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\Domain; |
11
|
|
|
|
12
|
|
|
use App\Domain\Models\Wiki\GoogleLivresTemplate; |
13
|
|
|
use App\Domain\Models\Wiki\OuvrageTemplate; |
14
|
|
|
use App\Domain\Utils\TextUtil; |
15
|
|
|
use App\Domain\Utils\WikiTextUtil; |
16
|
|
|
use App\Infrastructure\FileManager; |
17
|
|
|
use Exception; |
18
|
|
|
use function mb_strlen; |
19
|
|
|
use Throwable; |
20
|
|
|
|
21
|
|
|
/** |
22
|
|
|
* Legacy. |
23
|
|
|
* TODO move methods to OuvrageClean setters |
24
|
|
|
* TODO AbstractProcess |
25
|
|
|
* TODO observer/event (log, MajorEdition) |
26
|
|
|
* Class OuvrageProcess. |
27
|
|
|
*/ |
28
|
|
|
class OuvrageOptimize |
29
|
|
|
{ |
30
|
|
|
const CONVERT_GOOGLEBOOK_TEMPLATE = false; // change OuvrageOptimizeTest !! |
31
|
|
|
|
32
|
|
|
const WIKI_LANGUAGE = 'fr'; |
33
|
|
|
|
34
|
|
|
protected $original; |
35
|
|
|
|
36
|
|
|
private $wikiPageTitle; |
37
|
|
|
|
38
|
|
|
private $log = []; |
39
|
|
|
|
40
|
|
|
public $notCosmetic = false; |
41
|
|
|
|
42
|
|
|
public $major = false; |
43
|
|
|
|
44
|
|
|
private $ouvrage; |
45
|
|
|
|
46
|
|
|
private $currentTask; |
47
|
|
|
|
48
|
|
|
// todo inject TextUtil + ArticleVersion ou WikiRef |
49
|
|
|
public function __construct(OuvrageTemplate $ouvrage, $wikiPageTitle = null) |
50
|
|
|
{ |
51
|
|
|
$this->original = $ouvrage; |
52
|
|
|
$this->ouvrage = clone $ouvrage; |
53
|
|
|
$this->wikiPageTitle = ($wikiPageTitle) ?? null; |
54
|
|
|
} |
55
|
|
|
|
56
|
|
|
public function doTasks(): self |
57
|
|
|
{ |
58
|
|
|
$this->cleanAndPredictErrorParameters(); |
59
|
|
|
|
60
|
|
|
$this->processAuthors(); |
61
|
|
|
|
62
|
|
|
$this->processTitle(); |
63
|
|
|
$this->processEditeur(); |
64
|
|
|
$this->processDates(); |
65
|
|
|
$this->externalTemplates(); |
66
|
|
|
$this->predictFormatByPattern(); |
67
|
|
|
|
68
|
|
|
$this->processIsbn(); |
69
|
|
|
$this->processBnf(); |
70
|
|
|
|
71
|
|
|
$this->processLang(); |
72
|
|
|
$this->processLocation(); // 'lieu' |
73
|
|
|
|
74
|
|
|
$this->GoogleBookURL('lire en ligne'); |
75
|
|
|
$this->GoogleBookURL('présentation en ligne'); |
76
|
|
|
|
77
|
|
|
return $this; |
78
|
|
|
} |
79
|
|
|
|
80
|
|
|
/** |
81
|
|
|
* Todo: injection dep. |
82
|
|
|
* Todo : "[s. l.]" sans lieu "s.l.n.d." sans lieu ni date. |
83
|
|
|
* |
84
|
|
|
* @throws Exception |
85
|
|
|
*/ |
86
|
|
|
private function processLocation() |
87
|
|
|
{ |
88
|
|
|
$location = $this->getParam('lieu'); |
89
|
|
|
if (empty($location)) { |
90
|
|
|
return; |
91
|
|
|
} |
92
|
|
|
|
93
|
|
|
// typo and unwikify |
94
|
|
|
$memo = $location; |
95
|
|
|
$location = WikiTextUtil::unWikify($location); |
96
|
|
|
$location = TextUtil::mb_ucfirst($location); |
97
|
|
|
if ($memo !== $location) { |
98
|
|
|
$this->setParam('lieu', $location); |
99
|
|
|
$this->log('±lieu'); |
100
|
|
|
$this->notCosmetic = true; |
101
|
|
|
} |
102
|
|
|
|
103
|
|
|
// french translation : "London"->"Londres" |
104
|
|
|
$manager = new FileManager(); |
105
|
|
|
$row = $manager->findCSVline(__DIR__.'/resources/traduction_ville.csv', $location); |
106
|
|
|
if (!empty($row) && !empty($row[1])) { |
107
|
|
|
$this->setParam('lieu', $row[1]); |
108
|
|
|
$this->log('lieu francisé'); |
109
|
|
|
$this->notCosmetic = true; |
110
|
|
|
} |
111
|
|
|
} |
112
|
|
|
|
113
|
|
|
private function processBnf() |
114
|
|
|
{ |
115
|
|
|
$bnf = $this->getParam('bnf'); |
116
|
|
|
if (!$bnf) { |
117
|
|
|
return; |
118
|
|
|
} |
119
|
|
|
$bnf = str_ireplace('FRBNF', '', $bnf); |
120
|
|
|
