concordia search

Former-commit-id: 609c3a54e930ebae45a2e9a07f63991ec4abc9a6

TODO.txt

@@ -1,8 +1,8 @@
-- wyszukiwanie zdania: wyszukanie najdłuższych pasujących fragmentów Anubisem, 1D (approximate) bin packing. Nazwijmy to concordia search. Wyszukiwane są wszystkie najdłuższe dopasowania patternu dzięki LCP search. Zwracany jest wynik w postaci listy najdłuższych dopasowanych fragmentów, posortowanych malejąco po długości, z maksymalnie 3 przedstawicielami każdej długości. Dodatkowo obliczany jest zestaw optymalnego pokrycia patternu.
+- wyłączyć stopWords
+- Przy concordia searh dodatkowo obliczany ma być zestaw optymalnego pokrycia patternu. Może siłowo? (jeśli przyjąć max dł. zdania 500 tokenów, to nie powinno być źle)
 - Prawdopodobnie długość example w markers będzie potrzebna tylko anubisowi (który, jak się okazuje, jest wolny). Pomyśleć, do czego można wykorzystać markery, bo ich idea wydaje się niezła.
 - Multi-threading?
 - concordia-server (zastanowić się, czy nie napisać CAT-a)
-- wyłączyć stopWords
 - zastanowić się nad empty hash examples (rozwiązanie: w ogóle nie szukać fraz o pustym hashu, rzucać wyjątek).
 - puścić 100% search test na jrc
 
@@ -10,6 +10,7 @@
 
 ---------------------------- Archive -----------------------------
 
+DONE - wyszukiwanie zdania: wyszukanie najdłuższych pasujących fragmentów Anubisem, 1D (approximate) bin packing. Nazwijmy to concordia search. Wyszukiwane są wszystkie najdłuższe dopasowania patternu dzięki LCP search. Zwracany jest wynik w postaci listy najdłuższych dopasowanych fragmentów, posortowanych malejąco po długości, z maksymalnie 3 przedstawicielami każdej długości. 
 
 DONE 1. lokalizowane to_lower (wykorzystać utf8case, naprawić testy)
 DONE 2. anonimizacja zdań

concordia-concordiasearch-jrc.sh

@@ -0,0 +1,5 @@
+#!/bin/sh
+
+
+./build/concordia-console/concordia-console -c prod/resources/concordia-config/concordia.cfg -x "$1" 
+

concordia-console/concordia-console.cpp

@@ -25,9 +25,12 @@ int main(int argc, char** argv) {
                                  "Concordia configuration file (required)")
         ("simple-search,s", boost::program_options::value<std::string>(),
                                  "Pattern to be searched in the index")
-        ("silent,n", "While searching with simple-search, do not output search results")
+        ("silent,n",
+            "While searching with simple-search, do not output search results")
         ("anubis-search,a", boost::program_options::value<std::string>(),
-                                 "Pattern to be searched by anubis search in the index")
+                        "Pattern to be searched by anubis search in the index")
+        ("concordia-search,x", boost::program_options::value<std::string>(),
+                     "Pattern to be searched by concordia search in the index")
         ("read-file,r", boost::program_options::value<std::string>(),
                                  "File to be read and added to index");
 
