diff --git a/aarbology2/src/aarbology.cpp b/aarbology2/src/aarbology.cpp
index 6a7fef4fca30be83c2b99e7f38091d75be407aa1..64f74261fa7e82505b43efc5c49b79b06b54b2d9 100644
--- a/aarbology2/src/aarbology.cpp
+++ b/aarbology2/src/aarbology.cpp
@@ -20,6 +20,7 @@
#include <arbology/exact/BoyerMooreHorspool.h>
#include <arbology/exact/ReversedBoyerMooreHorspool.h>
#include <arbology/exact/KnuthMorrisPratt.h>
+#include <arbology/exact/DeadZoneUsingBadCharacterShiftAndBorderArray.h>
#include <arbology/exact/ExactSubtreeMatchingAutomaton.h>
#include <arbology/exact/ExactPatternMatchingAutomaton.h>
#include <arbology/exact/ExactSubtreeAutomaton.h>
@@ -37,6 +38,7 @@ int main ( int argc, char * argv[] ) {
allowed.push_back ( "boyerMooreHorspool" );
allowed.push_back ( "reversedBoyerMooreHorspool" );
allowed.push_back ( "knuthMorrisPratt" );
+ allowed.push_back ( "deadZoneUsingBadCharacterShiftAndBorderArray" );
allowed.push_back ( "exactSubtreeMatchingAutomaton" );
allowed.push_back ( "exactPatternMatchingAutomaton" );
allowed.push_back ( "exactSubtreeAutomaton" );
@@ -75,6 +77,8 @@ int main ( int argc, char * argv[] ) {
needPattern = needSubject = 1;
} else if ( algorithm.getValue ( ) == "exactSubtreeMatchingAutomaton" ) {
needPattern = 1;
+ } else if ( algorithm.getValue ( ) == "deadZoneUsingBadCharacterShiftAndBorderArray" ) {
+ needPattern = needSubject = 1;
} else {
}
@@ -193,6 +197,21 @@ int main ( int argc, char * argv[] ) {
std::chrono::measurements::end ( );
std::chrono::measurements::start ( "Output write", std::chrono::measurements::Type::AUXILARY );
+ alib::XmlDataFactory::toStdout ( res );
+ } else if ( algorithm.getValue ( ) == "deadZoneUsingBadCharacterShiftAndBorderArray" ) {
+ tree::Tree subject = alib::XmlDataFactory::fromTokens < tree::Tree > ( subjectTokens.front ( ) );
+ tree::Tree pattern = alib::XmlDataFactory::fromTokens < tree::Tree > ( patternTokens.front ( ) );
+
+ std::chrono::measurements::end ( );
+ std::chrono::measurements::start ( "Algorithm", std::chrono::measurements::Type::MAIN );
+
+ std::set < unsigned > res = arbology::exact::DeadZoneUsingBadCharacterShiftAndBorderArray::match ( subject, pattern );
+ if( ends.isSet ( ) )
+ res = arbology::transform::BeginToEndIndex::transform(subject, res);
+
+ std::chrono::measurements::end ( );
+ std::chrono::measurements::start ( "Output write", std::chrono::measurements::Type::AUXILARY );
+
alib::XmlDataFactory::toStdout ( res );
} else if ( algorithm.getValue ( ) == "exactSubtreeMatchingAutomaton" ) {
tree::Tree pattern = alib::XmlDataFactory::fromTokens < tree::Tree > ( patternTokens.front ( ) );
diff --git a/alib2algo/src/arbology/exact/DeadZoneUsingBadCharacterShiftAndBorderArray.cpp b/alib2algo/src/arbology/exact/DeadZoneUsingBadCharacterShiftAndBorderArray.cpp
new file mode 100644
index 0000000000000000000000000000000000000000..6592f0b27da2b1cc81017903184c6bfb9875c0ad
--- /dev/null
+++ b/alib2algo/src/arbology/exact/DeadZoneUsingBadCharacterShiftAndBorderArray.cpp
@@ -0,0 +1,81 @@
+/*
+ * DeadZoneUsingBadCharacterShiftAndBorderArray.cpp
+ *
+ * Created on: 5. 11. 2014
+ * Author: Jan Travnicek
+ */
+
+#include "DeadZoneUsingBadCharacterShiftAndBorderArray.h"
+#include "ReversedBadCharacterShiftTable.h"
+#include "BorderArrayNaive.h"
+#include "SubtreeJumpTable.h"
+
+#include <exception/AlibException.h>
+#include <tree/Tree.h>
