author | Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com> |
Mon, 03 May 2010 13:33:22 +0300 | |
changeset 1 | 6f2c1c46032b |
parent 0 | 671dee74050a |
child 2 | 6c1a2771f4b7 |
permissions | -rw-r--r-- |
0
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
1 |
/* |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
2 |
* Copyright (c) 2010 Nokia Corporation and/or its subsidiary(-ies). |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
3 |
* All rights reserved. |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
4 |
* This component and the accompanying materials are made available |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
5 |
* under the terms of "Eclipse Public License v1.0" |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
6 |
* which accompanies this distribution, and is available |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
7 |
* at the URL "http://www.eclipse.org/legal/epl-v10.html". |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
8 |
* |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
9 |
* Initial Contributors: |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
10 |
* Nokia Corporation - initial contribution. |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
11 |
* |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
12 |
* Contributors: |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
13 |
* |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
14 |
* Description: |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
15 |
* |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
16 |
*/ |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
17 |
|
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
18 |
|
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
19 |
#include "CLucene.h" |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
20 |
#include "CLucene/analysis/AnalysisHeader.h" |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
21 |
#include "CLucene/analysis/Analyzers.h" |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
22 |
|
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
23 |
#include "analyzer.h" |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
24 |
#include "analyzerexp.h" |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
25 |
#include "cpixanalyzer.h" |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
26 |
#include "cluceneext.h" |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
27 |
|
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
28 |
#include "cpixexc.h" |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
29 |
#include "cpixparsetools.h" |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
30 |
|
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
31 |
#include "wchar.h" |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
32 |
#include <string> |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
33 |
#include <vector> |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
34 |
#include <sstream> |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
35 |
#include <iostream> |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
36 |
|
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
37 |
#include "document.h" |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
38 |
|
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
39 |
#include "indevicecfg.h" |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
40 |
|
1
6f2c1c46032b
Revision: 201015
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
0
diff
changeset
|
41 |
#include "initparams.h" |
0
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
42 |
namespace |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
43 |
{ |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
44 |
const char AGGR_NONFILEREADERPROXY_ERR[] |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
45 |
= "Aggregated reader field should be FileReaderProxy instance"; |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
46 |
|
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
47 |
const char AGGR_STREAMREADER_ERR[] |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
48 |
= "Aggregating streamValue-fields not implemented"; |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
49 |
} |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
50 |
|
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
51 |
|
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
52 |
namespace Cpix { |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
53 |
|
1
6f2c1c46032b
Revision: 201015
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
0
diff
changeset
|
54 |
PrefixGenerator::PrefixGenerator( |
6f2c1c46032b
Revision: 201015
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
0
diff
changeset
|
55 |
lucene::analysis::TokenStream* in, |
6f2c1c46032b
Revision: 201015
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
0
diff
changeset
|
56 |
bool deleteTS, |
6f2c1c46032b
Revision: 201015
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
0
diff
changeset
|
57 |
size_t maxPrefixLength) |
6f2c1c46032b
Revision: 201015
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
0
diff
changeset
|
58 |
: TokenFilter(in, deleteTS), |
6f2c1c46032b
Revision: 201015
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
0
diff
changeset
|
59 |
token_(), |
6f2c1c46032b
Revision: 201015
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
0
diff
changeset
|
60 |
prefixLength_(0), |
6f2c1c46032b
Revision: 201015
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
0
diff
changeset
|
61 |
maxPrefixLength_(maxPrefixLength) {} |
6f2c1c46032b
Revision: 201015
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
0
diff
changeset
|
62 |
|
6f2c1c46032b
Revision: 201015
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
0
diff
changeset
|
63 |
|
6f2c1c46032b
Revision: 201015
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
0
diff
changeset
|
64 |
PrefixGenerator::~PrefixGenerator() { |
6f2c1c46032b
Revision: 201015
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
0
diff
changeset
|
65 |
} |
6f2c1c46032b
Revision: 201015
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
0
diff
changeset
|
66 |
|
6f2c1c46032b
Revision: 201015
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
0
diff
changeset
|
67 |
|
6f2c1c46032b
Revision: 201015
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
0
diff
changeset
|
68 |
bool PrefixGenerator::next(lucene::analysis::Token* token) { |
6f2c1c46032b
Revision: 201015
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
0
diff
changeset
|
69 |
token_.setPositionIncrement(0); |
6f2c1c46032b
Revision: 201015
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
0
diff
changeset
|
70 |
|
6f2c1c46032b
Revision: 201015
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
0
diff
changeset
|
71 |
while (prefixLength_ == 0) { |
6f2c1c46032b
Revision: 201015
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
0
diff
changeset
|
72 |
token_.setPositionIncrement(1); // default position increment |
6f2c1c46032b
Revision: 201015
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
0
diff
changeset
|
73 |
if (!input->next(&token_)) { |
6f2c1c46032b
Revision: 201015
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
0
diff
changeset
|
74 |
return false; |
6f2c1c46032b
Revision: 201015
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
0
diff
changeset
|
75 |
} |
6f2c1c46032b
Revision: 201015
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
0
diff
changeset
|
76 |
prefixLength_ = std::min(token_.termTextLength(), maxPrefixLength_); |
6f2c1c46032b
Revision: 201015
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
0
diff
changeset
|
77 |
} |
6f2c1c46032b
Revision: 201015
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
0
diff
changeset
|
78 |
|
6f2c1c46032b
Revision: 201015
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
0
diff
changeset
|
79 |
// Clip token |
6f2c1c46032b
Revision: 201015
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
0
diff
changeset
|
80 |
std::wstring clipped; |
6f2c1c46032b
Revision: 201015
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
0
diff
changeset
|
81 |
clipped = token_.termText(); |
6f2c1c46032b
Revision: 201015
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
0
diff
changeset
|
82 |