@@ -102,6 +105,38 @@ int main(int argc, char** argv) {
                               << std::endl;
                 }
             }
+        } else if  (cli.count("concordia-search")) {
+            std::string pattern = cli["concordia-search"].as<std::string>();
+            std::cout << "\tConcordia searching for pattern: \"" << pattern <<
+                                                          "\"" << std::endl;
+            time_start = boost::posix_time::microsec_clock::local_time();
+            boost::shared_ptr<ConcordiaSearchResult> result =
+                                             concordia.concordiaSearch(pattern);
+            time_end = boost::posix_time::microsec_clock::local_time();
+            msdiff = time_end - time_start;
+
+            std::cout << "\tPattern used: " << std::endl << "\t\t";
+            BOOST_FOREACH(std::string token, result->getTokenVector()) {
+                std::cout << token << " ";
+            }
+            std::cout << std::endl;
+
+            std::cout << "\tFound: " << result->getFragments().size()
+                << " matches. "  << "Search took: " <<
+                          msdiff.total_milliseconds() << "ms." << std::endl;
+            if (!cli.count("silent")) {
+                BOOST_FOREACH(MatchedPatternFragment fragment,
+                              result->getFragments()) {
+                    std::cout << "\t\tfound matching fragment "
+                              << "(exampleId, exampleOffset,"
+                              << " patternOffset, length): "
+                              << fragment.getExampleId() << ","
+                              << fragment.getExampleOffset() << ","
+                              << fragment.getPatternOffset() << ","
+                              << fragment.getMatchedLength() << ","
+                              << std::endl;
+                }
+            }
         } else if (cli.count("read-file")) {
             std::string filePath = cli["read-file"].as<std::string>();
             std::cout << "\tReading sentences from file: " << filePath <<

concordia-runner-jrc.sh

@@ -12,10 +12,10 @@ echo "CONCORDIA RUNNER: reading from file"
 ./build/concordia-console/concordia-console -c prod/resources/concordia-config/concordia.cfg -r prod/resources/text-files/jrc_smaller.txt
 
 
-echo "CONCORDIA RUNNER: anubis searching for pattern: \"Współpraca Państw Członkowskich i Komisji Europejskiej\""
+echo "CONCORDIA RUNNER: concordia searching for pattern: \"Współpraca Państw Członkowskich i Komisji Europejskiej\""
-./build/concordia-console/concordia-console -c prod/resources/concordia-config/concordia.cfg -a "Współpraca Państw Członkowskich i Komisji Europejskiej"
+./build/concordia-console/concordia-console -c prod/resources/concordia-config/concordia.cfg -x "Współpraca Państw Członkowskich i Komisji Europejskiej"
-echo "CONCORDIA RUNNER: anubis searching for pattern: \"8. W odniesieniu do artykułu 45 ustęp 12\""
+echo "CONCORDIA RUNNER: concordia searching for pattern: \"8. W odniesieniu do artykułu 45 ustęp 12\""
-./build/concordia-console/concordia-console -c prod/resources/concordia-config/concordia.cfg -a "8. W odniesieniu do artykułu 45 ustęp 12"
+./build/concordia-console/concordia-console -c prod/resources/concordia-config/concordia.cfg -x "8. W odniesieniu do artykułu 45 ustęp 12"
 
 echo "CONCORDIA RUNNER: searching for pattern: \"Parlamentu Europejskiego\""
 ./build/concordia-console/concordia-console -c prod/resources/concordia-config/concordia.cfg -s "Parlamentu Europejskiego" -n

concordia/CMakeLists.txt

@@ -6,6 +6,8 @@ foreach(dir ${ALL_DIRECTORIES})
 endforeach(dir)
 
 add_library(concordia SHARED
+  concordia_search_result.cpp
+  matched_pattern_fragment.cpp
   anubis_searcher.cpp
   regex_replacement.cpp
   sentence_anonymizer.cpp

concordia/anubis_searcher.cpp

@@ -11,6 +11,42 @@ AnubisSearcher::AnubisSearcher() {
 AnubisSearcher::~AnubisSearcher() {
 }
 