$this->setParam('bnf', $bnf); |
121
|
|
|
} |
122
|
|
|
|
123
|
|
|
/** |
124
|
|
|
* @throws Exception |
125
|
|
|
*/ |
126
|
|
|
private function processAuthors() |
127
|
|
|
{ |
128
|
|
|
$this->distinguishAuthors(); |
129
|
|
|
//$this->fusionFirstNameAndName(); // desactived : no consensus |
130
|
|
|
} |
131
|
|
|
|
132
|
|
|
/** |
133
|
|
|
* Detect and correct multiple authors in same parameter. |
134
|
|
|
* Like "auteurs=J. M. Waller, M. Bigger, R. J. Hillocks". |
135
|
|
|
* |
136
|
|
|
* @throws Exception |
137
|
|
|
*/ |
138
|
|
|
private function distinguishAuthors() |
139
|
|
|
{ |
140
|
|
|
// merge params of author 1 |
141
|
|
|
$auteur1 = $this->getParam('auteur') ?? ''; |
142
|
|
|
$auteur1 .= $this->getParam('auteurs') ?? ''; |
143
|
|
|
$auteur1 .= $this->getParam('prénom1') ?? ''; |
144
|
|
|
$auteur1 .= ' '.$this->getParam('nom1') ?? ''; |
145
|
|
|
$auteur1 = trim($auteur1); |
146
|
|
|
// of authors 2 |
147
|
|
|
$auteur2 = $this->getParam('auteur2') ?? ''; |
148
|
|
|
$auteur2 .= $this->getParam('prénom2') ?? ''; |
149
|
|
|
$auteur2 .= ' '.$this->getParam('nom2') ?? ''; |
150
|
|
|
$auteur2 = trim($auteur2); |
151
|
|
|
|
152
|
|
|
// skip if wikilink in author |
153
|
|
|
if (empty($auteur1) || WikiTextUtil::isWikify($auteur1)) { |
154
|
|
|
return; |
155
|
|
|
} |
156
|
|
|
|
157
|
|
|
$machine = new PredictAuthors(); |
158
|
|
|
$res = $machine->predictAuthorNames($auteur1); |
159
|
|
|
|
160
|
|
|
if (1 === count($res)) { |
161
|
|
|
// auteurs->auteur? |
162
|
|
|
return; |
163
|
|
|
} |
164
|
|
|
// Many authors... and empty "auteur2" |
165
|
|
|
if (count($res) >= 2 && empty($auteur2)) { |
166
|
|
|
// delete author-params |
167
|
|
|
array_map( |
168
|
|
|
function ($param) { |
169
|
|
|
$this->unsetParam($param); |
170
|
|
|
}, |
171
|
|
|
['auteur', 'auteurs', 'prénom1', 'nom1'] |
172
|
|
|
); |
173
|
|
|
// iterate and edit new values |
174
|
|
|
for ($i = 0; $i < count($res); ++$i) { |
|
|
|
|
175
|
|
|
$this->setParam(sprintf('auteur%s', $i + 1), $res[$i]); |
176
|
|
|
} |
177
|
|
|
$this->log('distinction auteurs'); |
178
|
|
|
$this->major = true; |
179
|
|
|
$this->notCosmetic = true; |
180
|
|
|
} |
181
|
|
|
} |
182
|
|
|
|
183
|
|
|
/** |
184
|
|
|
* todo: move/implement. |
185
|
|
|
* |
186
|
|
|
* @throws Exception |
187
|
|
|
*/ |
188
|
|
|
private function processLang() |
189
|
|
|
{ |
190
|
|
|
$lang = $this->getParam('langue') ?? null; |
191
|
|
|
|
192
|
|
|
if ($lang) { |
193
|
|
|
$replace = [ |
194
|
|
|
'french' => 'fr', |
195
|
|
|
'française' => 'fr', |
196
|
|
|
'français' => 'fr', |
197
|
|
|
'anglaise' => 'en', |
198
|
|
|
'anglais' => 'en', |
199
|
|
|
'english' => 'en', |
200
|
|
|
'japonais' => 'ja', |
201
|
|
|
'japanese' => 'ja', |
202
|
|
|
'allemand' => 'de', |
203
|
|
|
'deutsch' => 'de', |
204
|
|
|
'german' => 'de', |
205
|
|
|
'espagnol' => 'es', |
206
|
|
|
'spanish' => 'es', |
207
|
|
|
'español' => 'es', |
208
|
|
|
]; |
209
|
|
|
$lang2 = str_ireplace( |
210
|
|
|
array_keys($replace), |
211
|
|
|
array_values($replace), |
212
|
|
|
mb_strtolower($lang) |
213
|
|
|
); |
214
|
|
|
|
215
|
|
|
if ($lang !== $lang2) { |
216
|
|
|
$this->setParam('langue', $lang2); |
217
|
|
|
if (self::WIKI_LANGUAGE !== $lang2) { |
218
|
|
|
$this->log('±langue'); |
219
|
|
|
} |
220
|
|
|
} |
221
|
|
|
} |
222
|
|
|
} |
223
|
|
|
|
224
|
|
|
/** |
225
|
|
|
* Validate or correct ISBN. |
226
|
|
|
* |
227
|
|
|
* @throws Exception |
228
|
|
|
*/ |
229
|
|
|
private function processIsbn() |
230
|
|
|
{ |
231
|
|
|
