liblangUtil: Extends CharStream to translate from line/column to offset.

liblangutil/CharStream.cpp

@@ -144,3 +144,37 @@ string CharStream::singleLineSnippet(string const& _sourceCode, SourceLocation c
 
 	return cut;
 }
+
+optional<int> CharStream::translateLineColumnToPosition(int _line, int _column) const
+{
+	return translateLineColumnToPosition(m_source, _line, _column);
+}
+
+optional<int> CharStream::translateLineColumnToPosition(std::string const& _text, int _line, int _column)
+{
+	size_t offset = 0;
+
+	while (_line > 0)
+	{
+		offset = _text.find('\n', offset);
+		if (offset == _text.npos)
+			return nullopt;
+		offset++; // Skip linefeed.
+		_line--;
+	}
+
+	auto const nextEndOfLineOffset = _text.find('\n', offset);
+	if (nextEndOfLineOffset != string::npos)
+	{
+		if (offset + static_cast<size_t>(_column) < nextEndOfLineOffset)
+			return static_cast<int>(offset) + _column; // Column fitting current non-last line.
+		else
+			return nullopt; // Column-overflow in current non-last line.
+	}
+
+	if (offset + static_cast<size_t>(_column) < _text.size())
+		return static_cast<int>(offset) + _column; // Column fitting into last line
+
+	return nullopt;
+}
+

liblangutil/CharStream.h

@@ -51,6 +51,7 @@
 #pragma once
 
 #include <cstdint>
+#include <optional>
 #include <string>
 #include <tuple>
 #include <utility>
@@ -97,9 +98,19 @@ public:
 	/// Functions that help pretty-printing parse errors
 	/// Do only use in error cases, they are quite expensive.
 	std::string lineAtPosition(int _position) const;
+
+	/// @returns (1-based) line and column that matches to the byte offset @p _position.
 	std::tuple<int, int> translatePositionToLineColumn(int _position) const;
 	///@}
 
+	/// Translates a line:column to the absolute position.
+	/// Parameters @p _line and @p _column are 0-based indices.
+	std::optional<int> translateLineColumnToPosition(int _line, int _column) const;
+
+	/// Translates a line:column to the absolute position for the given input text.
+	/// Parameters @p _line and @p _column are 0-based indices.
+	static std::optional<int> translateLineColumnToPosition(std::string const& _text, int _line, int _column);
+
 	/// Tests whether or not given octet sequence is present at the current position in stream.
 	/// @returns true if the sequence could be found, false otherwise.
 	bool prefixMatch(std::string_view _sequence)

test/liblangutil/CharStream.cpp

@@ -48,6 +48,62 @@ BOOST_AUTO_TEST_CASE(test_fail)
 	);
 }
 
+#define testLineColumnToPositionTranslationFail(_line, _column, _text) \
+	do                                                                       \
+	{                                                                        \
+		auto const source = std::make_shared<CharStream>((_text), "source"); \
+		auto const actualPosition =                                          \
+			source->translateLineColumnToPosition((_line), (_column));       \
+		BOOST_CHECK_EQUAL(actualPosition.value_or(-1), -1);                  \
+	}                                                                        \
+	while (0)
+
+#define testLineColumnToPositionTranslation(_line, _column, _expectedPosition, _text) \
+	do                                                                       \
+	{                                                                        \
+		auto const source = std::make_shared<CharStream>((_text), "source"); \
+		auto const actualPosition =                                          \
+			source->translateLineColumnToPosition((_line), (_column));       \
+		BOOST_CHECK(actualPosition.has_value());                             \
+		BOOST_CHECK_EQUAL(*actualPosition, (_expectedPosition));             \
+	}                                                                        \
+	while (0)
+
+BOOST_AUTO_TEST_CASE(translateLineColumnToPosition)
+{
+	testLineColumnToPositionTranslationFail(0, 0, "");
+
+	// With last line containing no LF
+	testLineColumnToPositionTranslation(0, 0, 0, "ABC");
+	testLineColumnToPositionTranslation(0, 1, 1, "ABC");
+	testLineColumnToPositionTranslation(0, 2, 2, "ABC");
+
+	testLineColumnToPositionTranslation(1, 0, 4, "ABC\nDEF");
+	testLineColumnToPositionTranslation(1, 1, 5, "ABC\nDEF");
+	testLineColumnToPositionTranslation(1, 2, 6, "ABC\nDEF");
+
+	// With last line containing LF
+	testLineColumnToPositionTranslation(0, 0, 0, "ABC\nDEF\n");
+	testLineColumnToPositionTranslation(0, 1, 1, "ABC\nDEF\n");
+	testLineColumnToPositionTranslation(0, 2, 2, "ABC\nDEF\n");
+
+	testLineColumnToPositionTranslation(1, 0, 4, "ABC\nDEF\n");
+	testLineColumnToPositionTranslation(1, 1, 5, "ABC\nDEF\n");
+	testLineColumnToPositionTranslation(1, 2, 6, "ABC\nDEF\n");
+
+	testLineColumnToPositionTranslation(2, 0, 8, "ABC\nDEF\nGHI\n");
+	testLineColumnToPositionTranslation(2, 1, 9, "ABC\nDEF\nGHI\n");
+	testLineColumnToPositionTranslation(2, 2, 10, "ABC\nDEF\nGHI\n");
+
+	// Column overflows.
+	testLineColumnToPositionTranslationFail(0, 3, "ABC\nDEF\n");
+	testLineColumnToPositionTranslationFail(1, 3, "ABC\nDEF\n");
+	testLineColumnToPositionTranslationFail(2, 3, "ABC\nDEF\nGHI\n");
+
+	// Line overflow.
+	testLineColumnToPositionTranslationFail(3, 0, "ABC\nDEF\nGHI\n");
+}
+
 BOOST_AUTO_TEST_SUITE_END()
 
 } // end namespaces