+void AnubisSearcher::concordiaSearch(
+            boost::shared_ptr<ConcordiaSearchResult> result,
+            boost::shared_ptr<std::vector<sauchar_t> > T,
+            boost::shared_ptr<std::vector<SUFFIX_MARKER_TYPE> > markers,
+            boost::shared_ptr<std::vector<saidx_t> > SA,
+            const std::vector<INDEX_CHARACTER_TYPE> & pattern)
+                                                 throw(ConcordiaException) {
+    // add fragments to result and sort them
+
+    std::vector<sauchar_t> patternVector =
+        Utils::indexVectorToSaucharVector(pattern);
+
+    if (patternVector.size() !=
+        pattern.size() * sizeof(INDEX_CHARACTER_TYPE)) {
+        throw ConcordiaException("Increasing pattern resolution went wrong.");
+    }
+
+    for (int offset = 0; offset < pattern.size(); offset++) {
+        int highResOffset = offset * sizeof(INDEX_CHARACTER_TYPE);
+        std::vector<sauchar_t> currentPattern(
+            patternVector.begin()+highResOffset, patternVector.end());
+        SUFFIX_MARKER_TYPE lcpLength;
+        std::vector<SubstringOccurence> occurences =
+            lcpSearch(T, markers, SA, currentPattern, lcpLength);
+
+        BOOST_FOREACH(SubstringOccurence occurence, occurences) {
+            result->addFragment(MatchedPatternFragment(
+                                    occurence.getId(),
+                                    occurence.getOffset(),
+                                    offset,
+                                    lcpLength / sizeof(INDEX_CHARACTER_TYPE)));
+        }
+    }
+
+    result->sortFragments();
+}
 
 std::vector<AnubisSearchResult> AnubisSearcher::anubisSearch(
                 boost::shared_ptr<ConcordiaConfig> config,
@@ -26,21 +62,21 @@ std::vector<AnubisSearchResult> AnubisSearcher::anubisSearch(
     // 2. calculate score for each tmMatches
     // 3. create AnubisSearchResult from tmMatches with scores over threshold
     // 4. sort the AnubisSearchResult vector decending
-    
+
     std::vector<AnubisSearchResult> result;
-    for(TmMatchesMapIterator iterator = tmMatchesMap->begin();
+    for (TmMatchesMapIterator iterator = tmMatchesMap->begin();
-        iterator != tmMatchesMap->end(); iterator++) {
+        iterator != tmMatchesMap->end(); ++iterator) {
         TmMatches * tmMatches = iterator->second;
         tmMatches->calculateScore();
-        
+
         if (tmMatches->getScore() >= config->getAnubisThreshold()) {
             result.push_back(AnubisSearchResult(tmMatches->getExampleId(),
                                                 tmMatches->getScore()));
         }
     }
-    
+
     std::sort(result.begin(), result.end(), std::greater<AnubisSearchResult>());
-        
+
     return result;
 }
 
@@ -175,6 +211,7 @@ void AnubisSearcher::_collectResults(
                  boost::shared_ptr<std::vector<SUFFIX_MARKER_TYPE> > markers,
                  boost::shared_ptr<std::vector<saidx_t> > SA,
                  saidx_t left, saidx_t size) {
+    int resultsCount = 0;
     for (saidx_t i = 0; i < size; i++) {
         saidx_t resultPos = SA->at(left + i);
 
@@ -182,6 +219,12 @@ void AnubisSearcher::_collectResults(
             SUFFIX_MARKER_TYPE marker =
                 markers->at(resultPos / sizeof(INDEX_CHARACTER_TYPE));
             result.push_back(SubstringOccurence(marker));
+
+            // truncate results,
+            // we don't need too many identical pattern overlays
+            if (++resultsCount >= CONCORDIA_SEARCH_MAX_RESULTS) {
+                break;
+            }
         }
     }
 }

concordia/anubis_searcher.hpp

@@ -8,10 +8,11 @@
 #include "concordia/substring_occurence.hpp"
 #include "concordia/concordia_exception.hpp"
 #include "concordia/concordia_config.hpp"
+#include "concordia/concordia_search_result.hpp"
 #include "concordia/anubis_search_result.hpp"
 #include "concordia/tm_matches.hpp"
 
-#include<vector>
+#include <vector>
 #include <divsufsort.h>
 
 /*!
@@ -27,6 +28,14 @@ public:
     */
     virtual ~AnubisSearcher();
 
