From b9bfb0bbb05264d4cf9ff96b96f638a9d991b85c Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Radovan=20=C4=8Cerven=C3=BD?= <radovan.cerveny@gmail.com>
Date: Fri, 1 Apr 2016 18:19:30 +0200
Subject: [PATCH] somewhat cleaner implementation of BNDM using bitset
---
.../BackwardNondeterministicDAWGMatching.cpp | 82 +++++++++++++------
.../BackwardNondeterministicDAWGMatching.hpp | 20 ++++-
...ckwardNondeterministicDAWGMatchingTest.cpp | 7 +-
3 files changed, 78 insertions(+), 31 deletions(-)
diff --git a/alib2algo/src/stringology/exact/BackwardNondeterministicDAWGMatching.cpp b/alib2algo/src/stringology/exact/BackwardNondeterministicDAWGMatching.cpp
index 89a808fcb9..f170318045 100644
--- a/alib2algo/src/stringology/exact/BackwardNondeterministicDAWGMatching.cpp
+++ b/alib2algo/src/stringology/exact/BackwardNondeterministicDAWGMatching.cpp
@@ -9,68 +9,67 @@
#include <alphabet/Symbol.h>
#include <map>
+#include <bitset>
#include <measure>
namespace stringology {
namespace exact {
-std::set < unsigned > BackwardNondeterministicDAWGMatching::match ( const string::String & subject, const string::String & pattern ) {
+template < size_t BitmaskBitCount >
+std::set < unsigned > BackwardNondeterministicDAWGMatching::matchTemplate ( const string::String & subject, const string::String & pattern ) {
return getInstance ( ).dispatch ( subject.getData ( ), pattern.getData ( ) );
}
-std::set < unsigned > BackwardNondeterministicDAWGMatching::match ( const string::LinearString & subject, const string::LinearString & pattern ) {
- using Bitmask = unsigned long long int;
-
+template < size_t BitmaskBitCount >
+std::set < unsigned > BackwardNondeterministicDAWGMatching::matchTemplate ( const string::LinearString & subject, const string::LinearString & pattern ) {
std::set < unsigned > occ;
- measurements::start ( "Preprocess", measurements::Type::PREPROCESS );
-
- std::map < alphabet::Symbol, Bitmask > symbolBitmaskLookupTable;
- bool longPattern;
- size_t bitmaskLength;
- Bitmask highestBitBitmask;
- Bitmask allOnesBitmask;
-
// Setup helper variables
- longPattern = 64 < pattern.getContent ( ).size ( );
- bitmaskLength = longPattern ? 64 : pattern.getContent ( ).size ( );
+ using BitmaskType = std::bitset < BitmaskBitCount >;
+ bool patternIsLong = BitmaskBitCount < pattern.getContent ( ).size ( );
+ size_t bitmaskLength = patternIsLong ? BitmaskBitCount : pattern.getContent ( ).size ( );
+
+ measurements::start ( "Preprocess", measurements::Type::PREPROCESS );
- highestBitBitmask = 1ULL << ( bitmaskLength - 1 );
- allOnesBitmask = ( highestBitBitmask << 1 ) - 1;
+ std::map < alphabet::Symbol, BitmaskType > symbolBitmaskLookupTable;
// Initialize the bitmasks with zeros for each symbol in the alphabet
for ( const auto & symbol : pattern.getAlphabet ( ) )
- symbolBitmaskLookupTable[symbol] = 0ULL;
+ symbolBitmaskLookupTable[symbol] = BitmaskType ( 0 );
// Mark the position in the bitmask for each symbol in the pattern
for ( size_t i = 0; i < bitmaskLength; i++ )
- symbolBitmaskLookupTable[pattern.getContent ( ).at ( i )] |= 1ULL << ( bitmaskLength - i - 1 );
+ symbolBitmaskLookupTable[pattern.getContent ( ).at ( i )].set ( bitmaskLength - i - 1 );
measurements::end ( );
measurements::start ( "Algorithm", measurements::Type::ALGORITHM );
size_t posInSubject = 0;
+ BitmaskType currentBitmask;
while ( posInSubject <= subject.getContent ( ).size ( ) - pattern.getContent ( ).size ( ) ) {
size_t posInPattern = bitmaskLength;
size_t lastPosOfFactor = bitmaskLength;
- Bitmask currentBitmask = ~0ULL;
- while ( currentBitmask != 0ULL ) {
- currentBitmask = currentBitmask & symbolBitmaskLookupTable[subject.getContent ( ).at ( posInSubject + posInPattern - 1 )];
+ // Set the bitmask to all ones
+ currentBitmask.set ( );
+
+ while ( posInPattern > 0 && currentBitmask.any ( ) ) {
+ currentBitmask &= symbolBitmaskLookupTable[subject.getContent ( ).at ( posInSubject + posInPattern - 1 )];
posInPattern--;
- if ( ( currentBitmask & highestBitBitmask ) != 0 ) {
+ // Test whether the most significant bit is set
+ if ( currentBitmask.test ( bitmaskLength - 1 ) ) {
if ( posInPattern > 0 ) {
lastPosOfFactor = posInPattern;
} else {
- if ( !longPattern ) {
+ if ( !patternIsLong ) {
// Yay, there is match!!!
