| 1 | /****************************************************************************
|
|---|
| 2 | **
|
|---|
| 3 | ** Copyright (C) 2003-2006 Ben van Klinken and the CLucene Team.
|
|---|
| 4 | ** All rights reserved.
|
|---|
| 5 | **
|
|---|
| 6 | ** Portion Copyright (C) 2011 Nokia Corporation and/or its subsidiary(-ies).
|
|---|
| 7 | ** All rights reserved.
|
|---|
| 8 | **
|
|---|
| 9 | ** This file may be used under the terms of the GNU Lesser General Public
|
|---|
| 10 | ** License version 2.1 as published by the Free Software Foundation and
|
|---|
| 11 | ** appearing in the file LICENSE.LGPL included in the packaging of this file.
|
|---|
| 12 | ** Please review the following information to ensure the GNU Lesser General
|
|---|
| 13 | ** Public License version 2.1 requirements will be met:
|
|---|
| 14 | ** http://www.gnu.org/licenses/old-licenses/lgpl-2.1.html.
|
|---|
| 15 | **
|
|---|
| 16 | ****************************************************************************/
|
|---|
| 17 |
|
|---|
| 18 | #include "qanalyzer_p.h"
|
|---|
| 19 | #include "qclucene_global_p.h"
|
|---|
| 20 |
|
|---|
| 21 | #include <CLucene.h>
|
|---|
| 22 | #include <CLucene/analysis/AnalysisHeader.h>
|
|---|
| 23 |
|
|---|
| 24 | QT_BEGIN_NAMESPACE
|
|---|
| 25 |
|
|---|
| 26 | QCLuceneAnalyzerPrivate::QCLuceneAnalyzerPrivate()
|
|---|
| 27 | : QSharedData()
|
|---|
| 28 | {
|
|---|
| 29 | analyzer = 0;
|
|---|
| 30 | deleteCLuceneAnalyzer = true;
|
|---|
| 31 | }
|
|---|
| 32 |
|
|---|
| 33 | QCLuceneAnalyzerPrivate::QCLuceneAnalyzerPrivate(const QCLuceneAnalyzerPrivate &other)
|
|---|
| 34 | : QSharedData()
|
|---|
| 35 | {
|
|---|
| 36 | analyzer = _CL_POINTER(other.analyzer);
|
|---|
| 37 | deleteCLuceneAnalyzer = other.deleteCLuceneAnalyzer;
|
|---|
| 38 | }
|
|---|
| 39 |
|
|---|
| 40 | QCLuceneAnalyzerPrivate::~QCLuceneAnalyzerPrivate()
|
|---|
| 41 | {
|
|---|
| 42 | if (deleteCLuceneAnalyzer)
|
|---|
| 43 | _CLDECDELETE(analyzer);
|
|---|
| 44 | }
|
|---|
| 45 |
|
|---|
| 46 |
|
|---|
| 47 | QCLuceneAnalyzer::QCLuceneAnalyzer()
|
|---|
| 48 | : d(new QCLuceneAnalyzerPrivate())
|
|---|
| 49 | {
|
|---|
| 50 | //nothing todo, private
|
|---|
| 51 | }
|
|---|
| 52 |
|
|---|
| 53 | QCLuceneAnalyzer::~QCLuceneAnalyzer()
|
|---|
| 54 | {
|
|---|
| 55 | // nothing todo
|
|---|
| 56 | }
|
|---|
| 57 |
|
|---|
| 58 | qint32 QCLuceneAnalyzer::positionIncrementGap(const QString &fieldName) const
|
|---|
| 59 | {
|
|---|
| 60 | Q_UNUSED(fieldName);
|
|---|
| 61 | return 0;
|
|---|
| 62 | }
|
|---|
| 63 |
|
|---|
| 64 | QCLuceneTokenStream QCLuceneAnalyzer::tokenStream(const QString &fieldName,
|
|---|
| 65 | const QCLuceneReader &reader) const
|
|---|
| 66 | {
|
|---|
