From 1d66e5646f22ee3472688ad172ea8859d6e603ba Mon Sep 17 00:00:00 2001
From: Jan Travnicek <Jan.Travnicek@fit.cvut.cz>
Date: Fri, 5 May 2017 13:20:31 +0200
Subject: [PATCH] remove BNDM from experimental
---
.../BackwardNondeterministicDAWGMatching.cpp | 130 ------------------
.../BackwardNondeterministicDAWGMatching.hpp | 53 -------
...ckwardNondeterministicDAWGMatchingTest.cpp | 55 --------
...BackwardNondeterministicDAWGMatchingTest.h | 18 ---
4 files changed, 256 deletions(-)
delete mode 100644 alib2algo_experimental/src/stringology/exact/BackwardNondeterministicDAWGMatching.cpp
delete mode 100644 alib2algo_experimental/src/stringology/exact/BackwardNondeterministicDAWGMatching.hpp
delete mode 100644 alib2algo_experimental/test-src/stringology/exact/BackwardNondeterministicDAWGMatchingTest.cpp
delete mode 100644 alib2algo_experimental/test-src/stringology/exact/BackwardNondeterministicDAWGMatchingTest.h
diff --git a/alib2algo_experimental/src/stringology/exact/BackwardNondeterministicDAWGMatching.cpp b/alib2algo_experimental/src/stringology/exact/BackwardNondeterministicDAWGMatching.cpp
deleted file mode 100644
index f0262689ac..0000000000
--- a/alib2algo_experimental/src/stringology/exact/BackwardNondeterministicDAWGMatching.cpp
+++ /dev/null
@@ -1,130 +0,0 @@
-/*
- * Author: Radovan Cerveny
- */
-
-#include "BackwardNondeterministicDAWGMatching.hpp"
-
-#include <string/LinearString.h>
-#include <alphabet/Symbol.h>
-
-#include <map>
-#include <bitset>
-#include <measure>
-
-namespace stringology {
-
-namespace exact {
-
-template < size_t BitmaskBitCount >
-std::set < unsigned > BackwardNondeterministicDAWGMatching::matchTemplate ( const string::String & subject, const string::String & pattern ) {
- return dispatch ( subject.getData ( ), pattern.getData ( ) );
-}
-
-template < size_t BitmaskBitCount >
-std::set < unsigned > BackwardNondeterministicDAWGMatching::matchTemplate ( const string::LinearString < > & subject, const string::LinearString < > & pattern ) {
- std::set < unsigned > occ;
-
- // Setup helper variables
- 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 );
-
- std::map < DefaultSymbolType, BitmaskType > symbolBitmaskLookupTable;
-
- // Initialize the bitmasks with zeros for each symbol in the alphabet
- for ( const auto & symbol : pattern.getAlphabet ( ) )
- 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 )].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;
-
- // Set the bitmask to all ones
- currentBitmask.set ( );
-
- while ( posInPattern > 0 && currentBitmask.any ( ) ) {
- currentBitmask &= symbolBitmaskLookupTable[subject.getContent ( ).at ( posInSubject + posInPattern - 1 )];
- posInPattern--;
-
- // Test whether the most significant bit is set
- if ( currentBitmask.test ( bitmaskLength - 1 ) ) {
- if ( posInPattern > 0 ) {
- lastPosOfFactor = posInPattern;
- } else {
- if ( !patternIsLong ) {
- // Yay, there is match!!!
- occ.insert ( posInSubject );
- } else {
- // if the pattern is longer then BitmaskBitCount 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++;
-
- if ( k == pattern.getContent ( ).size ( ) )
- // Yay, there is match!!!
- occ.insert ( posInSubject );
- }
- }
- }
-
- currentBitmask <<= 1;
- }
-
- posInSubject += lastPosOfFactor;
- }
-
- measurements::end ( );
-
- 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::match );
-
-} /* namespace exact */
-
-} /* namespace stringology */
diff --git a/alib2algo_experimental/src/stringology/exact/BackwardNondeterministicDAWGMatching.hpp b/alib2algo_experimental/src/stringology/exact/BackwardNondeterministicDAWGMatching.hpp
deleted file mode 100644
index b3b6b1f339..0000000000
--- a/alib2algo_experimental/src/stringology/exact/BackwardNondeterministicDAWGMatching.hpp
+++ /dev/null
@@ -1,53 +0,0 @@
-/*
- * Author: Radovan Cerveny
- */
-
-#ifndef STRINGOLOGY_BACKWARD_NONDETERMINISTIC_DAWG_MATCHING_HPP__
-#define STRINGOLOGY_BACKWARD_NONDETERMINISTIC_DAWG_MATCHING_HPP__
-
-#include <string/String.h>
-#include <string/StringFeatures.h>
-#include <core/multipleDispatch.hpp>
-
-#include <set>
-
-namespace stringology {
-
-namespace exact {
-
-/**
- * Implementation of Backward Nondeterministic DAWG Matching using bit parallelism with 32/64/128bit bitmask and brute force switch for longer patterns.
- */
-class BackwardNondeterministicDAWGMatching : public std::DoubleDispatch < BackwardNondeterministicDAWGMatching, std::set < unsigned >, const string::StringBase &, const string::StringBase & > {
-private:
- /**
- * Search for pattern in linear string.
