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 Mediawiki\Api\UsageException; |
26
|
|
|
use Normalizer; |
27
|
|
|
use Psr\Log\LoggerInterface; |
28
|
|
|
use Psr\Log\NullLogger; |
29
|
|
|
use Throwable; |
30
|
|
|
|
31
|
|
|
/** |
32
|
|
|
* Legacy class, to be refactored. Too big, too many responsibilities. |
33
|
|
|
* todo use PageOuvrageCollectionDTO. |
34
|
|
|
* todo chain of responsibility pattern + log decorator (+ events ?) |
35
|
|
|
*/ |
36
|
|
|
class OuvrageEditWorker |
37
|
|
|
{ |
38
|
|
|
use OuvrageEditSummaryTrait, TalkPageEditTrait; |
39
|
|
|
|
40
|
|
|
public const TASK_NAME = '📗 Amélioration bibliographique'; // 📖📔📘📗 |
41
|
|
|
public const LUCKY_MESSAGE = ' 🇺🇦'; // ☘️ |
42
|
|
|
/** |
43
|
|
|
* poster ou pas le message en PD signalant les erreurs à résoudre |
44
|
|
|
*/ |
45
|
|
|
public const EDIT_SIGNALEMENT = true; |
46
|
|
|
public const CITATION_LIMIT = 150; |
47
|
|
|
public const DELAY_BOTFLAG_SECONDS = 60; |
48
|
|
|
public const DELAY_NO_BOTFLAG_SECONDS = 60; |
49
|
|
|
public const DELAY_MINUTES_AFTER_HUMAN_EDIT = 10; |
50
|
|
|
public const ERROR_MSG_TEMPLATE = __DIR__ . '/templates/message_errors.wiki'; |
51
|
|
|
|
52
|
|
|
/** |
53
|
|
|
* @var PageWorkStatus |
54
|
|
|
*/ |
55
|
|
|
protected $pageWorkStatus; |
56
|
|
|
|
57
|
|
|
/** |
58
|
|
|
* @var WikiPageAction |
59
|
|
|
*/ |
60
|
|
|
protected $wikiPageAction = null; |
61
|
|
|
|
62
|
|
|
protected $db; |
63
|
|
|
/** |
64
|
|
|
* @var WikiBotConfig |
65
|
|
|
*/ |
66
|
|
|
protected $bot; |
67
|
|
|
/** |
68
|
|
|
* @var MemoryInterface |
69
|
|
|
*/ |
70
|
|
|
protected $memory; |
71
|
|
|
|
72
|
|
|
/** |
73
|
|
|
* @var LoggerInterface |
74
|
|
|
*/ |
75
|
|
|
protected $log; |
76
|
|
|
|
77
|
|
|
public function __construct( |
78
|
|
|
DbAdapterInterface $dbAdapter, |
79
|
|
|
WikiBotConfig $bot, |
80
|
|
|
MemoryInterface $memory, |
81
|
|
|
?LoggerInterface $log = null |
82
|
|
|
) |
83
|
|
|
{ |
84
|
|
|
$this->db = $dbAdapter; |
85
|
|
|
$this->bot = $bot; |
86
|
|
|
$this->memory = $memory; |
87
|
|
|
$this->log = $log ?? new NullLogger(); |
88
|
|
|
} |
89
|
|
|
|
90
|
|
|
/** |
91
|
|
|
* @throws Exception |
92
|
|
|
*/ |
93
|
|
|
public function run(): void |
94
|
|
|
{ |
95
|
|
|
while (true) { |
96
|
|
|
echo "\n-------------------------------------\n\n"; |
97
|
|
|
echo date("Y-m-d H:i:s") . " "; |
98
|
|
|
$this->log->info($this->memory->getMemory(true)); |
99
|
|
|
$this->pageProcess(); |
100
|
|
|
sleep(2); // précaution boucle infinie |
101
|
|
|
} |
102
|
|
|
} |
103
|
|
|
|
104
|
|
|
/** |
105
|
|
|
* @throws UsageException |
106
|
|
|
* @throws Exception |
107
|
|
|
*/ |
108
|
|
|
protected function pageProcess(): bool |
109
|
|
|
{ |
110
|
|
|
$e = null; |
|
|
|
|
111
|
|
|
|
112
|
|
|
if ((new TalkStopValidator($this->bot))->validate() === false) { // move up ? |
