1
|
|
|
<?php |
2
|
|
|
/** |
3
|
|
|
* This file is part of dispositif/wikibot application |
4
|
|
|
* 2019 : Philippe M. <[email protected]> |
5
|
|
|
* For the full copyright and MIT license information, please view the LICENSE file. |
6
|
|
|
*/ |
7
|
|
|
|
8
|
|
|
declare(strict_types=1); |
9
|
|
|
|
10
|
|
|
namespace App\Application\Examples; |
11
|
|
|
|
12
|
|
|
use App\Application\Bot; |
13
|
|
|
use App\Application\Memory; |
14
|
|
|
use App\Application\WikiPageAction; |
15
|
|
|
use App\Domain\Utils\WikiTextUtil; |
16
|
|
|
use App\Infrastructure\DbAdapter; |
17
|
|
|
use App\Infrastructure\ServiceFactory; |
18
|
|
|
use Mediawiki\DataModel\EditInfo; |
19
|
|
|
use Throwable; |
20
|
|
|
|
21
|
|
|
//use App\Application\CLI; |
22
|
|
|
|
23
|
|
|
include __DIR__.'/../myBootstrap.php'; |
24
|
|
|
|
25
|
|
|
$process = new EditProcess(); |
26
|
|
|
$process->run(); |
27
|
|
|
|
28
|
|
|
/** |
29
|
|
|
* TODO refac |
30
|
|
|
* Class EditProcess |
31
|
|
|
* |
32
|
|
|
* @package App\Application\Examples |
33
|
|
|
*/ |
34
|
|
|
class EditProcess |
35
|
|
|
{ |
36
|
|
|
const TASK_NAME = 'Amélioration bibliographique'; |
37
|
|
|
const EDIT_SIGNALEMENT = false; |
38
|
|
|
|
39
|
|
|
const CITATION_LIMIT = 100; |
40
|
|
|
const DELAY_BOTFLAG_SECONDS = 30; |
41
|
|
|
const DELAY_NOBOT_IN_SECONDS = 150; |
42
|
|
|
const ERROR_MSG_TEMPLATE = __DIR__.'/../templates/message_errors.wiki'; |
43
|
|
|
|
44
|
|
|
private $db; |
45
|
|
|
private $bot; |
46
|
|
|
private $wiki; |
47
|
|
|
private $wikiText; |
48
|
|
|
private $citationSummary; |
49
|
|
|
private $citationVersion = ''; |
50
|
|
|
private $errorWarning = []; |
51
|
|
|
private $importantSummary = []; |
52
|
|
|
private $nbRows; |
53
|
|
|
|
54
|
|
|
// Minor flag on edit |
55
|
|
|
private $minorFlag = true; |
56
|
|
|
// Bot flag on edit |
57
|
|
|
private $botFlag = true; |
58
|
|
|
|
59
|
|
|
public function __construct() |
60
|
|
|
{ |
61
|
|
|
$this->db = new DbAdapter(); |
62
|
|
|
$this->bot = new Bot(); |
63
|
|
|
|
64
|
|
|
$this->wiki = ServiceFactory::wikiApi(); |
65
|
|
|
} |
66
|
|
|
|
67
|
|
|
public function run() |
68
|
|
|
{ |
69
|
|
|
$memory = new Memory(); |
70
|
|
|
while (true) { |
71
|
|
|
echo "\n-------------------------------------\n\n"; |
72
|
|
|
echo date("Y-m-d H:i")."\n"; |
73
|
|
|
$memory->echoMemory(true); |
74
|
|
|
|
75
|
|
|
$this->pageProcess(); |
76
|
|
|
} |
77
|
|
|
} |
78
|
|
|
|
79
|
|
|
private function pageProcess() |
80
|
|
|
{ |
81
|
|
|
$this->initialize(); |
82
|
|
|
|
83
|
|
|
// get a random queue line |
84
|
|
|
$json = $this->db->getAllRowsToEdit(self::CITATION_LIMIT); |
85
|
|
|
$data = json_decode($json, true); |
86
|
|
|
|
87
|
|
|
if (empty($data)) { |
88
|
|
|
