diff --git a/alib2algo/src/arbology/query/FullAndLinearIndexPatterns.cpp b/alib2algo/src/arbology/query/FullAndLinearIndexPatterns.cpp
new file mode 100644
index 0000000000000000000000000000000000000000..736457627943350fd455062d8d695b2e4af5b4d6
--- /dev/null
+++ b/alib2algo/src/arbology/query/FullAndLinearIndexPatterns.cpp
@@ -0,0 +1,24 @@
+/*
+ * FullAndLinearIndexPatterns.cpp
+ *
+ * Created on: 2. 1. 2017
+ * Author: Jan Travnicek
+ */
+
+#include "FullAndLinearIndexPatterns.h"
+
+#include <tree/ranked/PrefixRankedPattern.h>
+
+namespace arbology {
+
+namespace query {
+
+std::set < unsigned > FullAndLinearIndexPatterns::query ( const indexes::arbology::FullAndLinearIndex < std::ranked_symbol < DefaultSymbolType, DefaultRankType > > & fullAndLinearIndex, const tree::RankedTreeWrapper & tree ) {
+ return dispatch ( fullAndLinearIndex, tree.getData ( ) );
+}
+
+auto fullAndLinearIndexPatternsPrefixRankedPattern = FullAndLinearIndexPatterns::RegistratorWrapper < std::set < unsigned >, tree::PrefixRankedPattern < > > ( FullAndLinearIndexPatterns::query );
+
+} /* namespace query */
+
+} /* namespace arbology */
diff --git a/alib2algo/src/arbology/query/FullAndLinearIndexPatterns.h b/alib2algo/src/arbology/query/FullAndLinearIndexPatterns.h
new file mode 100644
index 0000000000000000000000000000000000000000..58a507d2b574de553848fb7d7dbaaa86aaa56a19
--- /dev/null
+++ b/alib2algo/src/arbology/query/FullAndLinearIndexPatterns.h
@@ -0,0 +1,105 @@
+/*
+ * FullAndLinearIndexPatterns.h
+ *
+ * Created on: 2. 1. 2017
+ * Author: Jan Travnicek
+ */
+
+#ifndef FULL_AND_LINEAR_INDEX_PATTERNS_H_
+#define FULL_AND_LINEAR_INDEX_PATTERNS_H_
+
+#include <indexes/arbology/FullAndLinearIndex.h>
+#include <tree/RankedTreeWrapper.h>
+#include <tree/ranked/PrefixRankedTree.h>
+#include <core/multipleDispatch.hpp>
+#include <global/GlobalData.h>
+
+#include <stringology/query/PositionHeapFactors.h>
+
+namespace arbology {
+
+namespace query {
+
+/**
+ * Query full and linear index for given tree.
+ *
+ */
+
+class FullAndLinearIndexPatterns : public std::SingleDispatchFirstStaticParam < FullAndLinearIndexPatterns, std::set < unsigned >, const indexes::arbology::FullAndLinearIndex < std::ranked_symbol < DefaultSymbolType, DefaultRankType > > &, const tree::RankedTreeBase & > {
+ template < class SymbolType, class RankType >
+ static std::vector < std::pair < unsigned, unsigned > > FindOccurrences ( const indexes::stringology::PositionHeap < std::ranked_symbol < SymbolType, RankType > > & stringIndex, const std::vector < std::ranked_symbol < SymbolType, RankType > > & string ) {
+ std::vector < std::pair < unsigned, unsigned > > res;
+ for ( unsigned occurrence : stringology::query::PositionHeapFactors::query ( stringIndex, string::LinearString < std::ranked_symbol < SymbolType, RankType > > ( string ) ) ) {
+ res.push_back ( std::make_pair ( occurrence, occurrence + string.size ( ) ) );
+ }
+ return res;
+ }
+
+ static std::vector < std::pair < unsigned, unsigned > > MergeOccurrences ( const std::vector < std::pair < unsigned, unsigned > > & prevOcc, const std::vector < std::pair < unsigned, unsigned > > & subOcc, std::vector < unsigned > & rev ) {
+ std::vector < std::pair < unsigned, unsigned > > res;
+
+ for ( const std::pair < unsigned, unsigned > & occurrence : prevOcc ) {
+ rev [ occurrence.second ] = occurrence.first;
+ }
+
+ for ( const std::pair < unsigned, unsigned > & subOccurrence : subOcc ) {
+ if ( rev [ subOccurrence.first ] != ( unsigned ) -1 )
+ res.push_back ( std::make_pair ( rev [ subOccurrence.first ], subOccurrence.second ) );
+ }
+
+ for ( const std::pair < unsigned, unsigned > & occurrence : prevOcc ) {
+ rev [ occurrence.second ] = ( unsigned ) -1;
+ }
+
+ return res;
+ }
+public:
+ /**
+ * Query a suffix trie
+ * @param suffix trie to query
+ * @param tree tree to query by
+ * @return occurences of factors
+ */
+ static std::set < unsigned > query ( const indexes::arbology::FullAndLinearIndex < std::ranked_symbol < DefaultSymbolType, DefaultRankType > > & fullAndLinearIndex, const tree::RankedTreeWrapper & pattern );
+
+ template < class SymbolType, class RankType >
+ static std::set < unsigned > query ( const indexes::arbology::FullAndLinearIndex < std::ranked_symbol < SymbolType, RankType > > & compressedBitParallelTreeIndex, const tree::PrefixRankedPattern < SymbolType, RankType > & pattern );
