1
|
|
|
<?php |
2
|
|
|
|
3
|
|
|
/** |
4
|
|
|
* This file is part of dispositif/wikibot application (@github) |
5
|
|
|
* 2019/2020 © Philippe M. <[email protected]> |
6
|
|
|
* For the full copyright and MIT license information, please view the license file. |
7
|
|
|
*/ |
8
|
|
|
|
9
|
|
|
declare(strict_types=1); |
10
|
|
|
|
11
|
|
|
namespace App\Application; |
12
|
|
|
|
13
|
|
|
use App\Domain\Utils\WikiTextUtil; |
14
|
|
|
use App\Infrastructure\DbAdapter; |
15
|
|
|
use App\Infrastructure\Memory; |
16
|
|
|
use App\Infrastructure\ServiceFactory; |
17
|
|
|
use Codedungeon\PHPCliColors\Color; |
18
|
|
|
use Exception; |
19
|
|
|
use LogicException; |
20
|
|
|
use Mediawiki\Api\UsageException; |
21
|
|
|
use Normalizer; |
22
|
|
|
use Psr\Log\LoggerInterface; |
23
|
|
|
use Psr\Log\NullLogger; |
24
|
|
|
use Throwable; |
25
|
|
|
|
26
|
|
|
/** |
27
|
|
|
* Class OuvrageEditWorker |
28
|
|
|
* |
29
|
|
|
* @package App\Application\Examples |
30
|
|
|
*/ |
31
|
|
|
class OuvrageEditWorker |
32
|
|
|
{ |
33
|
|
|
use EditSummaryTrait, TalkPageEditTrait; |
34
|
|
|
|
35
|
|
|
const TASK_NAME = 'Amélioration bibliographique'; |
36
|
|
|
/** |
37
|
|
|
* poster ou pas le message en PD signalant les erreurs à résoudre |
38
|
|
|
*/ |
39
|
|
|
const EDIT_SIGNALEMENT = true; |
40
|
|
|
|
41
|
|
|
const CITATION_LIMIT = 150; |
42
|
|
|
const DELAY_BOTFLAG_SECONDS = 20; |
43
|
|
|
const DELAY_NO_BOTFLAG_SECONDS = 50; |
44
|
|
|
const ERROR_MSG_TEMPLATE = __DIR__.'/templates/message_errors.wiki'; |
45
|
|
|
|
46
|
|
|
private $db; |
47
|
|
|
private $bot; |
48
|
|
|
private $wikiText; |
49
|
|
|
|
50
|
|
|
private $citationSummary; |
51
|
|
|
private $errorWarning = []; |
52
|
|
|
private $importantSummary = []; |
53
|
|
|
|
54
|
|
|
private $nbRows; |
55
|
|
|
|
56
|
|
|
// Minor flag on edit |
57
|
|
|
private $minorFlag = true; |
58
|
|
|
// WikiBotConfig flag on edit |
59
|
|
|
private $botFlag = true; |
60
|
|
|
|
61
|
|
|
/** |
62
|
|
|
* @var Memory |
63
|
|
|
*/ |
64
|
|
|
private $memory; |
65
|
|
|
|
66
|
|
|
/** |
67
|
|
|
* @var LoggerInterface|NullLogger |
68
|
|
|
*/ |
69
|
|
|
private $log; |
70
|
|
|
/** |
71
|
|
|
* @var mixed |
72
|
|
|
*/ |
73
|
|
|
private $citationVersion; |
74
|
|
|
|
75
|
|
|
public function __construct( |
76
|
|
|
DbAdapter $dbAdapter, |
77
|
|
|
WikiBotConfig $bot, |
78
|
|
|
Memory $memory, |
79
|
|
|
?LoggerInterface $log = null |
80
|
|
|
) { |
81
|
|
|
$this->db = $dbAdapter; |
82
|
|
|
$this->bot = $bot; |
83
|
|
|
$this->memory = $memory; |
84
|
|
|
$this->log = $log ?? new NullLogger(); |
85
|
|
|
} |
86
|
|
|
|
87
|
|
|
/** |
88
|
|
|
* @throws Exception |
89
|
|
|
*/ |
90
|
|
|
public function run(): void |
91
|
|
|
{ |
92
|
|
|
while (true) { |
93
|
|
|
echo "\n-------------------------------------\n\n"; |
94
|
|
|
echo date("Y-m-d H:i")." "; |
95
|
|
|
$this->log->info($this->memory->getMemory(true)); |
96
|
|
|
$this->pageProcess(); |
97
|
|
|
} |
98
|
|
|
} |
99
|
|
|
|
100
|