- * @return set set of occurences
- */
- 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:
-
- // Defaults to 32 bits
- 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 );
-
-};
-
-} /* namespace exact */
-
-} /* namespace stringology */
-
-#endif /* STRINGOLOGY_BACKWARD_NONDETERMINISTIC_DAWG_MATCHING_HPP__ */
diff --git a/alib2algo_experimental/test-src/stringology/exact/BackwardNondeterministicDAWGMatchingTest.cpp b/alib2algo_experimental/test-src/stringology/exact/BackwardNondeterministicDAWGMatchingTest.cpp
deleted file mode 100644
index e7fe488512..0000000000
--- a/alib2algo_experimental/test-src/stringology/exact/BackwardNondeterministicDAWGMatchingTest.cpp
+++ /dev/null
@@ -1,55 +0,0 @@
-#include "BackwardNondeterministicDAWGMatchingTest.h"
-
-#include "string/String.h"
-#include "stringology/exact/BackwardNondeterministicDAWGMatching.hpp"
-#include "stringology/exact/ExactFactorMatch.h"
-
-#include "string/generate/RandomStringFactory.h"
-#include "string/generate/RandomSubstringFactory.h"
-
-CPPUNIT_TEST_SUITE_NAMED_REGISTRATION ( BackwardNondeterministicDAWGMatchingTest, "stringology" );
-CPPUNIT_TEST_SUITE_REGISTRATION ( BackwardNondeterministicDAWGMatchingTest );
-
-void BackwardNondeterministicDAWGMatchingTest::setUp ( ) {
-}
-
-void BackwardNondeterministicDAWGMatchingTest::tearDown ( ) {
-}
-
-void BackwardNondeterministicDAWGMatchingTest::testBNDM ( ) {
-
- std::vector<std::string> subjects;
- std::vector<std::string> patterns;
- std::vector<std::set<unsigned>> expectedOccs;
-
- subjects.push_back("a"); patterns.push_back("a"); expectedOccs.push_back({0});
- subjects.push_back("a"); patterns.push_back("b"); expectedOccs.push_back({});
- subjects.push_back("alfalfalfa"); patterns.push_back("alfalfalfa"); expectedOccs.push_back({0});
- subjects.push_back("alfalfalfa"); patterns.push_back("blfalfalfa"); expectedOccs.push_back({});
- subjects.push_back("alfalfalfa"); patterns.push_back("alfalfalfb"); expectedOccs.push_back({});
- subjects.push_back("alfalfalfaalfalfalfaalfalfalfaalfalfalfaalfalfalfaalfalfalfaalfa"); patterns.push_back("alfalfalfaalfalfalfaalfalfalfaalfalfalfaalfalfalfaalfalfalfaalfa"); expectedOccs.push_back({0});
- subjects.push_back("alfalfalfaalfalfalfaabfalfalfaalfalfalfaalfalfalfaalfalfalfaalfa"); patterns.push_back("alfalfalfaalfalfalfaalfalfalfaalfalfalfaalfalfalfaalfalfalfaalfa"); expectedOccs.push_back({});
- 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});
-
-
- for(size_t i = 0; i < subjects.size(); ++i) {
- string::String subject = string::stringFrom ( subjects[i] );
- string::String pattern = string::stringFrom ( patterns[i] );
- std::set < unsigned > res = stringology::exact::BackwardNondeterministicDAWGMatching::match ( subject, pattern );
- 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);
- auto longPattern = string::generate::RandomSubstringFactory::generateSubstring(64 * 32 * 32, longSubject );
- std::set < unsigned > res = stringology::exact::BackwardNondeterministicDAWGMatching::match ( longSubject, longPattern );
- std::set < unsigned > resRef = stringology::exact::ExactFactorMatch::match ( longSubject, longPattern );
- std::cout << "long: " << res << std::endl;
- CPPUNIT_ASSERT ( res == resRef);
-}
diff --git a/alib2algo_experimental/test-src/stringology/exact/BackwardNondeterministicDAWGMatchingTest.h b/alib2algo_experimental/test-src/stringology/exact/BackwardNondeterministicDAWGMatchingTest.h
deleted file mode 100644
index 016be2e42a..0000000000
--- a/alib2algo_experimental/test-src/stringology/exact/BackwardNondeterministicDAWGMatchingTest.h
+++ /dev/null
@@ -1,18 +0,0 @@
-#ifndef BACKWARD_NONDETERMINISTIC_DAWG_MATCHING_TEST_H_
-#define BACKWARD_NONDETERMINISTIC_DAWG_MATCHING_TEST_H_
-
-#include <cppunit/extensions/HelperMacros.h>
-
-class BackwardNondeterministicDAWGMatchingTest : public CppUnit::TestFixture {
- CPPUNIT_TEST_SUITE ( BackwardNondeterministicDAWGMatchingTest );
- CPPUNIT_TEST ( testBNDM );
- CPPUNIT_TEST_SUITE_END ( );
-
-public:
- void setUp ( );
- void tearDown ( );
-
- void testBNDM ( );
-};
-
-#endif // BACKWARD_NONDETERMINISTIC_DAWG_MATCHING_TEST_H_
--
GitLab