Merge pull request #9 from OpenVicProject/node-line-column

7 files changed, 141 insertions, 36 deletions

diff --git a/src/openvic-dataloader/detail/BasicBufferHandler.hpp b/src/openvic-dataloader/detail/BasicBufferHandler.hpp
index 5f4f82d..72b85d3 100644
--- a/src/openvic-dataloader/detail/BasicBufferHandler.hpp
+++ b/src/openvic-dataloader/detail/BasicBufferHandler.hpp
@@ -41,6 +41,10 @@ namespace ovdl::detail {
 			return std::nullopt;
 		}
 
+		const auto& get_buffer() const {
+			return _buffer;
+		}
+
 	protected:
 		lexy::buffer<Encoding, MemoryResource> _buffer;
 	};
diff --git a/src/openvic-dataloader/detail/Errors.hpp b/src/openvic-dataloader/detail/Errors.hpp
index f53bedc..bf7c831 100644
--- a/src/openvic-dataloader/detail/Errors.hpp
+++ b/src/openvic-dataloader/detail/Errors.hpp
@@ -1,6 +1,6 @@
 #pragma once
 
-#include "openvic-dataloader/v2script/Parser.hpp"
+#include <openvic-dataloader/ParseError.hpp>
 
 namespace ovdl::errors {
 	inline const ParseError make_no_file_error(const char* file_path) {
diff --git a/src/openvic-dataloader/v2script/AbstractSyntaxTree.cpp b/src/openvic-dataloader/v2script/AbstractSyntaxTree.cpp
index f9fb716..8dc1800 100644
--- a/src/openvic-dataloader/v2script/AbstractSyntaxTree.cpp
+++ b/src/openvic-dataloader/v2script/AbstractSyntaxTree.cpp
@@ -1,3 +1,4 @@
+#include <concepts>
 #include <iomanip>
 #include <sstream>
 #include <string>
@@ -8,6 +9,8 @@
 
 #include <openvic-dataloader/v2script/AbstractSyntaxTree.hpp>
 
+#include <lexy/input_location.hpp>
+
 using namespace ovdl::v2script::ast;
 
 void ovdl::v2script::ast::copy_into_node_ptr_vector(const std::vector<NodePtr>& source, std::vector<NodeUPtr>& dest) {
@@ -18,13 +21,17 @@ void ovdl::v2script::ast::copy_into_node_ptr_vector(const std::vector<NodePtr>&
 	}
 }
 
-AbstractStringNode::AbstractStringNode(std::string&& name) : _name(std::move(name)) {}
+AbstractStringNode::AbstractStringNode(NodeLocation location, std::string&& name) : Node(location),
+																					_name(std::move(name)) {}
+AbstractStringNode::AbstractStringNode(std::string&& name) : AbstractStringNode({}, std::move(name)) {}
+
 std::ostream& AbstractStringNode::print(std::ostream& stream, size_t indent) const {
 	return stream << _name;
 }
 
-#define OVDL_AST_STRING_NODE_DEF(NAME, ...)                                 \
-	NAME::NAME(std::string&& name) : AbstractStringNode(std::move(name)) {} \
+#define OVDL_AST_STRING_NODE_DEF(NAME, ...)                                                                  \
+	NAME::NAME(std::string&& name) : AbstractStringNode(std::move(name)) {}                                  \
+	NAME::NAME(NodeLocation location, std::string&& name) : AbstractStringNode(location, std::move(name)) {} \
 	std::ostream& NAME::print(std::ostream& stream, size_t indent) const __VA_ARGS__
 
 OVDL_AST_STRING_NODE_DEF(IdentifierNode, {
@@ -73,8 +80,9 @@ OVDL_AST_STRING_NODE_DEF(IsTriggeredNode, {
 
 #undef OVDL_AST_STRING_NODE_DEF
 
-AssignNode::AssignNode(NodeCPtr name, NodePtr init)
-	: _initializer(std::move(init)) {
+AssignNode::AssignNode(NodeLocation location, NodeCPtr name, NodePtr init)
+	: Node(location),
+	  _initializer(std::move(init)) {
 	if (name->is_type<IdentifierNode>()) {
 		_name = cast_node_cptr<IdentifierNode>(name)._name;
 	}
@@ -101,9 +109,10 @@ static std::ostream& print_nodeuptr_vector(const std::vector<NodeUPtr>& nodes,
 	return stream;
 }
 
