util/tools/assistant/lib/fulltextsearch/qanalyzer.cpp
author eckhart.koppen@nokia.com
Wed, 31 Mar 2010 11:06:36 +0300
changeset 7 f7bc934e204c
permissions -rw-r--r--
5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
7
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
     1
/****************************************************************************
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
     2
**
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
     3
** Copyright (C) 2003-2006 Ben van Klinken and the CLucene Team.
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
     4
** All rights reserved.
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
     5
**
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
     6
** Portion Copyright (C) 2010 Nokia Corporation and/or its subsidiary(-ies).
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
     7
** All rights reserved.
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
     8
**
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
     9
** This file may be used under the terms of the GNU Lesser General Public
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
    10
** License version 2.1 as published by the Free Software Foundation and
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
    11
** appearing in the file LICENSE.LGPL included in the packaging of this file.
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
    12
** Please review the following information to ensure the GNU Lesser General
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
    13
** Public License version 2.1 requirements will be met:
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
    14
** http://www.gnu.org/licenses/old-licenses/lgpl-2.1.html.
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
    15
**
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
    16
****************************************************************************/
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
    17
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
    18
#include "qanalyzer_p.h"
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
    19
#include "qclucene_global_p.h"
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
    20
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
    21
#include <CLucene.h>
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
    22
#include <CLucene/analysis/AnalysisHeader.h>
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
    23
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
    24
QT_BEGIN_NAMESPACE
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
    25
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
    26
QCLuceneAnalyzerPrivate::QCLuceneAnalyzerPrivate()
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
    27
    : QSharedData()
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
    28
{
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
    29
    analyzer = 0;
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
    30
    deleteCLuceneAnalyzer = true;
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
    31
}
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
    32
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
    33
QCLuceneAnalyzerPrivate::QCLuceneAnalyzerPrivate(const QCLuceneAnalyzerPrivate &other)
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
    34
    : QSharedData()
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
    35
{
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
    36
    analyzer = _CL_POINTER(other.analyzer);
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
    37
    deleteCLuceneAnalyzer = other.deleteCLuceneAnalyzer;
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
    38
}
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
    39
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
    40
QCLuceneAnalyzerPrivate::~QCLuceneAnalyzerPrivate()
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
    41
{
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
    42
    if (deleteCLuceneAnalyzer)
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
    43
        _CLDECDELETE(analyzer);
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
    44
}
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
    45
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
    46
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
    47
QCLuceneAnalyzer::QCLuceneAnalyzer()
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
    48
    : d(new QCLuceneAnalyzerPrivate())
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
    49
{
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
    50
    //nothing todo, private
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
    51
}
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
    52
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
    53
QCLuceneAnalyzer::~QCLuceneAnalyzer()
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
    54
{
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
    55
    // nothing todo
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
    56
}
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
    57
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
    58
qint32 QCLuceneAnalyzer::positionIncrementGap(const QString &fieldName) const
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
    59
{
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
    60
    Q_UNUSED(fieldName);
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
    61
    return 0;
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
    62
}
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
    63
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
    64
QCLuceneTokenStream QCLuceneAnalyzer::tokenStream(const QString &fieldName,
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
    65
                                                  const QCLuceneReader &reader) const
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
    66
{
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
    67
    TCHAR *fName = QStringToTChar(fieldName);
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
    68
    QCLuceneTokenStream tokenStream;
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
    69
    tokenStream.d->tokenStream = d->analyzer->tokenStream(fName, reader.d->reader);
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
    70
    delete [] fName;
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
    71
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
    72
    return tokenStream;
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
    73
}
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
    74
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
    75
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
    76
QCLuceneStandardAnalyzer::QCLuceneStandardAnalyzer()
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
    77
    : QCLuceneAnalyzer()
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
    78
{
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
    79
    d->analyzer = new lucene::analysis::standard::StandardAnalyzer();
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
    80
}
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
    81
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
    82
QCLuceneStandardAnalyzer::~QCLuceneStandardAnalyzer()
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
    83
{
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
    84
    // nothing todo
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
    85
}
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
    86
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
    87
QCLuceneStandardAnalyzer::QCLuceneStandardAnalyzer(const QStringList &stopWords)
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
    88
{
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
    89
    const TCHAR **tArray = new const TCHAR*[stopWords.count() +1];
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
    90
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
    91
    for(int i = 0; i < stopWords.count(); ++i) {
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
    92
        TCHAR *stopWord = QStringToTChar(stopWords.at(i));
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
    93
        tArray[i] = STRDUP_TtoT(stopWord);
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
    94
        delete [] stopWord;
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
    95
    }
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
    96
    tArray[stopWords.count()] = 0;
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
    97
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
    98
    d->analyzer = new lucene::analysis::standard::StandardAnalyzer(tArray);
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
    99
}
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
   100
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
   101
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
   102
