WsParser_VS/libSyntax/libsyntax.cpp

#include "libsyntax.h"
#include "ast_basic.h"

using namespace lib_syntax;
using namespace std;
using namespace lib_token;
using namespace ast_basic;

TokenMatch::TokenMatch(shared_ptr<const ITokenDefine> define) : define_peer(define) {}

QList<std::shared_ptr<const BaseRule>> TokenMatch::children() const { return QList<std::shared_ptr<const BaseRule>>(); }

std::tuple<std::shared_ptr<const Expression>, std::shared_ptr<const IWordBase>> TokenMatch::parse(std::shared_ptr<ParseContext> rt_inst, std::shared_ptr<const IWordBase> head) const {
	if (!head)
		throw new InputTerminal(rt_inst->currentFile());

	auto match_result = define_peer->analysis(head);
	if (std::get<0>(match_result)) {
		rt_inst->currentInst()->addToken(std::get<0>(match_result));
	}
	else {
		throw new MismatchException(head);
	}

	if (std::get<1>(match_result)) {
		return std::make_tuple(nullptr, std::make_shared<WordImpl>(std::get<1>(match_result), head->nextWord()));
	}
	else {
		return std::make_tuple(nullptr, head->nextWord());
	}
}

QString TokenMatch::token_present() const {
	return QString(u8"<%1>").arg(this->define_peer->typeName());
}

Rept::Rept(std::shared_ptr<const BaseRule> rule, int min, int max) : rule_peer(rule), min_match(min), max_match(max) {}

QList<std::shared_ptr<const BaseRule>> Rept::children() const { return QList<std::shared_ptr<const BaseRule>>() << rule_peer; }

std::tuple<std::shared_ptr<const Expression>, std::shared_ptr<const IWordBase>> Rept::parse(std::shared_ptr<ParseContext> rt_inst, std::shared_ptr<const IWordBase> head) const {
	auto temp_head = head;

	// min-match
	for (auto idx = 0; idx < min_match; ++idx) {
		auto result_gen = rule_peer->parse(rt_inst, temp_head);
		if (std::get<0>(result_gen))
			rt_inst->currentInst()->addChild(std::get<0>(result_gen));

		temp_head = std::get<1>(result_gen);
	}

	// max-match
	for (auto idx = min_match; idx < max_match; ++idx) {
		try {
			auto result_gen = rule_peer->parse(rt_inst, temp_head);
			if (std::get<0>(result_gen))
				rt_inst->currentInst()->addChild(std::get<0>(result_gen));

			temp_head = std::get<1>(result_gen);
		}
		catch (SyntaxException* ex) {
			delete ex;
			return std::make_tuple(nullptr, temp_head);
		}
	}

	return std::make_tuple(nullptr, temp_head);
}

QString Rept::token_present() const
{
	return u8"(" + this->rule_peer->token_present() + QString(u8"{%1, %2}").arg(min_match).arg(max_match) + u8")";
}

Seqs::Seqs(const QList<std::shared_ptr<const BaseRule>> mbrs) : mbrs_store(mbrs) {}

QList<std::shared_ptr<const BaseRule>> Seqs::children() const { return mbrs_store; }

std::tuple<std::shared_ptr<const Expression>, std::shared_ptr<const IWordBase>> Seqs::parse(std::shared_ptr<ParseContext> rt_inst, std::shared_ptr<const IWordBase> head) const {
	auto temp_head = head;

	for (auto& r : mbrs_store) {
		auto rst_gene = r->parse(rt_inst, temp_head);
		temp_head = std::get<1>(rst_gene);

		if (std::get<0>(rst_gene))
			rt_inst->currentInst()->addChild(std::get<0>(rst_gene));
	}

	return std::make_tuple(nullptr, temp_head);
}

QString Seqs::token_present() const
{
	QString content;
	for (auto& it : children())
		content += it->token_present();
	return QString(u8"(%1)").arg(content);
}

//std::tuple<BaseRule::MatchResult, uint, std::shared_ptr<const BaseRule>, std::shared_ptr<const Token>>
Any::Any(const QList<std::shared_ptr<const BaseRule>> mbrs) : mbrs_store(mbrs) {}

QList<std::shared_ptr<const BaseRule>> Any::children() const { return mbrs_store; }

class words_span {
public:
	int row_span, column_span;
	words_span(int rspan, int cspan):row_span(rspan), column_span(cspan){}
	bool operator>(const words_span& other) {
		if(row_span > other.row_span)
			return true;
		if(row_span == other.row_span)
			return column_span > other.column_span;
		return false;
	}
};
std::tuple<std::shared_ptr<const Expression>, std::shared_ptr<const IWordBase>> Any::parse(std::shared_ptr<ParseContext> rt_inst, std::shared_ptr<const IWordBase> head) const {
	std::function<words_span(std::shared_ptr<const IWordBase>, std::shared_ptr<const IWordBase>)> measure_span =
		[&](std::shared_ptr<const IWordBase> anchor, std::shared_ptr<const IWordBase> head)->words_span {
		return words_span(anchor->row() - head->row(), anchor->column() - head->column());
	};

	std::tuple<std::shared_ptr<const BaseRule>, words_span> temp_result = std::make_tuple(mbrs_store.first(), words_span(0,0));
	for (auto& fork : mbrs_store) {
		try {
			auto gen = fork->parse(rt_inst, head);

