WsParser_VS/libSyntax/libsyntax.cpp

#include "libsyntax.h"
#include "ast_basic.h"
#include <tuple>
#include <QDebug>

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

__anyone_impl::__anyone_impl(const QList<std::shared_ptr<const IBasicRule>> mbrs) : mbrs_store(mbrs) { }

QList<std::shared_ptr<const IBasicRule>> __anyone_impl::children() const {
	return mbrs_store;
}

QList<std::shared_ptr<const MatchCursor>> __anyone_impl::parse(std::shared_ptr<const MatchCursor> cursor) const {
	if (cursor->mustStop())
		return QList<std::shared_ptr<const MatchCursor>>() << cursor;

	QList<std::shared_ptr<const MatchCursor>> result_list;
	for (auto rx : this->children())
		result_list.append(rx->parse(cursor));

	// 完全匹配分支
	decltype(result_list) completely_list;
	std::copy_if(result_list.begin(), result_list.end(), std::back_inserter(completely_list),
		[&](std::shared_ptr<const MatchCursor> ins) { return cursor->totalErrorCount() == ins->totalErrorCount(); });
	if (completely_list.size())
		return completely_list;

	// 经过修正的分支
	decltype(result_list) modify_list;
	std::copy_if(result_list.begin(), result_list.end(), std::back_inserter(modify_list),
		[&](std::shared_ptr<const MatchCursor> ins) { return !ins->mustStop(); });
	if (modify_list.size())
		return modify_list;

	return result_list;
}

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

__sequence_impl::__sequence_impl(const QList<std::shared_ptr<const IBasicRule>> mbrs) : mbrs_store(mbrs) { }

QList<std::shared_ptr<const IBasicRule>> __sequence_impl::children() const {
	return mbrs_store;
}

QList<std::shared_ptr<const MatchCursor>> __sequence_impl::parse(std::shared_ptr<const MatchCursor> cursor) const {
	if (cursor->mustStop())
		return QList<std::shared_ptr<const MatchCursor>>() << cursor;

	QList<std::shared_ptr<const MatchCursor>> results;
	QList<std::shared_ptr<const MatchCursor>> bridge_list{ cursor };
	for (auto rule : this->children()) {
		QList<std::shared_ptr<const MatchCursor>> current_result;
		std::for_each(bridge_list.begin(), bridge_list.end(),
			[&](std::shared_ptr<const MatchCursor> vcurs) {
				if (!vcurs->mustStop()) {
					current_result.append(rule->parse(vcurs));
				}
				else {
					results.push_back(vcurs);
				}
			});

		bridge_list = current_result;
	}

	// 完全匹配分支集合
	decltype(bridge_list) completely_list;
	std::copy_if(bridge_list.begin(), bridge_list.end(), std::back_inserter(completely_list),
		[&](std::shared_ptr<const MatchCursor> ins) { return ins->totalErrorCount() == cursor->totalErrorCount(); });
	if (completely_list.size())
		return completely_list;

	// 经过修正的分支
	decltype(bridge_list) modify_list;
	std::copy_if(bridge_list.begin(), bridge_list.end(), std::back_inserter(modify_list),
		[](std::shared_ptr<const MatchCursor> ins) { return !ins->mustStop(); });
	if (modify_list.size())
		return modify_list;

	results.append(bridge_list);
	return results;
}

QString __sequence_impl::present() const {
	QString content;
	for (auto& it : children())
		content += it->present() + " ";
	return content.mid(0, content.size() - 1);
}

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

QList<std::shared_ptr<const IBasicRule>> __repeat_impl::children() const {
	return QList<std::shared_ptr<const IBasicRule>>() << rule_peer;
}

#include <algorithm>
QList<std::shared_ptr<const MatchCursor>> __repeat_impl::parse(std::shared_ptr<const MatchCursor> cursor) const {
	if (cursor->mustStop())
		return QList<std::shared_ptr<const MatchCursor>>() << cursor;

	QList<std::shared_ptr<const MatchCursor>> max_match_begin = { cursor };
	if (min_match) {
		QList<std::shared_ptr<const IBasicRule>> temp_rules;
		for (auto idx = 0; idx < min_match; ++idx)
			temp_rules << this->rule_peer;

		auto seqs_rule = std::make_shared<__sequence_impl>(temp_rules);
		max_match_begin = seqs_rule->parse(cursor);
	}

	// 如果不满足最小重复匹配次数要求，则返回
	int continue_count = std::count_if(max_match_begin.begin(), max_match_begin.end(),
		[](std::shared_ptr<const MatchCursor > ins) { return !ins->mustStop(); });
	if (!continue_count)
		return max_match_begin;

	// 最小匹配次数中所有错误分支都是无用的、需要舍弃
	for (auto idx = 0; idx < max_match_begin.size(); ++idx) {
		auto current_cursor = max_match_begin.at(idx);
		if (current_cursor->mustStop())
			max_match_begin.removeAt(idx--);
	}

