From d32f5c4808a76223035ea37e0718466b455e3299 Mon Sep 17 00:00:00 2001
From: Jan Travnicek <Jan.Travnicek@fit.cvut.cz>
Date: Wed, 4 Apr 2018 16:12:16 +0200
Subject: [PATCH] tune tree backward pattern matching
---
.../src/arbology/exact/BoyerMooreHorspool.h | 12 ++++++------
alib2algo/src/arbology/exact/QuickSearch.h | 16 ++++++++--------
.../src/tree/exact/BackwardOccurrenceTest.h | 4 ++--
3 files changed, 16 insertions(+), 16 deletions(-)
diff --git a/alib2algo/src/arbology/exact/BoyerMooreHorspool.h b/alib2algo/src/arbology/exact/BoyerMooreHorspool.h
index 7b5a58af94..05f76f513d 100644
--- a/alib2algo/src/arbology/exact/BoyerMooreHorspool.h
+++ b/alib2algo/src/arbology/exact/BoyerMooreHorspool.h
@@ -57,10 +57,10 @@ ext::set < unsigned > BoyerMooreHorspool::match ( const tree::PrefixRankedBarTre
ext::vector < int > subjectSubtreeJumpTable = tree::properties::SubtreeJumpTable::compute ( subject );
// index to the subject
- unsigned i = 0;
+ unsigned i = pattern.getContent ( ).size ( ) - 1;
// main loop of the algorithm over all possible indexes where the pattern can start
- while ( i + pattern.getContent ( ).size ( ) <= subject.getContent ( ).size ( ) ) {
+ while ( i < subject.getContent ( ).size ( ) ) {
// pair j and offset
ext::pair < int, int > jOffset = tree::exact::BackwardOccurrenceTest::occurrence ( subject, subjectSubtreeJumpTable, pattern, i );
@@ -68,7 +68,7 @@ ext::set < unsigned > BoyerMooreHorspool::match ( const tree::PrefixRankedBarTre
if ( jOffset.first == -1 ) occ.insert ( jOffset.second + 1);
// shift heuristics
- i += bcs[subject.getContent ( )[i + pattern.getContent ( ).size ( ) - 1]];
+ i += bcs[subject.getContent ( )[i]];
}
return occ;
@@ -84,10 +84,10 @@ ext::set < unsigned > BoyerMooreHorspool::match ( const tree::PrefixRankedBarTre
tree::PrefixRankedBarTree < unsigned, RankType > repeats = tree::properties::ExactSubtreeRepeatsNaive::repeats ( subject );
// index to the subject
- unsigned i = 0;
+ unsigned i = pattern.getContent ( ).size ( ) - 1;
// main loop of the algorithm over all possible indexes where the pattern can start
- while ( i + pattern.getContent ( ).size ( ) <= subject.getContent ( ).size ( ) ) {
+ while ( i < subject.getContent ( ).size ( ) ) {
// pair j and offset
ext::pair < int, int > jOffset = tree::exact::BackwardOccurrenceTest::occurrence ( subject, subjectSubtreeJumpTable, repeats, pattern, i );
@@ -95,7 +95,7 @@ ext::set < unsigned > BoyerMooreHorspool::match ( const tree::PrefixRankedBarTre
if ( jOffset.first == -1 ) occ.insert ( jOffset.second + 1);
// shift heuristics
- i += bcs[subject.getContent ( )[i + pattern.getContent ( ).size ( ) - 1]];
+ i += bcs[subject.getContent ( )[i]];
}
return occ;
diff --git a/alib2algo/src/arbology/exact/QuickSearch.h b/alib2algo/src/arbology/exact/QuickSearch.h
index 4d7470110e..654ca0427e 100644
--- a/alib2algo/src/arbology/exact/QuickSearch.h
+++ b/alib2algo/src/arbology/exact/QuickSearch.h
@@ -58,22 +58,22 @@ ext::set < unsigned > QuickSearch::match ( const tree::PrefixRankedBarTree < Sym