113
|
|
|
return false; |
114
|
|
|
} |
115
|
|
|
|
116
|
|
|
// get a random queue line |
117
|
|
|
$json = $this->db->getAllRowsOfOneTitleToEdit(self::CITATION_LIMIT); |
118
|
|
|
$pageCitationCollection = $json ? json_decode($json, true, 512, JSON_THROW_ON_ERROR) : []; |
119
|
|
|
|
120
|
|
|
if ((new CitationsNotEmptyValidator($pageCitationCollection, $this->log))->validate() === false) { |
121
|
|
|
return false; |
122
|
|
|
} |
123
|
|
|
|
124
|
|
|
$this->pageWorkStatus = new PageWorkStatus($pageCitationCollection[0]['page']); |
125
|
|
|
$this->printTitle($this->pageWorkStatus->getTitle()); |
126
|
|
|
|
127
|
|
|
// Find on wikipedia the page to edit |
128
|
|
|
try { |
129
|
|
|
$this->wikiPageAction = ServiceFactory::wikiPageAction($this->pageWorkStatus->getTitle()); // , true ? |
130
|
|
|
} catch (Exception $e) { |
131
|
|
|
$this->log->warning("*** WikiPageAction error : " . $this->pageWorkStatus->getTitle() . " \n"); |
132
|
|
|
sleep(20); |
133
|
|
|
|
134
|
|
|
return false; |
135
|
|
|
} |
136
|
|
|
$pageValidator = new PageValidatorComposite( |
137
|
|
|
$this->bot, $pageCitationCollection, $this->db, $this->wikiPageAction |
138
|
|
|
); |
139
|
|
|
if ($pageValidator->validate() === false) { |
140
|
|
|
return false; |
141
|
|
|
} |
142
|
|
|
|
143
|
|
|
$this->pageWorkStatus->wikiText = $this->wikiPageAction->getText(); |
144
|
|
|
$this->checkArticleLabels($this->pageWorkStatus->getTitle()); |
145
|
|
|
// or do that at the end of ArticleValidForEditValidator if PageWorkStatus injected ? |
146
|
|
|
|
147
|
|
|
// print total number of rows completed in db |
148
|
|
|
$rowNumber = is_countable($pageCitationCollection) ? count($pageCitationCollection) : 0; |
149
|
|
|
$this->log->info(sprintf("%s rows to process\n", $rowNumber)); |
150
|
|
|
|
151
|
|
|
// Make citations replacements |
152
|
|
|
if ($this->makeCitationsReplacements($pageCitationCollection) === false) { |
153
|
|
|
return false; |
154
|
|
|
} |
155
|
|
|
|
156
|
|
|
if ($this->editPage()) { |
157
|
|
|
$this->updateDb($pageCitationCollection); |
158
|
|
|
|
159
|
|
|
return true; |
160
|
|
|
} |
161
|
|
|
|
162
|
|
|
return false; |
163
|
|
|
} |
164
|
|
|
|
165
|
|
|
protected function printTitle(string $title): void |
166
|
|
|
{ |
167
|
|
|
echo Color::BG_CYAN . $title . Color::NORMAL . " \n"; |
168
|
|
|
} |
169
|
|
|
|
170
|
|
|
protected function checkArticleLabels($title): void |
171
|
|
|
{ |
172
|
|
|
// Featured/Good article (AdQ/BA) todo event listener |
173
|
|
|
if (preg_match('#{{ ?En-tête label ?\| ?AdQ#i', $this->pageWorkStatus->wikiText)) { |
174
|
|
|
$this->db->setLabel($title, 2); |
|
|
|
|
175
|
|
|
$this->log->warning("Article de Qualité !\n"); |
176
|
|
|
$this->pageWorkStatus->botFlag = false; |
177
|
|
|
