1
|
|
|
<?php |
2
|
|
|
|
3
|
|
|
/* |
4
|
|
|
* This file is part of dispositif/wikibot application (@github) |
5
|
|
|
* 2019-2023 © Philippe M./Irønie <[email protected]> |
6
|
|
|
* For the full copyright and MIT license information, view the license file. |
7
|
|
|
*/ |
8
|
|
|
|
9
|
|
|
declare(strict_types=1); |
10
|
|
|
|
11
|
|
|
namespace App\Application\OuvrageEdit; |
12
|
|
|
|
13
|
|
|
use App\Application\InfrastructurePorts\DbAdapterInterface; |
14
|
|
|
use App\Application\InfrastructurePorts\MemoryInterface; |
15
|
|
|
use App\Application\OuvrageEdit\Validators\CitationsNotEmptyValidator; |
16
|
|
|
use App\Application\OuvrageEdit\Validators\CitationValidator; |
17
|
|
|
use App\Application\OuvrageEdit\Validators\PageValidatorComposite; |
18
|
|
|
use App\Application\OuvrageEdit\Validators\TalkStopValidator; |
19
|
|
|
use App\Application\OuvrageEdit\Validators\WikiTextValidator; |
20
|
|
|
use App\Application\WikiBotConfig; |
21
|
|
|
use App\Application\WikiPageAction; |
22
|
|
|
use App\Infrastructure\ServiceFactory; |
23
|
|
|
use Codedungeon\PHPCliColors\Color; |
24
|
|
|
use Exception; |
25
|
|
|
use LogicException; |
26
|
|
|
use Mediawiki\Api\UsageException; |
27
|
|
|
use Normalizer; |
28
|
|
|
use Psr\Log\LoggerInterface; |
29
|
|
|
use Psr\Log\NullLogger; |
30
|
|
|
use Throwable; |
31
|
|
|
|
32
|
|
|
/** |
33
|
|
|
* Legacy class, to be refactored. Too big, too many responsibilities. |
34
|
|
|
* todo use PageOuvrageCollectionDTO. |
35
|
|
|
* todo chain of responsibility pattern + log decorator (+ events ?) |
36
|
|
|
*/ |
37
|
|
|
class OuvrageEditWorker |
38
|
|
|
{ |
39
|
|
|
use OuvrageEditSummaryTrait, TalkPageEditTrait; |
40
|
|
|
|
41
|
|
|
public const TASK_NAME = '📗 Amélioration bibliographique'; // 📖📔📘📗 |
42
|
|
|
public const LUCKY_MESSAGE = ' 🇺🇦'; // ☘️ |
43
|
|
|
/** |
44
|
|
|
* poster ou pas le message en PD signalant les erreurs à résoudre |
45
|
|
|
*/ |
46
|
|
|
public const EDIT_SIGNALEMENT = true; |
47
|
|
|
public const CITATION_LIMIT = 150; |
48
|
|
|
public const DELAY_BOTFLAG_SECONDS = 60; |
49
|
|
|
public const DELAY_NO_BOTFLAG_SECONDS = 60; |
50
|
|
|
public const DELAY_MINUTES_AFTER_HUMAN_EDIT = 10; |
51
|
|
|
public const ERROR_MSG_TEMPLATE = __DIR__ . '/templates/message_errors.wiki'; |
52
|
|
|
|
53
|
|
|
/** |
54
|
|
|
* @var PageWorkStatus |
55
|
|
|
*/ |
56
|
|
|
protected $pageWorkStatus; |
57
|
|
|
|
58
|
|
|
/** |
59
|
|
|
* @var WikiPageAction |
60
|
|
|
*/ |
61
|
|
|
protected $wikiPageAction = null; |
62
|
|
|
|
63
|
|
|
protected $db; |
64
|
|
|
/** |
65
|
|
|
* @var WikiBotConfig |
66
|
|
|
*/ |
67
|
|
|
protected $bot; |
68
|
|
|
/** |
69
|
|
|
* @var MemoryInterface |
70
|
|
|
*/ |