token_.setText(clipped.substr(0, prefixLength_).c_str()); |
6f2c1c46032b
Revision: 201015
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
0
diff
changeset
|
83 |
|
6f2c1c46032b
Revision: 201015
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
0
diff
changeset
|
84 |
// Copy |
6f2c1c46032b
Revision: 201015
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
0
diff
changeset
|
85 |
token->set(token_.termText(), token_.startOffset(), token_.endOffset(), token_.type()); |
6f2c1c46032b
Revision: 201015
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
0
diff
changeset
|
86 |
token->setPositionIncrement(token_.getPositionIncrement()); |
6f2c1c46032b
Revision: 201015
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
0
diff
changeset
|
87 |
|
6f2c1c46032b
Revision: 201015
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
0
diff
changeset
|
88 |
// Reduce prefixLength_ |
6f2c1c46032b
Revision: 201015
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
0
diff
changeset
|
89 |
prefixLength_--; |
6f2c1c46032b
Revision: 201015
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
0
diff
changeset
|
90 |
return true; |
6f2c1c46032b
Revision: 201015
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
0
diff
changeset
|
91 |
} |
0
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
92 |
|
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
93 |
AggregateFieldTokenStream::AggregateFieldTokenStream(lucene::analysis::Analyzer& analyzer, |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
94 |
DocumentFieldIterator* fields) |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
95 |
: stream_(), analyzer_( analyzer ), reader_(), fields_( fields ) { |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
96 |
getNextStream(); |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
97 |
} |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
98 |
|
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
99 |
AggregateFieldTokenStream::~AggregateFieldTokenStream() { |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
100 |
_CLDELETE( stream_ ); |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
101 |
delete fields_; |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
102 |
} |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
103 |
|
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
104 |
bool AggregateFieldTokenStream::next(lucene::analysis::Token* token) { |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
105 |
while ( stream_ ) { |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
106 |
if ( stream_->next( token ) ) { |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
107 |
return true; |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
108 |
} |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
109 |
getNextStream(); |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
110 |
} |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
111 |
return false; |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
112 |
} |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
113 |
|
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
114 |
void AggregateFieldTokenStream::close() { |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
115 |
if (stream_) stream_->close(); |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
116 |
_CLDELETE( stream_ ); |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
117 |
_CLDELETE( reader_ ); |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
118 |
} |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
119 |
|
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
120 |
void AggregateFieldTokenStream::getNextStream() |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
121 |
{ |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
122 |
using namespace lucene::document; |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
123 |
using namespace lucene::util; |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
124 |
|
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
125 |
if ( stream_ ) stream_->close(); |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
126 |
_CLDELETE( stream_ ); |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
127 |
_CLDELETE( reader_ ); |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
128 |
|
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
129 |
Field* field = 0; |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
130 |
while (*fields_ && field == NULL) |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
131 |
{ |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
132 |
field = (*fields_)++; |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
133 |
|
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
134 |
if (!field->isAggregated()) |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
135 |
{ |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
136 |
field = 0; |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
137 |
} |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
138 |
} |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
139 |
if (field) { |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
140 |
if (field->stringValue() != NULL) |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
141 |
{ |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
142 |
reader_ = _CLNEW CL_NS(util)::StringReader(field->stringValue(),_tcslen(field->stringValue()),false); |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
143 |
} |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
144 |
else if (field->native().readerValue() != NULL) |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
145 |
{ |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
146 |
Reader* r = field->native().readerValue(); |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
147 |
FileReaderProxy |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
148 |
* frp = |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
149 |
dynamic_cast<FileReaderProxy*>(r); |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
150 |
if (frp == NULL) |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
151 |
{ |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
152 |
_CLTHROWA(CL_ERR_IO, AGGR_NONFILEREADERPROXY_ERR); |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
153 |
} |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
154 |
else |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
155 |
{ |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
156 |
reader_ = frp->clone(); |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
157 |
} |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
158 |
} |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
159 |
else |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
160 |
{ |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
161 |
_CLTHROWA(CL_ERR_IO, AGGR_STREAMREADER_ERR); |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
162 |
} |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
163 |
stream_ = analyzer_.tokenStream( field->name(), reader_ ); |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
164 |
} |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
165 |
} |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
166 |
|
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
167 |
|
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
168 |
AggregateFieldAnalyzer::AggregateFieldAnalyzer(Cpix::Document& document, |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
169 |
lucene::analysis::Analyzer& analyzer) |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
170 |
: analyzer_(analyzer), document_(document) |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
171 |
{ |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
172 |
} |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
173 |
|
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
174 |
lucene::analysis::TokenStream* AggregateFieldAnalyzer::tokenStream(const TCHAR * fieldName, |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
175 |
lucene::util::Reader * reader) { |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
176 |
if ( wcscmp( fieldName, LCPIX_DEFAULT_FIELD ) == 0 ) { |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
177 |
return new AggregateFieldTokenStream( analyzer_, document_.fields()); |