-AbstractListNode::AbstractListNode(const std::vector<NodePtr>& statements) {
+AbstractListNode::AbstractListNode(NodeLocation location, const std::vector<NodePtr>& statements) : Node(location) {
 	copy_into_node_ptr_vector(statements, _statements);
 }
+AbstractListNode::AbstractListNode(const std::vector<NodePtr>& statements) : AbstractListNode({}, statements) {}
 std::ostream& AbstractListNode::print(std::ostream& stream, size_t indent) const {
 	stream << '{';
 	if (!_statements.empty()) {
@@ -113,8 +122,9 @@ std::ostream& AbstractListNode::print(std::ostream& stream, size_t indent) const
 	return stream << "}";
 }
 
-#define OVDL_AST_LIST_NODE_DEF(NAME, ...)                                                \
-	NAME::NAME(const std::vector<NodePtr>& statements) : AbstractListNode(statements) {} \
+#define OVDL_AST_LIST_NODE_DEF(NAME, ...)                                                                                 \
+	NAME::NAME(const std::vector<NodePtr>& statements) : AbstractListNode(statements) {}                                  \
+	NAME::NAME(NodeLocation location, const std::vector<NodePtr>& statements) : AbstractListNode(location, statements) {} \
 	std::ostream& NAME::print(std::ostream& stream, size_t indent) const __VA_ARGS__
 
 OVDL_AST_LIST_NODE_DEF(FileNode, {
@@ -178,9 +188,11 @@ OVDL_AST_LIST_NODE_DEF(DecisionListNode, {
 
 #undef OVDL_AST_LIST_NODE_DEF
 
-EventNode::EventNode(Type type, const std::vector<NodePtr>& statements) : _type(type) {
+EventNode::EventNode(NodeLocation location, Type type, const std::vector<NodePtr>& statements) : Node(location),
+																								 _type(type) {
 	copy_into_node_ptr_vector(statements, _statements);
 }
+EventNode::EventNode(Type type, const std::vector<NodePtr>& statements) : EventNode({}, type, statements) {}
 std::ostream& EventNode::print(std::ostream& stream, size_t indent) const {
 	switch (_type) {
 		case Type::Country: stream << "country_event = "; break;
@@ -194,9 +206,11 @@ std::ostream& EventNode::print(std::ostream& stream, size_t indent) const {
 	return stream << '}';
 }
 
-DecisionNode::DecisionNode(NodePtr name, const std::vector<NodePtr>& statements) : _name(std::move(name)) {
+DecisionNode::DecisionNode(NodeLocation location, NodePtr name, const std::vector<NodePtr>& statements) : Node(location),
+																										  _name(std::move(name)) {
 	copy_into_node_ptr_vector(statements, _statements);
 }
+DecisionNode::DecisionNode(NodePtr name, const std::vector<NodePtr>& statements) : DecisionNode({}, name, statements) {}
 std::ostream& DecisionNode::print(std::ostream& stream, size_t indent) const {
 	print_ptr(stream, _name.get(), indent) << " = {";
 	if (!_statements.empty()) {
@@ -206,10 +220,12 @@ std::ostream& DecisionNode::print(std::ostream& stream, size_t indent) const {
 	return stream << '}';
 }
 
-ExecutionNode::ExecutionNode(Type type, NodePtr name, NodePtr init) : _type(type),
-																	  _name(std::move(name)),
-																	  _initializer(std::move(init)) {
+ExecutionNode::ExecutionNode(NodeLocation location, Type type, NodePtr name, NodePtr init) : Node(location),
+																							 _type(type),
+																							 _name(std::move(name)),
+																							 _initializer(std::move(init)) {
 }
+ExecutionNode::ExecutionNode(Type type, NodePtr name, NodePtr init) : ExecutionNode({}, type, name, init) {}
 std::ostream& ExecutionNode::print(std::ostream& stream, size_t indent) const {
 	print_ptr(stream, _name.get(), indent) << " = ";
 	if (_initializer) {
@@ -218,9 +234,11 @@ std::ostream& ExecutionNode::print(std::ostream& stream, size_t indent) const {
 	return stream;
 }
 
