpr 'and' EqualityExpr - xpath_ast_node* parse_and_expression() - { - xpath_ast_node* n = parse_equality_expression(); - - while (_lexer.current() == lex_string && _lexer.contents() == PUGIXML_TEXT("and")) - { - _lexer.next(); - - xpath_ast_node* expr = parse_equality_expression(); - - n = new (alloc_node()) xpath_ast_node(ast_op_and, xpath_type_boolean, n, expr); - } - - return n; - } - - // OrExpr ::= AndExpr | OrExpr 'or' AndExpr - xpath_ast_node* parse_or_expression() - { - xpath_ast_node* n = parse_and_expression(); - - while (_lexer.current() == lex_string && _lexer.contents() == PUGIXML_TEXT("or")) - { - _lexer.next(); - - xpath_ast_node* expr = parse_and_expression(); - - n = new (alloc_node()) xpath_ast_node(ast_op_or, xpath_type_boolean, n, expr); - } - - return n; - } - - // Expr ::= OrExpr - xpath_ast_node* parse_expression() - { - return parse_or_expression(); - } - - xpath_parser(const char_t* query, xpath_variable_set* variables, xpath_allocator* alloc, xpath_parse_result* result): _alloc(alloc), _lexer(query), _query(query), _variables(variables), _result(result) - { - } - - xpath_ast_node* parse() - { - xpath_ast_node* result = parse_expression(); - - if (_lexer.current() != lex_eof) - { - // there are still unparsed tokens left, error - throw_error("Incorrect query"); - } - - return result; - } - - static xpath_ast_node* parse(const char_t* query, xpath_variable_set* variables, xpath_allocator* alloc, xpath_parse_result* result) - { - xpath_parser parser(query, variables, alloc, result); - - #ifdef PUGIXML_NO_EXCEPTIONS - int error = setjmp(parser._error_handler); - - return (error == 0) ? parser.parse() : 0; - #else - return parser.parse(); - #endif - } - }; - - struct xpath_query_impl - { - static xpath_query_impl* create() - { - void* memory = xml_memory::allocate(sizeof(xpath_query_impl)); - - return new (memory) xpath_query_impl(); - } - - static void destroy(void* ptr) - { - if (!ptr) return; - - // free all allocated pages - static_cast(ptr)->alloc.release(); - - // free allocator memory (with the first page) - xml_memory::deallocate(ptr); - } - - xpath_query_impl(): root(0), alloc(&block) - { - block.next = 0; - } - - xpath_ast_node* root; - xpath_allocator alloc; - xpath_memory_block block; - }; - - PUGI__FN xpath_string evaluate_string_impl(xpath_query_impl* impl, const xpath_node& n, xpath_stack_data& sd) - { - if (!impl) return xpath_string(); - - #ifdef PUGIXML_NO_EXCEPTIONS - if (setjmp(sd.error_handler)) return xpath_string(); - #endif - - xpath_context c(n, 1, 1); - - return impl->root->eval_string(c, sd.stack); - } -PUGI__NS_END - -namespace pugi -{ -#ifndef PUGIXML_NO_EXCEPTIONS - PUGI__FN xpath_exception::xpath_exception(const xpath_parse_result& result_): _result(result_) - { - assert(_result.error); - } - - PUGI__FN const char* xpath_exception::what() const throw() - { - return _result.error; - } - - PUGI__FN const xpath_parse_result& xpath_exception::result() const - { - return _result; - } -#endif - - PUGI__FN xpath_node::xpath_node() - { - } - - PUGI__FN xpath_node::xpath_node(const xml_node& node_): _node(node_) - { - } - - PUGI__FN xpath_node::xpath_node(const xml_attribute& attribute_, const xml_node& parent_): _node(attribute_ ? parent_ : xml_node()), _attribute(attribute_) - { - } - - PUGI__FN xml_node xpath_node::node() const - { - return _attribute ? xml_node() : _node; - } - - PUGI__FN xml_attribute xpath_node::attribute() const - { - return _attribute; - } - - PUGI__FN xml_node xpath_node::parent() const - { - return _attribute ? _node : _node.parent(); - } - - PUGI__FN static void unspecified_bool_xpath_node(xpath_node***) - { - } - - PUGI__FN xpath_node::operator xpath_node::unspecified_bool_type() const - { - return (_node || _attribute) ? unspecified_bool_xpath_node : 0; - } - - PUGI__FN bool xpath_node::operator!() const - { - return !(_node || _attribute); - } - - PUGI__FN bool xpath_node::operator==(const xpath_node& n) const - { - return _node == n._node && _attribute == n._attribute; - } - - PUGI__FN bool xpath_node::operator!=(const xpath_node& n) const - { - return _node != n._node || _attribute != n._attribute; - } - -#ifdef __BORLANDC__ - PUGI__FN bool operator&&(const xpath_node& lhs, bool rhs) - { - return (bool)lhs && rhs; - } - - PUGI__FN bool operator||(const xpath_node& lhs, bool rhs) - { - return (bool)lhs || rhs; - } -#endif - - PUGI__FN void xpath_node_set::_assign(const_iterator begin_, const_iterator end_) - { - assert(begin_ <= end_); - - size_t size_ = static_cast(end_ - begin_); - - if (size_ <= 1) - { - // deallocate old buffer - if (_begin != &_storage) impl::xml_memory::deallocate(_begin); - - // use internal buffer - if (begin_ != end_) _storage = *begin_; - - _begin = &_storage; - _end = &_storage + size_; - } - else - { - // make heap copy - xpath_node* storage = static_cast(impl::xml_memory::allocate(size_ * sizeof(xpath_node))); - - if (!storage) - { - #ifdef PUGIXML_NO_EXCEPTIONS - return; - #else - throw std::bad_alloc(); - #endif - } - - memcpy(storage, begin_, size_ * sizeof(xpath_node)); - - // deallocate old buffer - if (_begin != &_storage) impl::xml_memory::deallocate(_begin); - - // finalize - _begin = storage; - _end = storage + size_; - } - } - - PUGI__FN xpath_node_set::xpath_node_set(): _type(type_unsorted), _begin(&_storage), _end(&_storage) - { - } - - PUGI__FN xpath_node_set::xpath_node_set(const_iterator begin_, const_iterator end_, type_t type_): _type(type_), _begin(&_storage), _end(&_storage) - { - _assign(begin_, end_); - } - - PUGI__FN xpath_node_set::~xpath_node_set() - { - if (_begin != &_storage) impl::xml_memory::deallocate(_begin); - } - - PUGI__FN xpath_node_set::xpath_node_set(const xpath_node_set& ns): _type(ns._type), _begin(&_storage), _end(&_storage) - { - _assign(ns._begin, ns._end); - } - - PUGI__FN xpath_node_set& xpath_node_set::operator=(const xpath_node_set& ns) - { - if (this == &ns) return *this; - - _type = ns._type; - _assign(ns._begin, ns._end); - - return *this; - } - - PUGI__FN xpath_node_set::type_t xpath_node_set::type() const - { - return _type; - } - - PUGI__FN size_t xpath_node_set::size() const - { - return _end - _begin; - } - - PUGI__FN bool xpath_node_set::empty() const - { - return _begin == _end; - } - - PUGI__FN const xpath_node& xpath_node_set::operator[](size_t index) const - { - assert(index < size()); - return _begin[index]; - } - - PUGI__FN xpath_node_set::const_iterator xpath_node_set::begin() const - { - return _begin; - } - - PUGI__FN xpath_node_set::const_iterator xpath_node_set::end() const - { - return _end; - } - - PUGI__FN void xpath_node_set::sort(bool reverse) - { - _type = impl::xpath_sort(_begin, _end, _type, reverse); - } - - PUGI__FN xpath_node xpath_node_set::first() const - { - return impl::xpath_first(_begin, _end, _type); - } - - PUGI__FN xpath_parse_result::xpath_parse_result(): error("Internal error"), offset(0) - { - } - - PUGI__FN xpath_parse_result::operator bool() const - { - return error == 0; - } - - PUGI__FN const char* xpath_parse_result::description() const - { - return error ? error : "No error"; - } - - PUGI__FN xpath_variable::xpath_variable() - { - } - - PUGI__FN const char_t* xpath_variable::name() const - { - switch (_type) - { - case xpath_type_node_set: - return static_cast(this)->name; - - case xpath_type_number: - return static_cast(this)->name; - - case xpath_type_string: - return static_cast(this)->name; - - case xpath_type_boolean: - return