echo "SKIP : no rows to process\n"; |
89
|
|
|
|
90
|
|
|
return false; |
91
|
|
|
} |
92
|
|
|
|
93
|
|
|
try { |
94
|
|
|
$title = $data[0]['page']; |
95
|
|
|
echo "$title \n"; |
96
|
|
|
$page = new WikiPageAction($this->wiki, $title); |
97
|
|
|
} catch (\Exception $e) { |
98
|
|
|
echo "*** WikiPageAction error : $title \n"; |
99
|
|
|
sleep(60); |
100
|
|
|
|
101
|
|
|
return false; |
102
|
|
|
} |
103
|
|
|
|
104
|
|
|
// TODO : HACK |
105
|
|
|
if (in_array($page->getLastEditor(), [getenv('BOT_NAME'), getenv('BOT_OWNER')])) { |
106
|
|
|
echo "SKIP : édité recemment par bot/dresseur.\n"; |
107
|
|
|
$this->db->skipRow($title); |
108
|
|
|
|
109
|
|
|
return false; |
110
|
|
|
} |
111
|
|
|
$this->wikiText = $page->getText(); |
112
|
|
|
|
113
|
|
|
if ($this->bot->minutesSinceLastEdit($title) < 15) { |
114
|
|
|
echo "SKIP : édition humaine dans les dernières 15 minutes.\n"; |
115
|
|
|
|
116
|
|
|
return false; |
117
|
|
|
} |
118
|
|
|
|
119
|
|
|
// Skip AdQ |
120
|
|
|
if (preg_match('#{{ ?En-tête label#i', $this->wikiText) > 0) { |
121
|
|
|
echo "SKIP : AdQ ou BA.\n"; |
122
|
|
|
$this->db->skipRow($title); |
123
|
|
|
|
124
|
|
|
return false; |
125
|
|
|
} |
126
|
|
|
|
127
|
|
|
// GET all article lines from db |
128
|
|
|
echo sprintf(">> %s rows to process\n", count($data)); |
129
|
|
|
|
130
|
|
|
// foreach line |
131
|
|
|
$changed = false; |
132
|
|
|
foreach ($data as $dat) { |
133
|
|
|
// hack temporaire pour éviter articles dont CompleteProcess incomplet |
134
|
|
|
if (empty($dat['opti']) || empty($dat['optidate']) || $dat['optidate'] < DbAdapter::OPTI_VALID_DATE) { |
135
|
|
|
echo "SKIP : Complètement incomplet de l'article \n"; |
136
|
|
|
|
137
|
|
|
return false; |
138
|
|
|
} |
139
|
|
|
$success = $this->dataProcess($dat); |
140
|
|
|
$changed = ($success) ? true : $changed; |
141
|
|
|
} |
142
|
|
|
if (!$changed) { |
143
|
|
|
echo "Rien à changer...\n\n"; |
144
|
|
|
|
145
|
|
|
return false; |
146
|
|
|
} |
147
|
|
|
|
148
|
|
|
// EDIT THE PAGE |
149
|
|
|
if (!$this->wikiText) { |
150
|
|
|
return false; |
151
|
|
|
} |
152
|
|
|
|
153
|
|
|
$miniSummary = $this->generateSummary(); |
154
|
|
|
echo "Edition ?\n".$miniSummary."\n\n"; |
155
|
|
|
echo "sleep 30...\n"; |
156
|
|
|
sleep(30); |
157
|
|
|
|
158
|
|
|
$editInfo = new EditInfo($miniSummary, $this->minorFlag, $this->botFlag); |
159
|
|
|
$success = $page->editPage(\Normalizer::normalize($this->wikiText), $editInfo); |
160
|
|
|
echo ($success) ? "Ok\n" : "***** Erreur edit\n"; |
161
|
|
|
|
162
|
|
|
if ($success) { |
163
|
|
|
// updata DB |
164
|
|
|
foreach ($data as $dat) { |
165
|
|
|
$this->db->sendEditedData(['id' => $dat['id']]); |
166
|
|
|
} |
167
|
|
|
|
168
|
|
|
try { |
