From a07c64148eb60b886f92caa46a9c687240ec420c Mon Sep 17 00:00:00 2001
From: Spartan322 <Megacake1234@gmail.com>
Date: Sat, 20 Jul 2024 21:02:30 -0400
Subject: Add backslash identifier support to v2script Add buffer::char_type
 size multiplier to max file size of string intern buffer

Fix list grammar segfaults
Fix diagnostic logger intern segfaults from buffer reallocation
Fix non-string-supported CSV parser not supporting Victoria 2 CSV escaping behavior
---
 src/openvic-dataloader/csv/CsvGrammar.hpp | 28 ++++++++++++++++++----------
 src/openvic-dataloader/csv/Parser.cpp     | 12 ++++++++----
 2 files changed, 26 insertions(+), 14 deletions(-)

(limited to 'src/openvic-dataloader/csv')

diff --git a/src/openvic-dataloader/csv/CsvGrammar.hpp b/src/openvic-dataloader/csv/CsvGrammar.hpp
index 19aee54..91226c7 100644
--- a/src/openvic-dataloader/csv/CsvGrammar.hpp
+++ b/src/openvic-dataloader/csv/CsvGrammar.hpp
@@ -11,10 +11,7 @@
 
 #include <lexy/_detail/config.hpp>
 #include <lexy/callback.hpp>
-#include <lexy/callback/string.hpp>
 #include <lexy/dsl.hpp>
-#include <lexy/dsl/ascii.hpp>
-#include <lexy/dsl/option.hpp>
 #include <lexy/encoding.hpp>
 
 #include "detail/Convert.hpp"
@@ -73,6 +70,9 @@ namespace ovdl::csv::grammar {
 	constexpr auto escaped_quote = lexy::symbol_table<char> //
 									   .map<'"'>('"');
 
+	constexpr auto escaped_newline = lexy::symbol_table<char> //
+										 .map<'n'>('\n');
+
 	template<ParseOptions Options>
 	struct CsvGrammar {
 		struct StringValue : lexy::scan_production<std::string>,
@@ -118,6 +118,11 @@ namespace ovdl::csv::grammar {
 			template<auto character>
 			static constexpr auto _escape_check = character - (lexy::dsl::lit_b<Options.SepChar> / lexy::dsl::ascii::newline);
 
+			struct Backslash {
+				static constexpr auto rule = LEXY_LIT("\\n");
+				static constexpr auto value = lexy::constant('\n');
+			};
+
 			template<typename Context, typename Reader>
 			static constexpr scan_result scan(lexy::rule_scanner<Context, Reader>& scanner, detail::IsFileParseState auto& state) {
 				using encoding = typename Reader::encoding;
@@ -134,13 +139,16 @@ namespace ovdl::csv::grammar {
 					if constexpr (Options.SupportStrings) {
 						return lexy::dsl::identifier(character - (lexy::dsl::lit_b<Options.SepChar> / lexy::dsl::ascii::newline));
 					} else {
-						auto escape_check_char = _escape_check<character>;
-						auto id_check_char = escape_check_char - lexy::dsl::lit_b<'\\'>;
-						auto id_segment = lexy::dsl::identifier(id_check_char);
-						auto escape_segement = lexy::dsl::token(escape_check_char);
-						auto escape_sym = lexy::dsl::symbol<escaped_symbols>(escape_segement);
-						auto escape_rule = lexy::dsl::lit_b<'\\'> >> escape_sym;
-						return lexy::dsl::list(id_segment | escape_rule);
+						constexpr auto backslash = lexy::dsl::lit_b<'\\'>;
+
+						constexpr auto escape_check_char = _escape_check<character>;
+						constexpr auto escape_rule = lexy::dsl::p<Backslash>;
+
+						return lexy::dsl::list(
+							lexy::dsl::identifier(escape_check_char - backslash) |
+							escape_rule |
+							lexy::dsl::capture(escape_check_char) //
+						);
 					}
 				}();
 
diff --git a/src/openvic-dataloader/csv/Parser.cpp b/src/openvic-dataloader/csv/Parser.cpp
index 8fe8b17..bbd1be4 100644
--- a/src/openvic-dataloader/csv/Parser.cpp
+++ b/src/openvic-dataloader/csv/Parser.cpp
@@ -195,6 +195,10 @@ typename Parser::error_range Parser::get_errors() const {
 	return _parse_handler->get_errors();
 }
 
+std::string_view Parser::error(const ovdl::error::Error* error) const {
+	return error->message(_parse_handler->parse_state().logger().symbol_interner());
+}
+
 const FilePosition Parser::get_error_position(const error::Error* error) const {
 	if (!error || !error->is_linked_in_tree()) {
 		return {};
@@ -231,20 +235,20 @@ void Parser::print_errors_to(std::basic_ostream<char>& stream) const {
 		dryad::visit_tree(
 			error,
 			[&](const error::BufferError* buffer_error) {
-				stream << "buffer error: " << buffer_error->message() << '\n';
+				stream << "buffer error: " << this->error(buffer_error) << '\n';
 			},
 			[&](dryad::child_visitor<error::ErrorKind> visitor, const error::AnnotatedError* annotated_error) {
-				stream << annotated_error->message() << '\n';
+				stream << this->error(annotated_error) << '\n';
 				auto annotations = annotated_error->annotations();
 				for (auto annotation : annotations) {
 					visitor(annotation);
 				}
 			},
 			[&](const error::PrimaryAnnotation* primary) {
-				stream << primary->message() << '\n';
+				stream << this->error(primary) << '\n';
 			},
 			[&](const error::SecondaryAnnotation* secondary) {
-				stream << secondary->message() << '\n';
+				stream << this->error(secondary) << '\n';
 			});
 	}
 }
\ No newline at end of file
-- 
cgit v1.2.3-56-ga3b1