+    void concordiaSearch(
+                boost::shared_ptr<ConcordiaSearchResult> result,
+                boost::shared_ptr<std::vector<sauchar_t> > T,
+                boost::shared_ptr<std::vector<SUFFIX_MARKER_TYPE> > markers,
+                boost::shared_ptr<std::vector<saidx_t> > SA,
+                const std::vector<INDEX_CHARACTER_TYPE> & pattern)
+                                                     throw(ConcordiaException);
+
     std::vector<AnubisSearchResult> anubisSearch(
                 boost::shared_ptr<ConcordiaConfig> config,
                 boost::shared_ptr<std::vector<sauchar_t> > T,

concordia/common/config.hpp.in

@@ -28,3 +28,5 @@ typedef @SUFFIX_MARKER_TYPE@ SUFFIX_MARKER_TYPE;
 //The sentence marker is build as follows: its first bytes store the
 // sentence id. Next, SUFFIX_MARKER_SENTENCE_BYTES store the suffix offset
 // and the last SUFFIX_MARKER_SENTENCE_BYTES store the sentence length.
+
+#define CONCORDIA_SEARCH_MAX_RESULTS 3

concordia/concordia.cpp

@@ -156,3 +156,17 @@ std::vector<AnubisSearchResult> Concordia::anubisSearch(
     }
 }
 
+boost::shared_ptr<ConcordiaSearchResult> Concordia::concordiaSearch(
+                                             const std::string & pattern)
+                                             throw(ConcordiaException) {
+    if (_T->size() > 0) {
+        return _searcher->concordiaSearch(_hashGenerator, _T,
+                                       _markers, _SA, pattern);
+    } else {
+        std::vector<std::string> empty;
+        return boost::shared_ptr<ConcordiaSearchResult>(
+            new ConcordiaSearchResult(empty));
+    }
+}
+
+

concordia/concordia.hpp

@@ -12,6 +12,7 @@
 #include "concordia/concordia_config.hpp"
 #include "concordia/concordia_index.hpp"
 #include "concordia/index_searcher.hpp"
+#include "concordia/concordia_search_result.hpp"
 #include "concordia/anubis_search_result.hpp"
 #include <divsufsort.h>
 
@@ -49,6 +50,10 @@ public:
     std::vector<AnubisSearchResult> anubisSearch(const std::string & pattern)
                                                  throw(ConcordiaException);
 
+    boost::shared_ptr<ConcordiaSearchResult> concordiaSearch(
+                                                 const std::string & pattern)
+                                                 throw(ConcordiaException);
+
     void loadRAMIndexFromDisk() throw(ConcordiaException);
 
     void refreshSAfromRAM() throw(ConcordiaException);

concordia/concordia_config.cpp

@@ -1,5 +1,5 @@
 #include <sstream>
-#include <stdlib.h>     
+#include <stdlib.h>
 #include "concordia/concordia_config.hpp"
 #include "concordia/common/logging.hpp"
 
@@ -46,9 +46,8 @@ ConcordiaConfig::ConcordiaConfig(const std::string & configFilePath)
           ConcordiaConfig::_readConfigParameterStr(NAMED_ENTITIES_PARAM);
     _stopSymbolsFilePath =
           ConcordiaConfig::_readConfigParameterStr(STOP_SYMBOLS_PARAM);
-    _anubisThreshold =
+    _anubisThreshold = atof(ConcordiaConfig::_readConfigParameterStr(
-       atof(
+                                            ANUBIS_THRESHOLD_PARAM).c_str());
-         ConcordiaConfig::_readConfigParameterStr(ANUBIS_THRESHOLD_PARAM).c_str());
 }
 