169
|
|
|
if (self::EDIT_SIGNALEMENT) { |
170
|
|
|
$this->sendErrorMessage($data); |
171
|
|
|
} |
172
|
|
|
} catch (Throwable $e) { |
173
|
|
|
dump($e); |
174
|
|
|
unset($e); |
175
|
|
|
} |
176
|
|
|
|
177
|
|
|
if (!$this->botFlag) { |
178
|
|
|
echo "sleep ".self::DELAY_NOBOT_IN_SECONDS."\n"; |
179
|
|
|
sleep(self::DELAY_NOBOT_IN_SECONDS); |
180
|
|
|
} |
181
|
|
|
echo "sleep ".self::DELAY_BOTFLAG_SECONDS."\n"; |
182
|
|
|
sleep(self::DELAY_BOTFLAG_SECONDS); |
183
|
|
|
} |
184
|
|
|
|
185
|
|
|
return $success; |
186
|
|
|
} |
187
|
|
|
|
188
|
|
|
private function dataProcess(array $data): bool |
189
|
|
|
{ |
190
|
|
|
$origin = $data['raw']; |
191
|
|
|
$completed = $data['opti']; |
192
|
|
|
|
193
|
|
|
dump($origin, $completed, $data['modifs'], $data['version']); |
194
|
|
|
|
195
|
|
|
if (WikiTextUtil::isCommented($origin)) { |
196
|
|
|
echo "SKIP: template avec commentaire HTML\n"; |
197
|
|
|
|
198
|
|
|
return false; |
199
|
|
|
} |
200
|
|
|
|
201
|
|
|
$find = mb_strpos($this->wikiText, $origin); |
202
|
|
|
if (!$find) { |
203
|
|
|
echo "String non trouvée. \n\n"; |
204
|
|
|
|
205
|
|
|
return false; |
206
|
|
|
} |
207
|
|
|
|
208
|
|
|
$this->checkErrorWarning($data); |
209
|
|
|
|
210
|
|
|
// Replace text |
211
|
|
|
$newText = WikiPageAction::replaceTemplateInText($this->wikiText, $origin, $completed); |
212
|
|
|
|
213
|
|
|
if (!$newText || $newText === $this->wikiText) { |
214
|
|
|
echo "newText error\n"; |
215
|
|
|
|
216
|
|
|
return false; |
217
|
|
|
} |
218
|
|
|
$this->wikiText = $newText; |
219
|
|
|
$this->minorFlag = ('1' === $data['major']) ? false : $this->minorFlag; |
220
|
|
|
$this->citationVersion = $data['version']; |
221
|
|
|
$this->citationSummary[] = $data['modifs']; |
222
|
|
|
$this->nbRows++; |
223
|
|
|
|
224
|
|
|
return true; |
225
|
|
|
} |
226
|
|
|
|
227
|
|
|
/** |
228
|
|
|
* Generate wiki edition summary. |
229
|
|
|
* |
230
|
|
|
* @return string |
231
|
|
|
*/ |
232
|
|
|
public function generateSummary(): string |
233
|
|
|
{ |
234
|
|
|
// Start summary with "Bot" when using botflag, else "*" |
235
|
|
|
$prefix = ($this->botFlag) ? 'bot' : '☆'; |
236
|
|
|
// add "/!\" when errorWarning |
237
|
|
|
$prefix = (!empty($this->errorWarning) && !$this->botFlag) ? '⚠' : $prefix; |
238
|
|
|
|
239
|
|
|
|
240
|
|
|
// basic modifs |
241
|
|
|
$citeSummary = implode(' ', $this->citationSummary); |
242
|
|
|
// replace by list of modifs to verify by humans |
243
|
|
|
if (!empty($this->importantSummary)) { |
244
|
|
|
$citeSummary = implode(', ', $this->importantSummary); |
245
|
|
|
} |
246
|
|
|
|
247
|
|
|
$summary = sprintf( |
248
|
|
|
'%s [%s/%s] %s %s : %s', |
249
|
|
|
$prefix, |
250
|
|
|
