diff --git a/alib2algo/src/arbology/exact/QuickSearch.h b/alib2algo/src/arbology/exact/QuickSearch.h
index 3fde07a6308f4a59fb1c9722dec42e63714ae785..4d7470110ef60d50844ae94b09b14850c5d73c36 100644
--- a/alib2algo/src/arbology/exact/QuickSearch.h
+++ b/alib2algo/src/arbology/exact/QuickSearch.h
@@ -16,6 +16,7 @@
#include <tree/properties/QuickSearchBadCharacterShiftTable.h>
#include <tree/properties/SubtreeJumpTable.h>
#include <tree/properties/ExactSubtreeRepeatsNaive.h>
+#include <tree/exact/BackwardOccurrenceTest.h>
#include <tree/ranked/PrefixRankedBarTree.h>
#include <tree/ranked/PrefixRankedBarPattern.h>
@@ -50,7 +51,7 @@ ext::set < unsigned > QuickSearch::match ( const tree::PrefixRankedBarTree < Sym
return match ( subject, tree::PrefixRankedBarPattern < SymbolType, RankType > ( pattern ) );
}
-template < class SymbolType, class RankType > // CURRENTLY WORKING HERE
+template < class SymbolType, class RankType >
ext::set < unsigned > QuickSearch::match ( const tree::PrefixRankedBarTree < SymbolType, RankType > & subject, const tree::PrefixRankedBarPattern < SymbolType, RankType > & pattern ) {
ext::set < unsigned > occ;
ext::map < common::ranked_symbol < SymbolType, RankType >, size_t > bcs = tree::properties::QuickSearchBadCharacterShiftTable::bcs ( pattern ); // NOTE: the subjects alphabet must be a subset or equal to the pattern
@@ -61,36 +62,17 @@ ext::set < unsigned > QuickSearch::match ( const tree::PrefixRankedBarTree < Sym
// main loop of the algorithm over all possible indexes where the pattern can start
while ( i + pattern.getContent ( ).size ( ) <= subject.getContent ( ).size ( ) ) {
+ // pair j and offset
+ ext::pair < int, int > jOffset = tree::exact::BackwardOccurrenceTest::occurrence ( subject, subjectSubtreeJumpTable, pattern, i );
- // index to the pattern
- int j = pattern.getContent ( ).size ( ) - 1;
-
- // offset to the subject
- int offset = i + j;
-
- while ( ( j >= 0 ) && ( offset >= 0 ) ) {
- if ( subject.getContent ( )[offset] == pattern.getContent ( )[j] ) {
- // match of symbol
- offset = offset - 1;
- j = j - 1;
- } else if ( ( pattern.getContent ( )[j] == pattern.getVariablesBar ( ) ) /* && ( pattern.getBars ( ).count ( subject.getContent ( )[offset] ) ) */ ) {
- // match of variable with subtree
- offset = subjectSubtreeJumpTable[offset];
- j = j - 2;
- } else {
- break;
- }
- }
-
- // match was found
- if ( j == -1 ) occ.insert ( offset + 1);
+ // match was found
+ if ( jOffset.first == -1 ) occ.insert ( jOffset.second + 1);
if ( i + pattern.getContent ( ).size ( ) == subject.getContent ( ).size ( ) ) {
break;
}
// shift heuristics
- //std::cout << "At: " << i << ", shifting by: " << bcs[subject.getContent ( )[i + pattern.getContent ( ).size ( )]] << ", according to symb at: " << i + pattern.getContent ( ).size ( ) << "!" << std::endl;
i += bcs[subject.getContent ( )[i + pattern.getContent ( ).size ( )]];
}
@@ -101,7 +83,6 @@ template < class SymbolType, class RankType >
ext::set < unsigned > QuickSearch::match ( const tree::PrefixRankedBarTree < SymbolType, RankType > & subject, const tree::PrefixRankedBarNonlinearPattern < SymbolType, RankType > & pattern ) {
ext::set < unsigned > occ;
ext::map < common::ranked_symbol < SymbolType, RankType >, size_t > bcs = tree::properties::QuickSearchBadCharacterShiftTable::bcs ( pattern ); //NOTE: the subjects alphabet must be a subset or equal to the pattern
- ext::map < common::ranked_symbol < SymbolType, RankType >, unsigned > variablesSetting;
ext::vector < int > subjectSubtreeJumpTable = tree::properties::SubtreeJumpTable::compute ( subject );
tree::PrefixRankedBarTree < unsigned, RankType > repeats = tree::properties::ExactSubtreeRepeatsNaive::repeats ( subject );
@@ -111,41 +92,11 @@ ext::set < unsigned > QuickSearch::match ( const tree::PrefixRankedBarTree < Sym
// main loop of the algorithm over all possible indexes where the pattern can start
while ( i + pattern.getContent ( ).size ( ) <= subject.getContent ( ).size ( ) ) {