+};
+
+template < class SymbolType, class RankType >
+std::set < unsigned > FullAndLinearIndexPatterns::query ( const indexes::arbology::FullAndLinearIndex < std::ranked_symbol < SymbolType, RankType > > & fullAndLinearIndex, const tree::PrefixRankedPattern < SymbolType, RankType > & pattern ) {
+ std::vector < unsigned > rev ( fullAndLinearIndex.getString ( ).size ( ), ( unsigned ) -1 );
+
+ std::vector < std::vector < std::ranked_symbol < SymbolType, RankType > > > treePatternParts;
+ treePatternParts.push_back ( std::vector < std::ranked_symbol < SymbolType, RankType > > ( ) );
+ for ( const std::ranked_symbol < SymbolType, RankType > & symbol : pattern.getContent ( ) ) {
+ if ( pattern.getSubtreeWildcard ( ) == symbol ) {
+ treePatternParts.push_back ( std::vector < std::ranked_symbol < SymbolType, RankType > > ( ) );
+ } else {
+ treePatternParts.back ( ).push_back ( symbol );
+ }
+ }
+
+ std::vector < std::pair < unsigned, unsigned > > prevOcc = FindOccurrences ( fullAndLinearIndex.getStringIndex ( ) , treePatternParts [ 0 ] );
+
+ for ( unsigned i = 1; i < treePatternParts.size ( ); ++ i ) {
+ for ( std::pair < unsigned, unsigned > & occurrence : prevOcc )
+ occurrence.second = fullAndLinearIndex.getJumps ( ) [ occurrence.second ];
+
+ if ( ! treePatternParts [ i ].empty ( ) )
+ prevOcc = MergeOccurrences ( prevOcc, FindOccurrences ( fullAndLinearIndex.getStringIndex ( ), treePatternParts [ i ] ), rev );
+ }
+
+ std::set < unsigned > res;
+ for ( const std::pair < unsigned, unsigned > & occurrence : prevOcc ) {
+ res.insert ( occurrence.first );
+ }
+
+ return res;
+}
+
+} /* namespace query */
+
+} /* namespace arbology */
+
+#endif /* FULL_AND_LINEAR_INDEX_PATTERNS_H_ */
diff --git a/aquery2/src/aquery.cpp b/aquery2/src/aquery.cpp
index b5ec252f630917c719ec1614777c8c5f3a4d1ea2..e825f1a46ea1e05f1418ddc06fc831fa2efead27 100644
--- a/aquery2/src/aquery.cpp
+++ b/aquery2/src/aquery.cpp
@@ -21,6 +21,7 @@
#include <stringology/query/BitParallelismFactors.h>
#include <stringology/query/CompressedBitParallelismFactors.h>
#include <arbology/query/CompressedBitParallelismPatterns.h>
+#include <arbology/query/FullAndLinearIndexPatterns.h>
int main ( int argc, char * argv[] ) {
try {
@@ -36,6 +37,7 @@ int main ( int argc, char * argv[] ) {
allowed.push_back ( "bitParallelismFactors" );
allowed.push_back ( "compressedBitParallelismFactors" );
allowed.push_back ( "compressedBitParallelismPatterns" );
+ allowed.push_back ( "fullAndLinearIndexPatterns" );
TCLAP::ValuesConstraint < std::string > allowedVals ( allowed );
TCLAP::ValueArg < std::string > query ( "q", "query", "Query index", false, "exactFactorMatch", & allowedVals );
@@ -140,6 +142,19 @@ int main ( int argc, char * argv[] ) {
measurements::end ( );
measurements::start ( "Output write", measurements::Type::AUXILIARY );
+ alib::XmlDataFactory::toStdout ( res );
+ } else if ( query.getValue ( ) == "fullAndLinearIndexPatterns" ) {
+ indexes::arbology::FullAndLinearIndex < std::ranked_symbol < DefaultSymbolType, DefaultRankType > > fullAndLinearIndex = alib::XmlDataFactory::fromTokens < indexes::arbology::FullAndLinearIndex < std::ranked_symbol < DefaultSymbolType, DefaultRankType > > > ( sax::FromXMLParserHelper::parseInput ( indexInput ) );
+ tree::RankedTreeWrapper pattern = alib::XmlDataFactory::fromTokens < tree::RankedTreeWrapper > ( std::move ( sax::FromXMLParserHelper::parseInput(true, patternInput).front ( ) ) );
+
+ measurements::end ( );
+ measurements::start ( "Algorithm", measurements::Type::MAIN );
+
+ std::set < unsigned > res = arbology::query::FullAndLinearIndexPatterns::query ( fullAndLinearIndex, pattern );
+
+ measurements::end ( );
+ measurements::start ( "Output write", measurements::Type::AUXILIARY );
+
alib::XmlDataFactory::toStdout ( res );
} else {
throw exception::CommonException ( "Invalid algorithm" );
diff --git a/tests.aarbology.sh b/tests.aarbology.sh
index 4188340f3c1356297d5c62d9dc5a7a6d2e695bad..75bb84263f9065d1a0c96d5bd34970a542f23831 100755
--- a/tests.aarbology.sh
+++ b/tests.aarbology.sh
@@ -410,6 +410,8 @@ function runTestNonlinearPatternEnds {
clearResults
}
+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"
+
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"