author | eckhart.koppen@nokia.com |
Wed, 31 Mar 2010 11:06:36 +0300 | |
changeset 7 | f7bc934e204c |
permissions | -rw-r--r-- |
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 "qtokenizer_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 |
QCLuceneTokenizer::QCLuceneTokenizer() |
f7bc934e204c
5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff
changeset
|
27 |
: QCLuceneTokenStream() |
f7bc934e204c
5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff
changeset
|
28 |
{ |
f7bc934e204c
5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff
changeset
|
29 |
// nothing todo |
f7bc934e204c
5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff
changeset
|
30 |
} |
f7bc934e204c
5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff
changeset
|
31 |
|
f7bc934e204c
5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff
changeset
|
32 |
QCLuceneTokenizer::QCLuceneTokenizer(const QCLuceneReader &reader) |
f7bc934e204c
5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff
changeset
|
33 |
: QCLuceneTokenStream() |
f7bc934e204c
5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff
changeset
|
34 |
, reader(reader) |
f7bc934e204c
5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff
changeset
|
35 |
{ |
f7bc934e204c
5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff
changeset
|
36 |
// nothing todo |
f7bc934e204c
5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff
changeset
|
37 |
} |
f7bc934e204c
5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff
changeset
|
38 |
|
f7bc934e204c
5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff
changeset
|
39 |
QCLuceneTokenizer::~QCLuceneTokenizer() |
f7bc934e204c
5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff
changeset
|
40 |
{ |
f7bc934e204c
5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff
changeset
|
41 |
close(); |
f7bc934e204c
5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff
changeset
|
42 |
} |
f7bc934e204c
5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff
changeset
|
43 |
|
f7bc934e204c
5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff
changeset
|
44 |
void QCLuceneTokenizer::close() |
f7bc934e204c
5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff
changeset
|
45 |
{ |
f7bc934e204c
5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff
changeset
|
46 |
d->tokenStream->close(); |
f7bc934e204c
5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff
changeset
|
47 |
} |
f7bc934e204c
5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff
changeset
|
48 |
|
f7bc934e204c
5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff
changeset
|
49 |
bool QCLuceneTokenizer::next(QCLuceneToken &token) |
f7bc934e204c
5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff
changeset
|
50 |
{ |
f7bc934e204c
5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff
changeset
|
51 |
return d->tokenStream->next(token.d->token); |
f7bc934e204c
5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff
changeset
|
52 |
} |
f7bc934e204c
5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff
changeset
|
53 |
|
f7bc934e204c
5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff
changeset
|
54 |
|
f7bc934e204c
5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff
changeset
|
55 |
QCLuceneStandardTokenizer::QCLuceneStandardTokenizer(const QCLuceneReader &reader) |
f7bc934e204c
5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff
changeset
|
56 |
: QCLuceneTokenizer(reader) |
f7bc934e204c
5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff
changeset
|
57 |
{ |
f7bc934e204c
5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff
changeset
|
58 |
d->tokenStream = |
f7bc934e204c
5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff
changeset
|
59 |
new lucene::analysis::standard::StandardTokenizer(reader.d->reader); |
f7bc934e204c
5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff
changeset
|
60 |
} |
f7bc934e204c
5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff
changeset
|
61 |
|
f7bc934e204c
5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff
changeset
|
62 |
QCLuceneStandardTokenizer::~QCLuceneStandardTokenizer() |
f7bc934e204c
5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff
changeset
|
63 |
{ |
f7bc934e204c
5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff
changeset
|
64 |
// nothing todo |
f7bc934e204c
5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff
changeset
|
65 |
} |
f7bc934e204c
5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff
changeset
|
66 |
|
f7bc934e204c
5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff
changeset
|
67 |
bool QCLuceneStandardTokenizer::readApostrophe(const QString &string, |
f7bc934e204c
5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff
changeset
|
68 |
QCLuceneToken &token) |
f7bc934e204c
5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff
changeset
|
69 |
{ |
f7bc934e204c
5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff
changeset
|
70 |
lucene::analysis::standard::StandardTokenizer *stdTokenizer = |
f7bc934e204c
5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff
changeset
|
71 |
static_cast<lucene::analysis::standard::StandardTokenizer*> (d->tokenStream); |
f7bc934e204c