$isbn = $this->getParam('isbn') ?? ''; |
232
|
|
|
|
233
|
|
|
if (empty($isbn)) { |
234
|
|
|
return; |
235
|
|
|
} |
236
|
|
|
|
237
|
|
|
$isbnMachine = new IsbnFacade($isbn); |
238
|
|
|
|
239
|
|
|
try { |
240
|
|
|
$isbnMachine->validate(); |
241
|
|
|
$isbn13 = $isbnMachine->format('ISBN-13'); |
242
|
|
|
} catch (Throwable $e) { |
243
|
|
|
// ISBN not validated |
244
|
|
|
// TODO : bot ISBN invalide (queue, message PD...) |
245
|
|
|
$this->setParam( |
246
|
|
|
'isbn invalide', |
247
|
|
|
sprintf('%s %s', $isbn, $e->getMessage() ?? '') |
248
|
|
|
); |
249
|
|
|
$this->log(sprintf('ISBN invalide: %s', $e->getMessage())); |
250
|
|
|
$this->notCosmetic = true; |
251
|
|
|
|
252
|
|
|
// TODO log file ISBNinvalide |
253
|
|
|
return; |
254
|
|
|
} |
255
|
|
|
|
256
|
|
|
// ISBN 10 ? |
257
|
|
|
if (10 === mb_strlen(str_replace('-', '', $isbn)) && !$this->getParam('isbn10')) { |
258
|
|
|
$this->setParam('isbn10', $isbn); |
259
|
|
|
$this->log('isbn10'); |
260
|
|
|
$this->notCosmetic = true; |
261
|
|
|
} |
262
|
|
|
|
263
|
|
|
// ISBN correction |
264
|
|
|
if ($isbn13 !== $isbn) { |
265
|
|
|
$this->setParam('isbn', $isbn13); |
266
|
|
|
$this->log('ISBN'); |
267
|
|
|
$this->notCosmetic = true; |
268
|
|
|
} |
269
|
|
|
} |
270
|
|
|
|
271
|
|
|
private function processTitle() |
272
|
|
|
{ |
273
|
|
|
$this->currentTask = 'titres'; |
274
|
|
|
|
275
|
|
|
$oldtitre = $this->getParam('titre'); |
276
|
|
|
$this->langInTitle(); |
277
|
|
|
$this->deWikifyExternalLink('titre'); |
278
|
|
|
$this->upperCaseFirstLetter('titre'); |
279
|
|
|
$this->typoDeuxPoints('titre'); |
280
|
|
|
|
281
|
|
|
$this->extractSubTitle(); |
282
|
|
|
|
283
|
|
|
// 20-11-2019 : Retiré majuscule à sous-titre |
284
|
|
|
|
285
|
|
|
if ($this->getParam('titre') !== $oldtitre) { |
286
|
|
|
$this->log('±titre'); |
287
|
|
|
} |
288
|
|
|
|
289
|
|
|
$this->currentTask = 'titre chapitre'; |
290
|
|
|
$this->valideNumeroChapitre(); |
291
|
|
|
$this->deWikifyExternalLink('titre chapitre'); |
292
|
|
|
$this->upperCaseFirstLetter('titre chapitre'); |
293
|
|
|
$this->typoDeuxPoints('titre chapitre'); |
294
|
|
|
} |
295
|
|
|
|
296
|
|
|
private function detectColon($param): bool |
297
|
|
|
{ |
298
|
|
|
// > 0 don't count a starting colon ":bla" |
299
|
|
|
if (!empty($this->getParam($param)) && mb_strrpos($this->getParam('titre'), ':') > 0) { |
300
|
|
|
return true; |
301
|
|
|
} |
302
|
|
|
|
303
|
|
|
return false; |
304
|
|
|
} |
305
|
|
|
|
306
|
|
|
private function extractSubTitle(): void |
307
|
|
|
{ |
308
|
|
|
// FIXED bug [[fu:bar]] |
309
|
|
|
if (!$this->getParam('titre') || WikiTextUtil::isWikify($this->getParam('titre'))) { |
|
|
|
|
310
|
|
|
return; |
311
|
|
|
} |
312
|
|
|
|
313
|
|
|
if (!$this->detectColon('titre')) { |
314
|
|
|
return; |
315
|
|
|
} |
316
|
|
|
// Que faire si déjà un sous-titre ? |
317
|
|
|
if (!empty($this->getParam('sous-titre'))) { |
318
|
|
|
return; |
319
|
|
|
} |
320
|
|
|
|
321
|
|
|
// titre>5 and sous-titre>5 and sous-titre<40 |
322
|
|
|
if (preg_match('#^(?<titre>[^:]{5,}):(?<st>.{5,40})$#', $this->getParam('titre'), $matches) > 0) { |
323
|
|
|
$this->setParam('titre', trim($matches['titre'])); |
324
|
|
|
$this->setParam('sous-titre', trim($matches['st'])); |
325
|
|
|
$this->log('>sous-titre'); |
326
|
|
|
} |
327
|
|
|
} |
328
|
|
|
|
329
|
|
|
/** |
330
|
|
|
* Normalize a Google Book links. |
331
|
|
|
* Clean the useless URL parameters or transform into wiki-template. |
332
|
|
|
* |
333
|
|
|
* @param $param |
334
|
|
|
* |
335
|
|
|