1
6f2c1c46032b
Revision: 201015
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
0
diff
changeset
|
178 |
} else if ( wcscmp( fieldName, LCPIX_DEFAULT_PREFIX_FIELD ) == 0 ) { |
6f2c1c46032b
Revision: 201015
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
0
diff
changeset
|
179 |
return |
6f2c1c46032b
Revision: 201015
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
0
diff
changeset
|
180 |
new PrefixGenerator( |
6f2c1c46032b
Revision: 201015
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
0
diff
changeset
|
181 |
new AggregateFieldTokenStream( analyzer_, document_.fields()), |
6f2c1c46032b
Revision: 201015
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
0
diff
changeset
|
182 |
true, |
6f2c1c46032b
Revision: 201015
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
0
diff
changeset
|
183 |
OPTIMIZED_PREFIX_MAX_LENGTH); |
0
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
184 |
} else { |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
185 |
return analyzer_.tokenStream( fieldName, reader ); |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
186 |
} |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
187 |
} |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
188 |
|
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
189 |
SystemAnalyzer::SystemAnalyzer(lucene::analysis::Analyzer* analyzer) : analyzer_(analyzer) {} |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
190 |
SystemAnalyzer::~SystemAnalyzer() { _CLDELETE(analyzer_); } |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
191 |
|
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
192 |
lucene::analysis::TokenStream* SystemAnalyzer::tokenStream(const TCHAR * fieldName, |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
193 |
lucene::util::Reader * reader) { |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
194 |
using namespace lucene::analysis; |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
195 |
if ( wcscmp( fieldName, LCPIX_DEFAULT_FIELD ) == 0 ) { |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
196 |
// Use standard analyzer without stop filter for this task |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
197 |
TokenStream* ret = _CLNEW standard::StandardTokenizer(reader); |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
198 |
ret = _CLNEW standard::StandardFilter(ret,true); |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
199 |
ret = _CLNEW LowerCaseFilter(ret,true); |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
200 |
return ret; |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
201 |
} else if (wcscmp( fieldName, LCPIX_DOCUID_FIELD) == 0){ |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
202 |
// Use standard analyzer without stop filter for this task |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
203 |
return _CLNEW KeywordTokenizer(reader); |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
204 |
} else if (wcscmp( fieldName, LCPIX_APPCLASS_FIELD ) == 0){ |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
205 |
// Use standard analyzer without stop filter for this task |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
206 |
TokenStream* ret = _CLNEW WhitespaceTokenizer(reader); |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
207 |
ret = _CLNEW LowerCaseFilter(ret,true); |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
208 |
return ret; |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
209 |
} else if (wcscmp( fieldName, LCPIX_MIMETYPE_FIELD ) == 0) { |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
210 |
TokenStream* ret = _CLNEW KeywordTokenizer(reader); |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
211 |
return ret; |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
212 |
} else { |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
213 |
return analyzer_->tokenStream( fieldName, reader ); |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
214 |
} |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
215 |
} |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
216 |
|
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
217 |
// |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
218 |
// Following sections provide the glue code for connecting the |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
219 |
// analyzer definition syntax with analyzer, tokenizers and filter |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
220 |
// implementations. |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
221 |
// |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
222 |
// The glue code is template heavy with the indent of providing |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
223 |
// automation for associating specific keywords with specific |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
224 |
// analyzers, tokenizers and filters implementing corresponding |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
225 |
// CLucene abstractions. Additional classes are needed only if |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
226 |
// filters, tokenizers, etc. accept parameters. |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
227 |
// |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
228 |
// NOTE: To understand the analyzers, it is sufficient to understand |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
229 |
// that an analyzer transforms characters stream into specific token streams |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
230 |
// (e.g. character stream 'foobarmetawords' can be transformed into token |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
231 |
// stream 'foo', 'bar' 'meta' 'words'). Analysis consist of two main |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
232 |
// parts which are tokenization and filtering. Tokenization converts |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
233 |
// the character stream into token stream (e.g. 'FoO bAr' -> 'FoO' 'bAr') |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
234 |
// and filtering modifies the tokens (e.g. lowercase filtering 'FoO' -> |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
235 |
// 'foo', 'bAr' -> 'bar'). Analyzer as an object is responsible for |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
236 |
// constructing a tokenizer and a sequence of filters to perform |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
237 |
// these required tasks. |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
238 |
// |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
239 |
// See the documentation around TokenizerClassEntries and |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
240 |
// FilterClassEntries to see how implementations not taking parameters |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
241 |
// can be easily added. |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
242 |
// |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
243 |
|
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
244 |
using namespace Cpix::AnalyzerExp; |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
245 |
|
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
246 |
/** |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
247 |
* Creates token stream for the given reader and fieldName. |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
248 |
* This class in in many ways similar to CLucene analyzer class |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
249 |
* definition. |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
250 |
*/ |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
251 |
class TokenStreamFactory { |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
252 |
public: |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
253 |
virtual ~TokenStreamFactory(); |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
254 |
virtual lucene::analysis::TokenStream* tokenStream(const wchar_t * fieldName, |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
255 |
lucene::util::Reader * reader) = 0; |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
256 |
}; |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
257 |
|
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
258 |
