ladybird/Userland/Shell/PosixParser.cpp
Ali Mohammad Pur 1c1aa2c0d0 Shell: Handle (most) errors in the parsers
This turns all errors into either "OOM" or a proper shell error (if
propagation is impossible or meaningless).
Fixes `echo -en '\xfe\x4a' | $SHELL` crashing.
2023-07-01 17:01:54 +03:30

2113 lines
74 KiB
C++

/*
* Copyright (c) 2022, Ali Mohammad Pur <mpfard@serenityos.org>
*
* SPDX-License-Identifier: BSD-2-Clause
*/
#include <AK/CharacterTypes.h>
#include <AK/Debug.h>
#include <AK/StringUtils.h>
#include <Shell/PosixParser.h>
#define TRY_OR_THROW_PARSE_ERROR_AT(expr, position) ({ \
/* Ignore -Wshadow to allow nesting the macro. */ \
AK_IGNORE_DIAGNOSTIC("-Wshadow", \
auto&& _value_or_error = expr;) \
if (_value_or_error.is_error()) { \
AK_IGNORE_DIAGNOSTIC("-Wshadow", \
auto _error = _value_or_error.release_error();) \
if (_error.is_errno() && _error.code() == ENOMEM) \
return make_ref_counted<AST::SyntaxError>(position, "OOM"_short_string); \
return make_ref_counted<AST::SyntaxError>(position, MUST(String::formatted("Error: {}", _error))); \
} \
_value_or_error.release_value(); \
})
static Shell::AST::Position empty_position()
{
return { 0, 0, { 0, 0 }, { 0, 0 } };
}
template<typename T, typename... Ts>
static inline bool is_one_of(T const& value, Ts const&... values)
{
return ((value == values) || ...);
}
static inline bool is_io_operator(Shell::Posix::Token const& token)
{
using namespace Shell::Posix;
return is_one_of(token.type,
Token::Type::Less, Token::Type::Great,
Token::Type::LessAnd, Token::Type::GreatAnd,
Token::Type::DoubleLess, Token::Type::DoubleGreat,
Token::Type::DoubleLessDash, Token::Type::LessGreat,
Token::Type::Clobber);
}
static inline bool is_separator(Shell::Posix::Token const& token)
{
using namespace Shell::Posix;
return is_one_of(token.type,
Token::Type::Semicolon, Token::Type::Newline,
Token::Type::AndIf, Token::Type::OrIf,
Token::Type::Pipe,
Token::Type::And);
}
static inline bool is_a_reserved_word_position(Shell::Posix::Token const& token, Optional<Shell::Posix::Token> const& previous_token, Optional<Shell::Posix::Token> const& previous_previous_token)
{
using namespace Shell::Posix;
auto is_start_of_command = !previous_token.has_value()
|| previous_token->value.is_empty()
|| is_separator(*previous_token)
|| is_one_of(previous_token->type,
Token::Type::OpenParen, Token::Type::CloseParen, Token::Type::Newline, Token::Type::DoubleSemicolon,
Token::Type::Semicolon, Token::Type::Pipe, Token::Type::OrIf, Token::Type::AndIf);
if (is_start_of_command)
return true;
if (!previous_token.has_value())
return false;
auto previous_is_reserved_word = is_one_of(previous_token->value,
"for"sv, "in"sv, "case"sv, "if"sv, "then"sv, "else"sv,
"elif"sv, "while"sv, "until"sv, "do"sv, "done"sv, "esac"sv,
"fi"sv, "!"sv, "{"sv, "}"sv);
if (previous_is_reserved_word)
return true;
if (!previous_previous_token.has_value())
return false;
auto is_third_in_case = previous_previous_token->value == "case"sv
&& token.type == Token::Type::Token && token.value == "in"sv;
if (is_third_in_case)
return true;
auto is_third_in_for = previous_previous_token->value == "for"sv
&& token.type == Token::Type::Token && is_one_of(token.value, "in"sv, "do"sv);
return is_third_in_for;
}
static inline bool is_reserved(Shell::Posix::Token const& token)
{
using namespace Shell::Posix;
return is_one_of(token.type,
Token::Type::If, Token::Type::Then, Token::Type::Else,
Token::Type::Elif, Token::Type::Fi, Token::Type::Do,
Token::Type::Done, Token::Type::Case, Token::Type::Esac,
Token::Type::While, Token::Type::Until, Token::Type::For,
Token::Type::In, Token::Type::OpenBrace, Token::Type::CloseBrace,
Token::Type::Bang);
}
static inline bool is_valid_name(StringView word)
{
// Dr.POSIX: a word consisting solely of underscores, digits, and alphabetics from the portable character set. The first character of a name is not a digit.
return !word.is_empty()
&& !is_ascii_digit(word[0])
&& all_of(word, [](auto ch) { return is_ascii_alphanumeric(ch) || ch == '_'; });
}
namespace Shell::Posix {
ErrorOr<void> Parser::fill_token_buffer(Optional<Reduction> starting_reduction)
{
for (;;) {
auto token = TRY(next_expanded_token(starting_reduction));
if (!token.has_value())
break;
#if SHELL_POSIX_PARSER_DEBUG
DeprecatedString position = "(~)";
if (token->position.has_value())
position = DeprecatedString::formatted("{}:{}", token->position->start_offset, token->position->end_offset);
DeprecatedString expansions = "";
for (auto& exp : token->resolved_expansions)
exp.visit(
[&](ResolvedParameterExpansion& x) { expansions = DeprecatedString::formatted("{}param({}),", expansions, x.to_deprecated_string()); },
[&](ResolvedCommandExpansion& x) { expansions = DeprecatedString::formatted("{}command({:p})", expansions, x.command.ptr()); },
[&](ResolvedArithmeticExpansion& x) { expansions = DeprecatedString::formatted("{}arith({})", expansions, x.source_expression); });
DeprecatedString rexpansions = "";
for (auto& exp : token->expansions)
exp.visit(
[&](ParameterExpansion& x) { rexpansions = DeprecatedString::formatted("{}param({}) from {} to {},", rexpansions, x.parameter.string_view(), x.range.start, x.range.length); },
[&](auto&) { rexpansions = DeprecatedString::formatted("{}...,", rexpansions); });
dbgln("Token @ {}: '{}' (type {}) - parsed expansions: {} - raw expansions: {}", position, token->value.replace("\n"sv, "\\n"sv, ReplaceMode::All), token->type_name(), expansions, rexpansions);
#endif
}
m_token_index = 0;
return {};
}
RefPtr<AST::Node> Parser::parse()
{
auto start_position = peek().position.value_or(empty_position());
return TRY_OR_THROW_PARSE_ERROR_AT(parse_complete_command(), start_position);
}
void Parser::handle_heredoc_contents()
{
while (!eof() && m_token_buffer[m_token_index].type == Token::Type::HeredocContents) {
auto& token = m_token_buffer[m_token_index++];
auto entry = m_unprocessed_heredoc_entries.get(token.relevant_heredoc_key.value());
if (!entry.has_value()) {
error(token, "Discarding unexpected heredoc contents for key '{}'", *token.relevant_heredoc_key);
continue;
}
auto& heredoc = **entry;
auto contents_or_error = [&]() -> ErrorOr<RefPtr<AST::Node>> {
if (heredoc.allow_interpolation()) {
Parser parser { token.value, m_in_interactive_mode, Reduction::HeredocContents };
return parser.parse_word();
}
return make_ref_counted<AST::StringLiteral>(
token.position.value_or(empty_position()),
TRY(String::from_utf8(token.value)),
AST::StringLiteral::EnclosureType::None);