 ConcordiaConfig::~ConcordiaConfig() {

concordia/concordia_search_result.cpp

@@ -0,0 +1,22 @@
+#include "concordia/concordia_search_result.hpp"
+
+#include <algorithm>
+
+ConcordiaSearchResult::ConcordiaSearchResult(
+                        const std::vector<std::string> & tokenVector):
+                                   _tokenVector(tokenVector) {
+}
+
+ConcordiaSearchResult::~ConcordiaSearchResult() {
+}
+
+void ConcordiaSearchResult::addFragment(
+                    const MatchedPatternFragment & fragment) {
+    _matchedPatternFragments.push_back(fragment);
+}
+
+void ConcordiaSearchResult::sortFragments() {
+    std::sort(_matchedPatternFragments.begin(),
+              _matchedPatternFragments.end(),
+              std::greater<MatchedPatternFragment>());
+}

concordia/concordia_search_result.hpp

@@ -0,0 +1,41 @@
+#ifndef CONCORDIA_SEARCH_RESULT_HDR
+#define CONCORDIA_SEARCH_RESULT_HDR
+
+#include "concordia/matched_pattern_fragment.hpp"
+
+#include <vector>
+#include <string>
+
+/*!
+  Class representing result of concordia search.
+
+*/
+
+class ConcordiaSearchResult {
+public:
+    explicit ConcordiaSearchResult(
+                const std::vector<std::string> & tokenVector);
+
+    /*! Destructor.
+    */
+    virtual ~ConcordiaSearchResult();
+
+    void addFragment(const MatchedPatternFragment & fragment);
+
+    void sortFragments();
+
+    std::vector<std::string> getTokenVector() const {
+        return _tokenVector;
+    }
+
+    std::vector<MatchedPatternFragment> getFragments() const {
+        return _matchedPatternFragments;
+    }
+
+private:
+    std::vector<std::string> _tokenVector;
+
+    std::vector<MatchedPatternFragment> _matchedPatternFragments;
+};
+
+#endif

concordia/index_searcher.cpp

@@ -59,3 +59,19 @@ std::vector<AnubisSearchResult> IndexSearcher::anubisSearch(
                                  hashGenerator->generateHash(pattern);
     return _anubisSearcher->anubisSearch(config, T, markers, SA, hash);
 }
+
+boost::shared_ptr<ConcordiaSearchResult> IndexSearcher::concordiaSearch(
+                  boost::shared_ptr<HashGenerator> hashGenerator,
+                  boost::shared_ptr<std::vector<sauchar_t> > T,
+                  boost::shared_ptr<std::vector<SUFFIX_MARKER_TYPE> > markers,
+                  boost::shared_ptr<std::vector<saidx_t> > SA,
+                  const std::string & pattern) throw(ConcordiaException) {
+    std::vector<INDEX_CHARACTER_TYPE> hash =
+                                 hashGenerator->generateHash(pattern);
+    boost::shared_ptr<ConcordiaSearchResult> result =
+     boost::shared_ptr<ConcordiaSearchResult>(
+       new ConcordiaSearchResult(hashGenerator->generateTokenVector(pattern)));
+
+    _anubisSearcher->concordiaSearch(result, T, markers, SA, hash);
+    return result;
+}

concordia/index_searcher.hpp

@@ -42,6 +42,14 @@ public:
                     boost::shared_ptr<std::vector<SUFFIX_MARKER_TYPE> > markers,
                     boost::shared_ptr<std::vector<saidx_t> > SA,
                     const std::string & pattern) throw(ConcordiaException);
+
+    boost::shared_ptr<ConcordiaSearchResult> concordiaSearch(
+                    boost::shared_ptr<HashGenerator> hashGenerator,
+                    boost::shared_ptr<std::vector<sauchar_t> > T,
+                    boost::shared_ptr<std::vector<SUFFIX_MARKER_TYPE> > markers,
+                    boost::shared_ptr<std::vector<saidx_t> > SA,
+                    const std::string & pattern) throw(ConcordiaException);
+
 private:
     boost::shared_ptr<AnubisSearcher> _anubisSearcher;
 };

concordia/matched_pattern_fragment.cpp

@@ -0,0 +1,16 @@
+#include "concordia/matched_pattern_fragment.hpp"
+
+MatchedPatternFragment::MatchedPatternFragment(
+                                   const SUFFIX_MARKER_TYPE & exampleId,
+                                   const SUFFIX_MARKER_TYPE & exampleOffset,
+                                   const SUFFIX_MARKER_TYPE & patternOffset,
+                                   const SUFFIX_MARKER_TYPE & matchedLength):
+                                   _exampleId(exampleId),
+                                   _exampleOffset(exampleOffset),
+                                   _patternOffset(patternOffset),
+                                   _matchedLength(matchedLength) {
+}
+
+MatchedPatternFragment::~MatchedPatternFragment() {
+}
+

concordia/matched_pattern_fragment.hpp

@@ -0,0 +1,51 @@
+#ifndef MATCHED_PATTERN_FRAGMENT_HDR
+#define MATCHED_PATTERN_FRAGMENT_HDR
+
+#include "concordia/common/config.hpp"
+
+/*!
+  Class representing matched pattern fragment in concordia search.
+
+*/
+
+class MatchedPatternFragment {
+public:
+    MatchedPatternFragment(const SUFFIX_MARKER_TYPE & exampleId,
+                           const SUFFIX_MARKER_TYPE & exampleOffset,
+                           const SUFFIX_MARKER_TYPE & patternOffset,
+                           const SUFFIX_MARKER_TYPE & matchedLength);
+    /*! Destructor.
+    */
+    virtual ~MatchedPatternFragment();
+
+    SUFFIX_MARKER_TYPE getExampleId() const {
+        return _exampleId;
+    }
+
+    SUFFIX_MARKER_TYPE getExampleOffset() const {
+        return _exampleOffset;
+    }
+
+    SUFFIX_MARKER_TYPE getPatternOffset() const {
+        return _patternOffset;
+    }
+
+    SUFFIX_MARKER_TYPE getMatchedLength() const {
+        return _matchedLength;
+    }
+
+    bool operator > (const MatchedPatternFragment & other) const {
+        return (_matchedLength > other.getMatchedLength());
+    }
+
+private:
+    SUFFIX_MARKER_TYPE _exampleId;
+
+    SUFFIX_MARKER_TYPE _exampleOffset;
+
+    SUFFIX_MARKER_TYPE _patternOffset;
+
+    SUFFIX_MARKER_TYPE _matchedLength;
+};
+
+#endif

concordia/t/test_concordia.cpp

@@ -191,16 +191,70 @@ BOOST_AUTO_TEST_CASE( ConcordiaAnubisSearch1 )
     boost::filesystem::remove(TestResourcesManager::getTestFilePath("temp",TEMP_WORD_MAP)); 
     boost::filesystem::remove(TestResourcesManager::getTestFilePath("temp",TEMP_MARKERS)); 
     boost::filesystem::remove(TestResourcesManager::getTestFilePath("temp",TEMP_HASHED_INDEX)); 
+}
+
+BOOST_AUTO_TEST_CASE( ConcordiaSearch1 )
+{
+    Concordia concordia = Concordia(TestResourcesManager::getTestConcordiaConfigFilePath("concordia.cfg"));
+    concordia.addExample(Example("Ala posiada kota",14));
+    concordia.addExample(Example("Ala posiada rysia",51));
+    concordia.addExample(Example("Marysia posiada rysia",123));
+    concordia.addExample(Example("Gosia chyba posiada rysia też",167));
+    concordia.addExample(Example("Ania od wczoraj posiada rysia",45));
+    concordia.refreshSAfromRAM();
+        
+    boost::shared_ptr<ConcordiaSearchResult> searchResult1 = concordia.concordiaSearch("posiada rysia chyba");
+    BOOST_CHECK_EQUAL(searchResult1->getFragments().size(), 7);
 