TokenStreamFactory::~TokenStreamFactory() {}; |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
259 |
|
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
260 |
/** |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
261 |
* Template class used to create CLucene tokenizers. Template |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
262 |
* parameter T must implement lucene::analysis::Tokenizer abstraction. |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
263 |
*/ |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
264 |
template<class T> |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
265 |
class TokenizerFactory : public TokenStreamFactory |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
266 |
{ |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
267 |
public: |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
268 |
TokenizerFactory(const Invokation& invokation) { |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
269 |
if (invokation.params().size() > 0) { |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
270 |
THROW_CPIXEXC(L"Tokenizer %S does not accept parameters", |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
271 |
invokation.id().c_str()); |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
272 |
} |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
273 |
} |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
274 |
virtual lucene::analysis::TokenStream* tokenStream(const TCHAR * /*fieldName*/, |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
275 |
lucene::util::Reader * reader) { |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
276 |
return _CLNEW T(reader); |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
277 |
} |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
278 |
}; |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
279 |
|
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
280 |
/** |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
281 |
* Template class wrapping CLucene analyzers. Template parameter T must |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
282 |
* implement lucene::analysis::Analyzer abstraction. |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
283 |
*/ |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
284 |
template<class T> |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
285 |
class AnalyzerWrap : public TokenStreamFactory |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
286 |
{ |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
287 |
public: |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
288 |
AnalyzerWrap(const Invokation& invokation) : analyzer_() { |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
289 |
if (invokation.params().size() > 0) { |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
290 |
THROW_CPIXEXC(L"Tokenizer %S does not accept parameters", |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
291 |
invokation.id().c_str()); |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
292 |
} |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
293 |
} |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
294 |
virtual lucene::analysis::TokenStream* tokenStream(const TCHAR * fieldName, |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
295 |
lucene::util::Reader * reader) { |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
296 |
return analyzer_.tokenStream(fieldName, reader); |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
297 |
} |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
298 |
private: |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
299 |
T analyzer_; |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
300 |
}; |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
301 |
|
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
302 |
/** |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
303 |
* Template class associated with CLucene filter and a TokenStreamFactory. |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
304 |
* Uses TokenStreamFactory to transform given character stream into tokenstream |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
305 |
* and then applies the given Clucene filter to the token stream. |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
306 |
* The template parameter T must implement lucene::analysis::Filter abstraction. |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
307 |
*/ |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
308 |
template<class T> |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
309 |
class FilterFactory : public TokenStreamFactory |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
310 |
{ |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
311 |
public: |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
312 |
FilterFactory(const Invokation& invokation, auto_ptr<TokenStreamFactory> factory) : factory_(factory) { |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
313 |
if (invokation.params().size() > 0) { |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
314 |
THROW_CPIXEXC(L"Filter %S does not accept parameters", |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
315 |
invokation.id().c_str()); |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
316 |
} |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
317 |
} |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
318 |
virtual lucene::analysis::TokenStream* tokenStream(const TCHAR * fieldName, |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
319 |
lucene::util::Reader * reader) { |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
320 |
return _CLNEW T(factory_->tokenStream(fieldName, reader), true); |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
321 |
} |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
322 |
private: |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
323 |
std::auto_ptr<TokenStreamFactory> factory_; |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
324 |
}; |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
325 |
|
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
326 |
/** |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
327 |
* Specialized Analyzer wrap for CLucene's PerFieldAnalyzer. Specialized |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
328 |
* template is needed because perfield analyzer accepts parameters |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
329 |
* (specific analyzers for different field plus default analyzer) |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
330 |
*/ |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
331 |
template<> |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
332 |
class AnalyzerWrap<lucene::analysis::PerFieldAnalyzerWrapper> : public TokenStreamFactory { |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
333 |
public: |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
334 |
AnalyzerWrap(const Switch& sw) : analyzer_(0) { |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
335 |
using namespace Cpt::Parser; |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
336 |
using namespace lucene::analysis; |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
337 |
|
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
338 |
analyzer_ = _CLNEW PerFieldAnalyzerWrapper(_CLNEW CustomAnalyzer(sw.def())); |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
339 |
|
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
340 |
for (int i = 0; i < sw.cases().size(); i++) { |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
341 |
const Case& cs = *sw.cases()[i]; |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
342 |
for (int j = 0; j < cs.fields().size(); j++) { |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
343 |
analyzer_->addAnalyzer( cs.fields()[j].c_str(), _CLNEW CustomAnalyzer( cs.piping() ) ); |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
344 |
} |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
345 |
} |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
346 |
} |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
347 |
virtual ~AnalyzerWrap() { |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
348 |
_CLDELETE(analyzer_); |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
349 |
} |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