}();
if (contents_or_error.is_error()) {
warnln("Shell: Failed to parse heredoc contents: {}", contents_or_error.error());
continue;
}
if (auto contents = contents_or_error.release_value())
heredoc.set_contents(contents);
m_unprocessed_heredoc_entries.remove(*token.relevant_heredoc_key);
}
}
ErrorOr<Optional<Token>> Parser::next_expanded_token(Optional<Reduction> starting_reduction)
{
while (m_token_buffer.find_if([](auto& token) { return token.type == Token::Type::Eof; }).is_end()) {
auto tokens = TRY(m_lexer.batch_next(starting_reduction));
auto expanded = perform_expansions(move(tokens));
m_token_buffer.extend(expanded);
}
if (m_token_buffer.size() == m_token_index)
return OptionalNone {};
return m_token_buffer[m_token_index++];
}
Vector<Token> Parser::perform_expansions(Vector<Token> tokens)
{
if (tokens.is_empty())
return {};
Vector<Token> expanded_tokens;
auto previous_token = Optional<Token>();
auto previous_previous_token = Optional<Token>();
auto tokens_taken_from_buffer = 0;
expanded_tokens.ensure_capacity(tokens.size());
auto swap_expansions = [&] {
if (previous_previous_token.has_value())
expanded_tokens.append(previous_previous_token.release_value());
if (previous_token.has_value())
expanded_tokens.append(previous_token.release_value());
for (; tokens_taken_from_buffer > 0; tokens_taken_from_buffer--)
m_token_buffer.append(expanded_tokens.take_first());
swap(tokens, expanded_tokens);
expanded_tokens.clear_with_capacity();
};
// (1) join all consecutive newlines (this works around a grammar ambiguity)
auto previous_was_newline = !m_token_buffer.is_empty() && m_token_buffer.last().type == Token::Type::Newline;
for (auto& token : tokens) {
if (token.type == Token::Type::Newline) {
if (previous_was_newline)
continue;
previous_was_newline = true;
} else {
previous_was_newline = false;
}
expanded_tokens.append(move(token));
}
swap_expansions();
// (2) Detect reserved words
if (m_token_buffer.size() >= 1) {
previous_token = m_token_buffer.take_last();
tokens_taken_from_buffer++;
}
if (m_token_buffer.size() >= 1) {
previous_previous_token = m_token_buffer.take_last();
tokens_taken_from_buffer++;
}
auto check_reserved_word = [&](auto& token) {
if (is_a_reserved_word_position(token, previous_token, previous_previous_token)) {
if (token.value == "if"sv)
token.type = Token::Type::If;
else if (token.value == "then"sv)
token.type = Token::Type::Then;
else if (token.value == "else"sv)
token.type = Token::Type::Else;
else if (token.value == "elif"sv)
token.type = Token::Type::Elif;
else if (token.value == "fi"sv)
token.type = Token::Type::Fi;
else if (token.value == "while"sv)
token.type = Token::Type::While;
else if (token.value == "until"sv)
token.type = Token::Type::Until;
else if (token.value == "do"sv)
token.type = Token::Type::Do;
else if (token.value == "done"sv)
token.type = Token::Type::Done;
else if (token.value == "case"sv)
token.type = Token::Type::Case;
else if (token.value == "esac"sv)
token.type = Token::Type::Esac;
else if (token.value == "for"sv)
token.type = Token::Type::For;
else if (token.value == "in"sv)
token.type = Token::Type::In;
else if (token.value == "!"sv)
token.type = Token::Type::Bang;
else if (token.value == "{"sv)
token.type = Token::Type::OpenBrace;
else if (token.value == "}"sv)
token.type = Token::Type::CloseBrace;
else if (token.type == Token::Type::Token)
token.type = Token::Type::Word;
} else if (token.type == Token::Type::Token) {
token.type = Token::Type::Word;
}
};
for (auto& token : tokens) {
if (!previous_token.has_value()) {
check_reserved_word(token);
previous_token = token;
continue;
}
if (!previous_previous_token.has_value()) {
check_reserved_word(token);
previous_previous_token = move(previous_token);
previous_token = token;
continue;
}
check_reserved_word(token);
expanded_tokens.append(exchange(*previous_previous_token, exchange(*previous_token, move(token))));
}
swap_expansions();
// (3) Detect io_number tokens
previous_token = Optional<Token>();
tokens_taken_from_buffer = 0;
if (m_token_buffer.size() >= 1) {
previous_token = m_token_buffer.take_last();
tokens_taken_from_buffer++;
}
for (auto& token : tokens) {
if (!previous_token.has_value()) {
previous_token = token;
continue;
}
if (is_io_operator(token) && previous_token->type == Token::Type::Word && all_of(previous_token->value.bytes_as_string_view(), is_ascii_digit)) {
previous_token->type = Token::Type::IoNumber;
}
expanded_tokens.append(exchange(*previous_token, move(token)));
}
swap_expansions();
// (4) Try to identify simple commands
previous_token = Optional<Token>();
tokens_taken_from_buffer = 0;
if (m_token_buffer.size() >= 1) {
previous_token = m_token_buffer.take_last();
tokens_taken_from_buffer++;
}
for (auto& token : tokens) {
if (!previous_token.has_value()) {
token.could_be_start_of_a_simple_command = true;
previous_token = token;
continue;
}
token.could_be_start_of_a_simple_command = is_one_of(previous_token->type, Token::Type::OpenParen, Token::Type::CloseParen, Token::Type::Newline)
|| is_separator(*previous_token)
|| (!is_reserved(*previous_token) && is_reserved(token));
expanded_tokens.append(exchange(*previous_token, move(token)));
}
swap_expansions();
// (5) Detect assignment words
for (auto& token : tokens) {
if (token.could_be_start_of_a_simple_command)
m_disallow_command_prefix = false;
// Check if we're in a command prefix (could be an assignment)
if (!m_disallow_command_prefix && token.type == Token::Type::Word && token.value.contains('=')) {
// If the word before '=' is a valid name, this is an assignment
auto equal_offset = *token.value.find_byte_offset('=');
if (is_valid_name(token.value.bytes_as_string_view().substring_view(0, equal_offset)))
token.type = Token::Type::AssignmentWord;
else
m_disallow_command_prefix = true;
} else {
m_disallow_command_prefix = true;
}
expanded_tokens.append(move(token));
}
swap_expansions();
// (6) Parse expansions
for (auto& token : tokens) {
if (!is_one_of(token.type, Token::Type::Word, Token::Type::AssignmentWord)) {
expanded_tokens.append(move(token));
continue;
}
Vector<ResolvedExpansion> resolved_expansions;
for (auto& expansion : token.expansions) {
auto resolved = expansion.visit(
[&](ParameterExpansion const& expansion) -> ResolvedExpansion {
auto text = expansion.parameter.string_view();
// ${NUMBER}
if (all_of(text, is_ascii_digit)) {
return ResolvedParameterExpansion {
.parameter = expansion.parameter.to_string().release_value_but_fixme_should_propagate_errors(),
.argument = {},