71
|
|
|
protected $memory; |
72
|
|
|
|
73
|
|
|
/** |
74
|
|
|
* @var LoggerInterface |
75
|
|
|
*/ |
76
|
|
|
protected $log; |
77
|
|
|
|
78
|
|
|
public function __construct( |
79
|
|
|
DbAdapterInterface $dbAdapter, |
80
|
|
|
WikiBotConfig $bot, |
81
|
|
|
MemoryInterface $memory, |
82
|
|
|
?LoggerInterface $log = null |
83
|
|
|
) |
84
|
|
|
{ |
85
|
|
|
$this->db = $dbAdapter; |
86
|
|
|
$this->bot = $bot; |
87
|
|
|
$this->memory = $memory; |
88
|
|
|
$this->log = $log ?? new NullLogger(); |
89
|
|
|
} |
90
|
|
|
|
91
|
|
|
/** |
92
|
|
|
* @throws Exception |
93
|
|
|
*/ |
94
|
|
|
public function run(): void |
95
|
|
|
{ |
96
|
|
|
while (true) { |
97
|
|
|
echo "\n-------------------------------------\n\n"; |
98
|
|
|
echo date("Y-m-d H:i:s") . " "; |
99
|
|
|
$this->log->info($this->memory->getMemory(true)); |
100
|
|
|
$this->pageProcess(); |
101
|
|
|
sleep(2); // précaution boucle infinie |
102
|
|
|
} |
103
|
|
|
} |
104
|
|
|
|
105
|
|
|
/** |
106
|
|
|
* @throws UsageException |
107
|
|
|
* @throws Exception |
108
|
|
|
*/ |
109
|
|
|
protected function pageProcess(): bool |
110
|
|
|
{ |
111
|
|
|
$e = null; |
|
|
|
|
112
|
|
|
$this->pageWorkStatus = new PageWorkStatus(); |
113
|
|
|
|
114
|
|
|
if ((new TalkStopValidator($this->bot))->validate() === false) { |
|
|
|
|
115
|
|
|
return false; |
116
|
|
|
} |
117
|
|
|
|
118
|
|
|
// get a random queue line |
119
|
|
|
$json = $this->db->getAllRowsOfOneTitleToEdit(self::CITATION_LIMIT); |
120
|
|
|
$pageCitationCollection = json_decode($json, true, 512, JSON_THROW_ON_ERROR); |
|
|
|
|
121
|
|
|
|
122
|
|
|
if ((new CitationsNotEmptyValidator($pageCitationCollection, $this->log))->validate() === false) { |
|
|
|
|
123
|
|
|
return false; |
124
|
|
|
} |
125
|
|
|
|
126
|
|
|
$title = $pageCitationCollection[0]['page']; // not used by validators |
127
|
|
|
$this->pageWorkStatus->title = $title; |
128
|
|
|
$this->printTitle($title); |
129
|
|
|
|
130
|
|
|
// Find on wikipedia the page to edit |
131
|
|
|
try { |
132
|
|
|
$this->wikiPageAction = ServiceFactory::wikiPageAction($title); // , true ? |
133
|
|
|
} catch (Exception $e) { |
134
|
|
|
$this->log->warning("*** WikiPageAction error : " . $title . " \n"); |
135
|
|
|
sleep(20); |
136
|
|
|
|
137
|
|
|
return false; |
138
|
|
|
} |
139
|
|
|
$pageValidator = new PageValidatorComposite( |
140
|
|
|
$this->bot, |
141
|
|
|
$pageCitationCollection, |
142
|
|
|
$this->db, |
143
|
|
|
$this->wikiPageAction |
144
|
|
|
); |
145
|
|
|
if ($pageValidator->validate() === false) { |
146
|
|
|
return false; |
147
|
|
|
} |
148
|
|
|
|
149
|
|
|
$this->pageWorkStatus->wikiText = $this->wikiPageAction->getText(); |
150
|
|
|