occ.insert ( posInSubject );
} else {
- // if the pattern is longer then 64 characters switch to brute force check
+ // if the pattern is longer then BITMASK_BIT_COUNT characters switch to brute force check
size_t k = bitmaskLength;
while ( k < pattern.getContent ( ).size ( ) && pattern.getContent ( ).at ( k ) == subject.getContent ( ).at ( posInSubject + k ) ) k++;
@@ -83,9 +82,6 @@ std::set < unsigned > BackwardNondeterministicDAWGMatching::match ( const string
}
currentBitmask <<= 1;
-
- // We need to trim excess ones in case the pattern is shorter then 64 characters
- currentBitmask &= allOnesBitmask;
}
posInSubject += lastPosOfFactor;
@@ -96,6 +92,38 @@ std::set < unsigned > BackwardNondeterministicDAWGMatching::match ( const string
return occ;
}
+std::set < unsigned > BackwardNondeterministicDAWGMatching::match ( const string::String & subject, const string::String & pattern ) {
+ return BackwardNondeterministicDAWGMatching::match32 ( subject, pattern );
+}
+
+std::set < unsigned > BackwardNondeterministicDAWGMatching::match ( const string::LinearString & subject, const string::LinearString & pattern ) {
+ return BackwardNondeterministicDAWGMatching::match32 ( subject, pattern );
+}
+
+std::set < unsigned > BackwardNondeterministicDAWGMatching::match32 ( const string::String & subject, const string::String & pattern ) {
+ return BackwardNondeterministicDAWGMatching::matchTemplate < 32 > ( subject, pattern );
+}
+
+std::set < unsigned > BackwardNondeterministicDAWGMatching::match32 ( const string::LinearString & subject, const string::LinearString & pattern ) {
+ return BackwardNondeterministicDAWGMatching::matchTemplate < 32 > ( subject, pattern );
+}
+
+std::set < unsigned > BackwardNondeterministicDAWGMatching::match64 ( const string::String & subject, const string::String & pattern ) {
+ return BackwardNondeterministicDAWGMatching::matchTemplate < 64 > ( subject, pattern );
+}
+
+std::set < unsigned > BackwardNondeterministicDAWGMatching::match64 ( const string::LinearString & subject, const string::LinearString & pattern ) {
+ return BackwardNondeterministicDAWGMatching::matchTemplate < 64 > ( subject, pattern );
+}
+
+std::set < unsigned > BackwardNondeterministicDAWGMatching::match128 ( const string::String & subject, const string::String & pattern ) {
+ return BackwardNondeterministicDAWGMatching::matchTemplate < 128 > ( subject, pattern );
+}
+
+std::set < unsigned > BackwardNondeterministicDAWGMatching::match128 ( const string::LinearString & subject, const string::LinearString & pattern ) {
+ return BackwardNondeterministicDAWGMatching::matchTemplate < 128 > ( subject, pattern );
+}
+
auto BackwardNondeterministicDAWGMatchingLinearStringLinearString = BackwardNondeterministicDAWGMatching::RegistratorWrapper < std::set < unsigned >, string::LinearString, string::LinearString > ( BackwardNondeterministicDAWGMatching::getInstance ( ), BackwardNondeterministicDAWGMatching::match );
} /* namespace exact */
diff --git a/alib2algo/src/stringology/exact/BackwardNondeterministicDAWGMatching.hpp b/alib2algo/src/stringology/exact/BackwardNondeterministicDAWGMatching.hpp
index 9426e0d154..5b43fe87b1 100644
--- a/alib2algo/src/stringology/exact/BackwardNondeterministicDAWGMatching.hpp
+++ b/alib2algo/src/stringology/exact/BackwardNondeterministicDAWGMatching.hpp
@@ -20,15 +20,31 @@ namespace exact {
* Implementation of Backward Nondeterministic DAWG Matching using bit parallelism with 64bit bitmask and brute force switch for longer patterns.