.range = expansion.range,
.op = ResolvedParameterExpansion::Op::GetPositionalParameter,
};
}
if (text.length() == 1) {
ResolvedParameterExpansion::Op op;
switch (text[0]) {
case '!':
op = ResolvedParameterExpansion::Op::GetLastBackgroundPid;
break;
case '@':
op = ResolvedParameterExpansion::Op::GetPositionalParameterList;
break;
case '-':
op = ResolvedParameterExpansion::Op::GetCurrentOptionFlags;
break;
case '#':
op = ResolvedParameterExpansion::Op::GetPositionalParameterCount;
break;
case '?':
op = ResolvedParameterExpansion::Op::GetLastExitStatus;
break;
case '*':
op = ResolvedParameterExpansion::Op::GetPositionalParameterListAsString;
break;
case '$':
op = ResolvedParameterExpansion::Op::GetShellProcessId;
break;
default:
if (is_valid_name(text)) {
op = ResolvedParameterExpansion::Op::GetVariable;
} else {
error(token, "Unknown parameter expansion: {}", text);
return ResolvedParameterExpansion {
.parameter = expansion.parameter.to_string().release_value_but_fixme_should_propagate_errors(),
.argument = {},
.range = expansion.range,
.op = ResolvedParameterExpansion::Op::StringLength,
};
}
}
return ResolvedParameterExpansion {
.parameter = String::from_code_point(text[0]),
.argument = {},
.range = expansion.range,
.op = op,
};
}
if (text.starts_with('#')) {
return ResolvedParameterExpansion {
.parameter = String::from_utf8(text.substring_view(1)).release_value_but_fixme_should_propagate_errors(),
.argument = {},
.range = expansion.range,
.op = ResolvedParameterExpansion::Op::StringLength,
};
}
GenericLexer lexer { text };
auto parameter = lexer.consume_while([first = true](char c) mutable {
if (first) {
first = false;
return is_ascii_alpha(c) || c == '_';
}
return is_ascii_alphanumeric(c) || c == '_';
});
StringView argument;
ResolvedParameterExpansion::Op op;
switch (lexer.peek()) {
case ':':
lexer.ignore();
switch (lexer.is_eof() ? 0 : lexer.consume()) {
case '-':
argument = lexer.consume_all();
op = ResolvedParameterExpansion::Op::UseDefaultValue;
break;
case '=':
argument = lexer.consume_all();
op = ResolvedParameterExpansion::Op::AssignDefaultValue;
break;
case '?':
argument = lexer.consume_all();
op = ResolvedParameterExpansion::Op::IndicateErrorIfEmpty;
break;
case '+':
argument = lexer.consume_all();
op = ResolvedParameterExpansion::Op::UseAlternativeValue;
break;
default:
error(token, "Unknown parameter expansion: {}", text);
return ResolvedParameterExpansion {
.parameter = String::from_utf8(parameter).release_value_but_fixme_should_propagate_errors(),
.argument = {},
.range = expansion.range,
.op = ResolvedParameterExpansion::Op::StringLength,
};
}
break;
case '-':
lexer.ignore();
argument = lexer.consume_all();
op = ResolvedParameterExpansion::Op::UseDefaultValueIfUnset;
break;
case '=':
lexer.ignore();
argument = lexer.consume_all();
op = ResolvedParameterExpansion::Op::AssignDefaultValueIfUnset;
break;
case '?':
lexer.ignore();
argument = lexer.consume_all();
op = ResolvedParameterExpansion::Op::IndicateErrorIfUnset;
break;
case '+':
lexer.ignore();
argument = lexer.consume_all();
op = ResolvedParameterExpansion::Op::UseAlternativeValueIfUnset;
break;
case '%':
if (lexer.consume_specific('%'))
op = ResolvedParameterExpansion::Op::RemoveLargestSuffixByPattern;
else
op = ResolvedParameterExpansion::Op::RemoveSmallestSuffixByPattern;
argument = lexer.consume_all();
break;
case '#':
if (lexer.consume_specific('#'))
op = ResolvedParameterExpansion::Op::RemoveLargestPrefixByPattern;
else
op = ResolvedParameterExpansion::Op::RemoveSmallestPrefixByPattern;
argument = lexer.consume_all();
break;
default:
if (is_valid_name(text)) {
op = ResolvedParameterExpansion::Op::GetVariable;
} else {
error(token, "Unknown parameter expansion: {}", text);
return ResolvedParameterExpansion {
.parameter = String::from_utf8(parameter).release_value_but_fixme_should_propagate_errors(),
.argument = {},
.range = expansion.range,
.op = ResolvedParameterExpansion::Op::StringLength,
};
}
}
VERIFY(lexer.is_eof());
return ResolvedParameterExpansion {
.parameter = String::from_utf8(parameter).release_value_but_fixme_should_propagate_errors(),
.argument = String::from_utf8(argument).release_value_but_fixme_should_propagate_errors(),
.range = expansion.range,
.op = op,
.expand = ResolvedParameterExpansion::Expand::Word,
};
},
[&](ArithmeticExpansion const& expansion) -> ResolvedExpansion {
return ResolvedArithmeticExpansion { expansion.expression, expansion.range };
},
[&](CommandExpansion const& expansion) -> ResolvedExpansion {
Parser parser { expansion.command.string_view() };
auto node = parser.parse();
m_errors.extend(move(parser.m_errors));
return ResolvedCommandExpansion {
move(node),
expansion.range,
};
});
resolved_expansions.append(move(resolved));
}
token.resolved_expansions = move(resolved_expansions);
expanded_tokens.append(move(token));
}
swap_expansions();
// (7) Loop variables
previous_token = {};
tokens_taken_from_buffer = 0;
if (m_token_buffer.size() >= 1) {
previous_token = m_token_buffer.take_last();
tokens_taken_from_buffer++;
}
for (auto& token : tokens) {
if (!previous_token.has_value()) {
previous_token = token;
continue;
}
if (previous_token->type == Token::Type::For && token.type == Token::Type::Word && is_valid_name(token.value)) {
token.type = Token::Type::VariableName;
}
expanded_tokens.append(exchange(*previous_token, token));
}
swap_expansions();
// (8) Function names
previous_token = {};
previous_previous_token = {};
tokens_taken_from_buffer = 0;
if (m_token_buffer.size() >= 1) {
previous_token = m_token_buffer.take_last();
tokens_taken_from_buffer++;
}
if (m_token_buffer.size() >= 1) {
previous_previous_token = m_token_buffer.take_last();
tokens_taken_from_buffer++;
}
for (auto& token : tokens) {
if (!previous_token.has_value()) {
previous_token = token;
continue;
}
if (!previous_previous_token.has_value()) {
previous_previous_token = move(previous_token);
previous_token = token;
continue;
}
// NAME ( )
if (previous_previous_token->could_be_start_of_a_simple_command
&& previous_previous_token->type == Token::Type::Word
&& previous_token->type == Token::Type::OpenParen
&& token.type == Token::Type::CloseParen) {
previous_previous_token->type = Token::Type::VariableName;
}
expanded_tokens.append(exchange(*previous_previous_token, exchange(*previous_token, token)));
}
swap_expansions();
return tokens;
}
ErrorOr<RefPtr<AST::Node>> Parser::parse_complete_command()
{
auto list = TRY([&]() -> ErrorOr<RefPtr<AST::Node>> {
// separator...