+#include <tree/ranked/PrefixRankedBarTree.h>
+#include <tree/ranked/PrefixRankedBarPattern.h>
+#include <alphabet/RankedSymbol.h>
+
+#include <map>
+
+namespace arbology {
+
+namespace exact {
+
+std::set < unsigned > DeadZoneUsingBadCharacterShiftAndBorderArray::match ( const tree::Tree & subject, const tree::Tree & pattern ) {
+ return getInstance ( ).dispatch ( subject.getData ( ), pattern.getData ( ) );
+}
+
+std::set < unsigned > DeadZoneUsingBadCharacterShiftAndBorderArray::match ( const tree::PrefixRankedBarTree & subject, const tree::PrefixRankedBarTree & pattern ) {
+ return match ( subject, tree::PrefixRankedBarPattern ( pattern ) );
+}
+
+auto DeadZoneUsingBadCharacterShiftAndBorderArrayPrefixRankedBarTreePrefixRankedBarTree = DeadZoneUsingBadCharacterShiftAndBorderArray::RegistratorWrapper < std::set < unsigned >, tree::PrefixRankedBarTree, tree::PrefixRankedBarTree > ( DeadZoneUsingBadCharacterShiftAndBorderArray::getInstance ( ), DeadZoneUsingBadCharacterShiftAndBorderArray::match );
+
+std::set < unsigned > DeadZoneUsingBadCharacterShiftAndBorderArray::match ( const tree::PrefixRankedBarTree & subject, const tree::PrefixRankedBarPattern & pattern ) {
+ std::set < unsigned > occ;
+ std::map < alphabet::RankedSymbol, size_t > bbcs = ReversedBadCharacterShiftTable::bcs ( pattern ); // NOTE: the subjects alphabet must be a subset or equal to the pattern
+ std::vector < size_t > fba = BorderArrayNaive::ba ( pattern );
+ std::vector < int > subjectSubtreeJumpTable = SubtreeJumpTable::compute ( subject );
+
+ match_rec ( occ, subject, pattern, fba, bbcs, subjectSubtreeJumpTable, 0, subject.getContent ( ).size ( ) - pattern.getContent ( ).size ( ) + 1 );
+ return occ;
+}
+
+void DeadZoneUsingBadCharacterShiftAndBorderArray::match_rec ( std::set < unsigned > & occ, const tree::PrefixRankedBarTree & subject, const tree::PrefixRankedBarPattern & pattern, std::vector < size_t > & fba, std::map < alphabet::RankedSymbol, size_t > & bbcs, std::vector < int > & subjectSubtreeJumpTable, int low, int high ) {
+ if ( low >= high ) return;
+
+ int i = ( low + high ) / 2;
+
+ // index to the pattern
+ unsigned j = 0;
+
+ // offset to the subject
+ unsigned offset = i;
+
+ while ( ( j < pattern.getContent ( ).size ( ) ) && ( offset < subject.getContent ( ).size ( ) ) ) {
+ if ( subject.getContent ( )[offset] == pattern.getContent ( )[j] ) {
+ // match of symbol
+ offset++;
+ j++;
+ } else if ( ( pattern.getContent ( )[j] == pattern.getSubtreeWildcard ( ) ) && ( subject.getContent ( )[offset].getSymbol ( ) != pattern.getBarSymbol ( ) ) ) {
+ // match of variable with subtree
+ offset = subjectSubtreeJumpTable[offset];
+ j += 2;
+ } else {
+ break;
+ }
+ }
+
+ // match was found
+ if ( j >= pattern.getContent ( ).size ( ) ) occ.insert ( i );
+
+ match_rec ( occ, subject, pattern, fba, bbcs, subjectSubtreeJumpTable, low, i - bbcs[subject.getContent ( )[i]] + 1 );
+ match_rec ( occ, subject, pattern, fba, bbcs, subjectSubtreeJumpTable, i + j - fba[j], high );
+}
+
+auto DeadZoneUsingBadCharacterShiftAndBorderArrayPrefixRankedBarTreePrefixRankedBarPattern = DeadZoneUsingBadCharacterShiftAndBorderArray::RegistratorWrapper < std::set < unsigned >, tree::PrefixRankedBarTree, tree::PrefixRankedBarPattern > ( DeadZoneUsingBadCharacterShiftAndBorderArray::getInstance ( ), DeadZoneUsingBadCharacterShiftAndBorderArray::match );