static_cast(this)->name; - - default: - assert(!"Invalid variable type"); - return 0; - } - } - - PUGI__FN xpath_value_type xpath_variable::type() const - { - return _type; - } - - PUGI__FN bool xpath_variable::get_boolean() const - { - return (_type == xpath_type_boolean) ? static_cast(this)->value : false; - } - - PUGI__FN double xpath_variable::get_number() const - { - return (_type == xpath_type_number) ? static_cast(this)->value : impl::gen_nan(); - } - - PUGI__FN const char_t* xpath_variable::get_string() const - { - const char_t* value = (_type == xpath_type_string) ? static_cast(this)->value : 0; - return value ? value : PUGIXML_TEXT(""); - } - - PUGI__FN const xpath_node_set& xpath_variable::get_node_set() const - { - return (_type == xpath_type_node_set) ? static_cast(this)->value : impl::dummy_node_set; - } - - PUGI__FN bool xpath_variable::set(bool value) - { - if (_type != xpath_type_boolean) return false; - - static_cast(this)->value = value; - return true; - } - - PUGI__FN bool xpath_variable::set(double value) - { - if (_type != xpath_type_number) return false; - - static_cast(this)->value = value; - return true; - } - - PUGI__FN bool xpath_variable::set(const char_t* value) - { - if (_type != xpath_type_string) return false; - - impl::xpath_variable_string* var = static_cast(this); - - // duplicate string - size_t size = (impl::strlength(value) + 1) * sizeof(char_t); - - char_t* copy = static_cast(impl::xml_memory::allocate(size)); - if (!copy) return false; - - memcpy(copy, value, size); - - // replace old string - if (var->value) impl::xml_memory::deallocate(var->value); - var->value = copy; - - return true; - } - - PUGI__FN bool xpath_variable::set(const xpath_node_set& value) - { - if (_type != xpath_type_node_set) return false; - - static_cast(this)->value = value; - return true; - } - - PUGI__FN xpath_variable_set::xpath_variable_set() - { - for (size_t i = 0; i < sizeof(_data) / sizeof(_data[0]); ++i) _data[i] = 0; - } - - PUGI__FN xpath_variable_set::~xpath_variable_set() - { - for (size_t i = 0; i < sizeof(_data) / sizeof(_data[0]); ++i) - { - xpath_variable* var = _data[i]; - - while (var) - { - xpath_variable* next = var->_next; - - impl::delete_xpath_variable(var->_type, var); - - var = next; - } - } - } - - PUGI__FN xpath_variable* xpath_variable_set::find(const char_t* name) const - { - const size_t hash_size = sizeof(_data) / sizeof(_data[0]); - size_t hash = impl::hash_string(name) % hash_size; - - // look for existing variable - for (xpath_variable* var = _data[hash]; var; var = var->_next) - if (impl::strequal(var->name(), name)) - return var; - - return 0; - } - - PUGI__FN xpath_variable* xpath_variable_set::add(const char_t* name, xpath_value_type type) - { - const size_t hash_size = sizeof(_data) / sizeof(_data[0]); - size_t hash = impl::hash_string(name) % hash_size; - - // look for existing variable - for (xpath_variable* var = _data[hash]; var; var = var->_next) - if (impl::strequal(var->name(), name)) - return var->type() == type ? var : 0; - - // add new variable - xpath_variable* result = impl::new_xpath_variable(type, name); - - if (result) - { - result->_type = type; - result->_next = _data[hash]; - - _data[hash] = result; - } - - return result; - } - - PUGI__FN bool xpath_variable_set::set(const char_t* name, bool value) - { - xpath_variable* var = add(name, xpath_type_boolean); - return var ? var->set(value) : false; - } - - PUGI__FN bool xpath_variable_set::set(const char_t* name, double value) - { - xpath_variable* var = add(name, xpath_type_number); - return var ? var->set(value) : false; - } - - PUGI__FN bool xpath_variable_set::set(const char_t* name, const char_t* value) - { - xpath_variable* var = add(name, xpath_type_string); - return var ? var->set(value) : false; - } - - PUGI__FN bool xpath_variable_set::set(const char_t* name, const xpath_node_set& value) - { - xpath_variable* var = add(name, xpath_type_node_set); - return var ? var->set(value) : false; - } - - PUGI__FN xpath_variable* xpath_variable_set::get(const char_t* name) - { - return find(name); - } - - PUGI__FN const xpath_variable* xpath_variable_set::get(const char_t* name) const - { - return find(name); - } - - PUGI__FN xpath_query::xpath_query(const char_t* query, xpath_variable_set* variables): _impl(0) - { - impl::xpath_query_impl* qimpl = impl::xpath_query_impl::create(); - - if (!qimpl) - { - #ifdef PUGIXML_NO_EXCEPTIONS - _result.error = "Out of memory"; - #else - throw std::bad_alloc(); - #endif - } - else - { - impl::buffer_holder impl_holder(qimpl, impl::xpath_query_impl::destroy); - - qimpl->root = impl::xpath_parser::parse(query, variables, &qimpl->alloc, &_result); - - if (qimpl->root) - { - _impl = static_cast(impl_holder.release()); - _result.error = 0; - } - } - } - - PUGI__FN xpath_query::~xpath_query() - { - impl::xpath_query_impl::destroy(_impl); - } - - PUGI__FN xpath_value_type xpath_query::return_type() const - { - if (!_impl) return xpath_type_none; - - return static_cast(_impl)->root->rettype(); - } - - PUGI__FN bool xpath_query::evaluate_boolean(const xpath_node& n) const - { - if (!_impl) return false; - - impl::xpath_context c(n, 1, 1); - impl::xpath_stack_data sd; - - #ifdef PUGIXML_NO_EXCEPTIONS - if (setjmp(sd.error_handler)) return false; - #endif - - return static_cast(_impl)->root->eval_boolean(c, sd.stack); - } - - PUGI__FN double xpath_query::evaluate_number(const xpath_node& n) const - { - if (!_impl) return impl::gen_nan(); - - impl::xpath_context c(n, 1, 1); - impl::xpath_stack_data sd; - - #ifdef PUGIXML_NO_EXCEPTIONS - if (setjmp(sd.error_handler)) return impl::gen_nan(); - #endif - - return static_cast(_impl)->root->eval_number(c, sd.stack); - } - -#ifndef PUGIXML_NO_STL - PUGI__FN string_t xpath_query::evaluate_string(const xpath_node& n) const - { - impl::xpath_stack_data sd; - - return impl::evaluate_string_impl(static_cast(_impl), n, sd).c_str(); - } -#endif - - PUGI__FN size_t xpath_query::evaluate_string(char_t* buffer, size_t capacity, const xpath_node& n) const - { - impl::xpath_stack_data sd; - - impl::xpath_string r = impl::evaluate_string_impl(static_cast(_impl), n, sd); - - size_t full_size = r.length() + 1; - - if (capacity > 0) - { - size_t size = (full_size < capacity) ? full_size : capacity; - assert(size > 0); - - memcpy(buffer, r.c_str(), (size - 1) * sizeof(char_t)); - buffer[size - 1] = 0; - } - - return full_size; - } - - PUGI__FN xpath_node_set xpath_query::evaluate_node_set(const xpath_node& n) const - { - if (!_impl) return xpath_node_set(); - - impl::xpath_ast_node* root = static_cast(_impl)->root; - - if (root->rettype() != xpath_type_node_set) - { - #ifdef PUGIXML_NO_EXCEPTIONS - return xpath_node_set(); - #else - xpath_parse_result res; - res.error = "Expression does not evaluate to node set"; - - throw xpath_exception(res); - #endif - } - - impl::xpath_context c(n, 1, 1); - impl::xpath_stack_data sd; - - #ifdef PUGIXML_NO_EXCEPTIONS - if (setjmp(sd.error_handler)) return xpath_node_set(); - #endif - - impl::xpath_node_set_raw r = root->eval_node_set(c, sd.stack); - - return xpath_node_set(r.begin(), r.end(), r.type()); - } - - PUGI__FN const xpath_parse_result& xpath_query::result() const - { - return _result; - } - - PUGI__FN static void unspecified_bool_xpath_query(xpath_query***) - { - } - - PUGI__FN xpath_query::operator xpath_query::unspecified_bool_type() const - { - return _impl ? unspecified_bool_xpath_query : 0; - } - - PUGI__FN bool xpath_query::operator!