350 |
virtual lucene::analysis::TokenStream* tokenStream(const TCHAR * fieldName, |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
351 |
lucene::util::Reader * reader) { |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
352 |
return analyzer_->tokenStream(fieldName, reader); |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
353 |
} |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
354 |
private: |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
355 |
lucene::analysis::PerFieldAnalyzerWrapper* analyzer_; |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
356 |
}; |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
357 |
|
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
358 |
|
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
359 |
|
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
360 |
/** |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
361 |
* Specialized StopFilter factory. Specialized filter is needed |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
362 |
* because StopFilter needs parameters (stop word list or a language) |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
363 |
*/ |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
364 |
template<> |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
365 |
class FilterFactory<lucene::analysis::StopFilter> : public TokenStreamFactory |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
366 |
{ |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
367 |
public: |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
368 |
FilterFactory(const Invokation& invokation, |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
369 |
auto_ptr<TokenStreamFactory> factory) |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
370 |
:words_(0), ownWords_(0), factory_(factory) { |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
371 |
using namespace Cpt::Parser; |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
372 |
if (invokation.params().size() == 1 && dynamic_cast<Identifier*>(invokation.params()[0])) { |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
373 |
Identifier* id = dynamic_cast<Identifier*>(invokation.params()[0]); |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
374 |
//cpix_LangCode lang; |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
375 |
if (id->id() == CPIX_WLANG_EN) { |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
376 |
words_ = lucene::analysis::StopAnalyzer::ENGLISH_STOP_WORDS; |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
377 |
} else { |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
378 |
THROW_CPIXEXC(L"No prepared stopword list for language code '%S'", |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
379 |
id->id().c_str()); |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
380 |
} |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
381 |
} else { |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
382 |
ownWords_ = new wchar_t*[invokation.params().size()+1]; |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
383 |
memset(ownWords_, 0, sizeof(wchar_t*)*(invokation.params().size()+1)); |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
384 |
// FIXE: args may leak |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
385 |
for (int i = 0; i < invokation.params().size(); i++) { |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
386 |
StringLit* lit = dynamic_cast<StringLit*>(invokation.params()[i]); |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
387 |
if (lit) { |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
388 |
const wstring& str = lit->text(); |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
389 |
ownWords_[i] = new wchar_t[str.length()+1]; |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
390 |
wcscpy(ownWords_[i], str.c_str()); |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
391 |
} else { |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
392 |
THROW_CPIXEXC(L"StopFilter accepts only language identifer or list of strings as a parameters."); |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
393 |
} |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
394 |
} |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
395 |
} |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
396 |
|
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
397 |
} |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
398 |
virtual ~FilterFactory() { |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
399 |
if (ownWords_) { |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
400 |
for (int i = 0; ownWords_[i]; i++) { |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
401 |
delete[] ownWords_[i]; |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
402 |
} |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
403 |
delete[] ownWords_; |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
404 |
} |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
405 |
} |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
406 |
virtual lucene::analysis::TokenStream* tokenStream(const TCHAR * fieldName, |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
407 |
lucene::util::Reader * reader) { |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
408 |
return _CLNEW lucene::analysis::StopFilter(factory_->tokenStream(fieldName, reader), true, ownWords_ ? const_cast<const wchar_t**>(ownWords_) : words_); |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
409 |
} |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
410 |
private: |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
411 |
const wchar_t **words_; |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
412 |
wchar_t **ownWords_; // owned |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
413 |
std::auto_ptr<TokenStreamFactory> factory_; |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
414 |
}; |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
415 |
|
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
416 |
/** |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
417 |
* Specialized SnowballFilter factory is needed, because SnowballFilter |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
418 |
* accepts parameters (the language). |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
419 |
*/ |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
420 |
template<> |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
421 |
class FilterFactory<lucene::analysis::SnowballFilter> : public TokenStreamFactory |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
422 |
{ |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
423 |
public: |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
424 |
FilterFactory(const Invokation& invokation, |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
425 |
auto_ptr<TokenStreamFactory> factory) |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
426 |
: factory_(factory) { |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
427 |
using namespace Cpt::Parser; |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
428 |
if (invokation.params().size() != 1 || !dynamic_cast<Identifier*>(invokation.params()[0])) { |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
429 |
THROW_CPIXEXC(L"Snowball filter takes exactly one identifier as a parameter." ); |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
430 |
} |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
431 |
Identifier* id = dynamic_cast<Identifier*>(invokation.params()[0]); |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
432 |
if (id->id() == CPIX_WLANG_EN) { |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
433 |
lang_ = cpix_LANG_EN; |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
434 |
} else { |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
435 |
THROW_CPIXEXC(L"Language identifier %S is not supported for stemming", |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
436 |
id->id().c_str()); |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
437 |
} |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
438 |
} |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
439 |
virtual lucene::analysis::TokenStream* tokenStream(const TCHAR * fieldName, |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
