New experimental LSPL parser

common/CMakeCommon.cmake

@@ -50,4 +50,4 @@ else(WIN32)
 
     SET(MY_TARGETLIB_PREFIX "lib")
     SET(MY_TARGETLIB_SUFFIX ".so")
-endif(WIN32)
+endif(WIN32)

core/CMakeLists.txt

@@ -2,7 +2,7 @@ cmake_minimum_required(VERSION 2.6)
 project(LSPL_CORE)
 
 include(../common/CMakeCommon.cmake)
-find_package(Boost COMPONENTS system)
+find_package(Boost COMPONENTS system timer)
 
 if(WIN32)
 else(WIN32)
@@ -30,6 +30,7 @@ set(LSPL_CORE_SOURCES
     src/main/lspl/morphology/AotMorphology.cpp
 
     src/main/lspl/patterns/Alternative.cpp
+    src/main/lspl/patterns/matchers/BaseComparator.cpp
     src/main/lspl/patterns/matchers/Context.cpp
     src/main/lspl/patterns/matchers/LoopMatcher.cpp
     src/main/lspl/patterns/matchers/Matcher.cpp
@@ -51,8 +52,13 @@ set(LSPL_CORE_SOURCES
     src/main/lspl/patterns/restrictions/DictionaryRestriction.cpp
     src/main/lspl/patterns/restrictions/AgreementRestriction.cpp
     src/main/lspl/patterns/restrictions/AndRestriction.cpp
+    src/main/lspl/patterns/restrictions/NotRestriction.cpp
+    src/main/lspl/patterns/restrictions/OrRestriction.cpp
+    src/main/lspl/patterns/restrictions/SharedRestriction.cpp
 
+    src/main/lspl/patterns/parsers/CharacterSets.cpp
     src/main/lspl/patterns/parsers/Functions.cpp
+
     src/main/lspl/patterns/Pattern.cpp
     src/main/lspl/patterns/PatternBuilder.cpp
     src/main/lspl/utils/Conversion.cpp
@@ -150,7 +156,7 @@ set(LSPL_BENCHMARK_SOURCES
 
 add_executable(lspl-benchmark ${LSPL_BENCHMARK_SOURCES})
 
-target_link_libraries(lspl-benchmark lspl)
+target_link_libraries(lspl-benchmark lspl ${Boost_TIMER_LIBRARY})
 
 # Flags for effective error parsing
 

core/src/main/lspl/base/BaseInternal.h

@@ -12,4 +12,9 @@
 #define LSPL_REFCOUNT_CLASS(className) \
 template<> lspl::uint lspl::base::IdentifiedObject<className>::count = 0;
 
+namespace lspl {
+	typedef unsigned int uint;
+	typedef unsigned long long uint64;
+}
+
 #endif//_LSPL_BASE_BASEINTERNAL_H_