() const - { - return !_impl; - } - - PUGI__FN xpath_node xml_node::select_single_node(const char_t* query, xpath_variable_set* variables) const - { - xpath_query q(query, variables); - return select_single_node(q); - } - - PUGI__FN xpath_node xml_node::select_single_node(const xpath_query& query) const - { - xpath_node_set s = query.evaluate_node_set(*this); - return s.empty() ? xpath_node() : s.first(); - } - - PUGI__FN xpath_node_set xml_node::select_nodes(const char_t* query, xpath_variable_set* variables) const - { - xpath_query q(query, variables); - return select_nodes(q); - } - - PUGI__FN xpath_node_set xml_node::select_nodes(const xpath_query& query) const - { - return query.evaluate_node_set(*this); - } -} - -#endif - -#ifdef __BORLANDC__ -# pragma option pop -#endif - -// Intel C++ does not properly keep warning state for function templates, -// so popping warning state at the end of translation unit leads to warnings in the middle. -#if defined(_MSC_VER) && !defined(__INTEL_COMPILER) -# pragma warning(pop) -#endif - -// Undefine all local macros (makes sure we're not leaking macros in header-only mode) -#undef PUGI__NO_INLINE -#undef PUGI__STATIC_ASSERT -#undef PUGI__DMC_VOLATILE -#undef PUGI__MSVC_CRT_VERSION -#undef PUGI__NS_BEGIN -#undef PUGI__NS_END -#undef PUGI__FN -#undef PUGI__FN_NO_INLINE -#undef PUGI__IS_CHARTYPE_IMPL -#undef PUGI__IS_CHARTYPE -#undef PUGI__IS_CHARTYPEX -#undef PUGI__SKIPWS -#undef PUGI__OPTSET -#undef PUGI__PUSHNODE -#undef PUGI__POPNODE -#undef PUGI__SCANFOR -#undef PUGI__SCANWHILE -#undef PUGI__ENDSEG -#undef PUGI__THROW_ERROR -#undef PUGI__CHECK_ERROR - -#endif - -/** - * Copyright (c) 2006-2012 Arseny Kapoulkine - * - * Permission is hereby granted, free of charge, to any person - * obtaining a copy of this software and associated documentation - * files (the "Software"), to deal in the Software without - * restriction, including without limitation the rights to use, - * copy, modify, merge, publish, distribute, sublicense, and/or sell - * copies of the Software, and to permit persons to whom the - * Software is furnished to do so, subject to the following - * conditions: - * - * The above copyright notice and this permission notice shall be - * included in all copies or substantial portions of the Software. - * - * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, - * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES - * OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND - * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT - * HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, - * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING - * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR - * OTHER DEALINGS IN THE SOFTWARE. - */ diff --git a/phrase-extract/extract-mixed-syntax/pugixml.hpp b/phrase-extract/extract-mixed-syntax/pugixml.hpp deleted file mode 100644 index 77b4dcf47..000000000 --- a/phrase-extract/extract-mixed-syntax/pugixml.hpp +++ /dev/null @@ -1,1265 +0,0 @@ -/** - * pugixml parser - version 1.2 - * -------------------------------------------------------- - * Copyright (C) 2006-2012, by Arseny Kapoulkine (arseny.kapoulkine@gmail.com) - * Report bugs and download new versions at http://pugixml.org/ - * - * This library is distributed under the MIT License. See notice at the end - * of this file. - * - * This work is based on the pugxml parser, which is: - * Copyright (C) 2003, by Kristen Wegner (kristen@tima.net) - */ - -#ifndef PUGIXML_VERSION -// Define version macro; evaluates to major * 100 + minor so that it's safe to use in less-than comparisons -# define PUGIXML_VERSION 120 -#endif - -// Include user configuration file (this can define various configuration macros) -#include "pugiconfig.hpp" - -#ifndef HEADER_PUGIXML_HPP -#define HEADER_PUGIXML_HPP - -// Include stddef.h for size_t and ptrdiff_t -#include - -// Include exception header for XPath -#if !defined(PUGIXML_NO_XPATH) && !defined(PUGIXML_NO_EXCEPTIONS) -# include -#endif - -// Include STL headers -#ifndef PUGIXML_NO_STL -# include -# include -# include -#endif - -// Macro for deprecated features -#ifndef PUGIXML_DEPRECATED -# if defined(__GNUC__) -# define PUGIXML_DEPRECATED __attribute__((deprecated)) -# elif defined(_MSC_VER) && _MSC_VER >= 1300 -# define PUGIXML_DEPRECATED __declspec(deprecated) -# else -# define PUGIXML_DEPRECATED -# endif -#endif - -// If no API is defined, assume default -#ifndef PUGIXML_API -# define PUGIXML_API -#endif - -// If no API for classes is defined, assume default -#ifndef PUGIXML_CLASS -# define PUGIXML_CLASS PUGIXML_API -#endif - -// If no API for functions is defined, assume default -#ifndef PUGIXML_FUNCTION -# define PUGIXML_FUNCTION PUGIXML_API -#endif - -// Character interface macros -#ifdef PUGIXML_WCHAR_MODE -# define PUGIXML_TEXT(t) L ## t -# define PUGIXML_CHAR wchar_t -#else -# define PUGIXML_TEXT(t) t -# define PUGIXML_CHAR char -#endif - -namespace pugi -{ - // Character type used for all internal storage and operations; depends on PUGIXML_WCHAR_MODE - typedef PUGIXML_CHAR char_t; - -#ifndef PUGIXML_NO_STL - // String type used for operations that work with STL string; depends on PUGIXML_WCHAR_MODE - typedef std::basic_string, std::allocator > string_t; -#endif -} - -// The PugiXML namespace -namespace pugi -{ - // Tree node types - enum xml_node_type - { - node_null, // Empty (null) node handle - node_document, // A document tree's absolute root - node_element, // Element tag, i.e. '' - node_pcdata, // Plain character data, i.e. 'text' - node_cdata, // Character data, i.e. '' - node_comment, // Comment tag, i.e. '' - node_pi, // Processing instruction, i.e. '' - node_declaration, // Document declaration, i.e. '' - node_doctype // Document type declaration, i.e. '' - }; - - // Parsing options - - // Minimal parsing mode (equivalent to turning all other flags off). - // Only elements and PCDATA sections are added to the DOM tree, no text conversions are performed. - const unsigned int parse_minimal = 0x0000; - - // This flag determines if processing instructions (node_pi) are added to the DOM tree. This flag is off by default. - const unsigned int parse_pi = 0x0001; - - // This flag determines if comments (node_comment) are added to the DOM tree. This flag is off by default. - const unsigned int parse_comments = 0x0002; - - // This flag determines if CDATA sections (node_cdata) are added to the DOM tree. This flag is on by default. - const unsigned int parse_cdata = 0x0004; - - // This flag determines if plain character data (node_pcdata) that consist only of whitespace are added to the DOM tree. - // This flag is off by default; turning it on usually results in slower parsing and more memory consumption. - const unsigned int parse_ws_pcdata = 0x0008; - - // This flag determines if character and entity references are expanded during parsing. This flag is on by default. - const unsigned int parse_escapes = 0x0010; - - // This flag determines if EOL characters are normalized (converted to #xA) during parsing. This flag is on by default. - const unsigned int parse_eol = 0x0020; - - // This flag determines if attribute values are normalized using CDATA normalization rules during parsing. This flag is on by default. - const unsigned int parse_wconv_attribute = 0x0040; - - // This flag determines if attribute values are normalized using NMTOKENS normalization rules during parsing. This flag is off by