AsmParser: parse source comment using scanner instead of regex

libyul/AsmParser.cpp

@@ -27,11 +27,10 @@
 #include <liblangutil/ErrorReporter.h>
 #include <liblangutil/Exceptions.h>
 #include <liblangutil/Scanner.h>
+#include <liblangutil/Common.h>
 #include <libsolutil/Common.h>
 #include <libsolutil/Visitor.h>
 
-#include <range/v3/view/subrange.hpp>
-
 #include <boost/algorithm/string.hpp>
 
 #include <algorithm>
@@ -58,6 +57,11 @@ std::optional<int> toInt(std::string const& _value)
  }
 }
 
+constexpr bool isNonBreakingWhitespace(char c)
+{
+ return c == ' ' || c == '\t' || c == '\r';
+}
+
 }
 
 langutil::DebugData::ConstPtr Parser::createDebugData() const
@@ -192,13 +196,67 @@ std::optional<std::pair<std::string_view, SourceLocation>> Parser::parseSrcComme
  langutil::SourceLocation const& _commentLocation
 )
 {
- static std::regex const argsRegex = std::regex(
- R"~~(^(-1|\d+):(-1|\d+):(-1|\d+)(?:\s+|$))~~" // index and location, e.g.: 1:234:-1
- R"~~(("(?:[^"\\]|\\.)*"?)?)~~", // optional code snippet, e.g.: "string memory s = \"abc\";..."
- std::regex_constants::ECMAScript | std::regex_constants::optimize
- );
- std::match_results<std::string_view::const_iterator> match;
- if (!regex_search(_arguments.cbegin(), _arguments.cend(), match, argsRegex))
+ std::string const s_args (_arguments);
+ CharStream argumentStream (s_args, "");
+ Scanner scanner (argumentStream);
+
+ int sourceIndex { -1 };
+ int start { -1 };
+ int end { -1 };
+ bool indexOutOfRange { false };
+ std::optional<std::string_view> snippet { std::nullopt };
+ std::string_view tail { _arguments.substr(_arguments.size()) };
+ auto const parseIndex = [](Scanner& _scanner) -> std::tuple<std::optional<int>, bool>
+ {
+ int sgn = 1;
+ if (_scanner.currentToken() == Token::Sub)
+ {
+ sgn = -1;
+ _scanner.next();
+ }
+ if (_scanner.currentToken() != Token::Number)
+ return std::make_tuple(std::nullopt, false);
+ try
+ {
+ return std::make_tuple(sgn * std::stoi(_scanner.currentLiteral()), false);
+ }
+ catch(std::out_of_range const&)
+ {
+ return std::make_tuple(std::nullopt, true);
+ }
+ };
+ bool success = [&]()
+ {
+ auto [parsed, exception] = parseIndex(scanner);
+ indexOutOfRange |= exception;
+ if (!parsed)
+ return false;
+ sourceIndex = *parsed;
+
+ if (scanner.next() != Token::Colon)
+ return false;
+ scanner.next();
+ std::tie(parsed, exception) = parseIndex(scanner);
+ indexOutOfRange |= exception;
+ if (!parsed)
+ return false;
+ start = *parsed;
+
+ if (scanner.next() != Token::Colon)
+ return false;
+ scanner.next();
+ std::tie(parsed, exception) = parseIndex(scanner);
+ indexOutOfRange |= exception;
+ if (!parsed)
+ return false;
+ end = *parsed;
+
+ size_t const endIndex = static_cast<size_t>(scanner.currentLocation().end);
+ return scanner.peekNextToken() == Token::EOS || langutil::isWhiteSpace(_arguments[endIndex]);
+ }();
+
+ // index out of range is handled by error 6367
+ if (!success && !indexOutOfRange)
  {
  m_errorReporter.syntaxError(
  8387_error,
@@ -208,13 +266,34 @@ std::optional<std::pair<std::string_view, SourceLocation>> Parser::parseSrcComme
  return std::nullopt;
  }
 