440 |
lucene::util::Reader * reader) { |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
441 |
return _CLNEW lucene::analysis::SnowballFilter(factory_->tokenStream(fieldName, reader), true, lang_); |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
442 |
} |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
443 |
private: |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
444 |
cpix_LangCode lang_; |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
445 |
std::auto_ptr<TokenStreamFactory> factory_; |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
446 |
}; |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
447 |
|
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
448 |
/** |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
449 |
* Specialized LengthFilter factory is needed, because length filter |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
450 |
* accepts parameters (minimum length and maximum length) |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
451 |
*/ |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
452 |
template<> |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
453 |
class FilterFactory<lucene::analysis::LengthFilter> : public TokenStreamFactory |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
454 |
{ |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
455 |
public: |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
456 |
FilterFactory(const Invokation& invokation, |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
457 |
auto_ptr<TokenStreamFactory> factory) |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
458 |
: factory_(factory) { |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
459 |
using namespace Cpt::Parser; |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
460 |
if (invokation.params().size() != 2 || |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
461 |
!dynamic_cast<IntegerLit*>(invokation.params()[0]) || |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
462 |
!dynamic_cast<IntegerLit*>(invokation.params()[1])) { |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
463 |
THROW_CPIXEXC("Length filter takes exactly two integer parameters"); |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
464 |
} |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
465 |
min_ = dynamic_cast<IntegerLit*>(invokation.params()[0])->value(); |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
466 |
max_ = dynamic_cast<IntegerLit*>(invokation.params()[1])->value(); |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
467 |
} |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
468 |
virtual lucene::analysis::TokenStream* tokenStream(const TCHAR * fieldName, |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
469 |
lucene::util::Reader * reader) { |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
470 |
return _CLNEW lucene::analysis::LengthFilter(factory_->tokenStream(fieldName, reader), true, min_, max_ ); |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
471 |
} |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
472 |
private: |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
473 |
int min_, max_; |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
474 |
std::auto_ptr<TokenStreamFactory> factory_; |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
475 |
}; |
1
6f2c1c46032b
Revision: 201015
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
0
diff
changeset
|
476 |
|
6f2c1c46032b
Revision: 201015
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
0
diff
changeset
|
477 |
/** |
6f2c1c46032b
Revision: 201015
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
0
diff
changeset
|
478 |
* Specialized PrefixGenerator factory is needed, because PrefixGenerator |
6f2c1c46032b
Revision: 201015
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
0
diff
changeset
|
479 |
* requires the max prefix size. |
6f2c1c46032b
Revision: 201015
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
0
diff
changeset
|
480 |
*/ |
6f2c1c46032b
Revision: 201015
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
0
diff
changeset
|
481 |
template<> |
6f2c1c46032b
Revision: 201015
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
0
diff
changeset
|
482 |
class FilterFactory<PrefixGenerator> : public TokenStreamFactory |
6f2c1c46032b
Revision: 201015
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
0
diff
changeset
|
483 |
{ |
6f2c1c46032b
Revision: 201015
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
0
diff
changeset
|
484 |
public: |
6f2c1c46032b
Revision: 201015
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
0
diff
changeset
|
485 |
FilterFactory(const Invokation& invokation, |
6f2c1c46032b
Revision: 201015
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
0
diff
changeset
|
486 |
auto_ptr<TokenStreamFactory> factory) |
6f2c1c46032b
Revision: 201015
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
0
diff
changeset
|
487 |
: factory_(factory) { |
6f2c1c46032b
Revision: 201015
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
0
diff
changeset
|
488 |
using namespace Cpt::Parser; |
6f2c1c46032b
Revision: 201015
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
0
diff
changeset
|
489 |
if (invokation.params().size() != 1 || |
6f2c1c46032b
Revision: 201015
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
0
diff
changeset
|
490 |
!dynamic_cast<IntegerLit*>(invokation.params()[0])) { |
6f2c1c46032b
Revision: 201015
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
0
diff
changeset
|
491 |
THROW_CPIXEXC("Prefix generator takes exactly one integer parameter"); |
6f2c1c46032b
Revision: 201015
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
0
diff
changeset
|
492 |
} |
6f2c1c46032b
Revision: 201015
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
0
diff
changeset
|
493 |
maxPrefixLength_ = dynamic_cast<IntegerLit*>(invokation.params()[0])->value(); |
6f2c1c46032b
Revision: 201015
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
0
diff
changeset
|
494 |
} |
6f2c1c46032b
Revision: 201015
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
0
diff
changeset
|
495 |
virtual lucene::analysis::TokenStream* tokenStream(const TCHAR * fieldName, |
6f2c1c46032b
Revision: 201015
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
0
diff
changeset
|
496 |
lucene::util::Reader * reader) { |
6f2c1c46032b
Revision: 201015
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
0
diff
changeset
|
497 |
return _CLNEW PrefixGenerator(factory_->tokenStream(fieldName, reader), true, maxPrefixLength_ ); |
6f2c1c46032b
Revision: 201015
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
0
diff
changeset
|
498 |
} |
6f2c1c46032b
Revision: 201015
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
0
diff
changeset
|
499 |
private: |
6f2c1c46032b
Revision: 201015
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
0
diff
changeset
|
500 |
int maxPrefixLength_; |
6f2c1c46032b
Revision: 201015
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
0
diff
changeset
|
501 |
std::auto_ptr<TokenStreamFactory> factory_; |
6f2c1c46032b
Revision: 201015
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
0
diff
changeset
|
502 |
}; |
6f2c1c46032b
Revision: 201015
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
0
diff
changeset
|
503 |
|
0
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
504 |
|
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
505 |
typedef auto_ptr<TokenStreamFactory> (*TokenizerFactoryCreator)(const Invokation& invokation); |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
506 |
typedef auto_ptr<TokenStreamFactory> (*FilterFactoryCreator)(const Invokation& invokation, |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
507 |
auto_ptr<TokenStreamFactory> factory); |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
508 |
/** |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
509 |
* Sets up a tokenizer factory with given invokation parameters |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
510 |
*/ |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
511 |
template<class T> |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
512 |
struct TokenizerFactoryCtor |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
513 |
{ |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