* @throws Exception |
336
|
|
|
*/ |
337
|
|
|
private function googleBookUrl(string $param): void |
338
|
|
|
{ |
339
|
|
|
$url = $this->getParam($param); |
340
|
|
|
if (empty($url) |
341
|
|
|
|| !GoogleLivresTemplate::isGoogleBookURL($url) |
342
|
|
|
) { |
343
|
|
|
return; |
344
|
|
|
} |
345
|
|
|
|
346
|
|
|
if (self::CONVERT_GOOGLEBOOK_TEMPLATE) { |
347
|
|
|
$template = GoogleLivresTemplate::createFromURL($url); |
348
|
|
|
if ($template) { |
|
|
|
|
349
|
|
|
$this->setParam($param, $template->serialize()); |
350
|
|
|
$this->log('{Google}'); |
351
|
|
|
$this->notCosmetic = true; |
352
|
|
|
|
353
|
|
|
return; |
354
|
|
|
} |
355
|
|
|
} |
356
|
|
|
|
357
|
|
|
$goo = GoogleLivresTemplate::simplifyGoogleUrl($url); |
358
|
|
|
if (!empty($goo)) { |
359
|
|
|
$this->setParam($param, $goo); |
360
|
|
|
$this->log('Google'); |
361
|
|
|
$this->notCosmetic = true; |
362
|
|
|
} |
363
|
|
|
} |
364
|
|
|
|
365
|
|
|
/** |
366
|
|
|
* - {{lang|...}} dans titre => langue=... puis titre nettoyé |
367
|
|
|
* langue=L’utilisation de ce paramètre permet aussi aux synthétiseurs vocaux de reconnaître la langue du titre de |
368
|
|
|
* l’ouvrage. |
369
|
|
|
* Il est possible d'afficher plusieurs langues, en saisissant le nom séparé par des espaces ou des virgules. La première langue doit être celle du titre. |
370
|
|
|
* |
371
|
|
|
* @throws Exception |
372
|
|
|
*/ |
373
|
|
|
private function langInTitle() |
374
|
|
|
{ |
375
|
|
|
if (preg_match( |
376
|
|
|
'#^{{ ?(?:lang|langue) ?\| ?([a-z-]{2,5}) ?\| ?(?:texte=)?([^{}=]+)(?:\|dir=rtl)?}}$#i', |
377
|
|
|
$this->getParam('titre'), |
378
|
|
|
$matches |
379
|
|
|
) > 0 |
380
|
|
|
) { |
381
|
|
|
$lang = trim($matches[1]); |
382
|
|
|
$newtitre = str_replace($matches[0], trim($matches[2]), $this->getParam('titre')); |
383
|
|
|
$this->setParam('titre', $newtitre); |
384
|
|
|
$this->log('°titre'); |
385
|
|
|
if (self::WIKI_LANGUAGE !== $lang && empty($this->getParam('langue'))) { |
386
|
|
|
$this->setParam('langue', $lang); |
387
|
|
|
$this->log('+langue='.$lang); |
388
|
|
|
} |
389
|
|
|
} |
390
|
|
|
} |
391
|
|
|
|
392
|
|
|
private function processDates() |
393
|
|
|
{ |
394
|
|
|
// dewikification |
395
|
|
|
$params = ['date', 'année', 'mois', 'jour']; |
396
|
|
|
foreach ($params as $param) { |
397
|
|
|
if (WikiTextUtil::isWikify(' '.$this->getParam($param))) { |
398
|
|
|
$this->setParam($param, WikiTextUtil::unWikify($this->getParam($param))); |
|
|
|
|
399
|
|
|
} |
400
|
|
|
} |
401
|
|
|
|
402
|
|
|
try { |
403
|
|
|
$this->moveDate2Year(); |
404
|
|
|
} catch (Exception $e) { |
405
|
|
|
dump($e); |
406
|
|
|
} |
407
|
|
|
} |
408
|
|
|
|
409
|
|
|
/** |
410
|
|
|
* todo: move to AbstractWikiTemplate ? |
411
|
|
|
* Correction des parametres rejetés à l'hydratation données. |
412
|
|
|
* |
413
|
|
|
* @throws Exception |
414
|
|
|
*/ |
415
|
|
|
private function cleanAndPredictErrorParameters() |
416
|
|
|
{ |
417
|
|
|
if (empty($this->ouvrage->parametersErrorFromHydrate)) { |
418
|
|
|
return; |
419
|
|
|
} |
420
|
|
|
$allParamsAndAlias = $this->ouvrage->getParamsAndAlias(); |
421
|
|
|
|
422
|
|
|
foreach ($this->ouvrage->parametersErrorFromHydrate as $name => $value) { |
423
|
|
|
if (!is_string($name)) { |
424
|
|
|
// example : 1 => "ouvrage collectif" from |ouvrage collectif| |
425
|
|
|
continue; |
426
|
|
|
} |
427
|
|
|
|
428
|
|
|
// delete error parameter if no value |
429
|
|
|
if (empty($value)) { |
430
|
|
|
unset($this->ouvrage->parametersErrorFromHydrate[$name]); |
431
|
|
|
|
432
|
|