|
|
/** |
101
|
|
|
* @return bool |
102
|
|
|
* @throws UsageException |
103
|
|
|
* @throws Exception |
104
|
|
|
* @throws Exception |
105
|
|
|
*/ |
106
|
|
|
private function pageProcess() |
107
|
|
|
{ |
108
|
|
|
$this->initialize(); |
109
|
|
|
|
110
|
|
|
// get a random queue line |
111
|
|
|
$json = $this->db->getAllRowsToEdit(self::CITATION_LIMIT); |
112
|
|
|
$data = json_decode($json, true); |
113
|
|
|
|
114
|
|
|
if (empty($data)) { |
115
|
|
|
$this->log->alert("SKIP : no row to process\n"); |
116
|
|
|
throw new Exception('no row to process'); |
117
|
|
|
} |
118
|
|
|
|
119
|
|
|
try { |
120
|
|
|
$title = $data[0]['page']; |
121
|
|
|
echo Color::BG_CYAN.$title.Color::NORMAL." \n"; |
122
|
|
|
$page = ServiceFactory::wikiPageAction($title, false); // , true ? |
123
|
|
|
} catch (Exception $e) { |
124
|
|
|
$this->log->warning("*** WikiPageAction error : ".$title." \n"); |
125
|
|
|
sleep(20); |
126
|
|
|
|
127
|
|
|
return false; |
128
|
|
|
} |
129
|
|
|
|
130
|
|
|
// Page supprimée ? |
131
|
|
|
if($page->getLastRevision() === null) { |
132
|
|
|
$this->log->warning("SKIP : page supprimée !\n"); |
133
|
|
|
$this->db->deleteArticle($title); |
134
|
|
|
|
135
|
|
|
return false; |
136
|
|
|
} |
137
|
|
|
|
138
|
|
|
// HACK |
139
|
|
|
if (in_array($page->getLastEditor(), [getenv('BOT_NAME'), getenv('BOT_OWNER')])) { |
140
|
|
|
$this->log->notice("SKIP : édité recemment par bot/dresseur.\n"); |
141
|
|
|
$this->db->skipArticle($title); |
142
|
|
|
|
143
|
|
|
return false; |
144
|
|
|
} |
145
|
|
|
if ($page->getNs() !== 0) { |
146
|
|
|
$this->log->notice("SKIP : page n'est pas dans Main (ns 0)\n"); |
147
|
|
|
$this->db->skipArticle($title); |
148
|
|
|
|
149
|
|
|
return false; |
150
|
|
|
} |
151
|
|
|
$this->wikiText = $page->getText(); |
152
|
|
|
|
153
|
|
|
if (empty($this->wikiText)) { |
154
|
|
|
return false; |
155
|
|
|
} |
156
|
|
|
|
157
|
|
|
// Skip AdQ |
158
|
|
|
if (preg_match('#{{ ?En-tête label ?\| ?AdQ#i', $this->wikiText)) { |
159
|
|
|
$this->db->setLabel($title, 2); |
160
|
|
|
$this->log->info("SKIP : AdQ.\n"); // BA ?? |
161
|
|
|
$this->db->skipArticle($title); |
162
|
|
|
|
163
|
|
|
return false; |
164
|
|
|
} |
165
|
|
|
if (preg_match('#{{ ?En-tête label ?\| ?BA#i', $this->wikiText)) { |
166
|
|
|
$this->db->setLabel($title, 1); |
167
|
|
|
$this->botFlag = false; |
168
|
|
|
$this->log->warning("Bon article !!\n"); |
169
|
|
|
} |
170
|
|
|
|
171
|
|
|
if (WikiBotConfig::isEditionRestricted($this->wikiText)) { |
172
|
|
|
$this->log->info("SKIP : protection/3R/travaux.\n"); |
173
|
|
|
$this->db->skipArticle($title); |
174
|
|
|
|
175
|
|
|
return false; |
176
|
|
|
} |
177
|
|
|
|
178
|
|
|
if ($this->bot->minutesSinceLastEdit($title) < 20) { |
179
|
|
|
$this->log->info("SKIP : édition humaine dans les dernières 20 minutes.\n"); |
180
|
|
|
|
181
|
|
|
return false; |
182
|
|
|
} |
183
|
|
|
|
184
|
|
|
|
185
|
|
|
|
186
|
|
|
// GET all article lines from db |
187
|
|
|