str_replace('v', '', $this->bot::getGitVersion()), |
251
|
|
|
str_replace('v0.', '', $this->citationVersion), |
252
|
|
|
self::TASK_NAME, |
253
|
|
|
$this->nbRows, |
254
|
|
|
$citeSummary |
255
|
|
|
); |
256
|
|
|
|
257
|
|
|
// shrink long summary if no important details to verify |
258
|
|
|
if (empty($this->importantSummary)) { |
259
|
|
|
$length = strlen($summary); |
260
|
|
|
$summary = substr($summary, 0, 80); |
261
|
|
|
$summary .= ($length > strlen($summary)) ? '…' : ''; |
262
|
|
|
} |
263
|
|
|
|
264
|
|
|
return $summary; |
265
|
|
|
} |
266
|
|
|
|
267
|
|
|
/** |
268
|
|
|
* Vérifie alerte d'erreurs humaines. |
269
|
|
|
* |
270
|
|
|
* @param array $data |
271
|
|
|
* |
272
|
|
|
* @throws \Exception |
273
|
|
|
*/ |
274
|
|
|
private function checkErrorWarning(array $data): void |
275
|
|
|
{ |
276
|
|
|
if (!isset($data['opti'])) { |
277
|
|
|
throw new \LogicException('Opti NULL'); |
278
|
|
|
} |
279
|
|
|
|
280
|
|
|
// paramètre inconnu |
281
|
|
|
if (preg_match( |
282
|
|
|
"#\|[^|]+<!-- ?(PARAMETRE [^>]+ N'EXISTE PAS) ?-->#", |
283
|
|
|
$data['opti'], |
284
|
|
|
$matches |
285
|
|
|
) > 0 |
286
|
|
|
) { |
287
|
|
|
$this->errorWarning[$data['page']][] = $matches[0]; |
288
|
|
|
$this->botFlag = false; |
289
|
|
|
$this->addSummaryTag('paramètre non corrigé'); |
290
|
|
|
} |
291
|
|
|
|
292
|
|
|
// ISBN invalide |
293
|
|
|
if (preg_match("#isbn invalide ?=[^|}]+#i", $data['opti'], $matches) > 0) { |
294
|
|
|
$this->errorWarning[$data['page']][] = $matches[0]; |
295
|
|
|
$this->botFlag = false; |
296
|
|
|
$this->addSummaryTag('ISBN invalide'); |
297
|
|
|
} |
298
|
|
|
|
299
|
|
|
// Edits avec ajout conséquent de donnée |
300
|
|
|
if (preg_match('#distinction des auteurs#', $data['modifs']) > 0) { |
301
|
|
|
$this->botFlag = false; |
302
|
|
|
$this->addSummaryTag('distinction des auteurs'); |
303
|
|
|
} |
304
|
|
|
// prédiction paramètre correct |
305
|
|
|
if (preg_match('#[^,]+=>[^,]+#', $data['modifs'], $matches) > 0) { |
306
|
|
|
$this->botFlag = false; |
307
|
|
|
$this->addSummaryTag(sprintf('%s', $matches[0])); |
308
|
|
|
} |
309
|
|
|
if (preg_match('#\+\+sous-titre#', $data['modifs']) > 0) { |
310
|
|
|
$this->botFlag = false; |
311
|
|
|
$this->addSummaryTag('+sous-titre'); |
312
|
|
|
} |
313
|
|
|
if (preg_match('#\+lieu#', $data['modifs']) > 0) { |
314
|
|
|
$this->addSummaryTag('+lieu'); |
315
|
|
|
} |
316
|
|
|
if (preg_match('#\+éditeur#', $data['modifs']) > 0) { |
317
|
|
|
$this->addSummaryTag('éditeur'); |
318
|
|
|
} |
319
|
|
|
// if (preg_match('#\+langue#', $data['modifs']) > 0) { |
320
|
|
|
// $this->addSummaryTag('langue'); |
321
|
|
|
// } |
322
|
|
|
|
323
|
|
|
// mention BnF si ajout donnée + ajout identifiant bnf= |
324
|
|