default. - const unsigned int parse_wnorm_attribute = 0x0080; - - // This flag determines if document declaration (node_declaration) is added to the DOM tree. This flag is off by default. - const unsigned int parse_declaration = 0x0100; - - // This flag determines if document type declaration (node_doctype) is added to the DOM tree. This flag is off by default. - const unsigned int parse_doctype = 0x0200; - - // This flag determines if plain character data (node_pcdata) that is the only child of the parent node and that consists only - // of whitespace is added to the DOM tree. - // This flag is off by default; turning it on may result in slower parsing and more memory consumption. - const unsigned int parse_ws_pcdata_single = 0x0400; - - // The default parsing mode. - // Elements, PCDATA and CDATA sections are added to the DOM tree, character/reference entities are expanded, - // End-of-Line characters are normalized, attribute values are normalized using CDATA normalization rules. - const unsigned int parse_default = parse_cdata | parse_escapes | parse_wconv_attribute | parse_eol; - - // The full parsing mode. - // Nodes of all types are added to the DOM tree, character/reference entities are expanded, - // End-of-Line characters are normalized, attribute values are normalized using CDATA normalization rules. - const unsigned int parse_full = parse_default | parse_pi | parse_comments | parse_declaration | parse_doctype; - - // These flags determine the encoding of input data for XML document - enum xml_encoding - { - encoding_auto, // Auto-detect input encoding using BOM or < / class xml_object_range - { - public: - typedef It const_iterator; - - xml_object_range(It b, It e): _begin(b), _end(e) - { - } - - It begin() const { return _begin; } - It end() const { return _end; } - - private: - It _begin, _end; - }; - - // Writer interface for node printing (see xml_node::print) - class PUGIXML_CLASS xml_writer - { - public: - virtual ~xml_writer() {} - - // Write memory chunk into stream/file/whatever - virtual void write(const void* data, size_t size) = 0; - }; - - // xml_writer implementation for FILE* - class PUGIXML_CLASS xml_writer_file: public xml_writer - { - public: - // Construct writer from a FILE* object; void* is used to avoid header dependencies on stdio - xml_writer_file(void* file); - - virtual void write(const void* data, size_t size); - - private: - void* file; - }; - - #ifndef PUGIXML_NO_STL - // xml_writer implementation for streams - class PUGIXML_CLASS xml_writer_stream: public xml_writer - { - public: - // Construct writer from an output stream object - xml_writer_stream(std::basic_ostream >& stream); - xml_writer_stream(std::basic_ostream >& stream); - - virtual void write(const void* data, size_t size); - - private: - std::basic_ostream >* narrow_stream; - std::basic_ostream >* wide_stream; - }; - #endif - - // A light-weight handle for manipulating attributes in DOM tree - class PUGIXML_CLASS xml_attribute - { - friend class xml_attribute_iterator; - friend class xml_node; - - private: - xml_attribute_struct* _attr; - - typedef void (*unspecified_bool_type)(xml_attribute***); - - public: - // Default constructor. Constructs an empty attribute. - xml_attribute(); - - // Constructs attribute from internal pointer - explicit xml_attribute(xml_attribute_struct* attr); - - // Safe bool conversion operator - operator unspecified_bool_type() const; - - // Borland C++ workaround - bool operator!() const; - - // Comparison operators (compares wrapped attribute pointers) - bool operator==(const xml_attribute& r) const; - bool operator!=(const xml_attribute& r) const; - bool operator<(const xml_attribute& r) const; - bool operator>(const xml_attribute& r) const; - bool operator<=(const xml_attribute& r) const; - bool operator>=(const xml_attribute& r) const; - - // Check if attribute is empty - bool empty() const; - - // Get attribute name/value, or "" if attribute is empty - const char_t* name() const; - const char_t* value() const; - - // Get attribute value, or the default value if attribute is empty - const char_t* as_string(const char_t* def = PUGIXML_TEXT("")) const; - - // Get attribute value as a number, or the default value if conversion did not succeed or attribute is empty - int as_int(int def = 0) const; - unsigned int as_uint(unsigned int def = 0) const; - double as_double(double def = 0) const; - float as_float(float def = 0) const; - - // Get attribute value as bool (returns true if first character is in '1tTyY' set), or the default value if attribute is empty - bool as_bool(bool def = false) const; - - // Set attribute name/value (returns false if attribute is empty or there is not enough memory) - bool set_name(const char_t* rhs); - bool set_value(const char_t* rhs); - - // Set attribute value with type conversion (numbers are converted to strings, boolean is converted to "true"/"false") - bool set_value(int rhs); - bool set_value(unsigned int rhs); - bool set_value(double rhs); - bool set_value(bool rhs); - - // Set attribute value (equivalent to set_value without error checking) - xml_attribute& operator=(const char_t* rhs); - xml_attribute& operator=(int rhs); - xml_attribute& operator=(unsigned int rhs); - xml_attribute& operator=(double rhs); - xml_attribute& operator=(bool rhs); - - // Get next/previous attribute in the attribute list of the parent node - xml_attribute next_attribute() const; - xml_attribute previous_attribute() const; - - // Get hash value (unique for handles to the same object) - size_t hash_value() const; - - // Get internal pointer - xml_attribute_struct* internal_object() const; - }; - -#ifdef __BORLANDC__ - // Borland C++ workaround - bool PUGIXML_FUNCTION operator&&(const xml_attribute& lhs, bool rhs); - bool PUGIXML_FUNCTION operator||(const xml_attribute& lhs, bool rhs); -#endif - - // A light-weight handle for manipulating nodes in DOM tree - class PUGIXML_CLASS xml_node - { - friend class xml_attribute_iterator; - friend class xml_node_iterator; - friend class xml_named_node_iterator; - - protected: - xml_node_struct* _root; - - typedef void (*unspecified_bool_type)(xml_node***); - - public: - // Default constructor. Constructs an empty node. - xml_node(); - - // Constructs node from internal pointer - explicit xml_node(xml_node_struct* p); - - // Safe bool conversion operator - operator unspecified_bool_type() const; - - // Borland C++ workaround - bool operator!() const; - - // Comparison operators (compares wrapped node pointers) - bool operator==(const xml_node& r) const; - bool operator!