// or do that at the end of ArticleValidForEditValidator if PageWorkStatus injected ? |
151
|
|
|
|
152
|
|
|
$this->checkArticleLabels($title); |
153
|
|
|
// or do that at the end of ArticleValidForEditValidator if PageWorkStatus injected ? |
154
|
|
|
|
155
|
|
|
// print total number of rows completed in db |
156
|
|
|
$this->log->info(sprintf("%s rows to process\n", is_countable($pageCitationCollection) ? count($pageCitationCollection) : 0)); |
157
|
|
|
|
158
|
|
|
|
159
|
|
|
// foreach line |
160
|
|
|
$changed = false; |
161
|
|
|
foreach ($pageCitationCollection as $dat) { |
162
|
|
|
if (!$this->isAllCitationCompleted($dat)) { |
163
|
|
|
$this->log->notice("SKIP : Amélioration incomplet de l'article. sleep 10min"); |
164
|
|
|
sleep(600); |
165
|
|
|
|
166
|
|
|
return false; |
167
|
|
|
} |
168
|
|
|
$success = $this->processOneCitation($dat); |
169
|
|
|
$changed = ($success) ? true : $changed; |
170
|
|
|
} |
171
|
|
|
if (!$changed) { |
172
|
|
|
$this->log->debug("Rien à changer..."); |
173
|
|
|
$this->db->skipArticle($title); |
|
|
|
|
174
|
|
|
|
175
|
|
|
return false; |
176
|
|
|
} |
177
|
|
|
|
178
|
|
|
|
179
|
|
|
// EDIT THE PAGE |
180
|
|
|
if (!(new WikiTextValidator($this->pageWorkStatus->wikiText))->validate()) { |
181
|
|
|
return false; |
182
|
|
|
} |
183
|
|
|
|
184
|
|
|
$miniSummary = $this->generateFinalSummary(); |
185
|
|
|
$this->log->notice($miniSummary); |
186
|
|
|
$this->log->debug("sleep 2..."); |
187
|
|
|
sleep(2); // todo ??? |
188
|
|
|
|
189
|
|
|
dump($miniSummary); |
190
|
|
|
die("\n*** STOP ON OuvrageEditWorker:191 FOR REFACTORING ***\n"); |
|
|
|
|
191
|
|
|
|
192
|
|
|
$success = $this->editPage($miniSummary); |
|
|
|
|
193
|
|
|
if ($success) { |
194
|
|
|
// updata DB |
195
|
|
|
foreach ($pageCitationCollection as $dat) { |
196
|
|
|
$this->db->sendEditedData(['id' => $dat['id']]); |
197
|
|
|
} |
198
|
|
|
|
199
|
|
|
try { |
200
|
|
|
if (self::EDIT_SIGNALEMENT && !empty($this->pageWorkStatus->errorWarning[$title])) { |
201
|
|
|
$this->sendOuvrageErrorsOnTalkPage($pageCitationCollection, $this->log); |
202
|
|
|
} |
203
|
|
|
} catch (Throwable $e) { |
204
|
|
|
$this->log->warning('Exception in editPage() ' . $e->getMessage()); |
205
|
|
|
unset($e); |
206
|
|
|
} |
207
|
|
|
|
208
|
|
|
if (!$this->pageWorkStatus->botFlag) { |
209
|
|
|
$this->log->debug("sleep " . self::DELAY_NO_BOTFLAG_SECONDS); |
210
|
|
|
sleep(self::DELAY_NO_BOTFLAG_SECONDS); |
211
|
|
|
} |
212
|
|
|
if ($this->pageWorkStatus->botFlag) { |
213
|
|
|
$this->log->debug("sleep " . self::DELAY_BOTFLAG_SECONDS); |
214
|
|
|
sleep(self::DELAY_BOTFLAG_SECONDS); |
215
|
|
|
} |
216
|
|
|
} |
217
|
|
|
|
218
|
|
|
return $success; |
219
|
|
|
} |