514 |
static auto_ptr<TokenStreamFactory> create(const Invokation& invokation) { |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
515 |
return auto_ptr<TokenStreamFactory>(new TokenizerFactory<T>(invokation)); |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
516 |
} |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
517 |
}; |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
518 |
|
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
519 |
/** |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
520 |
* Sets up an analyzer wrap with given invokation parameters |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
521 |
*/ |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
522 |
template<class T> |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
523 |
struct AnalyzerWrapCtor |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
524 |
{ |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
525 |
static auto_ptr<TokenStreamFactory> create(const Invokation& invokation) { |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
526 |
return auto_ptr<TokenStreamFactory>(new AnalyzerWrap<T>(invokation)); |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
527 |
} |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
528 |
}; |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
529 |
|
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
530 |
/** |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
531 |
* Sets up a filter factory with given invokation parameters |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
532 |
*/ |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
533 |
template<class T> |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
534 |
struct FilterFactoryCtor |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
535 |
{ |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
536 |
static auto_ptr<TokenStreamFactory> create(const Invokation& invokation, |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
537 |
auto_ptr<TokenStreamFactory> factory) { |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
538 |
return auto_ptr<TokenStreamFactory>(new FilterFactory<T>(invokation, factory)); |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
539 |
} |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
540 |
}; |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
541 |
|
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
542 |
struct TokenizerClassEntry { |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
543 |
const wchar_t *id_; |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
544 |
TokenizerFactoryCreator createFactory_; |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
545 |
}; |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
546 |
|
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
547 |
// |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
548 |
// Following TokenizerClassEntries and FilterClassEntries contain |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
549 |
// the mapping from tokenizer/analyzer/filter names into glue code |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
550 |
// templates providing the implementations. |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
551 |
// |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
552 |
|
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
553 |
TokenizerClassEntry TokenizerClassEntries[] = { |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
554 |
{CPIX_TOKENIZER_STANDARD, TokenizerFactoryCtor<lucene::analysis::standard::StandardTokenizer>::create}, |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
555 |
{CPIX_TOKENIZER_WHITESPACE, TokenizerFactoryCtor<lucene::analysis::WhitespaceTokenizer>::create}, |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
556 |
{CPIX_TOKENIZER_LETTER, TokenizerFactoryCtor<lucene::analysis::LetterTokenizer>::create}, |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
557 |
{CPIX_TOKENIZER_KEYWORD, TokenizerFactoryCtor<lucene::analysis::KeywordTokenizer>::create}, |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
558 |
{CPIX_ANALYZER_STANDARD, AnalyzerWrapCtor<lucene::analysis::standard::StandardAnalyzer>::create}, |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
559 |
|
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
560 |
// TODO: Add more Tokenizers/Analyzers |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
561 |
|
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
562 |
// Example tokenizer (works as such if tokenizers don't take parameters) |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
563 |
// {CPIX_TOKENIZER_MYTOKENIZER,TokenizerFactoryCtor<MyTokenizer>::create}, |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
564 |
|
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
565 |
// Example analyzer (works as such if analyzer don't take parameters) |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
566 |
// {CPIX_ANALYZER_MYANALYZER, AnalyzerWrapCtor<MyAnalyzer>::create}, |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
567 |
|
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
568 |
{0, 0} |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
569 |
}; |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
570 |
|
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
571 |
struct FilterClassEntry { |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
572 |
const wchar_t *id_; |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
573 |
FilterFactoryCreator createFactory_; |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
574 |
}; |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
575 |
|
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
576 |
FilterClassEntry FilterClassEntries[] = { |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
577 |
{CPIX_FILTER_STANDARD, FilterFactoryCtor<lucene::analysis::standard::StandardFilter>::create}, |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
578 |
{CPIX_FILTER_LOWERCASE, FilterFactoryCtor<lucene::analysis::LowerCaseFilter>::create}, |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
579 |
{CPIX_FILTER_ACCENT, FilterFactoryCtor<lucene::analysis::ISOLatin1AccentFilter>::create}, |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
580 |
{CPIX_FILTER_STOP, FilterFactoryCtor<lucene::analysis::StopFilter>::create}, |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
581 |
{CPIX_FILTER_STEM, FilterFactoryCtor<lucene::analysis::SnowballFilter>::create}, |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
582 |
{CPIX_FILTER_LENGTH, FilterFactoryCtor<lucene::analysis::LengthFilter>::create}, |
1
6f2c1c46032b
Revision: 201015
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
0
diff
changeset
|
583 |
{CPIX_FILTER_PREFIXES, FilterFactoryCtor<PrefixGenerator>::create}, |
0
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
584 |
|
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
585 |
// TODO: Add more Filters |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
586 |
|
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
587 |
// Example filter (works as such if analyzer don't take parameters) |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
588 |
// {CPIX_FILTER_MYFILTER, FilterFactoryCtor<MyFilter>::create}, |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
589 |
|
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
590 |
{0, 0} |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
591 |
}; |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
592 |
|
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
593 |
CustomAnalyzer::CustomAnalyzer(const wchar_t* definition) |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
594 |
{ |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
595 |
using namespace Cpt::Lex; |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
596 |
using namespace Cpt::Parser; |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
597 |
|
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
598 |
|
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
599 |
try |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
600 |
{ |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
601 |
// 1. Setup an tokenizer |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