*/
class BackwardNondeterministicDAWGMatching : public std::DoubleDispatch < std::set < unsigned >, string::StringBase, string::StringBase > {
-public:
+private:
/**
* Search for pattern in linear string.
* @return set set of occurences
*/
- static std::set < unsigned > match ( const string::String & subject, const string::String & pattern );
+ template <size_t BitmaskBitCount>
+ static std::set < unsigned > matchTemplate ( const string::String & subject, const string::String & pattern );
+ template <size_t BitmaskBitCount >
+ static std::set < unsigned > matchTemplate ( const string::LinearString & subject, const string::LinearString & pattern );
+public:
+
+ static std::set < unsigned > match ( const string::String & subject, const string::String & pattern );
static std::set < unsigned > match ( const string::LinearString & subject, const string::LinearString & pattern );
+ static std::set < unsigned > match32 ( const string::String & subject, const string::String & pattern );
+ static std::set < unsigned > match32 ( const string::LinearString & subject, const string::LinearString & pattern );
+
+ static std::set < unsigned > match64 ( const string::String & subject, const string::String & pattern );
+ static std::set < unsigned > match64 ( const string::LinearString & subject, const string::LinearString & pattern );
+
+ static std::set < unsigned > match128 ( const string::String & subject, const string::String & pattern );
+ static std::set < unsigned > match128 ( const string::LinearString & subject, const string::LinearString & pattern );
+
+
static BackwardNondeterministicDAWGMatching & getInstance ( ) {
static BackwardNondeterministicDAWGMatching res;
diff --git a/alib2algo/test-src/stringology/exact/BackwardNondeterministicDAWGMatchingTest.cpp b/alib2algo/test-src/stringology/exact/BackwardNondeterministicDAWGMatchingTest.cpp
index 65d8c4ad06..4059bf1f7b 100644
--- a/alib2algo/test-src/stringology/exact/BackwardNondeterministicDAWGMatchingTest.cpp
+++ b/alib2algo/test-src/stringology/exact/BackwardNondeterministicDAWGMatchingTest.cpp
@@ -39,9 +39,12 @@ void BackwardNondeterministicDAWGMatchingTest::testBNDM ( ) {
string::String subject = string::stringFrom ( subjects[i] );
string::String pattern = string::stringFrom ( patterns[i] );
std::set < unsigned > res = stringology::exact::BackwardNondeterministicDAWGMatching::match ( subject, pattern );
-
- std::cout << subjects[i] << ' ' << patterns[i] << ' ' << res << std::endl;
CPPUNIT_ASSERT ( res == expectedOccs[i] );
+ res = stringology::exact::BackwardNondeterministicDAWGMatching::match64 ( subject, pattern );
+ CPPUNIT_ASSERT ( res == expectedOccs[i] );
+ res = stringology::exact::BackwardNondeterministicDAWGMatching::match128 ( subject, pattern );
+ CPPUNIT_ASSERT ( res == expectedOccs[i] );
+ std::cout << subjects[i] << ' ' << patterns[i] << ' ' << res << std::endl;
}
auto longSubject = string::generate::RandomStringFactory::generateLinearString (64 * 64 * 64, 512, false, true);
--
GitLab