|
if (!empty($this->importantSummary) && preg_match('#\+bnf#i', $data['modifs'], $matches) > 0) { |
325
|
|
|
$this->addSummaryTag('[[BnF]]'); |
326
|
|
|
} |
327
|
|
|
} |
328
|
|
|
|
329
|
|
|
/** |
330
|
|
|
* For substantive or ambiguous modifications done. |
331
|
|
|
* |
332
|
|
|
* @param string $tag |
333
|
|
|
*/ |
334
|
|
|
private function addSummaryTag(string $tag) |
335
|
|
|
{ |
336
|
|
|
if (!in_array($tag, $this->importantSummary)) { |
337
|
|
|
$this->importantSummary[] = $tag; |
338
|
|
|
} |
339
|
|
|
} |
340
|
|
|
|
341
|
|
|
/** |
342
|
|
|
* @param array $rows Collection of citations |
343
|
|
|
*/ |
344
|
|
|
private function sendErrorMessage(array $rows): void |
345
|
|
|
{ |
346
|
|
|
if (empty($this->errorWarning[$rows[0]['page']])) { |
347
|
|
|
return; |
348
|
|
|
} |
349
|
|
|
echo "** Send Error Message on talk page. Wait 10... \n"; |
350
|
|
|
sleep(10); |
351
|
|
|
|
352
|
|
|
// format wiki message |
353
|
|
|
$errorList = ''; |
354
|
|
|
foreach ($this->errorWarning[$rows[0]['page']] as $error) { |
355
|
|
|
$errorList .= sprintf("* <span style=\"background:#FCDFE8\"><nowiki>%s</nowiki></span> \n", $error); |
356
|
|
|
} |
357
|
|
|
$errorMessage = file_get_contents(self::ERROR_MSG_TEMPLATE); |
358
|
|
|
$errorMessage = str_replace('##ERROR LIST##', trim($errorList), $errorMessage); |
359
|
|
|
$errorMessage = str_replace('##ARTICLE##', $rows[0]['page'], $errorMessage); |
360
|
|
|
|
361
|
|
|
// Edit wiki talk page |
362
|
|
|
try { |
363
|
|
|
$talkPage = new WikiPageAction($this->wiki, 'Discussion:'.$rows[0]['page']); |
364
|
|
|
$editInfo = new EditInfo('Signalement erreur {ouvrage}', false, false, 5); |
|
|
|
|
365
|
|
|
$talkPage->addToBottomOfThePage($errorMessage, $editInfo); |
366
|
|
|
} catch (\Throwable $e) { |
367
|
|
|
unset($e); |
368
|
|
|
} |
369
|
|
|
} |
370
|
|
|
|
371
|
|
|
private function initialize(): void |
372
|
|
|
{ |
373
|
|
|
// initialisation vars |
374
|
|
|
$this->botFlag = true; |
375
|
|
|
$this->errorWarning = []; |
376
|
|
|
$this->wikiText = null; |
377
|
|
|
$this->citationSummary = []; |
378
|
|
|
$this->importantSummary = []; |
379
|
|
|
$this->minorFlag = true; |
380
|
|
|
$this->nbRows = 0; |
381
|
|
|
|
382
|
|
|
$this->bot->checkStopOnTalkpage(); |
383
|
|
|
} |
384
|
|
|
|
385
|
|
|
} |
386
|
|
|
|
This check compares calls to functions or methods with their respective definitions. If the call has more arguments than are defined, it raises an issue.
If a function is defined several times with a different number of parameters, the check may pick up the wrong definition and report false positives. One codebase where this has been known to happen is Wordpress. Please note the @ignore annotation hint above.