|
continue; |
433
|
|
|
} |
434
|
|
|
|
435
|
|
|
$maxDistance = 1; |
436
|
|
|
if (mb_strlen($name) >= 4) { |
437
|
|
|
$maxDistance = 2; |
438
|
|
|
} |
439
|
|
|
if (mb_strlen($name) >= 8) { |
440
|
|
|
$maxDistance = 3; |
441
|
|
|
} |
442
|
|
|
|
443
|
|
|
$predName = TextUtil::predictCorrectParam($name, $allParamsAndAlias, $maxDistance); |
444
|
|
|
if ($predName && mb_strlen($name) >= 5) { |
445
|
|
|
if (empty($this->getParam($predName))) { |
446
|
|
|
$predName = $this->ouvrage->getAliasParam($predName); |
447
|
|
|
$this->setParam($predName, $value); |
448
|
|
|
$this->log(sprintf('%s⇒%s ?', $name, $predName)); |
449
|
|
|
$this->notCosmetic = true; |
450
|
|
|
unset($this->ouvrage->parametersErrorFromHydrate[$name]); |
451
|
|
|
} |
452
|
|
|
} |
453
|
|
|
} |
454
|
|
|
} |
455
|
|
|
|
456
|
|
|
/** |
457
|
|
|
* TODO : return "" instead of null ? |
458
|
|
|
* |
459
|
|
|
* @param $name |
460
|
|
|
* |
461
|
|
|
* @return string|null |
462
|
|
|
* |
463
|
|
|
* @throws Exception |
464
|
|
|
*/ |
465
|
|
|
private function getParam(string $name): ?string |
466
|
|
|
{ |
467
|
|
|
return $this->ouvrage->getParam($name); |
468
|
|
|
} |
469
|
|
|
|
470
|
|
|
private function setParam($name, $value) |
471
|
|
|
{ |
472
|
|
|
// todo : overwrite setParam() ? |
473
|
|
|
if (!empty($value) || $this->ouvrage->getParam($name)) { |
474
|
|
|
$this->ouvrage->setParam($name, $value); |
475
|
|
|
} |
476
|
|
|
} |
477
|
|
|
|
478
|
|
|
private function log(string $string): void |
479
|
|
|
{ |
480
|
|
|
if (!empty($string)) { |
481
|
|
|
$this->log[] = trim($string); |
482
|
|
|
} |
483
|
|
|
} |
484
|
|
|
|
485
|
|
|
/** |
486
|
|
|
* Bool ? |
487
|
|
|
* déwikification du titre : consensus Bistro 27 août 2011 |
488
|
|
|
* idem 'titre chapitre'. |
489
|
|
|
* |
490
|
|
|
* @param string $param |
491
|
|
|
* |
492
|
|
|
* @throws Exception |
493
|
|
|
*/ |
494
|
|
|
private function deWikifyExternalLink(string $param): void |
495
|
|
|
{ |
496
|
|
|
if (empty($this->getParam($param))) { |
497
|
|
|
return; |
498
|
|
|
} |
499
|
|
|
if (preg_match('#^\[(http[^ \]]+) ([^]]+)]#i', $this->getParam($param), $matches) > 0) { |
500
|
|
|
$this->setParam($param, str_replace($matches[0], $matches[2], $this->getParam($param))); |
501
|
|
|
$this->log('±'.$param); |
502
|
|
|
|
503
|
|
|
if (in_array($param, ['titre', 'titre chapitre'])) { |
504
|
|
|
if (empty($this->getParam('lire en ligne'))) { |
505
|
|
|
$this->setParam('lire en ligne', $matches[1]); |
506
|
|
|
$this->log('+lire en ligne'); |
507
|
|
|
|
508
|
|
|
return; |
509
|
|
|
} |
510
|
|
|
$this->log('autre lien externe: '.$matches[1]); |
511
|
|
|
} |
512
|
|
|
} |
513
|
|
|
} |
514
|
|
|
|
515
|
|
|
private function upperCaseFirstLetter($param) |
516
|
|
|
{ |
517
|
|
|
if (empty($this->getParam($param))) { |
518
|
|
|
return; |
519
|
|
|
} |
520
|
|
|
$newValue = TextUtil::mb_ucfirst(trim($this->getParam($param))); |
521
|
|
|
$this->setParam($param, $newValue); |
522
|
|
|
} |
523
|
|
|
|
524
|
|
|
/** |
525
|
|
|
* Typo internationale 'titre : sous-titre'. |
526
|
|
|
* Fix fantasy typo of subtitle with '. ' or ' - '. |
527
|
|
|
* International Standard Bibliographic Description : |
528
|
|
|
* https://fr.wikipedia.org/wiki/Wikip%C3%A9dia:Le_Bistro/13_janvier_2016#Modif_du_mod%C3%A8le:Ouvrage. |
529
|
|
|
* |
530
|
|
|
* @param $param |
531
|
|
|
* |
532
|
|
|
* @throws Exception |
533
|
|
|
*/ |
534
|
|
|
private function typoDeuxPoints($param) |
535
|
|
|
{ |
536
|
|
|
$origin = $this->getParam($param) ?? ''; |
537