-ExecutionListNode::ExecutionListNode(ExecutionNode::Type type, const std::vector<NodePtr>& statements) : _type(type) {
+ExecutionListNode::ExecutionListNode(NodeLocation location, ExecutionNode::Type type, const std::vector<NodePtr>& statements) : Node(location),
+																																_type(type) {
 	copy_into_node_ptr_vector(statements, _statements);
 }
+ExecutionListNode::ExecutionListNode(ExecutionNode::Type type, const std::vector<NodePtr>& statements) : ExecutionListNode({}, type, statements) {}
 std::ostream& ExecutionListNode::print(std::ostream& stream, size_t indent) const {
 	// Only way to make a valid declared parsable file
 	stream << "{ ";
@@ -244,4 +262,4 @@ Node::operator std::string() const {
 std::ostream& AssignNode::print(std::ostream& stream, size_t indent) const {
 	stream << _name << " = ";
 	return Node::print_ptr(stream, _initializer.get(), indent);
-}
+}
\ No newline at end of file
diff --git a/src/openvic-dataloader/v2script/ModifierGrammar.hpp b/src/openvic-dataloader/v2script/ModifierGrammar.hpp
index ad0704a..632cfd5 100644
--- a/src/openvic-dataloader/v2script/ModifierGrammar.hpp
+++ b/src/openvic-dataloader/v2script/ModifierGrammar.hpp
@@ -27,7 +27,7 @@ namespace ovdl::v2script::grammar {
 			lexy::as_list<std::vector<ast::NodePtr>> >>
 			lexy::callback<ast::NodePtr>(
 				[](auto&& list) {
-					return make_node_ptr<ast::ModifierNode>(LEXY_MOV(list));
+					return ast::make_node_ptr<ast::ModifierNode>(LEXY_MOV(list));
 				});
 	};
 }
\ No newline at end of file
diff --git a/src/openvic-dataloader/v2script/Parser.cpp b/src/openvic-dataloader/v2script/Parser.cpp
index 07d6455..1258bca 100644
--- a/src/openvic-dataloader/v2script/Parser.cpp
+++ b/src/openvic-dataloader/v2script/Parser.cpp
@@ -12,6 +12,7 @@
 #include <openvic-dataloader/ParseWarning.hpp>
 #include <openvic-dataloader/detail/Concepts.hpp>
 #include <openvic-dataloader/v2script/AbstractSyntaxTree.hpp>
+#include <openvic-dataloader/v2script/NodeLocationMap.hpp>
 
 #include <lexy/action/parse.hpp>
 #include <lexy/encoding.hpp>
@@ -57,8 +58,14 @@ public:
 		return _root;
 	}
 
+	ast::NodeLocationMap<decltype(_buffer)>& get_location_map() {
+		return _location_map;
+	}
+
 private:
+	friend class ::ovdl::v2script::ast::Node;
 	std::unique_ptr<ast::Node> _root;
+	ast::NodeLocationMap<decltype(_buffer)> _location_map;
 };
 
 /// BufferHandler ///
@@ -221,4 +228,60 @@ bool Parser::decision_parse() {
 
 const FileNode* Parser::get_file_node() const {
 	return _file_node.get();
+}
+
+void Parser::generate_node_location_map() {
+	_buffer_handler->get_location_map().clear();
+	_buffer_handler->get_location_map().generate_location_map(_buffer_handler->get_buffer(), get_file_node());
+}
+
+const ast::Node::line_col Parser::get_node_begin(const ast::NodeCPtr node) const {
+	if (!node) return { 0, 0 };
+	return node->get_begin_line_col(*this);
+}
+
+const ast::Node::line_col Parser::get_node_end(const ast::NodeCPtr node) const {
+	if (!node) return { 0, 0 };
+	return node->get_end_line_col(*this);
+}
+
+const ast::Node::line_col ast::Node::get_begin_line_col(const Parser& parser) const {
+	if (!parser._buffer_handler->is_valid() || parser._buffer_handler->_location_map.empty()) return {};
+	line_col result;
+	auto [itr, range_end] = parser._buffer_handler->_location_map.equal_range(this);
+	if (itr != range_end) {
+		result.line = itr->second.line_nr();
+		result.column = itr->second.column_nr();
+	}
+	// Standard doesn't really guarantee the direction of the range's sequence, but only GCC goes backwards
+	// TODO: DON'T USE STANDARD UNORDERED_MULTIMAP
+#if defined(__GNUC__) && !defined(__clang__)
+	itr++;
+	if (itr != range_end) {
+		result.line = itr->second.line_nr();
+		result.column = itr->second.column_nr();
+	}
+#endif
+	return result;
+}
+
+const ast::Node::line_col ast::Node::get_end_line_col(const Parser& parser) const {
+	if (!parser._buffer_handler->is_valid() || parser._buffer_handler->_location_map.empty()) return {};
+	line_col result;
+	auto [itr, range_end] = parser._buffer_handler->_location_map.equal_range(this);
+	if (itr != range_end) {
+		result.line = itr->second.line_nr();
+		result.column = itr->second.column_nr();
+	}
+	// Standard doesn't really guarantee the direction of the range's sequence, but only GCC goes backwards
+	// TODO: DON'T USE STANDARD UNORDERED_MULTIMAP
+#if defined(__GNUC__) && !defined(__clang__)
+	return result;
+#endif
+	itr++;
+	if (itr != range_end) {
+		result.line = itr->second.line_nr();
+		result.column = itr->second.column_nr();
+	}
+	return result;
 }
\ No newline at end of file
diff --git a/src/openvic-dataloader/v2script/SimpleGrammar.hpp b/src/openvic-dataloader/v2script/SimpleGrammar.hpp
index c91935e..3e9a96e 100644
--- a/src/openvic-dataloader/v2script/SimpleGrammar.hpp
+++ b/src/openvic-dataloader/v2script/SimpleGrammar.hpp
@@ -32,7 +32,11 @@ namespace ovdl::v2script::grammar {
 