	QList<std::shared_ptr<const MatchCursor>> results = max_match_begin;
	decltype(results) bridge_list = max_match_begin;
	// 尝试重复匹配最大次数
	for (auto idx = min_match; idx < max_match && bridge_list.size(); ++idx) {
		QList<std::shared_ptr<const MatchCursor>> current_list;

		// 匹配迭代一次
		std::for_each(bridge_list.begin(), bridge_list.end(),
			[&](std::shared_ptr<const MatchCursor> ins) {
				current_list.append(this->rule_peer->parse(ins));
			});

		for (auto vdx = 0; vdx < current_list.size(); ++vdx) {
			auto rst_branch = current_list.at(vdx);
			if (rst_branch->mustStop()) {
				results.append(rst_branch);
				current_list.removeAt(vdx--);
			}
		}

		bridge_list = current_list;
	}
	results.append(bridge_list);


	std::sort(results.begin(), results.end(),
		[](std::shared_ptr<const MatchCursor> a, std::shared_ptr<const MatchCursor> b) {
			return a->currentToken()->position() > b->currentToken()->position();
		});

	// 提取完全匹配的分支
	decltype(results) rets_completely;
	std::for_each(results.begin(), results.end(),
		[&](std::shared_ptr<const MatchCursor> ins) {
			if (ins->totalErrorCount() == cursor->totalErrorCount()) {
				if (!rets_completely.size()) {
					rets_completely.append(ins);
				}
				else if (rets_completely.last()->currentToken()->position() == ins->currentToken()->position()) {
					rets_completely.append(ins);
				}
			}
		});

	// 提取经过修正的分支
	decltype(results) rets_modified;
	std::for_each(results.begin(), results.end(),
		[&](std::shared_ptr<const MatchCursor> ins) {
			if (!ins->mustStop()) {
				if (!rets_modified.size()) {
					rets_modified.append(ins);
				}
				else if (rets_modified.last()->currentToken()->position() == ins->currentToken()->position()) {
					rets_modified.append(ins);
				}
			}
		});

	// 允许持续的集合
	rets_completely.append(rets_modified);
	if (rets_completely.size())
		return rets_completely;

	return results;
}

QString __repeat_impl::present() const {
	if (min_match == 0 && max_match == INT_MAX)
		return "(" + this->rule_peer->present() + QString(")*");
	else if (min_match == 1 && max_match == INT_MAX)
		return "(" + this->rule_peer->present() + QString(")+");
	else if (min_match == 0 && max_match == 1)
		return "(" + this->rule_peer->present() + QString(")?");

	return "(" + this->rule_peer->present() + QString("){%1, %2}").arg(min_match).arg(max_match);
}

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

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

ExprRule::ExprRule(const QString& rule_name, int expr_mark)
	: name_store(rule_name), mark_store(expr_mark) { }

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

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

#include <ast_novel.h>
MatchCursor::MatchCursor(const QString& path) :_file_path(path) { }

MatchCursor::MatchCursor(std::shared_ptr<const MatchCursor> other_ptr)
	: _prev_cursor(other_ptr),
	_file_path(other_ptr->_file_path),
	_total_errors(other_ptr->_total_errors),
	_exprs_errors(other_ptr->_exprs_errors),
	_current_token(other_ptr->_current_token),
	_remains_word(other_ptr->_remains_word) { }

std::shared_ptr<const MatchCursor> MatchCursor::previous() const {
	return _prev_cursor;
}

QString MatchCursor::filePath() const {
	return _file_path;
}

void MatchCursor::enterExprs() {
	auto new_expr = std::make_shared<ErrsPack>();
	this->_exprs_errors.push_back(new_expr);
}

void MatchCursor::logExprsError(const QString& msg) {
	this->_total_errors.push_back(msg);
	this->_exprs_errors.last()->addError(msg);
}

void MatchCursor::quitExprs() {
	this->_exprs_errors.pop_back();
}

bool MatchCursor::mustStop() const {
	return exprsErrorCount() >= 2;
}

int MatchCursor::exprsErrorCount() const {
	if (this->_exprs_errors.size())
		return this->_exprs_errors.last()->errorCount();
	return 0;
}

int MatchCursor::totalErrorCount() const {
	return this->_total_errors.size();
}

QList<QString> MatchCursor::totalErrors() const {
	return this->_total_errors;
}

void MatchCursor::setCurrent(std::shared_ptr<const IActionToken> t, std::shared_ptr<const IPrimitiveWord> remains) {
	this->_current_token = t;
	this->_remains_word = remains;
}

std::shared_ptr<const IActionToken> MatchCursor::currentToken() const {
	return this->_current_token;
}

std::shared_ptr<const IPrimitiveWord> MatchCursor::currentWords() const {
	return this->_remains_word;
}

bool lib_syntax::MatchCursor::parse_stop() const {
	return !currentWords();
}

void MatchCursor::ErrsPack::addError(const QString& msg) {
	this->_error_collection.append(msg);
}

QList<QString> MatchCursor::ErrsPack::errors() const {
	return _error_collection;
}

uint64_t MatchCursor::ErrsPack::errorCount() const {
	return _error_collection.size();
}