+
+} /* namespace exact */
+
+} /* namespace arbology */
diff --git a/alib2algo/src/arbology/exact/DeadZoneUsingBadCharacterShiftAndBorderArray.h b/alib2algo/src/arbology/exact/DeadZoneUsingBadCharacterShiftAndBorderArray.h
new file mode 100644
index 0000000000000000000000000000000000000000..50a4354da861d009141b7a427f37e9ed4e9ce58c
--- /dev/null
+++ b/alib2algo/src/arbology/exact/DeadZoneUsingBadCharacterShiftAndBorderArray.h
@@ -0,0 +1,48 @@
+/*
+ * DeadZoneUsingBadCharacterShiftAndBorderArray.h
+ *
+ * Created on: 5. 11. 2014
+ * Author: Jan Travnicek
+ */
+
+#ifndef _DEAD_ZONE_USING_BAD_CHARACTER_SHIFT_AND_BORDER_ARRAY_H_
+#define _DEAD_ZONE_USING_BAD_CHARACTER_SHIFT_AND_BORDER_ARRAY_H_
+
+#include <set>
+#include <vector>
+#include <common/multipleDispatch.hpp>
+#include <tree/TreeFeatures.h>
+#include <alphabet/RankedSymbol.h>
+
+namespace arbology {
+
+namespace exact {
+
+/**
+ * Implementation of DeadZone matching using bad character shift as shifting method on one direction and border array on the other
+ */
+class DeadZoneUsingBadCharacterShiftAndBorderArray : public std::DoubleDispatch < std::set < unsigned >, tree::TreeBase, tree::TreeBase > {
+public:
+ /**
+ * Search for pattern in linear string.
+ * @return set set of occurences
+ */
+ static std::set < unsigned > match ( const tree::Tree & subject, const tree::Tree & pattern );
+
+ static std::set < unsigned > match ( const tree::PrefixRankedBarTree & subject, const tree::PrefixRankedBarTree & pattern );
+ static std::set < unsigned > match ( const tree::PrefixRankedBarTree & subject, const tree::PrefixRankedBarPattern & pattern );
+ static void match_rec ( std::set < unsigned > & occ, const tree::PrefixRankedBarTree & subject, const tree::PrefixRankedBarPattern & pattern, std::vector < size_t > & fba, std::map < alphabet::RankedSymbol, size_t > & bbcs, std::vector < int > & subjectSubtreeJumpTable, int low, int high );
+
+ static DeadZoneUsingBadCharacterShiftAndBorderArray & getInstance ( ) {
+ static DeadZoneUsingBadCharacterShiftAndBorderArray res;
+
+ return res;
+ }
+
+};
+
+} /* namespace exact */
+
+} /* namespace arbology */
+
+#endif /* _DEAD_ZONE_USING_BAD_CHARACTER_SHIFT_AND_BORDER_ARRAY_H_ */
diff --git a/tests.aarbology.sh b/tests.aarbology.sh
index 38be0dae539920c9d0eecd195f2ac166a2e56850..3912b4f00763590473a1c00431de1a47a9a036c2 100755
--- a/tests.aarbology.sh
+++ b/tests.aarbology.sh
@@ -257,6 +257,8 @@ function runTestPatternEnds {
outputResults
}
+runTestPattern "Exact Dead Zone Using Bad Character Shift And Border Array (Pattern PrefixRankedBar)" "./aarbology2 -a deadZoneUsingBadCharacterShiftAndBorderArray -s <( ./acast2 -t PrefixRankedBarTree -i \"\$SUBJECT_FILE\" ) -p <( ./acast2 -t PrefixRankedBarPattern -i <(./alphabetManip2 -o add -i \"\$PATTERN_FILE\" -a <(./alphabetManip2 -o get -i \"\$SUBJECT_FILE\"))) | ./astat2 -p size --set"
+
runTestPatternEnds "Exact Pattern Matching Automaton (PrefixRanked)" "./aarbology2 -a exactPatternMatchingAutomaton -p <(./acast2 -t PrefixRankedPattern -i <(./alphabetManip2 -o add -i \"\$PATTERN_FILE\" -a <( ./alphabetManip2 -o get -i \"\$SUBJECT_FILE\" ) ) ) | ./adeterminize2 | ./arun2 -t occurrences -a - -i <( ./acast2 -t PrefixRankedTree -i \"\$SUBJECT_FILE\" | ./acast2 -t LinearString ) | ./astat2 -p size --set"
RAND_SIZE_SUBJECT=100