$this->pageWorkStatus->featured_article = true; // to add star in edit summary |
178
|
|
|
} |
179
|
|
|
if (preg_match('#{{ ?En-tête label ?\| ?BA#i', $this->pageWorkStatus->wikiText)) { |
180
|
|
|
$this->db->setLabel($title, 1); |
181
|
|
|
$this->pageWorkStatus->botFlag = false; |
182
|
|
|
$this->pageWorkStatus->featured_article = true; // to add star in edit summary |
183
|
|
|
$this->log->warning("Bon article !!\n"); |
184
|
|
|
} |
185
|
|
|
} |
186
|
|
|
|
187
|
|
|
protected function makeCitationsReplacements(array $pageCitationCollection): bool |
188
|
|
|
{ |
189
|
|
|
$oldText = $this->pageWorkStatus->wikiText; |
190
|
|
|
foreach ($pageCitationCollection as $dat) { |
191
|
|
|
$this->processOneCitation($dat); // that modify PageWorkStatus->wikiText |
192
|
|
|
} |
193
|
|
|
$newWikiTextValidator = new WikiTextValidator( |
194
|
|
|
$this->pageWorkStatus->wikiText, $oldText, $this->log, $this->pageWorkStatus->getTitle(), $this->db |
195
|
|
|
); |
196
|
|
|
|
197
|
|
|
return $newWikiTextValidator->validate(); |
198
|
|
|
} |
199
|
|
|
|
200
|
|
|
/** |
201
|
|
|
* @throws Exception |
202
|
|
|
*/ |
203
|
|
|
protected function processOneCitation(array $ouvrageData): bool |
204
|
|
|
{ |
205
|
|
|
$origin = $ouvrageData['raw']; |
206
|
|
|
$completed = $ouvrageData['opti']; |
207
|
|
|
$this->printDebug($ouvrageData); |
208
|
|
|
|
209
|
|
|
$citationValidator = new CitationValidator( |
210
|
|
|
$ouvrageData, |
211
|
|
|
$this->pageWorkStatus->wikiText, |
212
|
|
|
$this->log, |
213
|
|
|
$this->db |
214
|
|
|
); |
215
|
|
|
if ($citationValidator->validate() === false) { |
216
|
|
|
return false; |
217
|
|
|
} |
218
|
|
|
|
219
|
|
|
$this->generateSummaryOnPageWorkStatus($ouvrageData); |
220
|
|
|
|
221
|
|
|
// Replace text |
222
|
|
|
$newText = WikiPageAction::replaceTemplateInText($this->pageWorkStatus->wikiText, $origin, $completed); |
223
|
|
|
|
224
|
|
|
if (!$newText || $newText === $this->pageWorkStatus->wikiText) { |
225
|
|
|
$this->log->warning("newText error"); |
226
|
|
|
|
227
|
|
|
return false; |
228
|
|
|
} |
229
|
|
|
$this->pageWorkStatus->wikiText = $newText; |
230
|
|
|
$this->pageWorkStatus->minorFlag = ('1' === $ouvrageData['major']) ? false : $this->pageWorkStatus->minorFlag; |
231
|
|
|
$this->pageWorkStatus->citationVersion = $ouvrageData['version']; // todo gérer versions différentes |
232
|
|
|
$this->pageWorkStatus->citationSummary[] = $ouvrageData['modifs']; |
233
|
|
|
$this->pageWorkStatus->nbRows++; |
234
|
|
|
|
235
|
|
|
return true; |
236
|
|
|
} |
237
|
|
|
|
238
|
|
|
protected function printDebug(array $data) |
239
|
|
|
{ |
240
|
|
|
$this->log->debug('origin: ' . $data['raw']); |
241
|
|
|
$this->log->debug('completed: ' . $data['opti']); |
242
|
|
|
$this->log->debug('modifs: ' . $data['modifs']); |
243
|
|
|
$this->log->debug('version: ' . $data['version']); |