$this->log->info(sprintf("%s rows to process\n", count($data))); |
188
|
|
|
|
189
|
|
|
// foreach line |
190
|
|
|
$changed = false; |
191
|
|
|
foreach ($data as $dat) { |
192
|
|
|
// hack temporaire pour éviter articles dont CompleteProcess incomplet |
193
|
|
|
if (empty($dat['opti']) || empty($dat['optidate']) || $dat['optidate'] < DbAdapter::OPTI_VALID_DATE) { |
194
|
|
|
$this->log->notice("SKIP : Complètement incomplet de l'article. sleep 10min"); |
195
|
|
|
sleep(600); |
196
|
|
|
return false; |
197
|
|
|
} |
198
|
|
|
$success = $this->dataProcess($dat); |
199
|
|
|
$changed = ($success) ? true : $changed; |
200
|
|
|
} |
201
|
|
|
if (!$changed) { |
202
|
|
|
$this->log->debug("Rien à changer..."); |
203
|
|
|
$this->db->skipArticle($title); |
204
|
|
|
|
205
|
|
|
return false; |
206
|
|
|
} |
207
|
|
|
|
208
|
|
|
// EDIT THE PAGE |
209
|
|
|
if (!$this->wikiText) { |
210
|
|
|
return false; |
211
|
|
|
} |
212
|
|
|
|
213
|
|
|
$miniSummary = $this->generateSummary(); |
214
|
|
|
$this->log->notice($miniSummary); |
215
|
|
|
$this->log->debug("sleep 2..."); |
216
|
|
|
sleep(2); // todo ??? |
217
|
|
|
|
218
|
|
|
pageEdit: |
219
|
|
|
|
220
|
|
|
try { |
221
|
|
|
// corona Covid :) |
222
|
|
|
//$miniSummary .= (date('H:i') === '20:00') ? ' 🏥' : ''; // 🏥🦠 |
223
|
|
|
|
224
|
|
|
$editInfo = ServiceFactory::editInfo($miniSummary, $this->minorFlag, $this->botFlag, 5); |
225
|
|
|
$success = $page->editPage(Normalizer::normalize($this->wikiText), $editInfo); |
226
|
|
|
} catch (Throwable $e) { |
227
|
|
|
// Invalid CSRF token. |
228
|
|
|
if (strpos($e->getMessage(), 'Invalid CSRF token') !== false) { |
229
|
|
|
$this->log->alert("*** Invalid CSRF token \n"); |
230
|
|
|
throw new Exception('Invalid CSRF token'); |
231
|
|
|
} else { |
232
|
|
|
$this->log->warning('Exception in editPage() '.$e->getMessage()); |
233
|
|
|
sleep(10); |
234
|
|
|
|
235
|
|
|
return false; |
236
|
|
|
} |
237
|
|
|
} |
238
|
|
|
|
239
|
|
|
$this->log->info($success ? "Edition Ok\n" : "***** Edition KO !\n"); |
240
|
|
|
|
241
|
|
|
if ($success) { |
242
|
|
|
// updata DB |
243
|
|
|
foreach ($data as $dat) { |
244
|
|
|
$this->db->sendEditedData(['id' => $dat['id']]); |
245
|
|
|
} |
246
|
|
|
|
247
|
|
|
try { |
248
|
|
|
if (self::EDIT_SIGNALEMENT && !empty($this->errorWarning[$title])) { |
249
|
|
|
$this->sendOuvrageErrorsOnTalkPage($data, $this->log); |
250
|
|
|
} |
251
|
|
|
} catch (Throwable $e) { |
252
|
|
|
$this->log->warning('Exception in editPage() '.$e->getMessage()); |
253
|
|
|
unset($e); |
254
|
|
|
} |
255
|
|
|
|
256
|
|
|
if (!$this->botFlag) { |
257
|
|
|
$this->log->debug("sleep ".self::DELAY_NO_BOTFLAG_SECONDS); |
258
|
|
|
sleep(self::DELAY_NO_BOTFLAG_SECONDS); |
259
|
|
|
} |
260
|
|
|
if ($this->botFlag) { |
261
|
|
|
$this->log->debug("sleep ".self::DELAY_BOTFLAG_SECONDS); |
262
|
|
|
sleep(self::DELAY_BOTFLAG_SECONDS); |
263
|
|
|
} |
264
|
|
|
} |
265
|
|
|
|
266
|
|
|
return $success; |
267
|
|
|