 	struct Identifier {
 		static constexpr auto rule = lexy::dsl::identifier(data_char_class);
-		static constexpr auto value = lexy::as_string<std::string> | lexy::new_<ast::IdentifierNode, ast::NodePtr>;
+		static constexpr auto value = lexy::callback<ast::NodePtr>(
+			[](auto lexeme) {
+				std::string str(lexeme.data(), lexeme.size());
+				return ast::make_node_ptr<ast::IdentifierNode>(ast::NodeLocation { lexeme.begin(), lexeme.end() }, LEXY_MOV(str));
+			});
 	};
 
 	struct StringExpression {
@@ -55,57 +59,73 @@ namespace ovdl::v2script::grammar {
 			// or a Unicode code point of the form uXXXX.
 			auto escape = lexy::dsl::backslash_escape //
 							  .symbol<escaped_symbols>();
-			return lexy::dsl::quoted(c, escape);
+			return lexy::dsl::delimited(lexy::dsl::position(lexy::dsl::lit_b<'"'>))(c, escape);
 		}();
 
-		static constexpr auto value = lexy::as_string<std::string> >> lexy::new_<ast::StringNode, ast::NodePtr>;
+		static constexpr auto value =
+			lexy::as_string<std::string> >>
+			lexy::callback<ast::NodePtr>(
+				[](const char* begin, auto&& str, const char* end) {
+					return ast::make_node_ptr<ast::StringNode>(ast::NodeLocation::make_from(begin, end), LEXY_MOV(str));
+				});
 	};
 
 	struct SimpleAssignmentStatement {
 		static constexpr auto rule =
-			lexy::dsl::p<Identifier> >>
+			lexy::dsl::position(lexy::dsl::p<Identifier>) >>
 			lexy::dsl::equal_sign +
 				(lexy::dsl::p<Identifier> | lexy::dsl::p<StringExpression> | lexy::dsl::recurse_branch<StatementListBlock>);
 
 		static constexpr auto value = lexy::callback<ast::NodePtr>(
-			[](auto name, auto&& initalizer) {
-				return make_node_ptr<ast::AssignNode>(LEXY_MOV(name), LEXY_MOV(initalizer));
+			[](const char* pos, auto name, auto&& initalizer) {
+				return ast::make_node_ptr<ast::AssignNode>(pos, LEXY_MOV(name), LEXY_MOV(initalizer));
 			});
 	};
 
 	struct AssignmentStatement {
 		static constexpr auto rule =
-			lexy::dsl::p<Identifier> >>
+			lexy::dsl::position(lexy::dsl::p<Identifier>) >>
 				(lexy::dsl::equal_sign >>
 						(lexy::dsl::p<Identifier> | lexy::dsl::p<StringExpression> | lexy::dsl::recurse_branch<StatementListBlock>) |
 					lexy::dsl::else_ >> lexy::dsl::return_) |
 			lexy::dsl::p<StringExpression>;
 