     /*
-    BOOST_CHECK_EQUAL(searchResult1.at(0).getId(), 123);
+    addFragment 45,2,0,2
-    BOOST_CHECK_EQUAL(searchResult1.at(0).getOffset(), 1);
+    addFragment 51,1,0,2
-    BOOST_CHECK_EQUAL(searchResult1.at(1).getId(), 51);
+    addFragment 123,1,0,2
-    BOOST_CHECK_EQUAL(searchResult1.at(1).getOffset(), 1);
+    addFragment 45,3,1,1
-    
+    addFragment 51,2,1,1
-    // Checking pattern spanning over 2 segments
+    addFragment 123,2,1,1
-    BOOST_CHECK_EQUAL(searchResult2.size(), 0);
+    addFragment 167,1,2,1
     */
+    
+    BOOST_CHECK_EQUAL(searchResult1->getFragments().at(0).getExampleId(), 45);
+    BOOST_CHECK_EQUAL(searchResult1->getFragments().at(0).getExampleOffset(), 2);
+    BOOST_CHECK_EQUAL(searchResult1->getFragments().at(0).getPatternOffset(), 0);
+    BOOST_CHECK_EQUAL(searchResult1->getFragments().at(0).getMatchedLength(), 2);
+
+    BOOST_CHECK_EQUAL(searchResult1->getFragments().at(1).getExampleId(), 51);
+    BOOST_CHECK_EQUAL(searchResult1->getFragments().at(1).getExampleOffset(), 1);
+    BOOST_CHECK_EQUAL(searchResult1->getFragments().at(1).getPatternOffset(), 0);
+    BOOST_CHECK_EQUAL(searchResult1->getFragments().at(1).getMatchedLength(), 2);
+
+    BOOST_CHECK_EQUAL(searchResult1->getFragments().at(2).getExampleId(), 123);
+    BOOST_CHECK_EQUAL(searchResult1->getFragments().at(2).getExampleOffset(), 1);
+    BOOST_CHECK_EQUAL(searchResult1->getFragments().at(2).getPatternOffset(), 0);
+    BOOST_CHECK_EQUAL(searchResult1->getFragments().at(2).getMatchedLength(), 2);
+
+    BOOST_CHECK_EQUAL(searchResult1->getFragments().at(3).getExampleId(), 45);
+    BOOST_CHECK_EQUAL(searchResult1->getFragments().at(3).getExampleOffset(), 3);
+    BOOST_CHECK_EQUAL(searchResult1->getFragments().at(3).getPatternOffset(), 1);
+    BOOST_CHECK_EQUAL(searchResult1->getFragments().at(3).getMatchedLength(), 1);
+
+    BOOST_CHECK_EQUAL(searchResult1->getFragments().at(4).getExampleId(), 51);
+    BOOST_CHECK_EQUAL(searchResult1->getFragments().at(4).getExampleOffset(), 2);
+    BOOST_CHECK_EQUAL(searchResult1->getFragments().at(4).getPatternOffset(), 1);
+    BOOST_CHECK_EQUAL(searchResult1->getFragments().at(4).getMatchedLength(), 1);
+
+    BOOST_CHECK_EQUAL(searchResult1->getFragments().at(5).getExampleId(), 123);
+    BOOST_CHECK_EQUAL(searchResult1->getFragments().at(5).getExampleOffset(), 2);
+    BOOST_CHECK_EQUAL(searchResult1->getFragments().at(5).getPatternOffset(), 1);
+    BOOST_CHECK_EQUAL(searchResult1->getFragments().at(5).getMatchedLength(), 1);
+
+
+    BOOST_CHECK_EQUAL(searchResult1->getFragments().at(6).getExampleId(), 167);
+    BOOST_CHECK_EQUAL(searchResult1->getFragments().at(6).getExampleOffset(), 1);
+    BOOST_CHECK_EQUAL(searchResult1->getFragments().at(6).getPatternOffset(), 2);
+    BOOST_CHECK_EQUAL(searchResult1->getFragments().at(6).getMatchedLength(), 1);
+
+    boost::filesystem::remove(TestResourcesManager::getTestFilePath("temp",TEMP_WORD_MAP)); 
+    boost::filesystem::remove(TestResourcesManager::getTestFilePath("temp",TEMP_MARKERS)); 
+    boost::filesystem::remove(TestResourcesManager::getTestFilePath("temp",TEMP_HASHED_INDEX)); 
 }
 
 BOOST_AUTO_TEST_SUITE_END()