From b0c3ba3f91926b0c95625bdbf4aab69269130b13 Mon Sep 17 00:00:00 2001 From: Spartan322 Date: Thu, 9 May 2024 10:06:02 -0400 Subject: Add runtime encoding detection and conversion Win-1251/1252 detection is a reduced C++ version of https://github.com/hsivonen/chardetng Add manually-specified encoding fallback Add default system encoding fallback Add error recovery to v2script Add unknown encoding detection warning Remove csv::Parser templating Fix lua files dropping data Update lexy to foonathan/lexy@1e5d99fa3826b1c3c8628d3a11117fb4fb4cc0d0 Remove exclusive reliance on lexy::default_encoding for v2script Move internal concepts to src/openvic-detail/InternalConcepts.hpp Move contents of DetectUtf8.hpp to src/detail/Detect.hpp Move openvic-dataloader/AbstractSyntaxTree.hpp to src Move DiagnosticLogger.hpp to src Move File.hpp to src Move openvic-dataloader/detail/utlity files to openvic-dataloader/detail Add ovdl::utility::type_concat Add ovdl::utility::type_prepend Add ovdl::utility::is_instance_of Overhaul parse error messages --- src/openvic-dataloader/ParseState.hpp | 105 ++++++++++++++++++++++++++++++++++ 1 file changed, 105 insertions(+) create mode 100644 src/openvic-dataloader/ParseState.hpp (limited to 'src/openvic-dataloader/ParseState.hpp') diff --git a/src/openvic-dataloader/ParseState.hpp b/src/openvic-dataloader/ParseState.hpp new file mode 100644 index 0000000..806829c --- /dev/null +++ b/src/openvic-dataloader/ParseState.hpp @@ -0,0 +1,105 @@ +#pragma once + +#include + +#include + +#include +#include + +#include + +#include "DiagnosticLogger.hpp" +#include "detail/InternalConcepts.hpp" + +namespace ovdl { + struct BasicParseState { + explicit BasicParseState(detail::Encoding encoding = detail::Encoding::Unknown) : _encoding(encoding) {} + + detail::Encoding encoding() const { + return _encoding; + } + + protected: + detail::Encoding _encoding; + }; + + template + struct ParseState : BasicParseState { + using ast_type = AstT; + using file_type = typename ast_type::file_type; + using diagnostic_logger_type = BasicDiagnosticLogger; + + ParseState(typename ast_type::file_type&& file, detail::Encoding encoding) + : _ast { std::move(file) }, + _logger { this->ast().file() }, + BasicParseState(encoding) {} + + template + ParseState(lexy::buffer&& buffer, detail::Encoding encoding) + : ParseState(typename ast_type::file_type { std::move(buffer) }, encoding) {} + + template + ParseState(const char* path, lexy::buffer&& buffer, detail::Encoding encoding) + : ParseState(typename ast_type::file_type { path, std::move(buffer) }, encoding) {} + + ast_type& ast() { + return _ast; + } + + const ast_type& ast() const { + return _ast; + } + + diagnostic_logger_type& logger() { + return _logger; + } + + const diagnostic_logger_type& logger() const { + return _logger; + } + + private: + ast_type _ast; + diagnostic_logger_type _logger; + }; + + template + struct FileParseState : BasicParseState { + using file_type = FileT; + using diagnostic_logger_type = BasicDiagnosticLogger; + + FileParseState(file_type&& file, detail::Encoding encoding) + : _file { std::move(file) }, + _logger { this->file() }, + BasicParseState(encoding) {} + + template + FileParseState(lexy::buffer&& buffer, detail::Encoding encoding) + : FileParseState(file_type { std::move(buffer) }, encoding) {} + + template + FileParseState(const char* path, lexy::buffer&& buffer, detail::Encoding encoding) + : FileParseState(file_type { path, std::move(buffer) }, encoding) {} + + file_type& file() { + return _file; + } + + const file_type& file() const { + return _file; + } + + diagnostic_logger_type& logger() { + return _logger; + } + + const diagnostic_logger_type& logger() const { + return _logger; + } + + private: + file_type _file; + diagnostic_logger_type _logger; + }; +} \ No newline at end of file -- cgit v1.2.3-56-ga3b1