 		static constexpr auto value = lexy::callback<ast::NodePtr>(
+			[](const char* pos, auto name, lexy::nullopt = {}) {
+				return LEXY_MOV(name);
+			},
 			[](auto name, lexy::nullopt = {}) {
 				return LEXY_MOV(name);
 			},
-			[](auto name, auto&& initalizer) {
-				return make_node_ptr<ast::AssignNode>(LEXY_MOV(name), LEXY_MOV(initalizer));
+			[](const char* pos, auto name, auto&& initalizer) {
+				return ast::make_node_ptr<ast::AssignNode>(pos, LEXY_MOV(name), LEXY_MOV(initalizer));
 			});
 	};
 
 	struct StatementListBlock {
 		static constexpr auto rule =
-			lexy::dsl::curly_bracketed(
-				lexy::dsl::opt(lexy::dsl::list(lexy::dsl::p<AssignmentStatement>)) + lexy::dsl::opt(lexy::dsl::semicolon));
+			lexy::dsl::position(lexy::dsl::curly_bracketed.open()) >>
+			lexy::dsl::opt(lexy::dsl::list(lexy::dsl::p<AssignmentStatement>)) +
+				lexy::dsl::opt(lexy::dsl::semicolon) >>
+			lexy::dsl::position(lexy::dsl::curly_bracketed.close());
 
 		static constexpr auto value =
 			lexy::as_list<std::vector<ast::NodePtr>> >>
 			lexy::callback<ast::NodePtr>(
-				[](lexy::nullopt = {}, lexy::nullopt = {}) {
-					return ast::make_node_ptr<ast::ListNode>();
+				[](const char* begin, lexy::nullopt, const char* end) {
+					return ast::make_node_ptr<ast::ListNode>(ast::NodeLocation::make_from(begin, end));
+				},
+				[](const char* begin, auto&& list, const char* end) {
+					return ast::make_node_ptr<ast::ListNode>(ast::NodeLocation::make_from(begin, end), LEXY_MOV(list));
+				},
+				[](const char* begin, lexy::nullopt, lexy::nullopt, const char* end) {
+					return ast::make_node_ptr<ast::ListNode>(ast::NodeLocation::make_from(begin, end));
 				},
-				[](auto&& list, lexy::nullopt = {}) {
-					return make_node_ptr<ast::ListNode>(LEXY_MOV(list));
+				[](const char* begin, auto&& list, lexy::nullopt, const char* end) {
+					return ast::make_node_ptr<ast::ListNode>(ast::NodeLocation::make_from(begin, end), LEXY_MOV(list));
 				},
-				[](auto& list) {
-					return make_node_ptr<ast::ListNode>(list);
+				[](const char* begin, auto& list, const char* end) {
+					return ast::make_node_ptr<ast::ListNode>(ast::NodeLocation::make_from(begin, end), list);
 				});
 	};
 
@@ -113,7 +133,7 @@ namespace ovdl::v2script::grammar {
 		// Allow arbitrary spaces between individual tokens.
 		static constexpr auto whitespace = whitespace_specifier | comment_specifier;
 
-		static constexpr auto rule = lexy::dsl::terminator(lexy::dsl::eof).list(lexy::dsl::p<AssignmentStatement>);
+		static constexpr auto rule = lexy::dsl::position + lexy::dsl::terminator(lexy::dsl::eof).list(lexy::dsl::p<AssignmentStatement>);
 
 		static constexpr auto value = lexy::as_list<std::vector<ast::NodePtr>> >> lexy::new_<ast::FileNode, ast::NodePtr>;
 	};
diff --git a/src/openvic-dataloader/v2script/TriggerGrammar.hpp b/src/openvic-dataloader/v2script/TriggerGrammar.hpp
index 290fb70..6d964db 100644
--- a/src/openvic-dataloader/v2script/TriggerGrammar.hpp
+++ b/src/openvic-dataloader/v2script/TriggerGrammar.hpp
@@ -13,7 +13,7 @@ namespace ovdl::v2script::grammar {
 
 		static constexpr auto value = lexy::callback<ast::NodePtr>(
 			[](auto name, auto&& initalizer) {
-				return ast::make_node_ptr<ast::ExecutionNode>(ast::ExecutionNode::Type::Trigger, LEXY_MOV(name), LEXY_MOV(initalizer));
+				return ast::make_node_ptr<ast::ExecutionNode>({}, ast::ExecutionNode::Type::Trigger, LEXY_MOV(name), LEXY_MOV(initalizer));
 			});
 	};