=(const xml_node& r) const; - bool operator<(const xml_node& r) const; - bool operator>(const xml_node& r) const; - bool operator<=(const xml_node& r) const; - bool operator>=(const xml_node& r) const; - - // Check if node is empty. - bool empty() const; - - // Get node type - xml_node_type type() const; - - // Get node name/value, or "" if node is empty or it has no name/value - const char_t* name() const; - const char_t* value() const; - - // Get attribute list - xml_attribute first_attribute() const; - xml_attribute last_attribute() const; - - // Get children list - xml_node first_child() const; - xml_node last_child() const; - - // Get next/previous sibling in the children list of the parent node - xml_node next_sibling() const; - xml_node previous_sibling() const; - - // Get parent node - xml_node parent() const; - - // Get root of DOM tree this node belongs to - xml_node root() const; - - // Get text object for the current node - xml_text text() const; - - // Get child, attribute or next/previous sibling with the specified name - xml_node child(const char_t* name) const; - xml_attribute attribute(const char_t* name) const; - xml_node next_sibling(const char_t* name) const; - xml_node previous_sibling(const char_t* name) const; - - // Get child value of current node; that is, value of the first child node of type PCDATA/CDATA - const char_t* child_value() const; - - // Get child value of child with specified name. Equivalent to child(name).child_value(). - const char_t* child_value(const char_t* name) const; - - // Set node name/value (returns false if node is empty, there is not enough memory, or node can not have name/value) - bool set_name(const char_t* rhs); - bool set_value(const char_t* rhs); - - // Add attribute with specified name. Returns added attribute, or empty attribute on errors. - xml_attribute append_attribute(const char_t* name); - xml_attribute prepend_attribute(const char_t* name); - xml_attribute insert_attribute_after(const char_t* name, const xml_attribute& attr); - xml_attribute insert_attribute_before(const char_t* name, const xml_attribute& attr); - - // Add a copy of the specified attribute. Returns added attribute, or empty attribute on errors. - xml_attribute append_copy(const xml_attribute& proto); - xml_attribute prepend_copy(const xml_attribute& proto); - xml_attribute insert_copy_after(const xml_attribute& proto, const xml_attribute& attr); - xml_attribute insert_copy_before(const xml_attribute& proto, const xml_attribute& attr); - - // Add child node with specified type. Returns added node, or empty node on errors. - xml_node append_child(xml_node_type type = node_element); - xml_node prepend_child(xml_node_type type = node_element); - xml_node insert_child_after(xml_node_type type, const xml_node& node); - xml_node insert_child_before(xml_node_type type, const xml_node& node); - - // Add child element with specified name. Returns added node, or empty node on errors. - xml_node append_child(const char_t* name); - xml_node prepend_child(const char_t* name); - xml_node insert_child_after(const char_t* name, const xml_node& node); - xml_node insert_child_before(const char_t* name, const xml_node& node); - - // Add a copy of the specified node as a child. Returns added node, or empty node on errors. - xml_node append_copy(const xml_node& proto); - xml_node prepend_copy(const xml_node& proto); - xml_node insert_copy_after(const xml_node& proto, const xml_node& node); - xml_node insert_copy_before(const xml_node& proto, const xml_node& node); - - // Remove specified attribute - bool remove_attribute(const xml_attribute& a); - bool remove_attribute(const char_t* name); - - // Remove specified child - bool remove_child(const xml_node& n); - bool remove_child(const char_t* name); - - // Find attribute using predicate. Returns first attribute for which predicate returned true. - template xml_attribute find_attribute(Predicate pred) const - { - if (!_root) return xml_attribute(); - - for (xml_attribute attrib = first_attribute(); attrib; attrib = attrib.next_attribute()) - if (pred(attrib)) - return attrib; - - return xml_attribute(); - } - - // Find child node using predicate. Returns first child for which predicate returned true. - template xml_node find_child(Predicate pred) const - { - if (!_root) return xml_node(); - - for (xml_node node = first_child(); node; node = node.next_sibling()) - if (pred(node)) - return node; - - return xml_node(); - } - - // Find node from subtree using predicate. Returns first node from subtree (depth-first), for which predicate returned true. - template xml_node find_node(Predicate pred) const - { - if (!_root) return xml_node(); - - xml_node cur = first_child(); - - while (cur._root && cur._root != _root) - { - if (pred(cur)) return cur; - - if (cur.first_child()) cur = cur.first_child(); - else if (cur.next_sibling()) cur = cur.next_sibling(); - else - { - while (!cur.next_sibling() && cur._root != _root) cur = cur.parent(); - - if (cur._root != _root) cur = cur.next_sibling(); - } - } - - return xml_node(); - } - - // Find child node by attribute name/value - xml_node find_child_by_attribute(const char_t* name, const char_t* attr_name, const char_t* attr_value) const; - xml_node find_child_by_attribute(const char_t* attr_name, const char_t* attr_value) const; - - #ifndef PUGIXML_NO_STL - // Get the absolute node path from root as a text string. - string_t path(char_t delimiter = '/') const; - #endif - - // Search for a node by path consisting of node names and . or .. elements. - xml_node first_element_by_path(const char_t* path, char_t delimiter = '/') const; - - // Recursively traverse subtree with xml_tree_walker - bool traverse(xml_tree_walker& walker); - - #ifndef PUGIXML_NO_XPATH - // Select single node by evaluating XPath query. Returns first node from the resulting node set. - xpath_node select_single_node(const char_t* query, xpath_variable_set* variables = 0) const; - xpath_node select_single_node(const xpath_query& query) const; - - // Select node set by evaluating XPath query - xpath_node_set select_nodes(const char_t* query, xpath_variable_set* variables = 0) const; - xpath_node_set select_nodes(const xpath_query& query) const; - #endif - - // Print subtree using a writer object - void print(xml_writer& writer, const char_t* indent = PUGIXML_TEXT("\t"), unsigned int flags = format_default, xml_encoding encoding = encoding_auto, unsigned int depth = 0) const; - - #ifndef PUGIXML_NO_STL - // Print subtree to stream - void print(std::basic_ostream >& os, const char_t* indent = PUGIXML_TEXT("\t"), unsigned int flags = format_default, xml_encoding encoding = encoding_auto, unsigned int depth = 0) const; - void print(std::basic_ostream >& os, const char_t* indent = PUGIXML_TEXT("\t"), unsigned int flags = format_default, unsigned int depth = 0) const; - #endif - - // Child nodes iterators - typedef xml_node_iterator iterator; - - iterator begin() const; - iterator end() const; - - // Attribute iterators - typedef xml_attribute_iterator attribute_iterator; - - attribute_iterator attributes_begin() const; - attribute_iterator attributes_end() const; - - // Range-based for support - xml_object_range children() const; - xml_object_range children(const char_t* name) const; - xml_object_range attributes() const; - - // Get node offset in parsed file/string (in char_t units) for debugging purposes - ptrdiff_t offset_debug() const; - - // Get hash value (unique for handles to the same object) - size_t hash_value() const; - - // Get internal pointer - xml_node_struct* internal_object() const; - }; - -#ifdef __BORLANDC__ - // Borland C++ workaround - bool PUGIXML_FUNCTION operator&&(const xml_node& lhs, bool rhs); - bool PUGIXML_FUNCTION operator||(const xml_node& lhs, bool rhs); -#endif - - // A helper for working with text inside PCDATA nodes - class PUGIXML_CLASS xml_text - { - friend class xml_node; - - xml_node_struct* _root; - - typedef void (*unspecified_bool_type)(xml_text***); - - explicit xml_text(xml_node_struct* root); - - xml_node_struct* _data_new(); - xml_node_struct* _data() const; - - public: - // Default constructor. Constructs an empty object. - xml_text(); - - // Safe bool conversion operator - operator unspecified_bool_type() const; - - // Borland C++ workaround - bool operator!