core/src/main/lspl/morphology/Morphology.cpp

@@ -25,66 +25,77 @@ Morphology::~Morphology() {
 
 }
 
+struct AttributeInfo {
+	int bitsCount;
+	int bits[10];
+	const AttributeValue* values[10];
+};
+
+static AttributeInfo attributeInfos[10] = {
+	{
+		8,
+		{ rNominativ, rGenitiv, rDativ,	rAccusativ, rInstrumentalis, rLocativ, rVocativ, rIndeclinable },
+		{ &AttributeValue::NOMINATIVE, &AttributeValue::GENITIVE, &AttributeValue::DATIVE, &AttributeValue::ACCUSATIVE,	&AttributeValue::INSTRUMENTAL, &AttributeValue::PREPOSITIONAL, &AttributeValue::PREPOSITIONAL, &AttributeValue::UNINFLECTED }
+	},
+	{
+		2,
+		{ rSingular, rPlural },
+		{ &AttributeValue::SINGULAR, &AttributeValue::PLURAL }
+	},
+	{
+		3,
+		{ rMasculinum, rFeminum, rNeutrum },
+		{ &AttributeValue::MASCULINE, &AttributeValue::FEMININE, &AttributeValue::NEUTER }
+	},
+	{
+		2,
+		{ rComparative, rSuperlative },
+		{ &AttributeValue::COMPARATIVE, &AttributeValue::SUPERLATIVE }
+	},
+	{
+		3,
+		{ rPresentTense, rPastTense, rFutureTense },
+		{ &AttributeValue::PRESENT, &AttributeValue::PAST, &AttributeValue::FUTURE }
+	},
+	{
+		2,
+		{ rAnimative, rNonAnimative },
+		{ &AttributeValue::ANIMATE, &AttributeValue::INANIMATE }
+	},
+	{ // Форма
+		0,
+		{},
+		{}
+	},
+	{
+		1,
+		{ rImperative },
+		{ &AttributeValue::IMPERATIVE }
+	},
+	{
+		3,
+		{ rFirstPerson, rSecondPerson, rThirdPerson },
+		{ &AttributeValue::FIRST, &AttributeValue::SECOND, &AttributeValue::THIRD }
+	},
+	{
+		0,
+		{},
+		{}
+	}
+};
+
+AttributeKey Morphology::getAttributeKeyByValue(AttributeValue value) const {
+	if (value.type != AttributeType::INDEXED)
+		return AttributeKey::UNDEFINED;
+	for (uint i = 0; i < 10; ++i) {
+		for (uint j = 0; j < attributeInfos[i].bitsCount; ++j)
+			if ((*attributeInfos[i].values[j]) == value)
+				return AttributeKey(i + 1);
+	}
+	return AttributeKey::UNDEFINED;
+}
+
 AttributeValue Morphology::extractAttributeValue( uint64 attributeSet, AttributeKey key ) {
-	struct AttributeInfo {
-		int bitsCount;
-
-		int bits[10];
-		AttributeValue values[10];
-	};
-
-	static AttributeInfo attributeInfos[10] = {
-		{
-			8,
-			{ rNominativ, rGenitiv, rDativ,	rAccusativ, rInstrumentalis, rLocativ, rVocativ, rIndeclinable },
-			{ AttributeValue::NOMINATIVE, AttributeValue::GENITIVE, AttributeValue::DATIVE, AttributeValue::ACCUSATIVE,	AttributeValue::INSTRUMENTAL, AttributeValue::PREPOSITIONAL, AttributeValue::PREPOSITIONAL, AttributeValue::UNINFLECTED }
-		},
-		{
-			2,
-			{ rSingular, rPlural },
-			{ AttributeValue::SINGULAR, AttributeValue::PLURAL }
-		},
-		{
-			3,
-			{ rMasculinum, rFeminum, rNeutrum },
-			{ AttributeValue::MASCULINE, AttributeValue::FEMININE, AttributeValue::NEUTER }
-		},
-		{
-			2,
-			{ rComparative, rSuperlative },
-			{ AttributeValue::COMPARATIVE, AttributeValue::SUPERLATIVE }
-		},
-		{
-			3,
-			{ rPresentTense, rPastTense, rFutureTense },
-			{ AttributeValue::PRESENT, AttributeValue::PAST, AttributeValue::FUTURE }
-		},
-		{
-			2,
-			{ rAnimative, rNonAnimative },
-			{ AttributeValue::ANIMATE, AttributeValue::INANIMATE }
-		},
-		{ // Форма
-			0,
-			{},
-			{}
-		},
-		{
-			1,
-			{ rImperative },
-			{ AttributeValue::IMPERATIVE }
-		},
-		{
-			3,
-			{ rFirstPerson, rSecondPerson, rThirdPerson },
-			{ AttributeValue::FIRST, AttributeValue::SECOND, AttributeValue::THIRD }
-		},
-		{
-			0,
-			{},
-			{}
-		}
-	};
 
 	if ( key.index <= 0 || key.index > 10 ) // Если тип запрашиваемого аттрибута некорректен
 		return AttributeValue::UNDEFINED; // Возвращаем неопределенное значение
@@ -93,7 +104,7 @@ AttributeValue Morphology::extractAttributeValue( uint64 attributeSet, Attribute
 
 	for ( int i = 0; i < info.bitsCount; ++ i )
 		if ( attributeSet & ( ((uint64)1) << info.bits[ i ] ) )
-			return info.values[ i ];
+			return *(info.values[ i ]);
 
 	return AttributeValue::UNDEFINED;
 }

core/src/main/lspl/morphology/Morphology.h

@@ -57,6 +57,11 @@ class LSPL_EXPORT Morphology {
 		return result.release();
 	}
 
+	/**
+	 * По признаку получить его вид. Работает только для индексных (не составных и не строковых) признаков
+	 */
+	virtual text::attributes::AttributeKey getAttributeKeyByValue(text::attributes::AttributeValue value) const;
+
 	virtual std::string getAttributesString( uint64 attValues ) = 0;
 
 	std::string upcase( const std::string & str ) { return upcase( str.c_str() ); }

core/src/main/lspl/patterns/Alternative.cpp

@@ -107,7 +107,14 @@ void Alternative::dump( std::ostream & out, const std::string & tabs ) const {
 		getMatcher( i ).dump( out, tabs + "\t" );
 	}
 
-	out << "\n" << tabs << "] }";
+	out << "\n" << tabs << "], bindings = [\n";
+
+	for ( BindingMap::const_iterator it = bindings.begin(); it != bindings.end(); ++ it ) {
+		out << tabs << "\t" << it->first.getAbbrevation() << " : ";
+		it->second->dump(out);
+		out << '\n';
+	}
+	out << tabs << "] }";
 }
 
 } } // namespace lspl::patterns

core/src/main/lspl/patterns/Alternative.h

@@ -102,6 +102,8 @@ class LSPL_EXPORT Alternative : public matchers::MatcherContainer {
 
 public:
 
+	std::string transformSource;
+
 	/**
 	 * Добавить связывания аттрибутов
 	 */
@@ -152,8 +154,6 @@ class LSPL_EXPORT Alternative : public matchers::MatcherContainer {
 	 */
 	std::string source;
 