QCLuceneWhitespaceAnalyzer::QCLuceneWhitespaceAnalyzer()
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
   103
    : QCLuceneAnalyzer()
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
   104
{
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
   105
    d->analyzer = new lucene::analysis::WhitespaceAnalyzer();
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
   106
}
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
   107
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
   108
QCLuceneWhitespaceAnalyzer::~QCLuceneWhitespaceAnalyzer()
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
   109
{
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
   110
    // nothing todo
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
   111
}
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
   112
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
   113
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
   114
QCLuceneSimpleAnalyzer::QCLuceneSimpleAnalyzer()
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
   115
    : QCLuceneAnalyzer()
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
   116
{
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
   117
    d->analyzer = new lucene::analysis::SimpleAnalyzer();
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
   118
}
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
   119
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
   120
QCLuceneSimpleAnalyzer::~QCLuceneSimpleAnalyzer()
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
   121
{
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
   122
    // nothing todo
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
   123
}
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
   124
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
   125
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
   126
QCLuceneStopAnalyzer::QCLuceneStopAnalyzer()
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
   127
    : QCLuceneAnalyzer()
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
   128
{
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
   129
    d->analyzer = new lucene::analysis::StopAnalyzer();
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
   130
}
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
   131
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
   132
QCLuceneStopAnalyzer::~QCLuceneStopAnalyzer()
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
   133
{
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
   134
    // nothing todo
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
   135
}
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
   136
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
   137
QCLuceneStopAnalyzer::QCLuceneStopAnalyzer(const QStringList &stopWords)
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
   138
    : QCLuceneAnalyzer()
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
   139
{
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
   140
    const TCHAR **tArray = new const TCHAR*[stopWords.count() +1];
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
   141
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
   142
    for(int i = 0; i < stopWords.count(); ++i) {
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
   143
        TCHAR *stopWord = QStringToTChar(stopWords.at(i));
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
   144
        tArray[i] = STRDUP_TtoT(stopWord);
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
   145
        delete [] stopWord;
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
   146
    }
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
   147
    tArray[stopWords.count()] = 0;
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
   148
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
   149
    d->analyzer = new lucene::analysis::StopAnalyzer(tArray);
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
   150
}
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
   151
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
   152
QStringList QCLuceneStopAnalyzer::englishStopWords() const
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
   153
{
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
   154
    QStringList stopWordList;
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
   155
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
   156
    const TCHAR** stopWords = lucene::analysis::StopAnalyzer::ENGLISH_STOP_WORDS;
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
   157
    for (qint32 i = 0; stopWords[i] != 0; ++i)
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
   158
        stopWordList.append(TCharToQString(stopWords[i]));
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
   159
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
   160
    return stopWordList;
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
   161
}
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
   162
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
   163
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
   164
QCLuceneKeywordAnalyzer::QCLuceneKeywordAnalyzer()
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
   165
    : QCLuceneAnalyzer()
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
   166
{
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
   167
    d->analyzer = new lucene::analysis::KeywordAnalyzer();
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
   168
}
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
   169
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
   170
QCLuceneKeywordAnalyzer::~QCLuceneKeywordAnalyzer()
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
   171
{
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
   172
    // nothing todo
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
   173
}
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
   174
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
   175
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
   176
QCLucenePerFieldAnalyzerWrapper::QCLucenePerFieldAnalyzerWrapper(
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
   177
    QCLuceneAnalyzer *defaultAnalyzer) 
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
   178
    : QCLuceneAnalyzer()
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
   179
{
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
   180
    d->analyzer = new 
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
   181
        lucene::analysis::PerFieldAnalyzerWrapper(defaultAnalyzer->d->analyzer);
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
   182
    
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
   183
    analyzers.append(defaultAnalyzer);
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
   184
    defaultAnalyzer->d->deleteCLuceneAnalyzer = false;
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
   185
}
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
   186
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
   187
QCLucenePerFieldAnalyzerWrapper::~QCLucenePerFieldAnalyzerWrapper()
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
   188
{
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
   189
    qDeleteAll(analyzers);
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
   190
}
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
   191
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
   192
void QCLucenePerFieldAnalyzerWrapper::addAnalyzer(const QString &fieldName, 
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
   193
                                                  QCLuceneAnalyzer *analyzer)
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
   194
{
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
   195
    lucene::analysis::PerFieldAnalyzerWrapper *analyzerWrapper = 
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
   196
        static_cast<lucene::analysis::PerFieldAnalyzerWrapper*> (d->analyzer);
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
   197
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
   198
    if (analyzerWrapper == 0)
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
   199
        return;
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
   200
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
   201
    analyzers.append(analyzer);
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
   202
    analyzer->d->deleteCLuceneAnalyzer = false;
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
   203
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
   204
    TCHAR *fName = QStringToTChar(fieldName);
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
   205
    analyzerWrapper->addAnalyzer(fName, analyzer->d->analyzer);
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
   206
    delete [] fName;
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
   207
}
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
   208
f7bc934e204c 5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff changeset
   209
QT_END_NAMESPACE