() const; - - // Check if text object is empty - bool empty() const; - - // Get text, or "" if object is empty - const char_t* get() const; - - // Get text, or the default value if object is empty - const char_t* as_string(const char_t* def = PUGIXML_TEXT("")) const; - - // Get text as a number, or the default value if conversion did not succeed or object is empty - int as_int(int def = 0) const; - unsigned int as_uint(unsigned int def = 0) const; - double as_double(double def = 0) const; - float as_float(float def = 0) const; - - // Get text as bool (returns true if first character is in '1tTyY' set), or the default value if object is empty - bool as_bool(bool def = false) const; - - // Set text (returns false if object is empty or there is not enough memory) - bool set(const char_t* rhs); - - // Set text with type conversion (numbers are converted to strings, boolean is converted to "true"/"false") - bool set(int rhs); - bool set(unsigned int rhs); - bool set(double rhs); - bool set(bool rhs); - - // Set text (equivalent to set without error checking) - xml_text& operator=(const char_t* rhs); - xml_text& operator=(int rhs); - xml_text& operator=(unsigned int rhs); - xml_text& operator=(double rhs); - xml_text& operator=(bool rhs); - - // Get the data node (node_pcdata or node_cdata) for this object - xml_node data() const; - }; - -#ifdef __BORLANDC__ - // Borland C++ workaround - bool PUGIXML_FUNCTION operator&&(const xml_text& lhs, bool rhs); - bool PUGIXML_FUNCTION operator||(const xml_text& lhs, bool rhs); -#endif - - // Child node iterator (a bidirectional iterator over a collection of xml_node) - class PUGIXML_CLASS xml_node_iterator - { - friend class xml_node; - - private: - mutable xml_node _wrap; - xml_node _parent; - - xml_node_iterator(xml_node_struct* ref, xml_node_struct* parent); - - public: - // Iterator traits - typedef ptrdiff_t difference_type; - typedef xml_node value_type; - typedef xml_node* pointer; - typedef xml_node& reference; - - #ifndef PUGIXML_NO_STL - typedef std::bidirectional_iterator_tag iterator_category; - #endif - - // Default constructor - xml_node_iterator(); - - // Construct an iterator which points to the specified node - xml_node_iterator(const xml_node& node); - - // Iterator operators - bool operator==(const xml_node_iterator& rhs) const; - bool operator!=(const xml_node_iterator& rhs) const; - - xml_node& operator*() const; - xml_node* operator->() const; - - const xml_node_iterator& operator++(); - xml_node_iterator operator++(int); - - const xml_node_iterator& operator--(); - xml_node_iterator operator--(int); - }; - - // Attribute iterator (a bidirectional iterator over a collection of xml_attribute) - class PUGIXML_CLASS xml_attribute_iterator - { - friend class xml_node; - - private: - mutable xml_attribute _wrap; - xml_node _parent; - - xml_attribute_iterator(xml_attribute_struct* ref, xml_node_struct* parent); - - public: - // Iterator traits - typedef ptrdiff_t difference_type; - typedef xml_attribute value_type; - typedef xml_attribute* pointer; - typedef xml_attribute& reference; - - #ifndef PUGIXML_NO_STL - typedef std::bidirectional_iterator_tag iterator_category; - #endif - - // Default constructor - xml_attribute_iterator(); - - // Construct an iterator which points to the specified attribute - xml_attribute_iterator(const xml_attribute& attr, const xml_node& parent); - - // Iterator operators - bool operator==(const xml_attribute_iterator& rhs) const; - bool operator!=(const xml_attribute_iterator& rhs) const; - - xml_attribute& operator*() const; - xml_attribute* operator->() const; - - const xml_attribute_iterator& operator++(); - xml_attribute_iterator operator++(int); - - const xml_attribute_iterator& operator--(); - xml_attribute_iterator operator--(int); - }; - - // Named node range helper - class xml_named_node_iterator - { - public: - // Iterator traits - typedef ptrdiff_t difference_type; - typedef xml_node value_type; - typedef xml_node* pointer; - typedef xml_node& reference; - - #ifndef PUGIXML_NO_STL - typedef std::forward_iterator_tag iterator_category; - #endif - - // Default constructor - xml_named_node_iterator(); - - // Construct an iterator which points to the specified node - xml_named_node_iterator(const xml_node& node, const char_t* name); - - // Iterator operators - bool operator==(const xml_named_node_iterator& rhs) const; - bool operator!=(const xml_named_node_iterator& rhs) const; - - xml_node& operator*() const; - xml_node* operator->() const; - - const xml_named_node_iterator& operator++(); - xml_named_node_iterator operator++(int); - - private: - mutable xml_node _node; - const char_t* _name; - }; - - // Abstract tree walker class (see xml_node::traverse) - class PUGIXML_CLASS xml_tree_walker - { - friend class xml_node; - - private: - int _depth; - - protected: - // Get current traversal depth - int depth() const; - - public: - xml_tree_walker(); - virtual ~xml_tree_walker(); - - // Callback that is called when traversal begins - virtual bool begin(xml_node& node); - - // Callback that is called for each node traversed - virtual bool for_each(xml_node& node) = 0; - - // Callback that is called when traversal ends - virtual bool end(xml_node& node); - }; - - // Parsing status, returned as part of xml_parse_result object - enum xml_parse_status - { - status_ok = 0, // No error - - status_file_not_found, // File was not found during load_file() - status_io_error, // Error reading from file/stream - status_out_of_memory, // Could not allocate memory - status_internal_error, // Internal error occurred - - status_unrecognized_tag, // Parser could not determine tag type - - status_bad_pi, // Parsing error occurred while parsing document declaration/processing instruction - status_bad_comment, // Parsing error occurred while parsing comment - status_bad_cdata, // Parsing error occurred while parsing CDATA section - status_bad_doctype, // Parsing error occurred while parsing document type declaration - status_bad_pcdata, // Parsing error occurred while parsing PCDATA section - status_bad_start_element, // Parsing error occurred while parsing start element tag - status_bad_attribute, // Parsing error occurred while parsing element attribute - status_bad_end_element, // Parsing error occurred while parsing end element tag - status_end_element_mismatch // There was a mismatch of start-end tags (closing tag had incorrect name, some tag was not closed or there was an excessive closing tag) - }; - - // Parsing result - struct PUGIXML_CLASS xml_parse_result - { - // Parsing status (see xml_parse_status) - xml_parse_status status; - - // Last parsed offset (in char_t units from start of input data) - ptrdiff_t offset; - - // Source document encoding - xml_encoding encoding; - - // Default constructor, initializes object to failed state - xml_parse_result(); - - // Cast to bool operator - operator bool() const; - - // Get error description - const char* description() const; - }; - - // Document class (DOM tree root) - class PUGIXML_CLASS xml_document: public xml_node - { - private: - char_t* _buffer; - - char _memory[192]; - - // Non-copyable semantics - xml_document(const xml_document&); - const xml_document& operator=(const xml_document&); - - void create(); - void destroy(); - - xml_parse_result load_buffer_impl(void* contents, size_t size, unsigned int options, xml_encoding encoding, bool is_mutable, bool own); - - public: - // Default constructor, makes empty document - xml_document(); - - // Destructor, invalidates all node/attribute handles to this document - ~xml_document(); - - // Removes all nodes, leaving the empty document - void reset(); - - // Removes all nodes, then copies the entire contents of the specified document - void reset(const xml_document& proto); - - #ifndef PUGIXML_NO_STL - // Load document from stream. - xml_parse_result load(std::basic_istream >& stream, unsigned int options = parse_default, xml_encoding encoding = encoding_auto); - xml_parse_result load(std::basic_istream >& stream, unsigned int options = parse_default); - #endif - - // Load document from zero-terminated string. No encoding conversions are applied. - xml_parse_result load(const char_t* contents, unsigned int options = parse_default); - - // Load document from file - xml_parse_result load_file(const char* path, unsigned int options = parse_default, xml_encoding encoding = encoding_auto); - xml_parse_result load_file(const wchar_t* path, unsigned int options = parse_default, xml_encoding encoding = encoding_auto); - - // Load document from buffer. Copies/converts the buffer, so it may be deleted or changed after the function returns. - xml_parse_result load_buffer(const void* contents, size_t size, unsigned int options = parse_default, xml_encoding encoding = encoding_auto); - - // Load document from buffer, using the buffer for in-place parsing (the buffer is modified and used for storage of document data). - // You should ensure that buffer data will persist throughout the document's lifetime, and free the buffer memory manually once document is destroyed. - xml_parse_result load_buffer_inplace(void* contents, size_t size, unsigned int options = parse_default, xml_encoding encoding = encoding_auto); - - // Load document from buffer, using the buffer for in-place parsing (the buffer is modified and used for storage of document data). - // You should allocate the buffer with pugixml allocation function; document will free the buffer when it is no longer needed (you can't use it anymore). - xml_parse_result load_buffer_inplace_own(void* contents, size_t size, unsigned int options = parse_default, xml_encoding encoding = encoding_auto); - - // Save XML document to writer (semantics is slightly different from xml_node::print, see documentation for details). - void save(xml_writer& writer, const char_t* indent = PUGIXML_TEXT("\t"), unsigned int flags = format_default, xml_encoding encoding = encoding_auto) const; - - #ifndef PUGIXML_NO_STL - // Save XML document to stream (semantics is slightly different from xml_node::print, see documentation for details). - void save(std::basic_ostream >& stream, const char_t* indent = PUGIXML_TEXT("\t"), unsigned int flags = format_default, xml_encoding encoding = encoding_auto) const; - void save(std::basic_ostream >& stream, const char_t* indent = PUGIXML_TEXT("\t"), unsigned int flags = format_default) const; - #endif - - // Save XML to file - bool save_file(const char* path, const char_t* indent = PUGIXML_TEXT("\t"), unsigned int flags = format_default, xml_encoding encoding = encoding_auto) const; - bool save_file(const wchar_t* path, const char_t* indent = PUGIXML_TEXT("\t"), unsigned int flags = format_default, xml_encoding encoding = encoding_auto) const; - - // Get document element - xml_node document_element() const; - }; - -#ifndef PUGIXML_NO_XPATH - // XPath query return type - enum xpath_value_type - { - xpath_type_none, // Unknown type (query failed to compile) - xpath_type_node_set, // Node set (xpath_node_set) - xpath_type_number, // Number - xpath_type_string, // String - xpath_type_boolean // Boolean - }; - - // XPath parsing result - struct PUGIXML_CLASS xpath_parse_result - { - // Error message (0 if no error) - const char* error; - - // Last parsed offset (in char_t units from string start) - ptrdiff_t offset; - - // Default constructor, initializes object to failed state - xpath_parse_result(); - - // Cast to bool operator - operator bool() const; - - // Get error description - const char* description() const; - }; - - // A single XPath variable - class PUGIXML_CLASS xpath_variable - { - friend class xpath_variable_set; - - protected: - xpath_value_type _type; - xpath_variable* _next; - - xpath_variable(); - - // Non-copyable semantics - xpath_variable(const xpath_variable&); - xpath_variable& operator=(const xpath_variable&); - - public: - // Get variable name - const char_t* name() const; - - // Get variable type - xpath_value_type type() const; - - // Get variable value; no type conversion is performed, default value (false, NaN, empty string, empty node set) is returned on type mismatch error - bool get_boolean() const; - double get_number() const; - const char_t* get_string() const; - const xpath_node_set& get_node_set() const; - - // Set variable value; no type conversion is performed, false is returned on type mismatch error - bool set(bool value); - bool set(double value); - bool set(const char_t* value); - bool set(const xpath_node_set& value); - }; - - // A set of XPath variables - class PUGIXML_CLASS xpath_variable_set - { - private: - xpath_variable* _data[64]; - - // Non-copyable semantics - xpath_variable_set(const xpath_variable_set&); - xpath_variable_set& operator=(const xpath_variable_set&); - - xpath_variable* find(const char_t* name) const; - - public: - // Default constructor/destructor - xpath_variable_set(); - ~xpath_variable_set(); - - // Add a new variable or get the existing one, if the types match - xpath_variable* add(const char_t* name, xpath_value_type type); - - // Set value of an existing variable; no type conversion is performed, false is returned if there is no such variable or if types mismatch - bool set(const char_t* name, bool value); - bool set(const char_t* name, double value); - bool set(const char_t* name, const char_t* value); - bool set(const char_t* name, const xpath_node_set& value); - - // Get existing variable by name - xpath_variable* get(const char_t* name); - const xpath_variable* get(const char_t* name) const; - }; - - // A compiled XPath query object - class PUGIXML_CLASS xpath_query - { - private: - void* _impl; - xpath_parse_result _result; - - typedef void (*unspecified_bool_type)(xpath_query***); - - // Non-copyable semantics - xpath_query(const xpath_query&); - xpath_query& operator=(const xpath_query&); - - public: - // Construct a compiled object from XPath expression. - // If PUGIXML_NO_EXCEPTIONS is not defined, throws xpath_exception on compilation errors. - explicit xpath_query(const char_t* query, xpath_variable_set* variables = 0); - - // Destructor - ~xpath_query(); - - // Get query expression return type - xpath_value_type return_type() const; - - // Evaluate expression as boolean value in the specified context; performs type conversion if necessary. - // If PUGIXML_NO_EXCEPTIONS is not defined, throws std::bad_alloc on out of memory errors. - bool evaluate_boolean(const xpath_node& n) const; - - // Evaluate expression as double value in the specified context; performs type conversion if necessary. - // If PUGIXML_NO_EXCEPTIONS is not defined, throws std::bad_alloc on out of memory errors. - double evaluate_number(const xpath_node& n) const; - - #ifndef PUGIXML_NO_STL - // Evaluate expression as string value in the specified context; performs type conversion if necessary. - // If PUGIXML_NO_EXCEPTIONS is not defined, throws std::bad_alloc on out of memory errors. - string_t evaluate_string(const xpath_node& n) const; - #endif - - // Evaluate expression as string value in the specified context; performs type conversion if necessary. - // At most capacity characters are written to the destination buffer, full result size is returned (includes terminating zero). - // If PUGIXML_NO_EXCEPTIONS is not defined, throws std::bad_alloc on out of memory errors. - // If PUGIXML_NO_EXCEPTIONS is defined, returns empty set instead. - size_t evaluate_string(char_t* buffer, size_t capacity, const xpath_node& n) const; - - // Evaluate expression as node set in the specified context. - // If PUGIXML_NO_EXCEPTIONS is not defined, throws xpath_exception on type mismatch and std::bad_alloc on out of memory errors. - // If PUGIXML_NO_EXCEPTIONS is defined, returns empty node set instead. - xpath_node_set evaluate_node_set(const xpath_node& n) const; - - // Get parsing result (used to get compilation errors in PUGIXML_NO_EXCEPTIONS mode) - const xpath_parse_result& result() const; - - // Safe bool conversion operator - operator unspecified_bool_type() const; - - // Borland C++ workaround - bool operator!