220
|
|
|
|
221
|
|
|
protected function printTitle(string $title): void |
222
|
|
|
{ |
223
|
|
|
echo Color::BG_CYAN . $title . Color::NORMAL . " \n"; |
224
|
|
|
} |
225
|
|
|
|
226
|
|
|
protected function checkArticleLabels($title): void |
227
|
|
|
{ |
228
|
|
|
// Featured/Good article (AdQ/BA) todo event listener |
229
|
|
|
if (preg_match('#{{ ?En-tête label ?\| ?AdQ#i', $this->pageWorkStatus->wikiText)) { |
230
|
|
|
$this->db->setLabel($title, 2); |
|
|
|
|
231
|
|
|
$this->log->warning("Article de Qualité !\n"); |
232
|
|
|
$this->pageWorkStatus->botFlag = false; |
233
|
|
|
$this->pageWorkStatus->featured_article = true; // to add star in edit summary |
234
|
|
|
} |
235
|
|
|
if (preg_match('#{{ ?En-tête label ?\| ?BA#i', $this->pageWorkStatus->wikiText)) { |
236
|
|
|
$this->db->setLabel($title, 1); |
237
|
|
|
$this->pageWorkStatus->botFlag = false; |
238
|
|
|
$this->pageWorkStatus->featured_article = true; // to add star in edit summary |
239
|
|
|
$this->log->warning("Bon article !!\n"); |
240
|
|
|
} |
241
|
|
|
} |
242
|
|
|
|
243
|
|
|
protected function isAllCitationCompleted(array $dat): bool |
244
|
|
|
{ |
245
|
|
|
// hack temporaire pour éviter articles dont CompleteProcess incomplet |
246
|
|
|
if (empty($dat['opti']) || empty($dat['optidate']) || $dat['optidate'] < $this->db->getOptiValidDate()) { |
247
|
|
|
return false; |
248
|
|
|
} |
249
|
|
|
return true; |
250
|
|
|
} |
251
|
|
|
|
252
|
|
|
/** |
253
|
|
|
* @throws Exception |
254
|
|
|
*/ |
255
|
|
|
protected function processOneCitation(array $ouvrageData): bool |
256
|
|
|
{ |
257
|
|
|
$origin = $ouvrageData['raw']; |
258
|
|
|
$completed = $ouvrageData['opti']; |
259
|
|
|
$this->printDebug($ouvrageData); |
260
|
|
|
|
261
|
|
|
$citationValidator = new CitationValidator( |
262
|
|
|
$ouvrageData, |
263
|
|
|
$this->pageWorkStatus->wikiText, |
264
|
|
|
$this->log, |
265
|
|
|
$this->db |
266
|
|
|
); |
267
|
|
|
if ($citationValidator->validate() === false) { |
268
|
|
|
return false; |
269
|
|
|
} |
270
|
|
|
|
271
|
|
|
$this->generateSummaryOnPageWorkStatus($ouvrageData); |
272
|
|
|
|
273
|
|
|
// Replace text |
274
|
|
|
$newText = WikiPageAction::replaceTemplateInText($this->pageWorkStatus->wikiText, $origin, $completed); |
275
|
|
|
|
276
|
|
|
if (!$newText || $newText === $this->pageWorkStatus->wikiText) { |
277
|
|
|
$this->log->warning("newText error"); |
278
|
|
|
|
279
|
|
|
return false; |
280
|
|
|
} |
281
|
|
|
$this->pageWorkStatus->wikiText = $newText; |
282
|
|
|
$this->pageWorkStatus->minorFlag = ('1' === $ouvrageData['major']) ? false : $this->pageWorkStatus->minorFlag; |
283
|
|
|
$this->pageWorkStatus->citationVersion = $ouvrageData['version']; // todo gérer versions différentes |
