From 90e029aec88a7f76809745ba861c46dfdea4d0a0 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Radovan=20=C4=8Cerven=C3=BD?= <radovan.cerveny@gmail.com>
Date: Thu, 28 Apr 2016 15:53:13 +0200
Subject: [PATCH] fixed implementation of BDM search
---
alib2algo/src/stringology/exact/BackwardDAWGMatching.cpp | 8 ++++++--
.../test-src/stringology/exact/SuffixAutomatonTest.cpp | 1 +
2 files changed, 7 insertions(+), 2 deletions(-)
diff --git a/alib2algo/src/stringology/exact/BackwardDAWGMatching.cpp b/alib2algo/src/stringology/exact/BackwardDAWGMatching.cpp
index 84175a5c30..0a098b53c1 100644
--- a/alib2algo/src/stringology/exact/BackwardDAWGMatching.cpp
+++ b/alib2algo/src/stringology/exact/BackwardDAWGMatching.cpp
@@ -43,10 +43,10 @@ std::set < unsigned > BackwardDAWGMatching::match ( const string::LinearString &
size_t posInSubject = 0;
while ( posInSubject <= subject.getContent ( ).size ( ) - pattern.getContent ( ).size ( ) ) {
-
automaton::State currentState = suffixAutomaton.getInitialState ( );
size_t posInPattern = reversedPattern.getContent ( ).size ( );
+ size_t lastPrefixPos = posInPattern;
while ( posInPattern > 0 && currentState != failState ) {
auto transition = suffixAutomaton.getTransitions ( ).find ( { currentState, subject.getContent ( ).at ( posInSubject + posInPattern - 1 ) } );
@@ -57,13 +57,17 @@ std::set < unsigned > BackwardDAWGMatching::match ( const string::LinearString &
currentState = transition->second;
posInPattern--;
+
+ // found a prefix of nonreversed pattern that does not correspond to the entire pattern
+ if ( ( posInPattern != 0 ) && ( suffixAutomaton.getFinalStates ( ).find ( currentState ) != suffixAutomaton.getFinalStates ( ).end ( ) ) )
+ lastPrefixPos = posInPattern;
}
if ( currentState != failState )
// Yay, there is match!!!
occ.insert ( posInSubject );
- posInSubject += posInPattern + 1;
+ posInSubject += lastPrefixPos;
}
measurements::end ( );
diff --git a/alib2algo/test-src/stringology/exact/SuffixAutomatonTest.cpp b/alib2algo/test-src/stringology/exact/SuffixAutomatonTest.cpp
index c7485c8751..e73b48cc40 100644
--- a/alib2algo/test-src/stringology/exact/SuffixAutomatonTest.cpp
+++ b/alib2algo/test-src/stringology/exact/SuffixAutomatonTest.cpp
@@ -64,6 +64,7 @@ void SuffixAutomatonTest::testBackwardDAWGMatching ( ) {
subjects.push_back("alfalfalfaalfalfalfaalfalfalfaalfalfalfaalfalfalfaalfalfalfaalfaa"); patterns.push_back("alfalfalfaalfalfalfaalfalfalfaalfalfalfaalfalfalfaalfalfalfaalfaa"); expectedOccs.push_back({0});
subjects.push_back("atggccttgcc"); patterns.push_back("gcc"); expectedOccs.push_back({3,8});
subjects.push_back("aaaaaaaaaa"); patterns.push_back("a"); expectedOccs.push_back({0,1,2,3,4,5,6,7,8,9});
+ subjects.push_back("aaaaaaaaaa"); patterns.push_back("aa"); expectedOccs.push_back({0,1,2,3,4,5,6,7,8});
for(size_t i = 0; i < subjects.size(); ++i) {
--
GitLab