} |
268
|
|
|
|
269
|
|
|
/** |
270
|
|
|
* @throws UsageException |
271
|
|
|
*/ |
272
|
|
|
private function initialize(): void |
273
|
|
|
{ |
274
|
|
|
// initialisation vars |
275
|
|
|
$this->botFlag = true; |
276
|
|
|
$this->errorWarning = []; |
277
|
|
|
$this->wikiText = null; |
278
|
|
|
$this->citationSummary = []; |
279
|
|
|
$this->importantSummary = []; |
280
|
|
|
$this->minorFlag = true; |
281
|
|
|
$this->nbRows = 0; |
282
|
|
|
|
283
|
|
|
$this->bot->checkStopOnTalkpage(true); |
284
|
|
|
} |
285
|
|
|
|
286
|
|
|
/** |
287
|
|
|
* @param array $data |
288
|
|
|
* |
289
|
|
|
* @return bool |
290
|
|
|
* @throws Exception |
291
|
|
|
*/ |
292
|
|
|
private function dataProcess(array $data): bool |
293
|
|
|
{ |
294
|
|
|
$origin = $data['raw']; |
295
|
|
|
$completed = $data['opti']; |
296
|
|
|
|
297
|
|
|
$this->log->debug('origin: '.$origin); |
298
|
|
|
$this->log->debug('completed: '.$completed); |
299
|
|
|
$this->log->debug('modifs: '.$data['modifs']); |
300
|
|
|
$this->log->debug('version: '.$data['version']); |
301
|
|
|
|
302
|
|
|
if (WikiTextUtil::isCommented($origin)) { |
303
|
|
|
$this->log->notice("SKIP: template avec commentaire HTML."); |
304
|
|
|
$this->db->skipRow(intval($data['id'])); |
305
|
|
|
|
306
|
|
|
return false; |
307
|
|
|
} |
308
|
|
|
|
309
|
|
|
$find = mb_strpos($this->wikiText, $origin); |
310
|
|
|
if ($find === false) { |
311
|
|
|
$this->log->notice("String non trouvée."); |
312
|
|
|
$this->db->skipRow(intval($data['id'])); |
313
|
|
|
|
314
|
|
|
return false; |
315
|
|
|
} |
316
|
|
|
|
317
|
|
|
$this->checkErrorWarning($data); |
318
|
|
|
|
319
|
|
|
// Replace text |
320
|
|
|
$newText = WikiPageAction::replaceTemplateInText($this->wikiText, $origin, $completed); |
321
|
|
|
|
322
|
|
|
if (!$newText || $newText === $this->wikiText) { |
323
|
|
|
$this->log->warning("newText error"); |
324
|
|
|
|
325
|
|
|
return false; |
326
|
|
|
} |
327
|
|
|
$this->wikiText = $newText; |
328
|
|
|
$this->minorFlag = ('1' === $data['major']) ? false : $this->minorFlag; |
329
|
|
|
$this->citationVersion = $data['version']; |
330
|
|
|
$this->citationSummary[] = $data['modifs']; |
331
|
|
|
$this->nbRows++; |
332
|
|
|
|
333
|
|
|
return true; |
334
|
|
|
} |
335
|
|
|
|
336
|
|
|
/** |
337
|
|
|
* todo extract |
338
|
|
|
* Vérifie alerte d'erreurs humaines. |
339
|
|
|
* |
340
|
|
|
* @param array $data |
341
|
|
|
* |
342
|
|
|
* @throws Exception |
343
|
|
|
*/ |
344
|
|
|
private function checkErrorWarning(array $data): void |
345
|
|
|
{ |
346
|
|
|
if (!isset($data['opti'])) { |
347
|
|
|
throw new LogicException('Opti NULL'); |
348
|
|
|
} |
349
|
|
|
|
350
|
|
|
// paramètre inconnu |
351
|
|
|
if (preg_match_all( |
352
|
|
|
"#\|[^|]+<!-- ?(PARAMETRE [^>]+ N'EXISTE PAS|VALEUR SANS NOM DE PARAMETRE|ERREUR [^>]+) ?-->#", |
353
|
|
|
$data['opti'], |
354
|
|
|
$matches |
355
|
|
|
) > 0 |
356
|
|
|
) { |
357
|
|
|
foreach ($matches[0] as $line) { |
358
|
|
|