284
|
|
|
$this->pageWorkStatus->citationSummary[] = $ouvrageData['modifs']; |
285
|
|
|
$this->pageWorkStatus->nbRows++; |
286
|
|
|
|
287
|
|
|
return true; |
288
|
|
|
} |
289
|
|
|
|
290
|
|
|
protected function printDebug(array $data) |
291
|
|
|
{ |
292
|
|
|
$this->log->debug('origin: ' . $data['raw']); |
293
|
|
|
$this->log->debug('completed: ' . $data['opti']); |
294
|
|
|
$this->log->debug('modifs: ' . $data['modifs']); |
295
|
|
|
$this->log->debug('version: ' . $data['version']); |
296
|
|
|
} |
297
|
|
|
|
298
|
|
|
/** |
299
|
|
|
* todo extract. => responsability : pageWorkStatus + summary |
300
|
|
|
* Vérifie alerte d'erreurs humaines. |
301
|
|
|
*/ |
302
|
|
|
protected function generateSummaryOnPageWorkStatus(array $ouvrageData): void |
303
|
|
|
{ |
304
|
|
|
if (!isset($ouvrageData['opti'])) { // todo move to CitationValidator |
305
|
|
|
throw new LogicException('Opti NULL'); |
306
|
|
|
} |
307
|
|
|
|
308
|
|
|
// paramètre inconnu |
309
|
|
|
if (preg_match_all( |
310
|
|
|
"#\|[^|]+<!-- ?(PARAMETRE [^>]+ N'EXISTE PAS|VALEUR SANS NOM DE PARAMETRE|ERREUR [^>]+) ?-->#", |
311
|
|
|
$ouvrageData['opti'], |
312
|
|
|
$matches |
313
|
|
|
) > 0 |
314
|
|
|
) { |
315
|
|
|
foreach ($matches[0] as $line) { |
316
|
|
|
$this->addErrorWarning($ouvrageData['page'], $line); |
317
|
|
|
} |
318
|
|
|
// $this->pageWorkStatus->botFlag = false; |
319
|
|
|
$this->addSummaryTag('paramètre non corrigé'); |
320
|
|
|
} |
321
|
|
|
|
322
|
|
|
// ISBN invalide |
323
|
|
|
if (preg_match("#isbn invalide ?=[^|}]+#i", $ouvrageData['opti'], $matches) > 0) { |
324
|
|
|
$this->addErrorWarning($ouvrageData['page'], $matches[0]); |
325
|
|
|
$this->pageWorkStatus->botFlag = false; |
326
|
|
|
$this->addSummaryTag('ISBN invalide 💩'); |
327
|
|
|
} |
328
|
|
|
|
329
|
|
|
// Edits avec ajout conséquent de donnée |
330
|
|
|
if (preg_match('#distinction des auteurs#', $ouvrageData['modifs']) > 0) { |
331
|
|
|
$this->pageWorkStatus->botFlag = false; |
332
|
|
|
$this->addSummaryTag('distinction auteurs 🧠'); |
333
|
|
|
} |
334
|
|
|
// prédiction paramètre correct |
335
|
|
|
if (preg_match('#[^,]+(=>|⇒)[^,]+#', $ouvrageData['modifs'], $matches) > 0) { |
336
|
|
|
$this->pageWorkStatus->botFlag = false; |
337
|
|
|
$this->addSummaryTag($matches[0]); |
338
|
|
|
} |
339
|
|
|
if (preg_match('#\+\+sous-titre#', $ouvrageData['modifs']) > 0) { |
340
|
|
|
$this->pageWorkStatus->botFlag = false; |
341
|
|
|
$this->addSummaryTag('+sous-titre'); |
342
|
|
|
} |
343
|
|
|
if (preg_match('#\+lieu#', $ouvrageData['modifs']) > 0) { |
344
|
|
|
$this->addSummaryTag('+lieu'); |
345
|
|
|
} |
346
|
|
|
