Compressed bit parallel index ranked bar notation

alib2algo/src/arbology/indexing/CompressedBitParallelIndexConstruction.cpp

@@ -17,6 +17,8 @@ indexes::arbology::CompressedBitParallelTreeIndex < std::ranked_symbol < Default
 auto compressedBitParallelIndexConstructionPrefixRankedTree = CompressedBitParallelIndexConstruction::RegistratorWrapper < indexes::arbology::CompressedBitParallelTreeIndex < std::ranked_symbol < DefaultSymbolType, DefaultRankType > >, tree::PrefixRankedTree < > > ( CompressedBitParallelIndexConstruction::construct );
+auto compressedBitParallelIndexConstructionPrefixRankedBarTree = CompressedBitParallelIndexConstruction::RegistratorWrapper < indexes::arbology::CompressedBitParallelTreeIndex < std::ranked_symbol < DefaultSymbolType, DefaultRankType > >, tree::PrefixRankedBarTree < > > ( CompressedBitParallelIndexConstruction::construct );
+
 } /* namespace indexing */
 } /* namespace arbology */

alib2algo/src/arbology/indexing/CompressedBitParallelIndexConstruction.h

@@ -11,6 +11,7 @@
 #include <indexes/arbology/CompressedBitParallelTreeIndex.h>
 #include <tree/RankedTreeWrapper.h>
 #include <tree/ranked/PrefixRankedTree.h>
+#include <tree/ranked/PrefixRankedBarTree.h>
 #include <core/multipleDispatch.hpp>
 #include <exception/CommonException.h>
 #include <tree/properties/SubtreeJumpTable.h>
@@ -35,6 +36,9 @@ public:
 	template < class SymbolType, class RankType >
 	static indexes::arbology::CompressedBitParallelTreeIndex < std::ranked_symbol < SymbolType, RankType > > construct ( const tree::PrefixRankedTree < SymbolType, RankType > & tree );
+
+	template < class SymbolType, class RankType >
+	static indexes::arbology::CompressedBitParallelTreeIndex < std::ranked_symbol < SymbolType, RankType > > construct ( const tree::PrefixRankedBarTree < SymbolType, RankType > & tree );
 };
 template < class SymbolType, class RankType >
@@ -50,6 +54,19 @@ indexes::arbology::CompressedBitParallelTreeIndex < std::ranked_symbol < SymbolT
 	return indexes::arbology::CompressedBitParallelTreeIndex < std::ranked_symbol < SymbolType, RankType > > ( w.getAlphabet ( ), res, tree::properties::SubtreeJumpTable::compute ( w ) );
 }
+template < class SymbolType, class RankType >
+indexes::arbology::CompressedBitParallelTreeIndex < std::ranked_symbol < SymbolType, RankType > > CompressedBitParallelIndexConstruction::construct ( const tree::PrefixRankedBarTree < SymbolType, RankType > & w ) {
+	std::map < std::ranked_symbol < SymbolType, RankType >, common::SparseBoolVector > res;
+
+	for ( const std::ranked_symbol < SymbolType, RankType > & symbol : w.getAlphabet ( ) )
+		res [ symbol ].resize ( w.getContent ( ).size ( ) );
+
+	for ( unsigned i = 0; i < w.getContent ( ).size ( ); ++i )
+		res [ w.getContent ( ) [ i ] ] [ i ] = true;
+
+	return indexes::arbology::CompressedBitParallelTreeIndex < std::ranked_symbol < SymbolType, RankType > > ( w.getAlphabet ( ), res, tree::properties::SubtreeJumpTable::compute ( w ) );
+}
+
 } /* namespace indexing */
 } /* namespace arbology */

alib2algo/src/arbology/query/CompressedBitParallelismPatterns.cpp

@@ -18,6 +18,7 @@ std::set < unsigned > CompressedBitParallelismPatterns::query ( const indexes::a
 }
 auto CompressedBitParallelismPatternsPrefixRankedPattern = CompressedBitParallelismPatterns::RegistratorWrapper < std::set < unsigned >, tree::PrefixRankedPattern < > > ( CompressedBitParallelismPatterns::query );
+auto CompressedBitParallelismPatternsPrefixRankedBarPattern = CompressedBitParallelismPatterns::RegistratorWrapper < std::set < unsigned >, tree::PrefixRankedBarPattern < > > ( CompressedBitParallelismPatterns::query );
 } /* namespace query */