|
|
|
if (empty($origin)) { |
538
|
|
|
return; |
539
|
|
|
} |
540
|
|
|
// FIXED bug [[fu:bar]] |
541
|
|
|
if (WikiTextUtil::isWikify($origin)) { |
542
|
|
|
return; |
543
|
|
|
} |
544
|
|
|
|
545
|
|
|
$origin = TextUtil::replaceNonBreakingSpaces($origin); |
546
|
|
|
|
547
|
|
|
$strTitle = $origin; |
548
|
|
|
|
549
|
|
|
// CORRECTING TYPO FANTASY OF SUBTITLE |
550
|
|
|
|
551
|
|
|
// Replace first '.' by ':' if no ': ' and no numbers around (as PHP 7.3) |
552
|
|
|
// exlude pattern "blabla... blabla" |
553
|
|
|
// TODO: statistics |
554
|
|
|
|
555
|
|
|
// Replace ' - ' or ' / ' (spaced!) by ' : ' if no ':' and no numbers after (as PHP 7.3 or 1939-1945) |
556
|
|
|
if (!mb_strpos(':', $strTitle) && preg_match('#.{6,} ?[-/] ?[^0-9)]{6,}#', $strTitle) > 0) { |
557
|
|
|
$strTitle = preg_replace('#(.{6,}) [-/] ([^0-9)]{6,})#', '$1 : $2', $strTitle); |
558
|
|
|
} |
559
|
|
|
|
560
|
|
|
// international typo style " : " (first occurrence) |
561
|
|
|
$strTitle = preg_replace('#[ ]*:[ ]*#', ' : ', $strTitle); |
562
|
|
|
|
563
|
|
|
if ($strTitle !== $origin) { |
564
|
|
|
$this->setParam($param, $strTitle); |
565
|
|
|
$this->log(sprintf(':%s', $param)); |
566
|
|
|
} |
567
|
|
|
} |
568
|
|
|
|
569
|
|
|
private function valideNumeroChapitre() |
570
|
|
|
{ |
571
|
|
|
$value = $this->getParam('numéro chapitre'); |
572
|
|
|
if (empty($value)) { |
573
|
|
|
return; |
574
|
|
|
} |
575
|
|
|
// "12" ou "VI", {{II}}, II:3 |
576
|
|
|
if (preg_match('#^[0-9IVXL\-.:{}]+$#i', $value) > 0) { |
577
|
|
|
return; |
578
|
|
|
} |
579
|
|
|
// déplace vers "titre chapitre" ? |
580
|
|
|
if (!$this->getParam('titre chapitre')) { |
581
|
|
|
$this->unsetParam('numéro chapitre'); |
582
|
|
|
$this->setParam('titre chapitre', $value); |
583
|
|
|
} |
584
|
|
|
$this->log('≠numéro chapitre'); |
585
|
|
|
} |
586
|
|
|
|
587
|
|
|
private function unsetParam($name) |
588
|
|
|
{ |
589
|
|
|
$this->ouvrage->unsetParam($name); |
590
|
|
|
} |
591
|
|
|
|
592
|
|
|
/** |
593
|
|
|
* TODO move+refac |
594
|
|
|
* TODO PlumeTemplate CommentaireBiblioTemplate ExtraitTemplate |
595
|
|
|
* Probleme {{commentaire biblio}} <> {{commentaire biblio SRL}} |
596
|
|
|
* Generate supplementary templates from obsoletes params. |
597
|
|
|
* |
598
|
|
|
* @throws Exception |
599
|
|
|
*/ |
600
|
|
|
protected function externalTemplates() |
601
|
|
|
{ |
602
|
|
|
// "plume=bla" => {{plume}} |
603
|
|
|
// =oui selon doc, mais testé OK avec "non" |
604
|
|
|
// todo detect duplication ouvrage/plume dans externalTemplate ? |
605
|
|
|
if (!empty($this->getParam('plume'))) { |
606
|
|
|
$plumeValue = $this->getParam('plume'); |
607
|
|
|
$this->ouvrage->externalTemplates[] = (object) [ |
608
|
|
|
'template' => 'plume', |
609
|
|
|
'1' => $plumeValue, |
610
|
|
|
'raw' => '{{nobr|. {{plume}}}}', |
611
|
|
|
]; |
612
|
|
|
$this->unsetParam('plume'); |
613
|
|
|
$this->log('{plume}'); |
614
|
|
|
} |
615
|
|
|
|
616
|
|
|
// "extrait=bla" => {{citation bloc|bla}} |
617
|
|
|
if (!empty($this->getParam('extrait'))) { |
618
|
|
|
$extrait = $this->getParam('extrait'); |
619
|
|
|
// todo bug {{citation bloc}} si "=" ou "|" dans texte de citation |
620
|
|
|
// Legacy : use {{début citation}} ... {{fin citation}} |
621
|
|
|
if (preg_match('#[=|]#', $extrait) > 0) { |
622
|
|
|
$this->ouvrage->externalTemplates[] = (object) [ |
623
|
|
|
'template' => 'début citation', |
624
|
|
|
'1' => '', |
625
|
|
|
'raw' => '{{début citation}}'.$extrait.'{{fin citation}}', |
626
|
|
|
]; |
627
|
|
|