if (preg_match('#tracking#', $ouvrageData['modifs']) > 0) { |
347
|
|
|
$this->addSummaryTag('tracking'); |
348
|
|
|
} |
349
|
|
|
if (preg_match('#présentation en ligne#', $ouvrageData['modifs']) > 0) { |
350
|
|
|
$this->addSummaryTag('+présentation en ligne✨'); |
351
|
|
|
} |
352
|
|
|
if (preg_match('#distinction auteurs#', $ouvrageData['modifs']) > 0) { |
353
|
|
|
$this->addSummaryTag('distinction auteurs 🧠'); |
354
|
|
|
} |
355
|
|
|
if (preg_match('#\+lire en ligne#', $ouvrageData['modifs']) > 0) { |
356
|
|
|
$this->addSummaryTag('+lire en ligne✨'); |
357
|
|
|
} |
358
|
|
|
if (preg_match('#\+lien #', $ouvrageData['modifs']) > 0) { |
359
|
|
|
$this->addSummaryTag('wikif'); |
360
|
|
|
} |
361
|
|
|
|
362
|
|
|
if (preg_match('#\+éditeur#', $ouvrageData['modifs']) > 0) { |
363
|
|
|
$this->addSummaryTag('éditeur'); |
364
|
|
|
} |
365
|
|
|
// if (preg_match('#\+langue#', $data['modifs']) > 0) { |
366
|
|
|
// $this->addSummaryTag('langue'); |
367
|
|
|
// } |
368
|
|
|
|
369
|
|
|
// mention BnF si ajout donnée + ajout identifiant bnf= |
370
|
|
|
if (!empty($this->pageWorkStatus->importantSummary) && preg_match('#BnF#i', $ouvrageData['modifs'], $matches) > 0) { |
371
|
|
|
$this->addSummaryTag('©[[BnF]]'); |
372
|
|
|
} |
373
|
|
|
} |
374
|
|
|
|
375
|
|
|
/** |
376
|
|
|
* todo extract |
377
|
|
|
* Pour éviter les doublons dans signalements d'erreur. |
378
|
|
|
*/ |
379
|
|
|
protected function addErrorWarning(string $title, string $text): void |
380
|
|
|
{ |
381
|
|
|
if (!isset($this->pageWorkStatus->errorWarning[$title]) || !in_array($text, $this->pageWorkStatus->errorWarning[$title])) { |
382
|
|
|
$this->pageWorkStatus->errorWarning[$title][] = $text; |
383
|
|
|
} |
384
|
|
|
} |
385
|
|
|
|
386
|
|
|
private function editPage(string $miniSummary): bool |
387
|
|
|
{ |
388
|
|
|
try { |
389
|
|
|
$editInfo = ServiceFactory::editInfo($miniSummary, $this->pageWorkStatus->minorFlag, $this->pageWorkStatus->botFlag); |
390
|
|
|
$success = $this->wikiPageAction->editPage(Normalizer::normalize($this->pageWorkStatus->wikiText), $editInfo); |
391
|
|
|
} catch (Throwable $e) { |
392
|
|
|
// Invalid CSRF token. |
393
|
|
|
if (strpos($e->getMessage(), 'Invalid CSRF token') !== false) { |
394
|
|
|
$this->log->alert("*** Invalid CSRF token \n"); |
395
|
|
|
throw new Exception('Invalid CSRF token', $e->getCode(), $e); |
396
|
|
|
} else { |
397
|
|
|
$this->log->warning('Exception in editPage() ' . $e->getMessage()); |
398
|
|
|
sleep(10); |
399
|
|
|
|
400
|
|
|
return false; |
401
|
|
|
} |
402
|
|
|
} |
403
|
|
|
$this->log->info($success ? "Edition Ok\n" : "***** Edition KO !\n"); |
404
|
|
|
|
405
|
|
|
return $success; |
406
|
|
|
} |
407
|
|
|
} |
408
|
|
|
|