5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff
changeset
|
72 |
|
f7bc934e204c
5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff
changeset
|
73 |
if (stdTokenizer == 0) |
f7bc934e204c
5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff
changeset
|
74 |
return false; |
f7bc934e204c
5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff
changeset
|
75 |
|
f7bc934e204c
5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff
changeset
|
76 |
TCHAR* value = QStringToTChar(string); |
f7bc934e204c
5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff
changeset
|
77 |
lucene::util::StringBuffer buffer(value); |
f7bc934e204c
5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff
changeset
|
78 |
bool retValue = stdTokenizer->ReadApostrophe(&buffer, token.d->token); |
f7bc934e204c
5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff
changeset
|
79 |
delete [] value; |
f7bc934e204c
5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff
changeset
|
80 |
|
f7bc934e204c
5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff
changeset
|
81 |
return retValue; |
f7bc934e204c
5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff
changeset
|
82 |
} |
f7bc934e204c
5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff
changeset
|
83 |
|
f7bc934e204c
5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff
changeset
|
84 |
bool QCLuceneStandardTokenizer::readAt(const QString &string, QCLuceneToken &token) |
f7bc934e204c
5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff
changeset
|
85 |
{ |
f7bc934e204c
5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff
changeset
|
86 |
lucene::analysis::standard::StandardTokenizer *stdTokenizer = |
f7bc934e204c
5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff
changeset
|
87 |
static_cast<lucene::analysis::standard::StandardTokenizer*> (d->tokenStream); |
f7bc934e204c
5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff
changeset
|
88 |
|
f7bc934e204c
5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff
changeset
|
89 |
if (stdTokenizer == 0) |
f7bc934e204c
5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff
changeset
|
90 |
return false; |
f7bc934e204c
5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff
changeset
|
91 |
|
f7bc934e204c
5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff
changeset
|
92 |
TCHAR* value = QStringToTChar(string); |
f7bc934e204c
5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff
changeset
|
93 |
lucene::util::StringBuffer buffer(value); |
f7bc934e204c
5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff
changeset
|
94 |
bool retValue = stdTokenizer->ReadAt(&buffer, token.d->token); |
f7bc934e204c
5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff
changeset
|
95 |
delete [] value; |
f7bc934e204c
5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff
changeset
|
96 |
|
f7bc934e204c
5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff
changeset
|
97 |
return retValue; |
f7bc934e204c
5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff
changeset
|
98 |
} |
f7bc934e204c
5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff
changeset
|
99 |
|
f7bc934e204c
5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff
changeset
|
100 |
bool QCLuceneStandardTokenizer::readCompany(const QString &string, |
f7bc934e204c
5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff
changeset
|
101 |
QCLuceneToken &token) |
f7bc934e204c
5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff
changeset
|
102 |
{ |
f7bc934e204c
5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff
changeset
|
103 |
lucene::analysis::standard::StandardTokenizer *stdTokenizer = |
f7bc934e204c
5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff
changeset
|
104 |
static_cast<lucene::analysis::standard::StandardTokenizer*> (d->tokenStream); |
f7bc934e204c
5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff
changeset
|
105 |
|
f7bc934e204c
5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff
changeset
|
106 |
if (stdTokenizer == 0) |
f7bc934e204c
5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff
changeset
|
107 |
return false; |
f7bc934e204c
5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff
changeset
|
108 |
|
f7bc934e204c
5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff
changeset
|
109 |
TCHAR* value = QStringToTChar(string); |
f7bc934e204c
5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff
changeset
|
110 |
lucene::util::StringBuffer buffer(value); |
f7bc934e204c
5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff
changeset
|
111 |
bool retValue = stdTokenizer->ReadCompany(&buffer, token.d->token); |
f7bc934e204c
5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff
changeset
|
112 |
delete [] value; |
f7bc934e204c
5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff
changeset
|
113 |
|
f7bc934e204c
5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff
changeset
|
114 |
return retValue; |
f7bc934e204c
5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff
changeset
|
115 |
} |
f7bc934e204c
5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff
changeset
|
116 |
|
f7bc934e204c
5cabc75a39ca2f064f70b40f72ed93c74c4dc19b
eckhart.koppen@nokia.com
parents:
diff
changeset
|
117 |
QT_END_NAMESPACE |