alib2algo/src/arbology/query/CompressedBitParallelismPatterns.h

@@ -10,7 +10,8 @@
 #include <indexes/arbology/CompressedBitParallelTreeIndex.h>
 #include <tree/RankedTreeWrapper.h>
-#include <tree/ranked/PrefixRankedTree.h>
+#include <tree/ranked/PrefixRankedPattern.h>
+#include <tree/ranked/PrefixRankedBarPattern.h>
 #include <core/multipleDispatch.hpp>
 #include <global/GlobalData.h>
@@ -36,6 +37,9 @@ public:
 	template < class SymbolType, class RankType >
 	static std::set < unsigned > query ( const indexes::arbology::CompressedBitParallelTreeIndex < std::ranked_symbol < SymbolType, RankType > > & compressedBitParallelTreeIndex, const tree::PrefixRankedPattern < SymbolType, RankType > & pattern );
+
+	template < class SymbolType, class RankType >
+	static std::set < unsigned > query ( const indexes::arbology::CompressedBitParallelTreeIndex < std::ranked_symbol < SymbolType, RankType > > & compressedBitParallelTreeIndex, const tree::PrefixRankedBarPattern < SymbolType, RankType > & pattern );
 };
 template < class SymbolType, class RankType >
@@ -76,6 +80,46 @@ std::set < unsigned > CompressedBitParallelismPatterns::query ( const indexes::a
 	return res;
 }
+template < class SymbolType, class RankType >
+std::set < unsigned > CompressedBitParallelismPatterns::query ( const indexes::arbology::CompressedBitParallelTreeIndex < std::ranked_symbol < SymbolType, RankType > > & compressedBitParallelIndex, const tree::PrefixRankedBarPattern < SymbolType, RankType > & pattern ) {
+	auto symbolIter = pattern.getContent ( ).begin ( );
+
+	typename std::map < std::ranked_symbol < SymbolType, RankType >, common::SparseBoolVector >::const_iterator symbolVectorIter = compressedBitParallelIndex.getData ( ).find ( * symbolIter );
+
+	if ( symbolVectorIter == compressedBitParallelIndex.getData ( ).end ( ) )
+		return { };
+
+	common::SparseBoolVector indexVector = symbolVectorIter->second;
+
+	for ( ++ symbolIter; symbolIter != pattern.getContent ( ).end ( ); ++ symbolIter ) {
+		if ( * symbolIter == pattern.getSubtreeWildcard ( ) ) {
+			common::SparseBoolVector newVector;
+			newVector.resize ( indexVector.size ( ) );
+
+			for ( unsigned i : ( indexVector << 1 ) )
+				newVector [ compressedBitParallelIndex.getJumps ( ) [ i ] - 1 ] = true;
+
+			indexVector = newVector;
+
+			++ symbolIter;
+		} else {
+			symbolVectorIter = compressedBitParallelIndex.getData ( ).find ( * symbolIter );
+
+			if ( symbolVectorIter == compressedBitParallelIndex.getData ( ).end ( ) )
+				return { };
+
+			indexVector = ( indexVector << 1 ) & symbolVectorIter->second;
+		}
+	}
+
+	std::set < unsigned > res;
+
+	for ( unsigned i : indexVector )
+		res.insert ( i + 1 );
+
+	return res;
+}
+
 } /* namespace query */
 } /* namespace arbology */

tests.aarbology.sh

@@ -412,6 +412,8 @@ function runTestNonlinearPatternEnds {
 runTestPattern "Exact Pattern Matching Using Full And Linear Index (PrefixRanked)" "./aarbology2 -a fullAndLinearIndex -s <(./acast2 -t PrefixRankedTree -i \"\$SUBJECT_FILE\" ) | ./aquery2 -q fullAndLinearIndexPatterns -i - -p <( ./acast2 -t PrefixRankedPattern -i \"\$PATTERN_FILE\" ) | ./astat2 -p size"
+runTestPattern "Exact Pattern Matching Using Compressed Bit Vectors (PrefixRankedBar)" "./aarbology2 -a compressedBitParallelIndex -s <(./acast2 -t PrefixRankedBarTree -i \"\$SUBJECT_FILE\" ) | ./aquery2 -q compressedBitParallelismPatterns -i - -p <( ./acast2 -t PrefixRankedBarPattern -i \"\$PATTERN_FILE\" ) | ./astat2 -p size"
+
 runTestPatternEnds "Exact Pattern Matching Using Compressed Bit Vectors (PrefixRanked)" "./aarbology2 -a compressedBitParallelIndex -s <(./acast2 -t PrefixRankedTree -i \"\$SUBJECT_FILE\" ) | ./aquery2 -q compressedBitParallelismPatterns -i - -p <( ./acast2 -t PrefixRankedPattern -i \"\$PATTERN_FILE\" ) | ./astat2 -p size"
 runTestPatternEnds "Exact Pattern Matching Automaton (PrefixRanked)" "./aarbology2 -a exactPatternMatchingAutomaton -p <(./acast2 -t PrefixRankedPattern -i <(./aaccess2 --tree alphabet -o add -i \"\$PATTERN_FILE\" -a <( ./aaccess2 --tree alphabet -o get -i \"\$SUBJECT_FILE\" ) ) ) | ./adeterminize2 | ./arun2 -t occurrences -a - -i <( ./acast2 -t PrefixRankedTree -i \"\$SUBJECT_FILE\" | ./acast2 -t LinearString ) | ./astat2 -p size"