244
|
|
|
} |
245
|
|
|
|
246
|
|
|
protected function editPage(): bool |
247
|
|
|
{ |
248
|
|
|
$miniSummary = $this->generateFinalSummary(); |
|
|
|
|
249
|
|
|
|
250
|
|
|
$this->log->debug("sleep 2..."); |
251
|
|
|
sleep(2); // todo ??? |
252
|
|
|
|
253
|
|
|
die("\n*** STOP ON OuvrageEditWorker:253 FOR REFACTORING ***\n"); |
|
|
|
|
254
|
|
|
|
255
|
|
|
try { |
|
|
|
|
256
|
|
|
$editInfo = ServiceFactory::editInfo($miniSummary, $this->pageWorkStatus->minorFlag, $this->pageWorkStatus->botFlag); |
257
|
|
|
$success = $this->wikiPageAction->editPage(Normalizer::normalize($this->pageWorkStatus->wikiText), $editInfo); |
258
|
|
|
} catch (Throwable $e) { |
259
|
|
|
// Invalid CSRF token. |
260
|
|
|
if (strpos($e->getMessage(), 'Invalid CSRF token') !== false) { |
261
|
|
|
$this->log->alert("*** Invalid CSRF token \n"); |
262
|
|
|
throw new Exception('Invalid CSRF token', $e->getCode(), $e); |
263
|
|
|
} else { |
264
|
|
|
$this->log->warning('Exception in editPage() ' . $e->getMessage()); |
265
|
|
|
sleep(10); |
266
|
|
|
|
267
|
|
|
return false; |
268
|
|
|
} |
269
|
|
|
} |
270
|
|
|
$this->log->info($success ? "Edition Ok\n" : "***** Edition KO !\n"); |
271
|
|
|
|
272
|
|
|
return $success; |
273
|
|
|
} |
274
|
|
|
|
275
|
|
|
protected function updateDb(array $pageOuvrageCollection) |
276
|
|
|
{ |
277
|
|
|
$title = $pageOuvrageCollection[0]['page']; |
278
|
|
|
foreach ($pageOuvrageCollection as $ouvrageData) { |
279
|
|
|
$this->db->sendEditedData(['id' => $ouvrageData['id']]); |
|
|
|
|
280
|
|
|
} |
281
|
|
|
try { |
282
|
|
|
if (self::EDIT_SIGNALEMENT && !empty($this->pageWorkStatus->errorWarning[$title])) { |
283
|
|
|
$this->sendOuvrageErrorsOnTalkPage($pageOuvrageCollection, $this->log); |
284
|
|
|
} |
285
|
|
|
} catch (Throwable $e) { |
286
|
|
|
$this->log->warning('Exception in editPage() ' . $e->getMessage()); |
287
|
|
|
unset($e); |
288
|
|
|
} |
289
|
|
|
|
290
|
|
|
if (!$this->pageWorkStatus->botFlag) { |
291
|
|
|
$this->log->debug("sleep " . self::DELAY_NO_BOTFLAG_SECONDS); |
292
|
|
|
sleep(self::DELAY_NO_BOTFLAG_SECONDS); |
293
|
|
|
} |
294
|
|
|
if ($this->pageWorkStatus->botFlag) { |
295
|
|
|
$this->log->debug("sleep " . self::DELAY_BOTFLAG_SECONDS); |
296
|
|
|
sleep(self::DELAY_BOTFLAG_SECONDS); |
297
|
|
|
} |
298
|
|
|
} |
299
|
|
|
|
300
|
|
|
/** |
301
|
|
|
* todo extract |
302
|
|
|
* Pour éviter les doublons dans signalements d'erreur. |
303
|
|
|
*/ |
304
|
|
|
protected function addErrorWarning(string $title, string $text): void |
305
|
|
|
{ |
306
|
|
|
if (!isset($this->pageWorkStatus->errorWarning[$title]) || !in_array($text, $this->pageWorkStatus->errorWarning[$title])) { |
307
|
|
|
$this->pageWorkStatus->errorWarning[$title][] = $text; |
308
|
|
|
} |
309
|
|
|
} |
310
|
|
|
} |
311
|
|
|
|