1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
|
<?php
/**
* Create language models for https://github.com/crodas/LanguageDetector based
* on translation data in your wiki.
*
* @author Niklas Laxström
*
* @copyright Copyright © 2013, Niklas Laxström
* @license GPL-2.0+
* @file
*/
// Standard boilerplate to define $IP
if ( getenv( 'MW_INSTALL_PATH' ) !== false ) {
$IP = getenv( 'MW_INSTALL_PATH' );
} else {
$dir = __DIR__;
$IP = "$dir/../../..";
}
require_once "$IP/maintenance/Maintenance.php";
class LanguageModelCreator extends Maintenance {
protected $changes = array();
public function __construct() {
parent::__construct();
$this->mDescription = <<<TXT
Create language models for https://github.com/crodas/LanguageDetector based
on translation data in your wiki. It is safe to kill and restart the script.
List of pages and filtered language data is cached for 24 hours. Json files
present will be used, so don't forget to delete before new run.
TXT;
}
public function execute() {
global $wgTranslateMessageNamespaces;
ini_set( 'memory_limit', -1 );
// How many messages per language to use.
// Language is skipped if it has less than 1000 translations.
$messages = 5000;
$languages = TranslateUtils::getLanguageNames( 'en' );
$cache = wfGetCache( CACHE_DB );
$key = wfMemcKey( __METHOD__, $messages );
$pages = $cache->get( $key );
if ( !is_array( $pages ) ) {
$dbr = wfGetDB( DB_SLAVE );
$conds = array();
$conds[] = 'page_title' . $dbr->buildLike( $dbr->anyString(), '/', $dbr->anyString() );
$conds['page_namespace'] = $wgTranslateMessageNamespaces;
echo "Before query\n";
$res = $dbr->select(
array( 'page' ),
array( 'page_title, page_id' ),
$conds,
__METHOD__
);
echo "After query\n";
$total = $res->numRows();
$index = 0;
foreach ( $res as $row ) {
$index++;
$code = substr( $row->page_title, strrpos( $row->page_title, '/' ) + 1 );
if ( isset( $languages[$code] ) ) {
$pages[$code][] = $row->page_id;
}
if ( $index % 10000 === 0 ) {
$progress = number_format( $index / $total * 100, 2 );
echo "$progress%\n";
}
}
echo "\n";
foreach ( array_keys( $pages ) as $code ) {
if ( count( $pages[$code] ) > $messages ) {
$pages[$code] = array_slice( $pages[$code], 0, $messages );
}
$pages[$code] = implode( '|', $pages[$code] );
}
echo "After code map\n";
ksort( $pages );
echo "After sort map\n";
$cache->set( $key, $pages, 3600*24 );
echo "After set map\n";
}
unset( $pages['qqq'] );
unset( $pages['de-formal'] );
unset( $pages['nl-informal'] );
unset( $pages['en-gb'] );
$pids = array();
$threads = 2;
foreach ( $pages as $code => $pageids ) {
$pid = ( $threads > 1 ) ? pcntl_fork() : -1;
if ( $pid === 0 ) {
// Child, reseed because there is no bug in PHP:
// http://bugs.php.net/bug.php?id=42465
mt_srand( getmypid() );
$this->analyzeLanguage( $code, $pageids );
exit();
} elseif ( $pid === -1 ) {
// Fork failed or one thread, do it serialized
$this->analyzeLanguage( $code, $pageids );
} else {
// Main thread
$pids[] = $pid;
}
// If we hit the thread limit, wait for any child to finish.
if ( count( $pids ) >= $threads ) {
$status = 0;
$pid = pcntl_wait( $status );
unset( $pids[$pid] );
}
}
foreach ( $pids as $pid ) {
$status = 0;
pcntl_waitpid( $pid, $status );
}
$this->output( "Combining languages\n" );
$huge = array();
foreach ( glob( 'temp-*.json' ) as $file ) {
$contents = file_get_contents( $file );
$json = FormatJson::decode( $contents, true );
$huge = array_merge( $json, $huge );
$huge['data'] = array_merge( $json['data'], $huge['data'] );
}
$json = FormatJson::encode( $huge, true, FormatJson::ALL_OK );
file_put_contents( 'translatewiki.net.json', $json );
}
protected function analyzeLanguage( $code, $ids ) {
if ( file_exists( "temp-$code.json" ) ) {
$this->output( "$code MODEL EXISTS\n" );
return;
}
$text = $this->cacheSourceText( $code, $ids );
if ( $text === '' ) {
return;
}
$config = new LanguageDetector\Config;
$config->useMb( true );
$c = new LanguageDetector\Learn( $config );
$c->addSample( $code, $text );
$c->addStepCallback( function( $lang, $status ) {
echo "Learning {$lang}: $status\n";
} );
$target = LanguageDetector\AbstractFormat::initFormatByPath( "temp-$code.json" );
$c->save( $target );
}
protected function cacheSourceText( $code, $ids ) {
$cache = wfGetCache( CACHE_DB );
$key = wfMemcKey( __CLASS__, 'cc', $code );
$text = $cache->get( $key );
if ( !is_string( $text ) ) {
$snippets = array();
$ids = explode( '|', $ids );
$len = count( $ids );
if ( $len < 1000 ) {
$this->output( "$code: $len SKIPPED\n" );
return '';
} else {
$this->output( "$code PROCESSING\n" );
}
$time = microtime( true );
foreach ( $ids as $id ) {
$params = new FauxRequest( array(
'pageid' => $id,
'action' => 'parse',
'prop' => 'text',
'disablepp' => 'true',
) );
$api = new ApiMain( $params );
$api->execute();
$result = $api->getResult()->getResultData(
null,
array( 'BC' => array() )
);
$text = $result['parse']['text']['*'];
$text = strip_tags( $text );
$text = str_replace( '!!FUZZY!!', '', $text );
$text = preg_replace( '/\$[0-9]/', '', $text );
$text = trim( $text );
$snippets[] = $text;
}
$text = implode( ' ', $snippets );
$cache->set( $key, $text, 3600*24 );
$delta = microtime( true ) - $time;
$this->output( "$code TOOK $delta\n" );
} else {
$this->output( "$code FROM CACHE\n" );
}
return $text;
}
}
$maintClass = 'LanguageModelCreator';
require_once RUN_MAINTENANCE_IF_MAIN;
|
