filter-rule-table: stopgap (non-) filtering for T2S/SCFG

phrase-extract/filter-rule-table/FilterRuleTable.cpp

@@ -22,6 +22,7 @@
 #include "StringCfgFilter.h"
 #include "StringForest.h"
 #include "StringForestParser.h"
+#include "TreeCfgFilter.h"
 #include "TreeTsgFilter.h"
 
 namespace MosesTraining
@@ -85,7 +86,9 @@ int FilterRuleTable::Main(int argc, char *argv[])
     ReadTestSet(testStream, testTrees);
     if (sourceSideRuleFormat == kCfg) {
       // TODO Implement TreeCfgFilter
-      Error("tree/cfg filtering algorithm not supported yet");
+      Warn("tree/cfg filtering algorithm not implemented: input will be copied unchanged to output");
+      TreeCfgFilter filter(testTrees);
+      filter.Filter(std::cin, std::cout);
     } else if (sourceSideRuleFormat == kTsg) {
       TreeTsgFilter filter(testTrees);
       filter.Filter(std::cin, std::cout);
@@ -227,10 +230,15 @@ void FilterRuleTable::ProcessOptions(int argc, char *argv[],
 
 void FilterRuleTable::Error(const std::string &msg) const
 {
-  std::cerr << GetName() << ": " << msg << std::endl;
+  std::cerr << GetName() << ": error: " << msg << std::endl;
   std::exit(1);
 }
 
+void FilterRuleTable::Warn(const std::string &msg) const
+{
+  std::cerr << GetName() << ": warning: " << msg << std::endl;
+}
+
 }  // namespace FilterRuleTable
 }  // namespace Syntax
 }  // namespace MosesTraining

phrase-extract/filter-rule-table/FilterRuleTable.h

@@ -52,6 +52,7 @@ private:
   void ReadTestSet(std::istream &,
                    std::vector<boost::shared_ptr<StringForest> > &);
 
+  void Warn(const std::string &) const;
 
   std::string m_name;
 };

phrase-extract/filter-rule-table/TreeCfgFilter.cpp

@@ -0,0 +1,30 @@
+#include "TreeCfgFilter.h"
+
+#include <algorithm>
+
+#include "util/string_piece_hash.hh"
+
+namespace MosesTraining
+{
+namespace Syntax
+{
+namespace FilterRuleTable
+{
+
+TreeCfgFilter::TreeCfgFilter(
+    const std::vector<boost::shared_ptr<StringTree> > &sentences)
+{
+}
+
+void TreeCfgFilter::Filter(std::istream &in, std::ostream &out)
+{
+  // TODO Implement filtering!
+  std::string line;
+  while (std::getline(in, line)) {
+    out << line << std::endl;
+  }
+}
+
+}  // namespace FilterRuleTable
+}  // namespace Syntax
+}  // namespace MosesTraining

phrase-extract/filter-rule-table/TreeCfgFilter.h

@@ -0,0 +1,38 @@
+#pragma once
+
+#include <istream>
+#include <ostream>
+#include <string>
+#include <vector>
+
+#include <boost/shared_ptr.hpp>
+#include <boost/unordered_map.hpp>
+
+#include "syntax-common/numbered_set.h"
+#include "syntax-common/string_tree.h"
+#include "syntax-common/tree.h"
+#include "syntax-common/tree_fragment_tokenizer.h"
+
+#include "CfgFilter.h"
+
+namespace MosesTraining
+{
+namespace Syntax
+{
+namespace FilterRuleTable
+{
+
+// Filters a rule table, discarding rules that cannot be applied to a given
+// test set.  The rule table must have a TSG source-side and the test sentences
+// must be parse trees.
+class TreeCfgFilter : public CfgFilter {
+ public:
+  // Initialize the filter for a given set of test sentences.
+  TreeCfgFilter(const std::vector<boost::shared_ptr<StringTree> > &);
+
+  void Filter(std::istream &in, std::ostream &out);
+};
+
+}  // namespace FilterRuleTable
+}  // namespace Syntax
+}  // namespace MosesTraining