			// <20><><EFBFBD><EFBFBD><EFBFBD>ɹ<EFBFBD><C9B9><EFBFBD>ֱ<EFBFBD>ӷ<EFBFBD><D3B7>ؽ<EFBFBD><D8BD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>
			if (std::get<0>(gen))
				rt_inst->currentInst()->addChild(std::get<0>(gen));
			return std::make_tuple(nullptr, std::get<1>(gen));
		}
		// <20>﷨<EFBFBD><EFB7A8><EFBFBD><EFBFBD><EFBFBD>Ļ<EFBFBD><C4BB><EFBFBD><EFBFBD>бȽ<D0B1>
		catch (MismatchException* ex) {
			auto current_span = measure_span(ex->targetWord(), head);

			if (current_span > std::get<1>(temp_result))
				temp_result = std::make_tuple(fork, current_span);

			delete ex;
		}
	}

	// <20><><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ƥ<EFBFBD><C6A5><EFBFBD>ķ<EFBFBD>֧
	return std::get<0>(temp_result)->parse(rt_inst, head);
}

QString Any::token_present() const
{
	QString members_content;
	for (auto& it : children()) {
		members_content += it->token_present() + u8"|";
	}
	return u8"(" + members_content.mid(0, members_content.size() - 1) + u8")";
}


SyntaxException::SyntaxException(const QString& message) { this->msg_store = message; }

QString SyntaxException::message() const { return msg_store; }

ExpressionRule::ExpressionRule(const QString& rule_name, int expr_mark) : name_store(rule_name) {
	this->filter_proc = [](const TokenSeqs& seqs) { return seqs; };
	this->mark_store = expr_mark;
}

std::shared_ptr<const ExpressionRule> ExpressionRule::reloadRule(std::function<TokenSeqs(const TokenSeqs&)> filter, std::shared_ptr<const BaseRule> rule) {
	auto ninst = makeCopy();
	ninst->child_store = rule;
	ninst->filter_proc = filter;
	return ninst;
}

QString ExpressionRule::name() const { return name_store; }

int ExpressionRule::typeMark() const { return this->mark_store; }

QList<std::shared_ptr<const BaseRule>> ExpressionRule::children() const {
	return QList<std::shared_ptr<const BaseRule>>() << this->child_store;
}

std::tuple<std::shared_ptr<const Expression>, std::shared_ptr<const IWordBase>> ExpressionRule::parse(std::shared_ptr<ParseContext> rt_inst, std::shared_ptr<const IWordBase> head) const {
	std::shared_ptr<Expression> elm_ast = this->newEmptyInstance();

	rt_inst->pushExpressionRule(this->shared_from_this());
	rt_inst->pushInst(elm_ast);

	try {
		auto rstg = child_store->parse(rt_inst, head);

		auto tokens_decl = this->filter_proc(elm_ast->tokens());
		elm_ast->tokensReset(tokens_decl);

		rt_inst->popInst();
		rt_inst->popExpressionRule();
		return std::make_tuple(elm_ast, std::get<1>(rstg));
	}
	catch (...) {
		rt_inst->popInst();
		rt_inst->popExpressionRule();
		throw;
	}
}

QString ExpressionRule::token_present() const {
	return QString(u8"(%1)").arg(child_store->token_present());
}

MismatchException::MismatchException(std::shared_ptr<const lib_token::IWordBase> inst) :SyntaxException(
	QString(u8"Syntax[0x00001]<5D>﷨ƥ<EFB7A8><C6A5><EFBFBD><EFBFBD><EFBFBD>󣬲<EFBFBD><F3A3ACB2><EFBFBD>ʶ<EFBFBD><CAB6>token<65><6E>%1<%2,%3>(%4)")
		.arg(inst->content()).arg(inst->row()).arg(inst->column()).arg(inst->file())), target(inst) {}

std::shared_ptr<const IWordBase>MismatchException::targetWord() const {
	return this->target;
}

InputTerminal::InputTerminal(const QString& file_path) 
	:SyntaxException(QString(u8"Syntax[0x0000]token<65><6E><EFBFBD><EFBFBD>%1<><31><EFBFBD><EFBFBD>ǰ<EFBFBD><C7B0>ֹ").arg(file_path)) {}