while (is_separator(peek()))
skip();
// list EOF
auto list = TRY(parse_list());
if (eof())
return list;
// list separator EOF
while (is_separator(peek()))
skip();
if (eof())
return list;
auto position = peek().position;
auto syntax_error = make_ref_counted<AST::SyntaxError>(
position.value_or(empty_position()),
TRY("Extra tokens after complete command"_string));
if (list)
list->set_is_syntax_error(*syntax_error);
else
list = syntax_error;
return list;
}());
if (!list)
return nullptr;
return make_ref_counted<AST::Execute>(list->position(), *list);
}
ErrorOr<RefPtr<AST::Node>> Parser::parse_list()
{
Vector<NonnullRefPtr<AST::Node>> nodes;
Vector<AST::Position> positions;
auto start_position = peek().position.value_or(empty_position());
for (;;) {
auto new_node = TRY(parse_and_or());
if (!new_node)
break;
if (peek().type == Token::Type::And) {
new_node = make_ref_counted<AST::Background>(
new_node->position(),
*new_node);
}
nodes.append(new_node.release_nonnull());
if (!is_separator(peek()) || eof())
break;
auto position = consume().position;
if (position.has_value())
positions.append(position.release_value());
}
auto end_position = peek().position.value_or(empty_position());
return make_ref_counted<AST::Sequence>(
AST::Position {
start_position.start_offset,
end_position.end_offset,
start_position.start_line,
end_position.end_line,
},
move(nodes),
move(positions));
}
ErrorOr<RefPtr<AST::Node>> Parser::parse_and_or()
{
while (peek().type == Token::Type::Newline)
skip();
auto node = TRY(parse_pipeline());
if (!node)
return RefPtr<AST::Node> {};
for (;;) {
if (peek().type == Token::Type::AndIf) {
auto and_token = consume();
while (peek().type == Token::Type::Newline)
skip();
auto rhs = TRY(parse_pipeline());
if (!rhs)
return RefPtr<AST::Node> {};
node = make_ref_counted<AST::And>(
node->position(),
*node,
rhs.release_nonnull(),
and_token.position.value_or(empty_position()));
continue;
}
if (peek().type == Token::Type::OrIf) {
auto or_token = consume();
while (peek().type == Token::Type::Newline)
skip();
auto rhs = TRY(parse_pipeline());
if (!rhs)
return RefPtr<AST::Node> {};
node = make_ref_counted<AST::And>(
node->position(),
*node,
rhs.release_nonnull(),
or_token.position.value_or(empty_position()));
continue;
}
break;
}
return node;
}
ErrorOr<RefPtr<AST::Node>> Parser::parse_pipeline()
{
return parse_pipe_sequence();
}
ErrorOr<RefPtr<AST::Node>> Parser::parse_pipe_sequence()
{
auto node = TRY(parse_command());
if (!node)
return RefPtr<AST::Node> {};
for (;;) {
if (peek().type != Token::Type::Pipe)
break;
consume();
while (peek().type == Token::Type::Newline)
skip();
auto rhs = TRY(parse_command());
if (!rhs)
return RefPtr<AST::Node> {};
node = make_ref_counted<AST::Pipe>(
node->position(),
*node,
rhs.release_nonnull());
}
return node;
}
ErrorOr<RefPtr<AST::Node>> Parser::parse_command()
{
auto node = TRY([this]() -> ErrorOr<RefPtr<AST::Node>> {
if (auto node = TRY(parse_function_definition()))
return node;
if (auto node = TRY(parse_simple_command()))
return make_ref_counted<AST::CastToCommand>(node->position(), *node);
auto node = TRY(parse_compound_command());
if (!node)
return node;
if (auto list = TRY(parse_redirect_list())) {
auto position = list->position();
node = make_ref_counted<AST::Join>(
node->position().with_end(position),
*node,
list.release_nonnull());
}
return node;
}());
if (!node)
return nullptr;
return node;
}
ErrorOr<RefPtr<AST::Node>> Parser::parse_function_definition()
{
// NAME OPEN_PAREN CLOSE_PAREN newline* function_body
auto start_index = m_token_index;
ArmedScopeGuard reset = [&] {
m_token_index = start_index;
};
if (peek().type != Token::Type::VariableName) {
return nullptr;
}
auto name = consume();
if (consume().type != Token::Type::OpenParen)
return nullptr;
if (consume().type != Token::Type::CloseParen)
return nullptr;
while (peek().type == Token::Type::Newline)
skip();
auto body = TRY(parse_function_body());
if (!body)
return nullptr;
reset.disarm();
return make_ref_counted<AST::FunctionDeclaration>(
name.position.value_or(empty_position()).with_end(peek().position.value_or(empty_position())),
AST::NameWithPosition { TRY(String::from_utf8(name.value)), name.position.value_or(empty_position()) },
Vector<AST::NameWithPosition> {},
body.release_nonnull());
}
ErrorOr<RefPtr<AST::Node>> Parser::parse_function_body()
{
// compound_command redirect_list?
auto node = TRY(parse_compound_command());
if (!node)
return nullptr;
if (auto list = TRY(parse_redirect_list())) {
auto position = list->position();
node = make_ref_counted<AST::Join>(
node->position().with_end(position),
*node,
list.release_nonnull());
}
return node;
}
ErrorOr<RefPtr<AST::Node>> Parser::parse_redirect_list()
{
// io_redirect*
RefPtr<AST::Node> node;
for (;;) {
auto new_node = TRY(parse_io_redirect());
if (!new_node)
break;
if (node) {
node = make_ref_counted<AST::Join>(
node->position().with_end(new_node->position()),
*node,
new_node.release_nonnull());
} else {
node = new_node;
}
}
return node;
}
ErrorOr<RefPtr<AST::Node>> Parser::parse_compound_command()
{
if (auto node = TRY(parse_brace_group()))
return node;
if (auto node = TRY(parse_subshell()))
return node;
if (auto node = TRY(parse_if_clause()))
return node;
if (auto node = TRY(parse_for_clause()))
return node;
if (auto node = TRY(parse_case_clause()))
return node;
if (auto node = TRY(parse_while_clause()))
return node;
if (auto node = TRY(parse_until_clause()))
return node;
return nullptr;
}
ErrorOr<RefPtr<AST::Node>> Parser::parse_while_clause()
{
if (peek().type != Token::Type::While)
return nullptr;
auto start_position = consume().position.value_or(empty_position());
auto condition = TRY(parse_compound_list());
if (!condition)
condition = make_ref_counted<AST::SyntaxError>(
peek().position.value_or(empty_position()),
TRY("Expected condition after 'while'"_string));
auto do_group = TRY(parse_do_group());
if (!do_group)
do_group = make_ref_counted<AST::SyntaxError>(
peek().position.value_or(empty_position()),
TRY("Expected 'do' after 'while'"_string));
// while foo; bar -> loop { if foo { bar } else { break } }
auto position = start_position.with_end(peek().position.value_or(empty_position()));
return make_ref_counted<AST::ForLoop>(
position,
Optional<AST::NameWithPosition> {},
Optional<AST::NameWithPosition> {},
nullptr,
make_ref_counted<AST::Execute>(position,
make_ref_counted<AST::IfCond>(
position,
Optional<AST::Position> {},
condition.release_nonnull(),
do_group.release_nonnull(),
make_ref_counted<AST::ContinuationControl>(
start_position,
AST::ContinuationControl::ContinuationKind::Break))));
}
ErrorOr<RefPtr<AST::Node>> Parser::parse_until_clause()
{
if (peek().type != Token::Type::Until)
return nullptr;
auto start_position = consume().position.value_or(empty_position());
auto condition = TRY(parse_compound_list());
if (!condition)
condition = make_ref_counted<AST::SyntaxError>(
peek().position.value_or(empty_position()),
TRY("Expected condition after 'until'"_string));
auto do_group = TRY(parse_do_group());
if (!do_group)
do_group = make_ref_counted<AST::SyntaxError>(
peek().position.value_or(empty_position()),
TRY("Expected 'do' after 'until'"_string));
// until foo; bar -> loop { if foo { break } else { bar } }
auto position = start_position.with_end(peek().position.value_or(empty_position()));
return make_ref_counted<AST::ForLoop>(
position,
Optional<AST::NameWithPosition> {},
Optional<AST::NameWithPosition> {},
nullptr,
make_ref_counted<AST::Execute>(position,
make_ref_counted<AST::IfCond>(