- solAssert(match.size() == 5, "");
- std::string_view tail = _arguments.substr(static_cast<size_t>(match.position() + match.length()));
+ if (scanner.peekNextToken() == Token::StringLiteral &&
+ isNonBreakingWhitespace(_arguments[static_cast<size_t>(scanner.currentLocation().end)]))
+ {
+ scanner.next();
+ tail = _arguments.substr(static_cast<size_t>(scanner.currentLocation().end));
+ }
+ else if (scanner.peekNextToken() != Token::EOS && isNonBreakingWhitespace(_arguments[static_cast<size_t>(scanner.currentLocation().end)]))
+ {
+ scanner.next();
+ if (_arguments[static_cast<size_t>(scanner.currentLocation().start)] == '"')
+ {
+ auto const endOfLine = _arguments.find_first_of('\n', static_cast<size_t>(scanner.currentLocation().end));
+ snippet = _arguments.substr(static_cast<size_t>(scanner.currentLocation().start), endOfLine);
+ while (!snippet->empty() && isNonBreakingWhitespace(snippet->back()))
+ snippet->remove_suffix(1);
+ if (endOfLine != std::string::npos)
+ tail = _arguments.substr(endOfLine);
+ }
+ else
+ tail = _arguments.substr(static_cast<size_t>(scanner.currentLocation().start));
+ }
+ else
+ tail = _arguments.substr(static_cast<size_t>(scanner.currentLocation().end));
 
- if (match[4].matched && (
- !boost::algorithm::ends_with(match[4].str(), "\"") ||
- boost::algorithm::ends_with(match[4].str(), "\\\"")
- ))
+ if (snippet && (
+  !boost::algorithm::ends_with(*snippet, "\"") ||
+  boost::algorithm::ends_with(*snippet, "\\\"")
+  ))
  {
  m_errorReporter.syntaxError(
  1544_error,
@@ -224,30 +303,25 @@ std::optional<std::pair<std::string_view, SourceLocation>> Parser::parseSrcComme
  return {{tail, SourceLocation{}}};
  }
 
- std::optional<int> const sourceIndex = toInt(match[1].str());
- std::optional<int> const start = toInt(match[2].str());
- std::optional<int> const end = toInt(match[3].str());
-
- if (!sourceIndex.has_value() || !start.has_value() || !end.has_value())
+ if (indexOutOfRange || (sourceIndex < 0 && sourceIndex != -1) || (start < 0 && start != -1) || (end < 0 && end != -1))
  m_errorReporter.syntaxError(
  6367_error,
  _commentLocation,
  "Invalid value in source location mapping. "
  "Expected non-negative integer values or -1 for source index and location."
  );
  else if (sourceIndex == -1)
- return {{tail, SourceLocation{start.value(), end.value(), nullptr}}};
- else if (!(sourceIndex >= 0 && m_sourceNames->count(static_cast<unsigned>(sourceIndex.value()))))
+ return {{tail, SourceLocation{start, end, nullptr}}};
+ else if (!(sourceIndex >= 0 && m_sourceNames->count(static_cast<unsigned>(sourceIndex))))
  m_errorReporter.syntaxError(
  2674_error,
  _commentLocation,
  "Invalid source mapping. Source index not defined via @use-src."
  );
  else
  {
- std::shared_ptr<std::string const> sourceName = m_sourceNames->at(static_cast<unsigned>(sourceIndex.value()));
- solAssert(sourceName, "");
- return {{tail, SourceLocation{start.value(), end.value(), std::move(sourceName)}}};
+ std::shared_ptr<std::string const> sourceName = m_sourceNames->at(static_cast<unsigned>(sourceIndex));
+ return {{tail, SourceLocation{start, end, std::move(sourceName)}}};
  }
  return {{tail, SourceLocation{}}};
 }