- // clear the current state of variable to subtree repeat
- variablesSetting.clear();
-
- // index to the pattern
- int j = pattern.getContent ( ).size ( ) - 1;
-
- // offset to the subject
- int offset = i + j;
-
- while ( ( j >= 0 ) && ( offset >= 0 ) ) {
- if ( subject.getContent ( )[offset] == pattern.getContent ( )[j] ) {
- // match of symbol
- offset = offset - 1;
- j = j - 1;
- } else if ( ( pattern.getContent ( )[j] == pattern.getVariablesBar ( ) ) /* && ( pattern.getBars ( ).count ( subject.getContent ( )[offset] ) ) */ ) {
- // else match of variable with subtree
- offset = subjectSubtreeJumpTable[offset];
- j = j - 2;
-
- // check nonlinear variable
- if ( pattern.getNonlinearVariables ( ).count ( pattern.getContent ( )[ j + 1 ] ) ) {
- auto setting = variablesSetting.find ( pattern.getContent ( )[ j + 1 ] );
-
- if ( setting != variablesSetting.end ( ) && repeats.getContent ( )[ offset + 1 ].getSymbol ( ) != setting->second )
- break;
-
- variablesSetting.insert ( std::make_pair ( pattern.getContent ( )[ j + 1 ], repeats.getContent( )[ offset + 1 ].getSymbol ( ) ) );
- }
- } else {
- break;
- }
- }
+ // pair j and offset
+ ext::pair < int, int > jOffset = tree::exact::BackwardOccurrenceTest::occurrence ( subject, subjectSubtreeJumpTable, repeats, pattern, i );
- // match was found
- if ( j == -1 ) occ.insert ( offset + 1);
+ // match was found
+ if ( jOffset.first == -1 ) occ.insert ( jOffset.second + 1);
if ( i + pattern.getContent ( ).size ( ) == subject.getContent ( ).size ( ) ) {
break;
diff --git a/alib2algo/src/tree/properties/QuickSearchBadCharacterShiftTable.h b/alib2algo/src/tree/properties/QuickSearchBadCharacterShiftTable.h
index d0cb8af17297f250213f93c932b07d7a892178c1..4895407e54b224c9cfe4e3b4a29e64c020f2b25e 100644
--- a/alib2algo/src/tree/properties/QuickSearchBadCharacterShiftTable.h
+++ b/alib2algo/src/tree/properties/QuickSearchBadCharacterShiftTable.h
@@ -12,6 +12,8 @@
#include <tree/ranked/PrefixRankedBarPattern.h>
#include <tree/ranked/PrefixRankedBarNonlinearPattern.h>
+#include "LastVariableOffsetBack.h"
+
#include <alib/set>
#include <alib/map>
@@ -52,11 +54,7 @@ ext::map < common::ranked_symbol < SymbolType, RankType >, size_t > QuickSearchB
// find the distance between the end of the pattern and the index
// of the last symbol representing the variable
- unsigned lastSOffset = pattern.getContent ( ).size ( );
-
- for ( unsigned i = 0; i < pattern.getContent ( ).size ( ); i++ )
- if ( pattern.getContent ( )[i] == pattern.getSubtreeWildcard ( ) || pattern.getNonlinearVariables ( ).count ( pattern.getContent ( )[i] ) )
- lastSOffset = pattern.getContent ( ).size ( ) - i;
+ size_t lastSOffset = LastVariableOffsetBack::offset ( pattern );
// limit the shift by occurrence of the last variable
@@ -64,13 +62,13 @@ ext::map < common::ranked_symbol < SymbolType, RankType >, size_t > QuickSearchB
if ( symbol == pattern.getSubtreeWildcard ( ) || pattern.getNonlinearVariables ( ).count ( symbol ) || symbol == pattern.getVariablesBar ( ) )
continue;
- size_t tmp = lastSOffset;
+ size_t tmp = lastSOffset + 1;
if ( ! pattern.getBars ( ).count ( symbol ) )
// size of the smallest subtree containing given terminal depend
// on the arity of the terminal
tmp += ( size_t ) symbol.getRank ( ) * 2;
- else if ( tmp >= 2 )
+ else
// bar symbols match the variable bar which is one symbol after
// the last variable, conditioned because of the case S S| where
// the -1 would cause shift by 0 -- illegal
@@ -81,7 +79,7 @@ ext::map < common::ranked_symbol < SymbolType, RankType >, size_t > QuickSearchB
}
// limit the shift by position of symbols within the pattern
- for ( unsigned i = 0; i < pattern.getContent ( ).size ( ); i++ ) { // last symbol is concerned here
+ for ( unsigned i = 0; i < pattern.getContent ( ).size ( ); i++ ) {
if ( pattern.getContent ( )[i] == pattern.getSubtreeWildcard ( ) || pattern.getNonlinearVariables ( ).count ( pattern.getContent ( )[i] ) || pattern.getContent ( )[i] == pattern.getVariablesBar ( ) )
continue;