position,
Optional<AST::Position> {},
condition.release_nonnull(),
make_ref_counted<AST::ContinuationControl>(
start_position,
AST::ContinuationControl::ContinuationKind::Break),
do_group.release_nonnull())));
}
ErrorOr<RefPtr<AST::Node>> Parser::parse_brace_group()
{
if (peek().type != Token::Type::OpenBrace)
return nullptr;
consume();
auto list = TRY(parse_compound_list());
RefPtr<AST::SyntaxError> error;
if (peek().type != Token::Type::CloseBrace) {
error = make_ref_counted<AST::SyntaxError>(
peek().position.value_or(empty_position()),
TRY(String::formatted("Expected '}}', not {}", peek().type_name())));
} else {
consume();
}
if (error) {
if (list)
list->set_is_syntax_error(*error);
else
list = error;
}
return make_ref_counted<AST::Execute>(list->position(), *list);
}
ErrorOr<RefPtr<AST::Node>> Parser::parse_case_clause()
{
auto start_position = peek().position.value_or(empty_position());
if (peek().type != Token::Type::Case)
return nullptr;
skip();
RefPtr<AST::SyntaxError> syntax_error;
auto expr = TRY(parse_word());
if (!expr)
expr = make_ref_counted<AST::SyntaxError>(
peek().position.value_or(empty_position()),
TRY(String::formatted("Expected a word, not {}", peek().type_name())));
if (peek().type != Token::Type::In) {
syntax_error = make_ref_counted<AST::SyntaxError>(
peek().position.value_or(empty_position()),
TRY(String::formatted("Expected 'in', not {}", peek().type_name())));
} else {
skip();
}
while (peek().type == Token::Type::Newline)
skip();
Vector<AST::MatchEntry> entries;
for (;;) {
if (eof() || peek().type == Token::Type::Esac)
break;
if (peek().type == Token::Type::Newline) {
skip();
continue;
}
// Parse a pattern list
auto needs_dsemi = true;
if (peek().type == Token::Type::OpenParen) {
skip();
needs_dsemi = false;
}
auto result = TRY(parse_case_list());
if (peek().type == Token::Type::CloseParen) {
skip();
} else {
if (!syntax_error)
syntax_error = make_ref_counted<AST::SyntaxError>(
peek().position.value_or(empty_position()),
TRY(String::formatted("Expected ')', not {}", peek().type_name())));
break;
}
while (peek().type == Token::Type::Newline)
skip();
auto compound_list = TRY(parse_compound_list());
if (peek().type == Token::Type::DoubleSemicolon) {
skip();
} else if (needs_dsemi) {
if (!syntax_error)
syntax_error = make_ref_counted<AST::SyntaxError>(
peek().position.value_or(empty_position()),
TRY(String::formatted("Expected ';;', not {}", peek().type_name())));
}
if (syntax_error) {
if (compound_list)
compound_list->set_is_syntax_error(*syntax_error);
else
compound_list = syntax_error;
syntax_error = nullptr;
}
auto position = compound_list->position();
entries.append(AST::MatchEntry {
.options = move(result.nodes),
.match_names = {},
.match_as_position = {},
.pipe_positions = move(result.pipe_positions),
.body = make_ref_counted<AST::Execute>(position, compound_list.release_nonnull()),
});
}
if (peek().type != Token::Type::Esac) {
syntax_error = make_ref_counted<AST::SyntaxError>(
peek().position.value_or(empty_position()),
TRY(String::formatted("Expected 'esac', not {}", peek().type_name())));
} else {
skip();
}
auto node = make_ref_counted<AST::MatchExpr>(
start_position.with_end(peek().position.value_or(empty_position())),
expr.release_nonnull(),
String {},
Optional<AST::Position> {},
move(entries));
if (syntax_error)
node->set_is_syntax_error(*syntax_error);
return node;
}
ErrorOr<Parser::CaseItemsResult> Parser::parse_case_list()
{
// Just a list of words split by '|', delimited by ')'
Vector<NonnullRefPtr<AST::Node>> nodes;
Vector<AST::Position> pipes;
for (;;) {
if (eof() || peek().type == Token::Type::CloseParen)
break;
if (peek().type != Token::Type::Word)
break;
auto node = TRY(parse_word());
if (!node)
node = make_ref_counted<AST::SyntaxError>(
peek().position.value_or(empty_position()),
TRY(String::formatted("Expected a word, not {}", peek().type_name())));
nodes.append(node.release_nonnull());
if (peek().type == Token::Type::Pipe) {
pipes.append(peek().position.value_or(empty_position()));
skip();
} else {
break;
}
}
if (nodes.is_empty())
nodes.append(make_ref_counted<AST::SyntaxError>(
peek().position.value_or(empty_position()),
TRY(String::formatted("Expected a word, not {}", peek().type_name()))));
return CaseItemsResult { move(pipes), move(nodes) };
}
ErrorOr<RefPtr<AST::Node>> Parser::parse_if_clause()
{
// If compound_list Then compound_list {Elif compound_list Then compound_list (Fi|Else)?} [(?=Else) compound_list] (?!=Fi) Fi
auto start_position = peek().position.value_or(empty_position());
if (peek().type != Token::Type::If)
return nullptr;
skip();
auto main_condition = TRY(parse_compound_list());
if (!main_condition)
main_condition = make_ref_counted<AST::SyntaxError>(empty_position(), TRY("Expected compound list after 'if'"_string));
RefPtr<AST::SyntaxError> syntax_error;
if (peek().type != Token::Type::Then) {
syntax_error = make_ref_counted<AST::SyntaxError>(
peek().position.value_or(empty_position()),
TRY(String::formatted("Expected 'then', not {}", peek().type_name())));
} else {
skip();
}
auto main_consequence = TRY(parse_compound_list());
if (!main_consequence)
main_consequence = make_ref_counted<AST::SyntaxError>(empty_position(), TRY("Expected compound list after 'then'"_string));
auto node = make_ref_counted<AST::IfCond>(start_position, Optional<AST::Position>(), main_condition.release_nonnull(), main_consequence.release_nonnull(), nullptr);
auto active_node = node;
while (peek().type == Token::Type::Elif) {
skip();
auto condition = TRY(parse_compound_list());
if (!condition)
condition = make_ref_counted<AST::SyntaxError>(empty_position(), TRY("Expected compound list after 'elif'"_string));
if (peek().type != Token::Type::Then) {
if (!syntax_error)
syntax_error = make_ref_counted<AST::SyntaxError>(
peek().position.value_or(empty_position()),
TRY(String::formatted("Expected 'then', not {}", peek().type_name())));
} else {
skip();
}
auto consequence = TRY(parse_compound_list());
if (!consequence)
consequence = make_ref_counted<AST::SyntaxError>(empty_position(), TRY("Expected compound list after 'then'"_string));
auto new_node = make_ref_counted<AST::IfCond>(start_position, Optional<AST::Position>(), condition.release_nonnull(), consequence.release_nonnull(), nullptr);
active_node->false_branch() = new_node;
active_node = move(new_node);
}
auto needs_fi = true;
switch (peek().type) {
case Token::Type::Else:
skip();
active_node->false_branch() = TRY(parse_compound_list());
if (!active_node->false_branch())
active_node->false_branch() = make_ref_counted<AST::SyntaxError>(empty_position(), TRY("Expected compound list after 'else'"_string));
break;
case Token::Type::Fi:
skip();
needs_fi = false;
break;
default:
if (!syntax_error)
syntax_error = make_ref_counted<AST::SyntaxError>(
peek().position.value_or(empty_position()),
TRY(String::formatted("Expected 'else' or 'fi', not {}", peek().type_name())));
break;
}
if (needs_fi) {
if (peek().type != Token::Type::Fi) {
if (!syntax_error)
syntax_error = make_ref_counted<AST::SyntaxError>(
peek().position.value_or(empty_position()),
TRY(String::formatted("Expected 'fi', not {}", peek().type_name())));
} else {
skip();
}
}
if (syntax_error)
node->set_is_syntax_error(*syntax_error);
return node;
}
ErrorOr<RefPtr<AST::Node>> Parser::parse_subshell()
{
auto start_position = peek().position.value_or(empty_position());
if (peek().type != Token::Type::OpenParen)
return nullptr;
skip();
RefPtr<AST::SyntaxError> error;
auto list = TRY(parse_compound_list());
if (!list)