$this->log('{début citation}'); |
628
|
|
|
$this->notCosmetic = true; |
629
|
|
|
} else { |
630
|
|
|
// StdClass |
631
|
|
|
$this->ouvrage->externalTemplates[] = (object) [ |
632
|
|
|
'template' => 'citation bloc', |
633
|
|
|
'1' => $extrait, |
634
|
|
|
'raw' => '{{extrait|'.$extrait.'}}', |
635
|
|
|
]; |
636
|
|
|
$this->log('{extrait}'); |
637
|
|
|
$this->notCosmetic = true; |
638
|
|
|
} |
639
|
|
|
|
640
|
|
|
$this->unsetParam('extrait'); |
641
|
|
|
$this->notCosmetic = true; |
642
|
|
|
} |
643
|
|
|
|
644
|
|
|
// "commentaire=bla" => {{Commentaire biblio|1=bla}} |
645
|
|
|
if (!empty($this->getParam('commentaire'))) { |
646
|
|
|
$commentaire = $this->getParam('commentaire'); |
647
|
|
|
$this->ouvrage->externalTemplates[] = (object) [ |
648
|
|
|
'template' => 'commentaire biblio', |
649
|
|
|
'1' => $commentaire, |
650
|
|
|
'raw' => '{{commentaire biblio|'.$commentaire.'}}', |
651
|
|
|
]; |
652
|
|
|
$this->unsetParam('commentaire'); |
653
|
|
|
$this->log('{commentaire}'); |
654
|
|
|
$this->notCosmetic = true; |
655
|
|
|
} |
656
|
|
|
} |
657
|
|
|
|
658
|
|
|
// ---------------------- |
659
|
|
|
// ---------------------- |
660
|
|
|
// ---------------------- |
661
|
|
|
|
662
|
|
|
/** |
663
|
|
|
* Date->année (nécessaire pour OuvrageComplete). |
664
|
|
|
* |
665
|
|
|
* @throws Exception |
666
|
|
|
*/ |
667
|
|
|
private function moveDate2Year() |
668
|
|
|
{ |
669
|
|
|
$date = $this->getParam('date') ?? false; |
670
|
|
|
if ($date) { |
671
|
|
|
if (preg_match('#^-?[12][0-9][0-9][0-9]$#', $date)) { |
672
|
|
|
$this->setParam('année', $date); |
673
|
|
|
$this->unsetParam('date'); |
674
|
|
|
$this->log('>année'); |
675
|
|
|
} |
676
|
|
|
} |
677
|
|
|
} |
678
|
|
|
|
679
|
|
|
private function predictFormatByPattern() |
680
|
|
|
{ |
681
|
|
|
if (($value = $this->getParam('format'))) { |
682
|
|
|
// predict if 'format électronique' |
683
|
|
|
// format electronique lié au champ 'lire en ligne' |
684
|
|
|
// 2015 https://fr.wikipedia.org/wiki/Discussion_mod%C3%A8le:Ouvrage#format,_format_livre,_format_%C3%A9lectronique |
685
|
|
|
// if (preg_match('#(pdf|epub|html|kindle|audio|\{\{aud|jpg)#i', $value) > 0) { |
686
|
|
|
// |
687
|
|
|
// $this->setParam('format électronique', $value); |
688
|
|
|
// $this->unsetParam('format'); |
689
|
|
|
// $this->log('format:électronique?'); |
690
|
|
|
// |
691
|
|
|
// return; |
692
|
|
|
// } |
693
|
|
|
if (preg_match( |
694
|
|
|
'#(ill\.|couv\.|in-[0-9]|in-fol|poche|broché|relié|{{unité|{{Dunité|[0-9]{2} ?cm|\|cm}}|vol\.|A4)#i', |
695
|
|
|
$value |
696
|
|
|
) > 0 |
697
|
|
|
) { |
698
|
|
|
$this->setParam('format livre', $value); |
699
|
|
|
$this->unsetParam('format'); |
700
|
|
|
$this->log('format:livre?'); |
701
|
|
|
$this->notCosmetic = true; |
702
|
|
|
} |
703
|
|
|
// Certainement 'format électronique'... |
704
|
|
|
} |
705
|
|
|
} |
706
|
|
|
|
707
|
|
|
/** |
708
|
|
|
* @return bool |
709
|
|
|
* |
710
|
|
|
* @throws Exception |
711
|
|
|
*/ |
712
|
|
|
public function checkMajorEdit(): bool |
713
|
|
|
{ |
714
|
|
|
// Correction paramètre |
715
|
|
|
if ($this->ouvrage->parametersErrorFromHydrate !== $this->original->parametersErrorFromHydrate) { |
716
|
|
|
return true; |
717
|
|
|
} |
718
|
|
|
// Complétion langue ? |
719
|
|
|
if (!empty($this->getParam('langue')) && empty($this->original->getParam('langue')) |
720
|
|
|
&& self::WIKI_LANGUAGE !== $this->getParam('langue') |
721
|
|
|
) { |
722
|
|
|
return true; |
723
|
|
|
} |
724
|
|
|
// TODO replace conditions ci-dessous par event flagMajor() |