$this->addErrorWarning($data['page'], $line); |
359
|
|
|
} |
360
|
|
|
// $this->botFlag = false; |
361
|
|
|
$this->addSummaryTag('paramètre non corrigé'); |
362
|
|
|
} |
363
|
|
|
|
364
|
|
|
// ISBN invalide |
365
|
|
|
if (preg_match("#isbn invalide ?=[^|}]+#i", $data['opti'], $matches) > 0) { |
366
|
|
|
$this->addErrorWarning($data['page'], $matches[0]); |
367
|
|
|
$this->botFlag = false; |
368
|
|
|
$this->addSummaryTag('ISBN invalide'); |
369
|
|
|
} |
370
|
|
|
|
371
|
|
|
// Edits avec ajout conséquent de donnée |
372
|
|
|
if (preg_match('#distinction des auteurs#', $data['modifs']) > 0) { |
373
|
|
|
$this->botFlag = false; |
374
|
|
|
$this->addSummaryTag('distinction des auteurs'); |
375
|
|
|
} |
376
|
|
|
// prédiction paramètre correct |
377
|
|
|
if (preg_match('#[^,]+(=>|⇒)[^,]+#', $data['modifs'], $matches) > 0) { |
378
|
|
|
$this->botFlag = false; |
379
|
|
|
$this->addSummaryTag(sprintf('%s', $matches[0])); |
380
|
|
|
} |
381
|
|
|
if (preg_match('#\+\+sous-titre#', $data['modifs']) > 0) { |
382
|
|
|
$this->botFlag = false; |
383
|
|
|
$this->addSummaryTag('+sous-titre'); |
384
|
|
|
} |
385
|
|
|
if (preg_match('#\+lieu#', $data['modifs']) > 0) { |
386
|
|
|
$this->addSummaryTag('+lieu'); |
387
|
|
|
} |
388
|
|
|
if (preg_match('#tracking#', $data['modifs']) > 0) { |
389
|
|
|
$this->addSummaryTag('tracking'); |
390
|
|
|
} |
391
|
|
|
if (preg_match('#présentation en ligne#', $data['modifs']) > 0) { |
392
|
|
|
$this->addSummaryTag('+présentation en ligne'); |
393
|
|
|
} |
394
|
|
|
if (preg_match('#distinction auteurs#', $data['modifs']) > 0) { |
395
|
|
|
$this->addSummaryTag('distinction auteurs'); |
396
|
|
|
} |
397
|
|
|
if (preg_match('#\+lire en ligne#', $data['modifs']) > 0) { |
398
|
|
|
$this->addSummaryTag('+lire en ligne'); |
399
|
|
|
} |
400
|
|
|
if (preg_match('#\+lien #', $data['modifs']) > 0) { |
401
|
|
|
$this->addSummaryTag('wikif'); |
402
|
|
|
} |
403
|
|
|
|
404
|
|
|
if (preg_match('#\+éditeur#', $data['modifs']) > 0) { |
405
|
|
|
$this->addSummaryTag('éditeur'); |
406
|
|
|
} |
407
|
|
|
// if (preg_match('#\+langue#', $data['modifs']) > 0) { |
408
|
|
|
// $this->addSummaryTag('langue'); |
409
|
|
|
// } |
410
|
|
|
|
411
|
|
|
// mention BnF si ajout donnée + ajout identifiant bnf= |
412
|
|
|
if (!empty($this->importantSummary) && preg_match('#BnF#i', $data['modifs'], $matches) > 0) { |
413
|
|
|
$this->addSummaryTag('©[[BnF]]'); |
414
|
|
|
} |
415
|
|
|
} |
416
|
|
|
|
417
|
|
|
/** |
418
|
|
|
* todo extract |
419
|
|
|
* Pour éviter les doublons dans signalements d'erreur. |
420
|
|
|
* |
421
|
|
|
* @param string $page |
422
|
|
|
* @param string $text |
423
|
|
|
*/ |
424
|
|
|
private function addErrorWarning(string $page, string $text): void |
425
|
|
|
{ |
426
|
|
|
if (!isset($this->errorWarning[$page]) || !in_array($text, $this->errorWarning[$page])) { |
427
|
|
|
$this->errorWarning[$page][] = $text; |
428
|
|
|
} |
429
|
|
|
} |
430
|
|
|
|
431
|
|
|
} |
432
|
|
|
|