-	std::string transformSource;
-
 	/**
 	 * Список связываний аттрибутов шаблона
 	 */

core/src/main/lspl/patterns/Pattern.cpp

@@ -21,14 +21,14 @@ void Pattern::dump( std::ostream & out, const std::string & tabs ) const {
 	out << "Pattern{ name = " << name << ", alternatives = [\n\t" << tabs;
 
 	bool first = true;
-	for( boost::ptr_vector<Alternative>::const_iterator altIt = alternatives.begin(); altIt != alternatives.end(); ++ altIt ) {
+	for( const std::unique_ptr<Alternative> &alt : alternatives ) {
 		if ( first ) {
 			first = false;
 		} else {
 			out << ",\n\t" << tabs;
 		}
 
-		altIt->dump( out, tabs + "\t" );
+		alt->dump( out, tabs + "\t" );
 	}
 
 	out << "\n" << tabs << "] }";
@@ -44,22 +44,29 @@ Alternative & Pattern::newAlternative( const std::string & source )  {
 
 void Pattern::addAlternative( Alternative * alt ) {
 	alt->pattern = this;
-	alternatives.push_back( alt );
+	alternatives.emplace_back( alt );
 }
 
-void Pattern::addAlternatives( boost::ptr_vector<Alternative> & r ) {
-	for( Alternative & alt : r ) {
-		alt.pattern = this;
+void Pattern::addAlternatives( std::vector<std::unique_ptr<Alternative>> & r ) {
+	for( const std::unique_ptr<Alternative> & alt : r ) {
+		alt->pattern = this;
 	}
-	alternatives.transfer( alternatives.end(), r.begin(), r.end(), r );
+	int oldSize = alternatives.size();
+	alternatives.resize(oldSize + r.size());
+	std::move(r.begin(), r.end(), alternatives.begin() + oldSize);
+}
+
+void Pattern::mergePattern ( Pattern &other ) {
+	addAlternatives(other.alternatives);
+	other.alternatives.clear();
 }
 
 void Pattern::updateDependencies() {
 	dependencies.clear();
 
-	for( const Alternative & alt : alternatives ) {
+	for( const std::unique_ptr<Alternative> & alt : alternatives ) {
 
-		for( const Pattern * ptr : alt.getDependencies() ) {
+		for( const Pattern * ptr : alt->getDependencies() ) {
 			bool found = false;
 
 			for( const Pattern * dep : dependencies ) {
@@ -82,7 +89,7 @@ void Pattern::removeDuplicateAlternatives() {
 	while ( i < alternatives.size() ) {
 		bool found = false;
 		for ( uint j = 0; j < i; ++ j ) {
-			if ( alternatives[i].equals( alternatives[j] ) ) {
+			if ( alternatives[i]->equals( *alternatives[j] ) ) {
 				found = true;
 				break;
 			}
@@ -137,14 +144,14 @@ std::string Pattern::getSource() const {
 	std::string result = "";
 
 	bool first = true;
-	for( boost::ptr_vector<Alternative>::const_iterator altIt = alternatives.begin(); altIt != alternatives.end(); ++ altIt ) {
+	for( const std::unique_ptr<Alternative> &alt : alternatives ) {
 		if ( first ) {
 			first = false;
 		} else {
 			result += " | ";
 		}
 
-		result += altIt->getSource();
+		result += alt->getSource();
 	}
 
 	return result;

core/src/main/lspl/patterns/Pattern.h

@@ -2,11 +2,10 @@
 #define _LSPL_PATTERNS_PATTERN_H_
 
 #include <string>
+#include <memory>
 #include "Forward.h"
 #include "../text/Forward.h"
 
-#include <boost/ptr_container/ptr_vector.hpp>
-
 namespace lspl { namespace patterns {
 
 /**
@@ -44,7 +43,13 @@ class LSPL_EXPORT Pattern : public base::RefCountObject, public base::Identified
 	/**
 	 * Добавить альтернативы к шаблону
 	 */
-	void addAlternatives( boost::ptr_vector<Alternative> & r );
+	void addAlternatives( std::vector<std::unique_ptr<Alternative>> & r );
+
+	/**
+	 * Перенести все альтернативы шаблона other к текущему (подразумевается, что
+	 * они имеют одно имя)
+	 */
+	void mergePattern ( Pattern &other );
 
 	/**
 	 * Обновить список зависимостей альтернативы
@@ -86,7 +91,7 @@ class LSPL_EXPORT Pattern : public base::RefCountObject, public base::Identified
 	/**
 	 * Получить список внешних альтернатив шаблона
 	 */
-	const boost::ptr_vector<Alternative> & getAlternatives() const {
+	const std::vector<std::unique_ptr<Alternative>> & getAlternatives() const {
 		return alternatives;
 	}
 
@@ -104,7 +109,7 @@ class LSPL_EXPORT Pattern : public base::RefCountObject, public base::Identified
 	/**
 	 * Список альтернатив шаблона
 	 */
-	boost::ptr_vector<Alternative> alternatives;
+	std::vector<std::unique_ptr<Alternative>> alternatives;
 
     /**
      * Количество объектов в памяти