ext::vector < int > subjectSubtreeJumpTable = tree::properties::SubtreeJumpTable::compute ( subject );
// index to the subject
- int i = 0;
+ unsigned i = pattern.getContent ( ).size ( ) - 1;
// main loop of the algorithm over all possible indexes where the pattern can start
- while ( i + pattern.getContent ( ).size ( ) <= subject.getContent ( ).size ( ) ) {
+ while ( i < subject.getContent ( ).size ( ) ) {
// pair j and offset
ext::pair < int, int > jOffset = tree::exact::BackwardOccurrenceTest::occurrence ( subject, subjectSubtreeJumpTable, pattern, i );
// match was found
if ( jOffset.first == -1 ) occ.insert ( jOffset.second + 1);
- if ( i + pattern.getContent ( ).size ( ) == subject.getContent ( ).size ( ) ) {
+ if ( i + 1 >= subject.getContent ( ).size ( ) ) {
break;
}
// shift heuristics
- i += bcs[subject.getContent ( )[i + pattern.getContent ( ).size ( )]];
+ i += bcs[subject.getContent ( )[i + 1]];
}
return occ;
@@ -88,22 +88,22 @@ ext::set < unsigned > QuickSearch::match ( const tree::PrefixRankedBarTree < Sym
tree::PrefixRankedBarTree < unsigned, RankType > repeats = tree::properties::ExactSubtreeRepeatsNaive::repeats ( subject );
// index to the subject
- unsigned i = 0;
+ unsigned i = pattern.getContent ( ).size ( ) - 1;
// main loop of the algorithm over all possible indexes where the pattern can start
- while ( i + pattern.getContent ( ).size ( ) <= subject.getContent ( ).size ( ) ) {
+ while ( i < subject.getContent ( ).size ( ) ) {
// pair j and offset
ext::pair < int, int > jOffset = tree::exact::BackwardOccurrenceTest::occurrence ( subject, subjectSubtreeJumpTable, repeats, pattern, i );
// match was found
if ( jOffset.first == -1 ) occ.insert ( jOffset.second + 1);
- if ( i + pattern.getContent ( ).size ( ) == subject.getContent ( ).size ( ) ) {
+ if ( i + 1 >= subject.getContent ( ).size ( ) ) {
break;
}
// shift heuristics
- i += bcs[subject.getContent ( )[i + pattern.getContent ( ).size ( )]];
+ i += bcs[subject.getContent ( )[i + 1]];
}
return occ;
diff --git a/alib2algo/src/tree/exact/BackwardOccurrenceTest.h b/alib2algo/src/tree/exact/BackwardOccurrenceTest.h
index 329eb6dd09..3a0e46afa5 100644
--- a/alib2algo/src/tree/exact/BackwardOccurrenceTest.h
+++ b/alib2algo/src/tree/exact/BackwardOccurrenceTest.h
@@ -43,7 +43,7 @@ ext::pair < int, int > BackwardOccurrenceTest::occurrence ( const PrefixRankedBa
int j = pattern.getContent ( ).size ( ) - 1;
// offset to the subject
- int offset = subjectPosition + j;
+ int offset = subjectPosition;
while ( ( j >= 0 ) && ( offset >= 0 ) ) {
if ( subject.getContent ( )[offset] == pattern.getContent ( )[j] ) {
@@ -72,7 +72,7 @@ ext::pair < int, int > BackwardOccurrenceTest::occurrence ( const PrefixRankedBa
int j = pattern.getContent ( ).size ( ) - 1;
// offset to the subject
- int offset = subjectPosition + j;
+ int offset = subjectPosition;
while ( ( j >= 0 ) && ( offset >= 0 ) ) {
if ( subject.getContent ( )[offset] == pattern.getContent ( )[j] ) {
--
GitLab