| 67 | TCHAR *fName = QStringToTChar(fieldName);
|
|---|
| 68 | QCLuceneTokenStream tokenStream;
|
|---|
| 69 | tokenStream.d->tokenStream = d->analyzer->tokenStream(fName, reader.d->reader);
|
|---|
| 70 | delete [] fName;
|
|---|
| 71 |
|
|---|
| 72 | return tokenStream;
|
|---|
| 73 | }
|
|---|
| 74 |
|
|---|
| 75 |
|
|---|
| 76 | QCLuceneStandardAnalyzer::QCLuceneStandardAnalyzer()
|
|---|
| 77 | : QCLuceneAnalyzer()
|
|---|
| 78 | {
|
|---|
| 79 | d->analyzer = new lucene::analysis::standard::StandardAnalyzer();
|
|---|
| 80 | }
|
|---|
| 81 |
|
|---|
| 82 | QCLuceneStandardAnalyzer::~QCLuceneStandardAnalyzer()
|
|---|
| 83 | {
|
|---|
| 84 | // nothing todo
|
|---|
| 85 | }
|
|---|
| 86 |
|
|---|
| 87 | QCLuceneStandardAnalyzer::QCLuceneStandardAnalyzer(const QStringList &stopWords)
|
|---|
| 88 | {
|
|---|
| 89 | const TCHAR **tArray = new const TCHAR*[stopWords.count() +1];
|
|---|
| 90 |
|
|---|
| 91 | for(int i = 0; i < stopWords.count(); ++i) {
|
|---|
| 92 | TCHAR *stopWord = QStringToTChar(stopWords.at(i));
|
|---|
| 93 | tArray[i] = STRDUP_TtoT(stopWord);
|
|---|
| 94 | delete [] stopWord;
|
|---|
| 95 | }
|
|---|
| 96 | tArray[stopWords.count()] = 0;
|
|---|
| 97 |
|
|---|
| 98 | d->analyzer = new lucene::analysis::standard::StandardAnalyzer(tArray);
|
|---|
| 99 | }
|
|---|
| 100 |
|
|---|
| 101 |
|
|---|
| 102 | QCLuceneWhitespaceAnalyzer::QCLuceneWhitespaceAnalyzer()
|
|---|
| 103 | : QCLuceneAnalyzer()
|
|---|
| 104 | {
|
|---|
| 105 | d->analyzer = new lucene::analysis::WhitespaceAnalyzer();
|
|---|
| 106 | }
|
|---|
| 107 |
|
|---|
| 108 | QCLuceneWhitespaceAnalyzer::~QCLuceneWhitespaceAnalyzer()
|
|---|
| 109 | {
|
|---|
| 110 | // nothing todo
|
|---|
| 111 | }
|
|---|
| 112 |
|
|---|
| 113 |
|
|---|
| 114 | QCLuceneSimpleAnalyzer::QCLuceneSimpleAnalyzer()
|
|---|
| 115 | : QCLuceneAnalyzer()
|
|---|
| 116 | {
|
|---|
| 117 | d->analyzer = new lucene::analysis::SimpleAnalyzer();
|
|---|
| 118 | }
|
|---|
| 119 |
|
|---|
| 120 | QCLuceneSimpleAnalyzer::~QCLuceneSimpleAnalyzer()
|
|---|
| 121 | {
|
|---|
| 122 | // nothing todo
|
|---|
| 123 | }
|
|---|
| 124 |
|
|---|
| 125 |
|
|---|
| 126 | QCLuceneStopAnalyzer::QCLuceneStopAnalyzer()
|
|---|
| 127 | : QCLuceneAnalyzer()
|
|---|
| 128 | {
|
|---|
| 129 | d->analyzer = new lucene::analysis::StopAnalyzer();
|
|---|
| 130 | }
|
|---|
| 131 |
|
|---|
| 132 | QCLuceneStopAnalyzer::~QCLuceneStopAnalyzer()
|
|---|
| 133 | {
|
|---|
| 134 | // nothing todo
|
|---|
| 135 | }
|
|---|
| 136 |
|
|---|
| 137 | QCLuceneStopAnalyzer::QCLuceneStopAnalyzer(const QStringList &stopWords)
|
|---|
| 138 | : QCLuceneAnalyzer()
|
|---|
| 139 | {
|
|---|
| 140 | const TCHAR **tArray = new const TCHAR*[stopWords.count() +1];
|
|---|