error = make_ref_counted<AST::SyntaxError>(peek().position.value_or(empty_position()), TRY("Expected compound list after ("_string));
if (peek().type != Token::Type::CloseParen)
error = make_ref_counted<AST::SyntaxError>(peek().position.value_or(empty_position()), TRY("Expected ) after compound list"_string));
else
skip();
if (!list)
return error;
return make_ref_counted<AST::Subshell>(
start_position.with_end(peek().position.value_or(empty_position())),
list.release_nonnull());
}
ErrorOr<RefPtr<AST::Node>> Parser::parse_compound_list()
{
while (peek().type == Token::Type::Newline)
skip();
auto term = TRY(parse_term());
if (!term)
return term;
if (is_separator(peek())) {
if (consume().type == Token::Type::And) {
term = make_ref_counted<AST::Background>(
term->position().with_end(peek().position.value_or(empty_position())),
*term);
}
}
return term;
}
ErrorOr<RefPtr<AST::Node>> Parser::parse_term()
{
Vector<NonnullRefPtr<AST::Node>> nodes;
Vector<AST::Position> positions;
auto start_position = peek().position.value_or(empty_position());
for (;;) {
auto new_node = TRY(parse_and_or());
if (!new_node)
break;
nodes.append(new_node.release_nonnull());
if (!is_separator(peek()))
break;
auto position = consume().position;
if (position.has_value())
positions.append(position.release_value());
}
auto end_position = peek().position.value_or(empty_position());
return make_ref_counted<AST::Sequence>(
start_position.with_end(end_position),
move(nodes),
move(positions));
}
ErrorOr<RefPtr<AST::Node>> Parser::parse_for_clause()
{
// FOR NAME newline+ do_group
// FOR NAME newline+ IN separator do_group
// FOR NAME IN separator do_group
// FOR NAME IN wordlist separator do_group
if (peek().type != Token::Type::For)
return nullptr;
auto start_position = consume().position.value_or(empty_position());
String name;
Optional<AST::Position> name_position;
if (peek().type == Token::Type::VariableName) {
name_position = peek().position;
name = consume().value;
} else {
name = "it"_short_string;
error(peek(), "Expected a variable name, not {}", peek().type_name());
}
auto saw_newline = false;
while (peek().type == Token::Type::Newline) {
saw_newline = true;
skip();
}
auto saw_in = false;
Optional<AST::Position> in_kw_position;
if (peek().type == Token::Type::In) {
saw_in = true;
in_kw_position = peek().position;
skip();
} else if (!saw_newline) {
error(peek(), "Expected 'in' or a newline, not {}", peek().type_name());
}
RefPtr<AST::Node> iterated_expression;
if (!saw_newline)
iterated_expression = parse_word_list();
if (saw_in) {
if (peek().type == Token::Type::Semicolon || peek().type == Token::Type::Newline)
skip();
else
error(peek(), "Expected a semicolon, not {}", peek().type_name());
}
while (peek().type == Token::Type::Newline)
skip();
auto body = TRY(parse_do_group());
return AST::make_ref_counted<AST::ForLoop>(
start_position.with_end(peek().position.value_or(empty_position())),
AST::NameWithPosition { name, name_position.value_or(empty_position()) },
Optional<AST::NameWithPosition> {},
move(iterated_expression),
move(body),
move(in_kw_position),
Optional<AST::Position> {});
}
RefPtr<AST::Node> Parser::parse_word_list()
{
Vector<NonnullRefPtr<AST::Node>> nodes;
auto start_position = peek().position.value_or(empty_position());
for (; peek().type == Token::Type::Word;) {
auto word = TRY_OR_THROW_PARSE_ERROR_AT(parse_word(), start_position);
nodes.append(word.release_nonnull());
}
return make_ref_counted<AST::ListConcatenate>(
start_position.with_end(peek().position.value_or(empty_position())),
move(nodes));
}
ErrorOr<RefPtr<AST::Node>> Parser::parse_word()
{
if (peek().type != Token::Type::Word)
return nullptr;
auto token = consume();
RefPtr<AST::Node> word;
enum class Quote {
None,
Single,
Double,
} in_quote { Quote::None };
auto append_bareword = [&](StringView string) -> ErrorOr<void> {
if (!word && string.starts_with('~')) {
GenericLexer lexer { string };
lexer.ignore();
auto user = lexer.consume_while(is_ascii_alphanumeric);
string = lexer.remaining();
word = make_ref_counted<AST::Tilde>(token.position.value_or(empty_position()), TRY(String::from_utf8(user)));
}
if (string.is_empty())
return {};
auto node = make_ref_counted<AST::BarewordLiteral>(
token.position.value_or(empty_position()),
TRY(String::from_utf8(string)));
if (word) {
word = make_ref_counted<AST::Juxtaposition>(
word->position().with_end(token.position.value_or(empty_position())),
*word,
move(node),
AST::Juxtaposition::Mode::StringExpand);
} else {
word = move(node);
}
return {};
};
auto append_string_literal = [&](StringView string) -> ErrorOr<void> {
auto node = make_ref_counted<AST::StringLiteral>(
token.position.value_or(empty_position()),
TRY(String::from_utf8(string)),
AST::StringLiteral::EnclosureType::SingleQuotes);
if (word) {
word = make_ref_counted<AST::Juxtaposition>(
word->position().with_end(token.position.value_or(empty_position())),
*word,
move(node),
AST::Juxtaposition::Mode::StringExpand);
} else {
word = move(node);
}
return {};
};
auto append_string_part = [&](StringView string) -> ErrorOr<void> {
auto node = make_ref_counted<AST::StringLiteral>(
token.position.value_or(empty_position()),
TRY(String::from_utf8(string)),
AST::StringLiteral::EnclosureType::DoubleQuotes);
if (word) {
word = make_ref_counted<AST::Juxtaposition>(
word->position().with_end(token.position.value_or(empty_position())),
*word,
move(node),
AST::Juxtaposition::Mode::StringExpand);
} else {
word = move(node);
}
return {};
};
auto append_arithmetic_expansion = [&](ResolvedArithmeticExpansion const& x) -> ErrorOr<void> {
auto node = make_ref_counted<AST::ImmediateExpression>(
token.position.value_or(empty_position()),
AST::NameWithPosition {
TRY("math"_string),
token.position.value_or(empty_position()),
},
Vector<NonnullRefPtr<AST::Node>> {
make_ref_counted<AST::ImmediateExpression>(
token.position.value_or(empty_position()),
AST::NameWithPosition {
TRY("reexpand"_string),
token.position.value_or(empty_position()),
},
Vector<NonnullRefPtr<AST::Node>> {
make_ref_counted<AST::StringLiteral>(
token.position.value_or(empty_position()),
TRY(String::from_utf8(x.source_expression)),
AST::StringLiteral::EnclosureType::DoubleQuotes),
},
Optional<AST::Position> {}) },
Optional<AST::Position> {});
if (word) {
word = make_ref_counted<AST::Juxtaposition>(
word->position().with_end(token.position.value_or(empty_position())),
*word,
move(node),
AST::Juxtaposition::Mode::StringExpand);
} else {
word = move(node);
}
return {};
};
auto append_parameter_expansion = [&](ResolvedParameterExpansion const& x) -> ErrorOr<void> {
StringView immediate_function_name;
RefPtr<AST::Node> node;
switch (x.op) {
case ResolvedParameterExpansion::Op::UseDefaultValue:
immediate_function_name = "value_or_default"sv;
break;
case ResolvedParameterExpansion::Op::AssignDefaultValue:
immediate_function_name = "assign_default"sv;
break;
case ResolvedParameterExpansion::Op::IndicateErrorIfEmpty:
immediate_function_name = "error_if_empty"sv;
break;
case ResolvedParameterExpansion::Op::UseAlternativeValue:
immediate_function_name = "null_or_alternative"sv;
break;
case ResolvedParameterExpansion::Op::UseDefaultValueIfUnset:
immediate_function_name = "defined_value_or_default"sv;
break;
case ResolvedParameterExpansion::Op::AssignDefaultValueIfUnset:
immediate_function_name = "assign_defined_default"sv;
break;
case ResolvedParameterExpansion::Op::IndicateErrorIfUnset:
immediate_function_name = "error_if_unset"sv;
break;
case ResolvedParameterExpansion::Op::UseAlternativeValueIfUnset:
immediate_function_name = "null_if_unset_or_alternative"sv;
break;
case ResolvedParameterExpansion::Op::RemoveLargestSuffixByPattern:
// FIXME: Implement this
case ResolvedParameterExpansion::Op::RemoveSmallestSuffixByPattern:
immediate_function_name = "remove_suffix"sv;
break;
case ResolvedParameterExpansion::Op::RemoveLargestPrefixByPattern:
// FIXME: Implement this
case ResolvedParameterExpansion::Op::RemoveSmallestPrefixByPattern:
immediate_function_name = "remove_prefix"sv;
break;
case ResolvedParameterExpansion::Op::StringLength:
immediate_function_name = "length_of_variable"sv;
break;
case ResolvedParameterExpansion::Op::GetPositionalParameter:
case ResolvedParameterExpansion::Op::GetVariable:
node = make_ref_counted<AST::SimpleVariable>(
token.position.value_or(empty_position()),
x.parameter);
break;
case ResolvedParameterExpansion::Op::GetLastBackgroundPid:
node = make_ref_counted<AST::SyntaxError>(
token.position.value_or(empty_position()),
TRY("$! not implemented"_string));
break;
case ResolvedParameterExpansion::Op::GetPositionalParameterList:
node = make_ref_counted<AST::SpecialVariable>(
token.position.value_or(empty_position()),
'*');
break;
case ResolvedParameterExpansion::Op::GetCurrentOptionFlags:
node = make_ref_counted<AST::SyntaxError>(
token.position.value_or(empty_position()),
TRY("The current option flags are not available in parameter expansions"_string));
break;
case ResolvedParameterExpansion::Op::GetPositionalParameterCount:
node = make_ref_counted<AST::SpecialVariable>(
token.position.value_or(empty_position()),
'#');
break;
case ResolvedParameterExpansion::Op::GetLastExitStatus:
node = make_ref_counted<AST::SpecialVariable>(
token.position.value_or(empty_position()),
'?');
break;
case ResolvedParameterExpansion::Op::GetPositionalParameterListAsString:
node = make_ref_counted<AST::SyntaxError>(
token.position.value_or(empty_position()),
TRY("$* not implemented"_string));
break;
case ResolvedParameterExpansion::Op::GetShellProcessId:
node = make_ref_counted<AST::SpecialVariable>(
token.position.value_or(empty_position()),
'$');
break;
}
if (!node) {
Vector<NonnullRefPtr<AST::Node>> arguments;
arguments.append(make_ref_counted<AST::BarewordLiteral>(
token.position.value_or(empty_position()),
x.parameter));
if (!x.argument.is_empty()) {
// dbgln("Will parse {}", x.argument);
arguments.append(*TRY(Parser { x.argument }.parse_word()));
}
node = make_ref_counted<AST::ImmediateExpression>(
token.position.value_or(empty_position()),
AST::NameWithPosition {
TRY(String::from_utf8(immediate_function_name)),
token.position.value_or(empty_position()),
},
move(arguments),
Optional<AST::Position> {});
}
if (x.expand == ResolvedParameterExpansion::Expand::Word) {
node = make_ref_counted<AST::ImmediateExpression>(
token.position.value_or(empty_position()),
AST::NameWithPosition {
TRY("reexpand"_string),
token.position.value_or(empty_position()),
},
Vector { node.release_nonnull() },
Optional<AST::Position> {});
}
if (word) {
word = make_ref_counted<AST::Juxtaposition>(
word->position().with_end(token.position.value_or(empty_position())),
*word,
node.release_nonnull(),
AST::Juxtaposition::Mode::StringExpand);
} else {
word = move(node);
}
return {};
};
auto append_command_expansion = [&](ResolvedCommandExpansion const& x) -> ErrorOr<void> {
if (!x.command)
return {};
RefPtr<AST::Execute> execute_node;
if (x.command->is_execute()) {
execute_node = const_cast<AST::Execute&>(static_cast<AST::Execute const&>(*x.command));
execute_node->capture_stdout();
} else {
execute_node = make_ref_counted<AST::Execute>(
word ? word->position() : empty_position(),
*x.command,
true);
}
if (word) {
word = make_ref_counted<AST::Juxtaposition>(
word->position(),
*word,
execute_node.release_nonnull(),
AST::Juxtaposition::Mode::StringExpand);
} else {
word = move(execute_node);
}
return {};
};
auto append_string = [&](StringView string) -> ErrorOr<void> {
if (string.is_empty())
return {};
Optional<size_t> run_start;
auto escape = false;
for (size_t i = 0; i < string.length(); ++i) {
auto ch = string[i];
switch (ch) {
case '\\':
if (!escape && i + 1 < string.length()) {
if (run_start.has_value())
TRY(append_string_literal(string.substring_view(*run_start, i - *run_start)));
run_start = i + 1;
if (is_one_of(string[i + 1], '"', '\'', '$', '`', '\\')) {
escape = in_quote != Quote::Single;
continue;
}
}
break;
case '\'':
if (in_quote == Quote::Single) {
in_quote = Quote::None;
TRY(append_string_literal(string.substring_view(*run_start, i - *run_start)));
run_start = i + 1;
continue;
}
if (in_quote == Quote::Double) {
escape = false;
continue;
}
[[fallthrough]];
case '"':
if (ch == '\'' && in_quote == Quote::Single) {
escape = false;
continue;
}
if (!escape) {
if (ch == '"' && in_quote == Quote::Double) {
in_quote = Quote::None;
if (run_start.has_value())
TRY(append_string_part(string.substring_view(*run_start, i - *run_start)));
run_start = i + 1;
continue;
}
if (run_start.has_value())
TRY(append_bareword(string.substring_view(*run_start, i - *run_start)));
in_quote = ch == '\'' ? Quote::Single : Quote::Double;
run_start = i + 1;
}
escape = false;
[[fallthrough]];
default:
if (!run_start.has_value())
run_start = i;
escape = false;
continue;
}
}
if (run_start.has_value())
TRY(append_bareword(string.substring_view(*run_start, string.length() - *run_start)));
return {};
};
if (!token.resolved_expansions.is_empty())
dbgln_if(SHELL_POSIX_PARSER_DEBUG, "Expanding '{}' with {} expansion entries", token.value, token.resolved_expansions.size());
size_t current_offset = 0;
auto value_bytes = token.value.bytes_as_string_view();
for (auto& expansion : token.resolved_expansions) {
TRY(expansion.visit(
[&](ResolvedParameterExpansion const& x) -> ErrorOr<void> {
dbgln_if(SHELL_POSIX_PARSER_DEBUG, " Expanding '{}' ({}+{})", x.to_deprecated_string(), x.range.start, x.range.length);
if (x.range.start >= value_bytes.length()) {
dbgln("Parameter expansion range {}-{} is out of bounds for '{}'", x.range.start, x.range.length, value_bytes);
return {};
}
if (x.range.start != current_offset) {
TRY(append_string(value_bytes.substring_view(current_offset, x.range.start - current_offset)));
current_offset = x.range.start;
}
current_offset += x.range.length;
return append_parameter_expansion(x);
},
[&](ResolvedArithmeticExpansion const& x) -> ErrorOr<void> {
if (x.range.start >= value_bytes.length()) {
dbgln("Parameter expansion range {}-{} is out of bounds for '{}'", x.range.start, x.range.length, value_bytes);
return {};
}
if (x.range.start != current_offset) {
TRY(append_string(value_bytes.substring_view(current_offset, x.range.start - current_offset)));
current_offset = x.range.start;
}
current_offset += x.range.length;
return append_arithmetic_expansion(x);
},
[&](ResolvedCommandExpansion const& x) -> ErrorOr<void> {
if (x.range.start >= value_bytes.length()) {
dbgln("Parameter expansion range {}-{} is out of bounds for '{}'", x.range.start, x.range.length, value_bytes);
return {};
}
if (x.range.start != current_offset) {
TRY(append_string(value_bytes.substring_view(current_offset, x.range.start - current_offset)));
current_offset = x.range.start;
}
current_offset += x.range.length;
return append_command_expansion(x);
}));
}
if (current_offset > value_bytes.length()) {
dbgln("Parameter expansion range {}- is out of bounds for '{}'", current_offset, value_bytes);
return word;
}
if (current_offset != value_bytes.length())
TRY(append_string(value_bytes.substring_view(current_offset)));