() const; - }; - - #ifndef PUGIXML_NO_EXCEPTIONS - // XPath exception class - class PUGIXML_CLASS xpath_exception: public std::exception - { - private: - xpath_parse_result _result; - - public: - // Construct exception from parse result - explicit xpath_exception(const xpath_parse_result& result); - - // Get error message - virtual const char* what() const throw(); - - // Get parse result - const xpath_parse_result& result() const; - }; - #endif - - // XPath node class (either xml_node or xml_attribute) - class PUGIXML_CLASS xpath_node - { - private: - xml_node _node; - xml_attribute _attribute; - - typedef void (*unspecified_bool_type)(xpath_node***); - - public: - // Default constructor; constructs empty XPath node - xpath_node(); - - // Construct XPath node from XML node/attribute - xpath_node(const xml_node& node); - xpath_node(const xml_attribute& attribute, const xml_node& parent); - - // Get node/attribute, if any - xml_node node() const; - xml_attribute attribute() const; - - // Get parent of contained node/attribute - xml_node parent() const; - - // Safe bool conversion operator - operator unspecified_bool_type() const; - - // Borland C++ workaround - bool operator!() const; - - // Comparison operators - bool operator==(const xpath_node& n) const; - bool operator!=(const xpath_node& n) const; - }; - -#ifdef __BORLANDC__ - // Borland C++ workaround - bool PUGIXML_FUNCTION operator&&(const xpath_node& lhs, bool rhs); - bool PUGIXML_FUNCTION operator||(const xpath_node& lhs, bool rhs); -#endif - - // A fixed-size collection of XPath nodes - class PUGIXML_CLASS xpath_node_set - { - public: - // Collection type - enum type_t - { - type_unsorted, // Not ordered - type_sorted, // Sorted by document order (ascending) - type_sorted_reverse // Sorted by document order (descending) - }; - - // Constant iterator type - typedef const xpath_node* const_iterator; - - // Default constructor. Constructs empty set. - xpath_node_set(); - - // Constructs a set from iterator range; data is not checked for duplicates and is not sorted according to provided type, so be careful - xpath_node_set(const_iterator begin, const_iterator end, type_t type = type_unsorted); - - // Destructor - ~xpath_node_set(); - - // Copy constructor/assignment operator - xpath_node_set(const xpath_node_set& ns); - xpath_node_set& operator=(const xpath_node_set& ns); - - // Get collection type - type_t type() const; - - // Get collection size - size_t size() const; - - // Indexing operator - const xpath_node& operator[](size_t index) const; - - // Collection iterators - const_iterator begin() const; - const_iterator end() const; - - // Sort the collection in ascending/descending order by document order - void sort(bool reverse = false); - - // Get first node in the collection by document order - xpath_node first() const; - - // Check if collection is empty - bool empty() const; - - private: - type_t _type; - - xpath_node _storage; - - xpath_node* _begin; - xpath_node* _end; - - void _assign(const_iterator begin, const_iterator end); - }; -#endif - -#ifndef PUGIXML_NO_STL - // Convert wide string to UTF8 - std::basic_string, std::allocator > PUGIXML_FUNCTION as_utf8(const wchar_t* str); - std::basic_string, std::allocator > PUGIXML_FUNCTION as_utf8(const std::basic_string, std::allocator >& str); - - // Convert UTF8 to wide string - std::basic_string, std::allocator > PUGIXML_FUNCTION as_wide(const char* str); - std::basic_string, std::allocator > PUGIXML_FUNCTION as_wide(const std::basic_string, std::allocator >& str); -#endif - - // Memory allocation function interface; returns pointer to allocated memory or NULL on failure - typedef void* (*allocation_function)(size_t size); - - // Memory deallocation function interface - typedef void (*deallocation_function)(void* ptr); - - // Override default memory management functions. All subsequent allocations/deallocations will be performed via supplied functions. - void PUGIXML_FUNCTION set_memory_management_functions(allocation_function allocate, deallocation_function deallocate); - - // Get current memory management functions - allocation_function PUGIXML_FUNCTION get_memory_allocation_function(); - deallocation_function PUGIXML_FUNCTION get_memory_deallocation_function(); -} - -#if !defined(PUGIXML_NO_STL) && (defined(_MSC_VER) || defined(__ICC)) -namespace std -{ - // Workarounds for (non-standard) iterator category detection for older versions (MSVC7/IC8 and earlier) - std::bidirectional_iterator_tag PUGIXML_FUNCTION _Iter_cat(const pugi::xml_node_iterator&); - std::bidirectional_iterator_tag PUGIXML_FUNCTION _Iter_cat(const pugi::xml_attribute_iterator&); - std::forward_iterator_tag PUGIXML_FUNCTION _Iter_cat(const pugi::xml_named_node_iterator&); -} -#endif - -#if !defined(PUGIXML_NO_STL) && defined(__SUNPRO_CC) -namespace std -{ - // Workarounds for (non-standard) iterator category detection - std::bidirectional_iterator_tag PUGIXML_FUNCTION __iterator_category(const pugi::xml_node_iterator&); - std::bidirectional_iterator_tag PUGIXML_FUNCTION __iterator_category(const pugi::xml_attribute_iterator&); - std::forward_iterator_tag PUGIXML_FUNCTION __iterator_category(const pugi::xml_named_node_iterator&); -} -#endif - -#endif - -/** - * Copyright (c) 2006-2012 Arseny Kapoulkine - * - * Permission is hereby granted, free of charge, to any person - * obtaining a copy of this software and associated documentation - * files (the "Software"), to deal in the Software without - * restriction, including without limitation the rights to use, - * copy, modify, merge, publish, distribute, sublicense, and/or sell - * copies of the Software, and to permit persons to whom the - * Software is furnished to do so, subject to the following - * conditions: - * - * The above copyright notice and this permission notice shall be - * included in all copies or substantial portions of the Software. - * - * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, - * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES - * OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND - * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT - * HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, - * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING - * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR - * OTHER DEALINGS IN THE SOFTWARE. - */