| 141 |
|
|---|
| 142 | for(int i = 0; i < stopWords.count(); ++i) {
|
|---|
| 143 | TCHAR *stopWord = QStringToTChar(stopWords.at(i));
|
|---|
| 144 | tArray[i] = STRDUP_TtoT(stopWord);
|
|---|
| 145 | delete [] stopWord;
|
|---|
| 146 | }
|
|---|
| 147 | tArray[stopWords.count()] = 0;
|
|---|
| 148 |
|
|---|
| 149 | d->analyzer = new lucene::analysis::StopAnalyzer(tArray);
|
|---|
| 150 | }
|
|---|
| 151 |
|
|---|
| 152 | QStringList QCLuceneStopAnalyzer::englishStopWords() const
|
|---|
| 153 | {
|
|---|
| 154 | QStringList stopWordList;
|
|---|
| 155 |
|
|---|
| 156 | const TCHAR** stopWords = lucene::analysis::StopAnalyzer::ENGLISH_STOP_WORDS;
|
|---|
| 157 | for (qint32 i = 0; stopWords[i] != 0; ++i)
|
|---|
| 158 | stopWordList.append(TCharToQString(stopWords[i]));
|
|---|
| 159 |
|
|---|
| 160 | return stopWordList;
|
|---|
| 161 | }
|
|---|
| 162 |
|
|---|
| 163 |
|
|---|
| 164 | QCLuceneKeywordAnalyzer::QCLuceneKeywordAnalyzer()
|
|---|
| 165 | : QCLuceneAnalyzer()
|
|---|
| 166 | {
|
|---|
| 167 | d->analyzer = new lucene::analysis::KeywordAnalyzer();
|
|---|
| 168 | }
|
|---|
| 169 |
|
|---|
| 170 | QCLuceneKeywordAnalyzer::~QCLuceneKeywordAnalyzer()
|
|---|
| 171 | {
|
|---|
| 172 | // nothing todo
|
|---|
| 173 | }
|
|---|
| 174 |
|
|---|
| 175 |
|
|---|
| 176 | QCLucenePerFieldAnalyzerWrapper::QCLucenePerFieldAnalyzerWrapper(
|
|---|
| 177 | QCLuceneAnalyzer *defaultAnalyzer)
|
|---|
| 178 | : QCLuceneAnalyzer()
|
|---|
| 179 | {
|
|---|
| 180 | d->analyzer = new
|
|---|
| 181 | lucene::analysis::PerFieldAnalyzerWrapper(defaultAnalyzer->d->analyzer);
|
|---|
| 182 |
|
|---|
| 183 | analyzers.append(defaultAnalyzer);
|
|---|
| 184 | defaultAnalyzer->d->deleteCLuceneAnalyzer = false;
|
|---|
| 185 | }
|
|---|
| 186 |
|
|---|
| 187 | QCLucenePerFieldAnalyzerWrapper::~QCLucenePerFieldAnalyzerWrapper()
|
|---|
| 188 | {
|
|---|
| 189 | qDeleteAll(analyzers);
|
|---|
| 190 | }
|
|---|
| 191 |
|
|---|
| 192 | void QCLucenePerFieldAnalyzerWrapper::addAnalyzer(const QString &fieldName,
|
|---|
| 193 | QCLuceneAnalyzer *analyzer)
|
|---|
| 194 | {
|
|---|
| 195 | lucene::analysis::PerFieldAnalyzerWrapper *analyzerWrapper =
|
|---|
| 196 | static_cast<lucene::analysis::PerFieldAnalyzerWrapper*> (d->analyzer);
|
|---|
| 197 |
|
|---|
| 198 | if (analyzerWrapper == 0)
|
|---|
| 199 | return;
|
|---|
| 200 |
|
|---|
| 201 | analyzers.append(analyzer);
|
|---|
| 202 | analyzer->d->deleteCLuceneAnalyzer = false;
|
|---|
| 203 |
|
|---|
| 204 | TCHAR *fName = QStringToTChar(fieldName);
|
|---|
| 205 | analyzerWrapper->addAnalyzer(fName, analyzer->d->analyzer);
|
|---|
| 206 | delete [] fName;
|
|---|
| 207 | }
|
|---|
| 208 |
|
|---|
| 209 | QT_END_NAMESPACE
|
|---|