return word;
}
ErrorOr<RefPtr<AST::Node>> Parser::parse_do_group()
{
if (peek().type != Token::Type::Do) {
return make_ref_counted<AST::SyntaxError>(
peek().position.value_or(empty_position()),
TRY(String::formatted("Expected 'do', not {}", peek().type_name())));
}
consume();
auto list = TRY(parse_compound_list());
RefPtr<AST::SyntaxError> error;
if (peek().type != Token::Type::Done) {
error = make_ref_counted<AST::SyntaxError>(
peek().position.value_or(empty_position()),
TRY(String::formatted("Expected 'done', not {}", peek().type_name())));
} else {
consume();
}
if (error) {
if (list)
list->set_is_syntax_error(*error);
else
list = error;
}
return make_ref_counted<AST::Execute>(list->position(), *list);
}
ErrorOr<RefPtr<AST::Node>> Parser::parse_simple_command()
{
auto start_position = peek().position.value_or(empty_position());
Vector<String> definitions;
Vector<NonnullRefPtr<AST::Node>> nodes;
for (;;) {
if (auto io_redirect = TRY(parse_io_redirect()))
nodes.append(*io_redirect);
else
break;
}
while (peek().type == Token::Type::AssignmentWord) {
definitions.append(peek().value);
if (nodes.is_empty()) {
// run_with_env -e*(assignments) -- (command)
nodes.append(make_ref_counted<AST::BarewordLiteral>(
empty_position(),
TRY("run_with_env"_string)));
}
auto position = peek().position.value_or(empty_position());
nodes.append(make_ref_counted<AST::ImmediateExpression>(
position,
AST::NameWithPosition {
TRY("reexpand"_string),
position,
},
Vector<NonnullRefPtr<AST::Node>> {
make_ref_counted<AST::StringLiteral>(
position,
TRY(String::formatted("-e{}", consume().value)),
AST::StringLiteral::EnclosureType::DoubleQuotes),
},
Optional<AST::Position> {}));
}
if (!definitions.is_empty()) {
nodes.append(make_ref_counted<AST::BarewordLiteral>(
empty_position(),
"--"_short_string));
}
// WORD or io_redirect: IO_NUMBER or io_file
if (!is_one_of(peek().type,
Token::Type::Word, Token::Type::IoNumber,
Token::Type::Less, Token::Type::LessAnd, Token::Type::Great, Token::Type::GreatAnd,
Token::Type::DoubleGreat, Token::Type::LessGreat, Token::Type::Clobber)) {
if (!nodes.is_empty()) {
Vector<AST::VariableDeclarations::Variable> variables;
for (auto& definition : definitions) {
auto equal_offset = definition.find_byte_offset('=');
auto split_offset = equal_offset.value_or(definition.bytes().size());
auto name = make_ref_counted<AST::BarewordLiteral>(
empty_position(),
TRY(definition.substring_from_byte_offset_with_shared_superstring(0, split_offset)));
auto position = peek().position.value_or(empty_position());
auto expanded_value = make_ref_counted<AST::ImmediateExpression>(
position,
AST::NameWithPosition {
TRY("reexpand"_string),
position,
},
Vector<NonnullRefPtr<AST::Node>> {
make_ref_counted<AST::StringLiteral>(
position,
TRY(definition.substring_from_byte_offset_with_shared_superstring(split_offset + 1)),
AST::StringLiteral::EnclosureType::DoubleQuotes),
},
Optional<AST::Position> {});
variables.append({ move(name), move(expanded_value) });
}
return make_ref_counted<AST::VariableDeclarations>(empty_position(), move(variables));
}
return nullptr;
}
// auto first = true;
for (;;) {
if (peek().type == Token::Type::Word) {
auto new_word = TRY(parse_word());
if (!new_word)
break;
// if (first) {
// first = false;
// new_word = make_ref_counted<AST::ImmediateExpression>(
// new_word->position(),
// AST::NameWithPosition {
// "substitute_aliases"sv,
// empty_position(),
// },
// Vector<NonnullRefPtr<AST::Node>> { *new_word },
// Optional<AST::Position> {});
// }
nodes.append(new_word.release_nonnull());
} else if (auto io_redirect = TRY(parse_io_redirect())) {
nodes.append(io_redirect.release_nonnull());
} else {
break;
}
}
auto node = make_ref_counted<AST::ListConcatenate>(
start_position.with_end(peek().position.value_or(empty_position())),
move(nodes));
return node;
}
ErrorOr<RefPtr<AST::Node>> Parser::parse_io_redirect()
{
auto start_position = peek().position.value_or(empty_position());
auto start_index = m_token_index;
// io_redirect: IO_NUMBER? io_file | IO_NUMBER? io_here
Optional<int> io_number;
if (peek().type == Token::Type::IoNumber)
io_number = consume().value.bytes_as_string_view().to_int();
if (auto io_file = TRY(parse_io_file(start_position, io_number)))
return io_file;
if (auto io_here = TRY(parse_io_here(start_position, io_number)))
return io_here;
m_token_index = start_index;
return nullptr;
}
ErrorOr<RefPtr<AST::Node>> Parser::parse_io_here(AST::Position start_position, Optional<int> fd)
{
// io_here: IO_NUMBER? (DLESS | DLESSDASH) WORD
auto io_operator = peek().type;
if (!is_one_of(io_operator, Token::Type::DoubleLess, Token::Type::DoubleLessDash))
return nullptr;
auto io_operator_token = consume();
auto redirection_fd = fd.value_or(0);
auto end_keyword = consume();
if (!is_one_of(end_keyword.type, Token::Type::Word, Token::Type::Token))
return make_ref_counted<AST::SyntaxError>(io_operator_token.position.value_or(start_position), TRY("Expected a heredoc keyword"_string), true);
auto [end_keyword_text, allow_interpolation] = Lexer::process_heredoc_key(end_keyword);
RefPtr<AST::SyntaxError> error;
auto position = start_position.with_end(peek().position.value_or(empty_position()));
auto result = make_ref_counted<AST::Heredoc>(
position,
end_keyword_text,
allow_interpolation,
io_operator == Token::Type::DoubleLessDash,
Optional<int> { redirection_fd });
m_unprocessed_heredoc_entries.set(end_keyword_text, result);
if (error)
result->set_is_syntax_error(*error);
return result;
}
ErrorOr<RefPtr<AST::Node>> Parser::parse_io_file(AST::Position start_position, Optional<int> fd)
{
auto start_index = m_token_index;
// io_file = (LESS | LESSAND | GREAT | GREATAND | DGREAT | LESSGREAT | CLOBBER) WORD
auto io_operator = peek().type;
if (!is_one_of(io_operator,
Token::Type::Less, Token::Type::LessAnd, Token::Type::Great, Token::Type::GreatAnd,
Token::Type::DoubleGreat, Token::Type::LessGreat, Token::Type::Clobber))
return nullptr;
auto io_operator_token = consume();
auto word = TRY(parse_word());
if (!word) {
m_token_index = start_index;
return nullptr;
}
auto position = start_position.with_end(peek().position.value_or(empty_position()));
switch (io_operator) {
case Token::Type::Less:
return make_ref_counted<AST::ReadRedirection>(
position,
fd.value_or(0),
word.release_nonnull());
case Token::Type::Clobber:
// FIXME: Add support for clobber (and 'noclobber')
case Token::Type::Great:
return make_ref_counted<AST::WriteRedirection>(
position,
fd.value_or(1),
word.release_nonnull());
case Token::Type::DoubleGreat:
return make_ref_counted<AST::WriteAppendRedirection>(
position,
fd.value_or(1),
word.release_nonnull());
case Token::Type::LessGreat:
return make_ref_counted<AST::ReadWriteRedirection>(
position,
fd.value_or(0),
word.release_nonnull());
case Token::Type::LessAnd:
case Token::Type::GreatAnd: {
auto is_less = io_operator == Token::Type::LessAnd;
auto source_fd = fd.value_or(is_less ? 0 : 1);
if (word->is_bareword()) {
auto maybe_target_fd = static_ptr_cast<AST::BarewordLiteral>(word)->text().bytes_as_string_view().to_int();
if (maybe_target_fd.has_value()) {
auto target_fd = maybe_target_fd.release_value();
if (is_less)
swap(source_fd, target_fd);
return make_ref_counted<AST::Fd2FdRedirection>(
position,
source_fd,
target_fd);
}
}
if (is_less) {
return make_ref_counted<AST::ReadRedirection>(
position,
source_fd,
word.release_nonnull());
}
return make_ref_counted<AST::WriteRedirection>(
position,
source_fd,
word.release_nonnull());
}
default:
VERIFY_NOT_REACHED();
}
}
}