602 |
Cpix::AnalyzerExp::Tokenizer |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
603 |
tokenizer; |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
604 |
StdLexer |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
605 |
lexer(tokenizer, definition); |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
606 |
|
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
607 |
// 2. Parse |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
608 |
std::auto_ptr<Piping> |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
609 |
def = ParsePiping(lexer); |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
610 |
lexer.eatEof(); |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
611 |
|
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
612 |
// 3. Setup this item based on parsed definition |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
613 |
setup(*def); |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
614 |
} |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
615 |
catch (Cpt::ITxtCtxtExc & exc) |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
616 |
{ |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
617 |
// provide addition info for thrown exception |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
618 |
exc.setContext(definition); |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
619 |
|
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
620 |
// throw it fwd |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
621 |
throw; |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
622 |
} |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
623 |
} |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
624 |
|
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
625 |
CustomAnalyzer::CustomAnalyzer(const Piping& definition) |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
626 |
{ |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
627 |
setup(definition); |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
628 |
} |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
629 |
using namespace Cpt::Parser; |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
630 |
|
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
631 |
void CustomAnalyzer::setup(const Piping& piping) { |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
632 |
|
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
633 |
// If the first item is invokation, create corresponding analyzer/tokenizer |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
634 |
if (dynamic_cast<const Invokation*>(&piping.tokenizer())) |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
635 |
{ |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
636 |
const Invokation& tokenizer = dynamic_cast<const Invokation&>(piping.tokenizer()); |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
637 |
TokenizerClassEntry& tokenizerEntry = getTokenizerEntry( tokenizer.id() ); |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
638 |
factory_ = tokenizerEntry.createFactory_( tokenizer ); |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
639 |
} else { |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
640 |
// If the first item is switch statement, create per-field analyzer |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
641 |
const Switch& tokenizer = dynamic_cast<const Switch&>(piping.tokenizer()); |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
642 |
factory_ = new AnalyzerWrap<lucene::analysis::PerFieldAnalyzerWrapper>( tokenizer ); |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
643 |
} |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
644 |
|
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
645 |
// Add filters |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
646 |
const std::vector<Invokation*>& filters = piping.filters(); |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
647 |
for (int i = 0; i < filters.size(); i++) { |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
648 |
FilterClassEntry& filterEntry = getFilterEntry( filters[i]->id() ); |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
649 |
factory_ = filterEntry.createFactory_( *filters[i], factory_ ); |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
650 |
} |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
651 |
} |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
652 |
|
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
653 |
TokenizerClassEntry& CustomAnalyzer::getTokenizerEntry(std::wstring id) { |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
654 |
|
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
655 |
// Looks for a match in the TokenizerClassEntries. After finding |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
656 |
// a match it returns a proper tokenizer/analyzer implementation provider |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
657 |
// |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
658 |
for (int i = 0; TokenizerClassEntries[i].id_; i++) { |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
659 |
if (id == std::wstring(TokenizerClassEntries[i].id_)) { |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
660 |
return TokenizerClassEntries[i]; |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
661 |
} |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
662 |
} |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
663 |
|
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
664 |
THROW_CPIXEXC(L"Unknown tokenizer '%S'.", |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
665 |
id.c_str()); |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
666 |
} |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
667 |
|
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
668 |
FilterClassEntry& CustomAnalyzer::getFilterEntry(std::wstring id) { |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
669 |
|
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
670 |
// Looks for a match in the FilterClassEntries. After finding |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
671 |
// a match it returns a proper tokenizer/analyzer implementation |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
672 |
// provider |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
673 |
// |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
674 |
for (int i = 0; FilterClassEntries[i].id_; i++) { |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
675 |
if (id == std::wstring(FilterClassEntries[i].id_)) { |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
676 |
return FilterClassEntries[i]; |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
677 |
} |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
678 |
} |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
679 |
|
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
680 |
THROW_CPIXEXC(L"Unknown filter '%S'.", |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
681 |
id.c_str()); |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
682 |
} |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
683 |
|
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
684 |
CustomAnalyzer::~CustomAnalyzer() {} |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
685 |
|
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
686 |
lucene::analysis::TokenStream* CustomAnalyzer::tokenStream(const wchar_t * fieldName, |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
687 |
lucene::util::Reader * reader) { |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
688 |
// Utilizes the the token stream factory to form token stream. |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
689 |
// token stream factory is prepared during custom analyzer construction |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
690 |
// and based on the analyzer definition string. |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
691 |
|
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
692 |
return factory_->tokenStream(fieldName, reader); |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
693 |
} |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
694 |
|
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
695 |
} |
671dee74050a
Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
696 |