725
|
|
|
// Retire le param/value 'langue' (pas major si conversion nom langue) |
726
|
|
|
$datOuvrage = $this->ouvrage->toArray(); |
727
|
|
|
$datOriginal = $this->original->toArray(); |
728
|
|
|
unset($datOriginal['langue'], $datOuvrage['langue']); |
729
|
|
|
|
730
|
|
|
// Modification données |
731
|
|
|
if ($datOriginal !== $datOuvrage) { |
732
|
|
|
return true; |
733
|
|
|
} |
734
|
|
|
|
735
|
|
|
return false; |
736
|
|
|
} |
737
|
|
|
|
738
|
|
|
/** |
739
|
|
|
* @return array |
740
|
|
|
*/ |
741
|
|
|
public function getLog(): array |
742
|
|
|
{ |
743
|
|
|
return $this->log; |
744
|
|
|
} |
745
|
|
|
|
746
|
|
|
/** |
747
|
|
|
* @return OuvrageTemplate |
748
|
|
|
*/ |
749
|
|
|
public function getOuvrage(): OuvrageTemplate |
750
|
|
|
{ |
751
|
|
|
return $this->ouvrage; |
752
|
|
|
} |
753
|
|
|
|
754
|
|
|
/** |
755
|
|
|
* todo : vérif lien rouge |
756
|
|
|
* todo 'lien éditeur' affiché 1x par page |
757
|
|
|
* opti : Suppression lien éditeur si c'est l'article de l'éditeur. |
758
|
|
|
* |
759
|
|
|
* @throws Exception |
760
|
|
|
*/ |
761
|
|
|
private function processEditeur() |
762
|
|
|
{ |
763
|
|
|
$editeur = $this->getParam('éditeur'); |
764
|
|
|
if (empty($editeur)) { |
765
|
|
|
return; |
766
|
|
|
} |
767
|
|
|
|
768
|
|
|
// FIX bug "GEO Art ([[Prisma Media]]) ; [[Le Monde]]" |
769
|
|
|
if (preg_match('#\[.*\[.*\[#', $editeur) > 0) { |
770
|
|
|
return; |
771
|
|
|
} |
772
|
|
|
// FIX bug "[[Fu|Bar]] bla" => [[Fu|Bar bla]] |
773
|
|
|
if (preg_match('#(.+\[\[|\]\].+)#', $editeur) > 0) { |
774
|
|
|
return; |
775
|
|
|
} |
776
|
|
|
|
777
|
|
|
// [[éditeur]] |
778
|
|
|
if (preg_match('#\[\[([^|]+)]]#', $editeur, $matches) > 0) { |
779
|
|
|
$editeurUrl = $matches[1]; |
780
|
|
|
} |
781
|
|
|
// [[bla|éditeur]] |
782
|
|
|
if (preg_match('#\[\[([^]|]+)\|.+]]#', $editeur, $matches) > 0) { |
783
|
|
|
$editeurUrl = $matches[1]; |
784
|
|
|
} |
785
|
|
|
|
786
|
|
|
// abréviations communes |
787
|
|
|
// ['éd. de ', 'éd. du ', 'éd.', 'ed.', 'Éd. de ', 'Éd.', 'édit.', 'Édit.', '(éd.)', '(ed.)', 'Ltd.'] |
788
|
|
|
|
789
|
|
|
$editeurStr = WikiTextUtil::unWikify($editeur); |
790
|
|
|
// On garde minuscule sur éditeur, pour nuance Éditeur/éditeur permettant de supprimer "éditeur" |
791
|
|
|
// ex: "éditions Milan" => "Milan" |
792
|
|
|
|
793
|
|
|
// $editeurStr = TextUtil::mb_ucfirst($editeurStr); |
794
|
|
|
|
795
|
|
|
// Déconseillé : 'lien éditeur' (obsolete 2019) |
796
|
|
|
if (!empty($this->getParam('lien éditeur'))) { |
797
|
|
|
if (empty($editeurUrl)) { |
798
|
|
|
$editeurUrl = $this->getParam('lien éditeur'); |
799
|
|
|
} |
800
|
|
|
$this->unsetParam('lien éditeur'); |
801
|
|
|
} |
802
|
|
|
|
803
|
|
|
// if (isset($editeurUrl)) { |
804
|
|
|
// $editeurUrl = TextUtil::mb_ucfirst($editeurUrl); |
805
|
|
|
// } |
806
|
|
|
$newEditeur = $editeurStr; |
807
|
|
|
if (isset($editeurUrl) && $editeurUrl !== $editeurStr) { |
808
|
|
|
$newEditeur = '[['.$editeurUrl.'|'.$editeurStr.']]'; |
809
|
|
|
} |
810
|
|
|
if (isset($editeurUrl) && $editeurUrl === $editeurStr) { |
811
|
|
|
$newEditeur = '[['.$editeurStr.']]'; |
812
|
|
|
} |
813
|
|
|
|
814
|
|
|
if ($newEditeur !== $editeur) { |
815
|
|
|
$this->setParam('éditeur', $newEditeur); |
816
|
|
|
$this->log('±éditeur'); |
817
|
|
|
$this->notCosmetic = true; |
818
|
|
|
} |
819
|
|
|
} |
820
|
|
|
} |
821
|
|
|
|
If the size of the collection does not change during the iteration, it is generally a good practice to compute it beforehand, and not on each iteration: