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\Enums\Language; |
13
|
|
|
use App\Domain\Models\Wiki\GoogleLivresTemplate; |
14
|
|
|
use App\Domain\Models\Wiki\OuvrageTemplate; |
15
|
|
|
use App\Domain\Utils\NumberUtil; |
16
|
|
|
use App\Domain\Utils\TextUtil; |
17
|
|
|
use App\Domain\Utils\WikiTextUtil; |
18
|
|
|
use App\Infrastructure\FileManager; |
19
|
|
|
use Exception; |
20
|
|
|
use Throwable; |
21
|
|
|
|
22
|
|
|
use function mb_strlen; |
23
|
|
|
|
24
|
|
|
/** |
25
|
|
|
* Legacy. |
26
|
|
|
* TODO move methods to OuvrageClean setters |
27
|
|
|
* TODO AbstractProcess |
28
|
|
|
* TODO observer/event (log, MajorEdition) |
29
|
|
|
* Class OuvrageProcess. |
30
|
|
|
*/ |
31
|
|
|
class OuvrageOptimize |
32
|
|
|
{ |
33
|
|
|
const CONVERT_GOOGLEBOOK_TEMPLATE = false; // change OuvrageOptimizeTest !! |
34
|
|
|
|
35
|
|
|
const WIKI_LANGUAGE = 'fr'; |
36
|
|
|
|
37
|
|
|
protected $original; |
38
|
|
|
|
39
|
|
|
private $wikiPageTitle; |
40
|
|
|
|
41
|
|
|
private $log = []; |
42
|
|
|
|
43
|
|
|
public $notCosmetic = false; |
44
|
|
|
|
45
|
|
|
public $major = false; |
46
|
|
|
|
47
|
|
|
private $ouvrage; |
48
|
|
|
|
49
|
|
|
private $currentTask; |
50
|
|
|
|
51
|
|
|
// todo inject TextUtil + ArticleVersion ou WikiRef |
52
|
|
|
public function __construct(OuvrageTemplate $ouvrage, $wikiPageTitle = null) |
53
|
|
|
{ |
54
|
|
|
$this->original = $ouvrage; |
55
|
|
|
$this->ouvrage = clone $ouvrage; |
56
|
|
|
$this->wikiPageTitle = ($wikiPageTitle) ?? null; |
57
|
|
|
} |
58
|
|
|
|
59
|
|
|
public function doTasks(): self |
60
|
|
|
{ |
61
|
|
|
$this->cleanAndPredictErrorParameters(); |
62
|
|
|
|
63
|
|
|
$this->processAuthors(); |
64
|
|
|
|
65
|
|
|
$this->processTitle(); |
66
|
|
|
$this->processEditeur(); |
67
|
|
|
$this->processDates(); |
68
|
|
|
$this->externalTemplates(); |
69
|
|
|
$this->predictFormatByPattern(); |
70
|
|
|
|
71
|
|
|
// $this->convertRoman('tome'); |
72
|
|
|
// $this->convertRoman('volume'); |
73
|
|
|
|
74
|
|
|
$this->processIsbn(); |
75
|
|
|
$this->processBnf(); |
76
|
|
|
|
77
|
|
|
$this->processLang(); |
78
|
|
|
$this->processLocation(); // 'lieu' |
79
|
|
|
|
80
|
|
|
$this->GoogleBookURL('lire en ligne'); |
81
|
|
|
$this->GoogleBookURL('présentation en ligne'); |
82
|
|
|
|
83
|
|
|
return $this; |
84
|
|
|
} |
85
|
|
|
|
86
|
|
|
/** |
87
|
|
|
* Todo: injection dep. |
88
|
|
|
* Todo : "[s. l.]" sans lieu "s.l.n.d." sans lieu ni date. |
89
|
|
|
* |
90
|
|
|
* @throws Exception |
91
|
|
|
*/ |
92
|
|
|
private function processLocation() |
93
|
|
|
{ |
94
|
|
|
$location = $this->getParam('lieu'); |
95
|
|
|
if (empty($location)) { |
96
|
|
|
return; |
97
|
|
|
} |
98
|
|
|
|
99
|
|
|
// typo and unwikify |
100
|
|
|
$memo = $location; |
101
|
|
|
$location = WikiTextUtil::unWikify($location); |
102
|
|
|
$location = TextUtil::mb_ucfirst($location); |
103
|
|
|
if ($memo !== $location) { |
104
|
|
|
$this->setParam('lieu', $location); |
105
|
|
|
$this->log('±lieu'); |
106
|
|
|
$this->notCosmetic = true; |
107
|
|
|
} |
108
|
|
|
|
109
|
|
|
// french translation : "London"->"Londres" |
110
|
|
|
$manager = new FileManager(); |
111
|
|
|
$row = $manager->findCSVline(__DIR__.'/resources/traduction_ville.csv', $location); |
112
|
|
|
if (!empty($row) && !empty($row[1])) { |
113
|
|
|
$this->setParam('lieu', $row[1]); |
114
|
|
|
$this->log('lieu francisé'); |
115
|
|
|
$this->notCosmetic = true; |
116
|
|
|
} |
117
|
|
|
} |
118
|
|
|
|
119
|
|
|
private function processBnf() |
120
|
|
|
{ |
121
|
|
|
$bnf = $this->getParam('bnf'); |
122
|
|
|
if (!$bnf) { |
123
|
|
|
return; |
124
|
|
|
} |
125
|
|
|
$bnf = str_ireplace('FRBNF', '', $bnf); |
126
|
|
|
$this->setParam('bnf', $bnf); |
127
|
|
|
} |
128
|
|
|
|
129
|
|
|
private function convertRoman(string $param): void |
|
|
|
|
130
|
|
|
{ |
131
|
|
|
$value = $this->getParam($param); |
132
|
|
|
// note : strval() condition because intval('4c') = 4 |
133
|
|
|
if ($value && intval($value) > 0 && intval($value) <= 10 && strval(intval($value)) === $value) { |
134
|
|
|
$number = abs(intval($value)); |
135
|
|
|
$roman = NumberUtil::arab2roman($number); |
|
|
|
|
136
|
|
|
// if ($number > 10) { |
137
|
|
|
// $roman = '{{'.$roman.'}}'; |
138
|
|
|
// } |
139
|
|
|
$this->setParam($param, $roman); |
140
|
|
|
$this->log('romain'); |
141
|
|
|
$this->notCosmetic = true; |
142
|
|
|
} |
143
|
|
|
} |
144
|
|
|
|
145
|
|
|
/** |
146
|
|
|
* @throws Exception |
147
|
|
|
*/ |
148
|
|
|
private function processAuthors() |
149
|
|
|
{ |
150
|
|
|
$this->distinguishAuthors(); |
151
|
|
|
//$this->fusionFirstNameAndName(); // desactived : no consensus |
152
|
|
|
} |
153
|
|
|
|
154
|
|
|
/** |
155
|
|
|
* Detect and correct multiple authors in same parameter. |
156
|
|
|
* Like "auteurs=J. M. Waller, M. Bigger, R. J. Hillocks". |
157
|
|
|
* |
158
|
|
|
* @throws Exception |
159
|
|
|
*/ |
160
|
|
|
private function distinguishAuthors() |
161
|
|
|
{ |
162
|
|
|
// merge params of author 1 |
163
|
|
|
$auteur1 = $this->getParam('auteur') ?? ''; |
164
|
|
|
$auteur1 .= $this->getParam('auteurs') ?? ''; |
165
|
|
|
$auteur1 .= $this->getParam('prénom1') ?? ''; |
166
|
|
|
$auteur1 .= ' '.$this->getParam('nom1') ?? ''; |
167
|
|
|
$auteur1 = trim($auteur1); |
168
|
|
|
// of authors 2 |
169
|
|
|
$auteur2 = $this->getParam('auteur2') ?? ''; |
170
|
|
|
$auteur2 .= $this->getParam('prénom2') ?? ''; |
171
|
|
|
$auteur2 .= ' '.$this->getParam('nom2') ?? ''; |
172
|
|
|
$auteur2 = trim($auteur2); |
173
|
|
|
|
174
|
|
|
// skip if wikilink in author |
175
|
|
|
if (empty($auteur1) || WikiTextUtil::isWikify($auteur1)) { |
176
|
|
|
return; |
177
|
|
|
} |
178
|
|
|
|
179
|
|
|
$machine = new PredictAuthors(); |
180
|
|
|
$res = $machine->predictAuthorNames($auteur1); |
181
|
|
|
|
182
|
|
|
if (1 === count($res)) { |
183
|
|
|
// auteurs->auteur? |
184
|
|
|
return; |
185
|
|
|
} |
186
|
|
|
// Many authors... and empty "auteur2" |
187
|
|
|
if (count($res) >= 2 && empty($auteur2)) { |
188
|
|
|
// delete author-params |
189
|
|
|
array_map( |
190
|
|
|
function ($param) { |
191
|
|
|
$this->unsetParam($param); |
192
|
|
|
}, |
193
|
|
|
['auteur', 'auteurs', 'prénom1', 'nom1'] |
194
|
|
|
); |
195
|
|
|
// iterate and edit new values |
196
|
|
|
for ($i = 0; $i < count($res); ++$i) { |
|
|
|
|
197
|
|
|
$this->setParam(sprintf('auteur%s', $i + 1), $res[$i]); |
198
|
|
|
} |
199
|
|
|
$this->log('distinction auteurs'); |
200
|
|
|
$this->major = true; |
201
|
|
|
$this->notCosmetic = true; |
202
|
|
|
} |
203
|
|
|
} |
204
|
|
|
|
205
|
|
|
/** |
206
|
|
|
* todo: move/implement. |
207
|
|
|
* |
208
|
|
|
* @throws Exception |
209
|
|
|
*/ |
210
|
|
|
private function processLang() |
211
|
|
|
{ |
212
|
|
|
$lang = $this->getParam('langue') ?? null; |
213
|
|
|
|
214
|
|
|
if ($lang) { |
215
|
|
|
$lang2 = Language::all2wiki($lang); |
216
|
|
|
|
217
|
|
|
if ($lang2 && $lang !== $lang2) { |
218
|
|
|
$this->setParam('langue', $lang2); |
219
|
|
|
if (self::WIKI_LANGUAGE !== $lang2) { |
220
|
|
|
$this->log('±langue'); |
221
|
|
|
} |
222
|
|
|
} |
223
|
|
|
} |
224
|
|
|
} |
225
|
|
|
|
226
|
|
|
/** |
227
|
|
|
* Validate or correct ISBN. |
228
|
|
|
* |
229
|
|
|
* @throws Exception |
230
|
|
|
*/ |
231
|
|
|
private function processIsbn() |
232
|
|
|
{ |
233
|
|
|
$isbn = $this->getParam('isbn') ?? ''; |
234
|
|
|
if (empty($isbn)) { |
235
|
|
|
return; |
236
|
|
|
} |
237
|
|
|
|
238
|
|
|
// ISBN-13 à partir de 2007 |
239
|
|
|
$year = $this->findBookYear(); |
240
|
|
|
if ($year && $year < 2007 && 10 === strlen($this->stripIsbn($isbn))) { |
|
|
|
|
241
|
|
|
// juste mise en forme ISBN-10 pour 'isbn' |
242
|
|
|
try { |
243
|
|
|
$isbnMachine = new IsbnFacade($isbn); |
244
|
|
|
$isbnMachine->validate(); |
245
|
|
|
$isbn10pretty = $isbnMachine->format('ISBN-10'); |
246
|
|
|
if ($isbn10pretty !== $isbn) { |
247
|
|
|
$this->setParam('isbn', $isbn10pretty); |
248
|
|
|
$this->log('ISBN10'); |
249
|
|
|
// $this->notCosmetic = true; |
250
|
|
|
} |
251
|
|
|
} catch (\Throwable $e) { |
252
|
|
|
// ISBN not validated |
253
|
|
|
$this->setParam( |
254
|
|
|
'isbn invalide', |
255
|
|
|
sprintf('%s %s', $isbn, $e->getMessage() ?? '') |
256
|
|
|
); |
257
|
|
|
$this->log(sprintf('ISBN invalide: %s', $e->getMessage())); |
258
|
|
|
$this->notCosmetic = true; |
259
|
|
|
} |
260
|
|
|
|
261
|
|
|
return; |
262
|
|
|
} |
263
|
|
|
|
264
|
|
|
try { |
265
|
|
|
$isbnMachine = new IsbnFacade($isbn); |
266
|
|
|
$isbnMachine->validate(); |
267
|
|
|
$isbn13 = $isbnMachine->format('ISBN-13'); |
268
|
|
|
} catch (Throwable $e) { |
269
|
|
|
// ISBN not validated |
270
|
|
|
// TODO : bot ISBN invalide (queue, message PD...) |
271
|
|
|
$this->setParam( |
272
|
|
|
'isbn invalide', |
273
|
|
|
sprintf('%s %s', $isbn, $e->getMessage() ?? '') |
274
|
|
|
); |
275
|
|
|
$this->log(sprintf('ISBN invalide: %s', $e->getMessage())); |
276
|
|
|
$this->notCosmetic = true; |
277
|
|
|
|
278
|
|
|
// TODO log file ISBNinvalide |
279
|
|
|
return; |
280
|
|
|
} |
281
|
|
|
|
282
|
|
|
// Si $isbn13 et 'isbn2' correspond à ISBN-13 => suppression |
283
|
|
|
if (isset($isbn13) |
284
|
|
|
&& $this->getParam('isbn2') |
285
|
|
|
&& $this->stripIsbn($this->getParam('isbn2')) === $this->stripIsbn($isbn13) |
|
|
|
|
286
|
|
|
) { |
287
|
|
|
$this->unsetParam('isbn2'); |
288
|
|
|
} |
289
|
|
|
|
290
|
|
|
|
291
|
|
|
// ISBN-10 ? |
292
|
|
|
$stripIsbn = $this->stripIsbn($isbn); |
293
|
|
|
if (10 === mb_strlen($stripIsbn)) { |
294
|
|
|
// ajout des tirets |
295
|
|
|
$isbn10pretty = $isbn; |
|
|
|
|
296
|
|
|
try { |
297
|
|
|
$isbn10pretty = $isbnMachine->format('ISBN-10'); |
298
|
|
|
if ($isbn10pretty !== $isbn) { |
299
|
|
|
// $this->notCosmetic = true; |
300
|
|
|
} |
301
|
|
|
} catch (\Throwable $e) { |
302
|
|
|
unset($e); |
303
|
|
|
} |
304
|
|
|
|
305
|
|
|
// archivage ISBN-10 dans 'isbn2' |
306
|
|
|
if (!$this->getParam('isbn2')) { |
307
|
|
|
$this->setParam('isbn2', $isbn10pretty); |
308
|
|
|
} |
309
|
|
|
// sinon dans 'isbn3' |
310
|
|
|
if (!empty($this->getParam('isbn2')) |
311
|
|
|
&& $this->stripIsbn($this->getParam('isbn2')) !== $stripIsbn |
312
|
|
|
&& empty($this->getParam('isbn3')) |
313
|
|
|
) { |
314
|
|
|
$this->setParam('isbn3', $isbn10pretty); |
315
|
|
|
} |
316
|
|
|
// delete 'isbn10' (en attendant modification modèle) |
317
|
|
|
if (!empty($this->getParam('isbn10')) && $this->stripIsbn($this->getParam('isbn10')) === $stripIsbn) { |
318
|
|
|
$this->unsetParam('isbn10'); |
319
|
|
|
} |
320
|
|
|
} |
321
|
|
|
|
322
|
|
|
// ISBN correction |
323
|
|
|
if ($isbn13 !== $isbn) { |
324
|
|
|
$this->setParam('isbn', $isbn13); |
325
|
|
|
$this->log('ISBN'); |
326
|
|
|
// $this->notCosmetic = true; |
327
|
|
|
} |
328
|
|
|
} |
329
|
|
|
|
330
|
|
|
/** |
331
|
|
|
* Find year of book publication. |
332
|
|
|
* |
333
|
|
|
* @return int|null |
334
|
|
|
* @throws Exception |
335
|
|
|
*/ |
336
|
|
|
private function findBookYear(): ?int |
337
|
|
|
{ |
338
|
|
|
$annee = $this->getParam('année'); |
339
|
|
|
if (!empty($annee) && is_numeric($annee)) { |
340
|
|
|
return intval($annee); |
341
|
|
|
} |
342
|
|
|
$date = $this->getParam('date'); |
343
|
|
|
if ($date && preg_match('#[^0-9]?([12][0-9][0-9][0-9])[^0-9]?#', $date, $matches) > 0) { |
344
|
|
|
return intval($matches[1]); |
345
|
|
|
} |
346
|
|
|
|
347
|
|
|
return null; |
348
|
|
|
} |
349
|
|
|
|
350
|
|
|
private function stripIsbn(string $isbn): string |
351
|
|
|
{ |
352
|
|
|
return trim(preg_replace('#[^0-9Xx]#', '', $isbn)); |
353
|
|
|
} |
354
|
|
|
|
355
|
|
|
private function processTitle() |
356
|
|
|
{ |
357
|
|
|
$this->currentTask = 'titres'; |
358
|
|
|
|
359
|
|
|
$oldtitre = $this->getParam('titre'); |
360
|
|
|
$this->langInTitle(); |
361
|
|
|
$this->deWikifyExternalLink('titre'); |
362
|
|
|
$this->upperCaseFirstLetter('titre'); |
363
|
|
|
$this->typoDeuxPoints('titre'); |
364
|
|
|
|
365
|
|
|
$this->extractSubTitle(); |
366
|
|
|
|
367
|
|
|
// 20-11-2019 : Retiré majuscule à sous-titre |
368
|
|
|
|
369
|
|
|
if ($this->getParam('titre') !== $oldtitre) { |
370
|
|
|
$this->log('±titre'); |
371
|
|
|
} |
372
|
|
|
|
373
|
|
|
$this->valideNumeroChapitre(); |
374
|
|
|
$this->deWikifyExternalLink('titre chapitre'); |
375
|
|
|
$this->upperCaseFirstLetter('titre chapitre'); |
376
|
|
|
} |
377
|
|
|
|
378
|
|
|
private function detectColon($param): bool |
379
|
|
|
{ |
380
|
|
|
// > 0 don't count a starting colon ":bla" |
381
|
|
|
if (!empty($this->getParam($param)) && mb_strrpos($this->getParam('titre'), ':') > 0) { |
382
|
|
|
return true; |
383
|
|
|
} |
384
|
|
|
|
385
|
|
|
return false; |
386
|
|
|
} |
387
|
|
|
|
388
|
|
|
private function extractSubTitle(): void |
389
|
|
|
{ |
390
|
|
|
// FIXED bug [[fu:bar]] |
391
|
|
|
if (!$this->getParam('titre') || WikiTextUtil::isWikify($this->getParam('titre'))) { |
|
|
|
|
392
|
|
|
return; |
393
|
|
|
} |
394
|
|
|
|
395
|
|
|
if (!$this->detectColon('titre')) { |
396
|
|
|
return; |
397
|
|
|
} |
398
|
|
|
// Que faire si déjà un sous-titre ? |
399
|
|
|
if (!empty($this->getParam('sous-titre'))) { |
400
|
|
|
return; |
401
|
|
|
} |
402
|
|
|
|
403
|
|
|
// titre>5 and sous-titre>5 and sous-titre<40 |
404
|
|
|
if (preg_match('#^(?<titre>[^:]{5,}):(?<st>.{5,40})$#', $this->getParam('titre'), $matches) > 0) { |
405
|
|
|
$this->setParam('titre', trim($matches['titre'])); |
406
|
|
|
$this->setParam('sous-titre', trim($matches['st'])); |
407
|
|
|
$this->log('>sous-titre'); |
408
|
|
|
} |
409
|
|
|
} |
410
|
|
|
|
411
|
|
|
/** |
412
|
|
|
* Normalize a Google Book links. |
413
|
|
|
* Clean the useless URL parameters or transform into wiki-template. |
414
|
|
|
* |
415
|
|
|
* @param $param |
416
|
|
|
* |
417
|
|
|
* @throws Exception |
418
|
|
|
*/ |
419
|
|
|
private function googleBookUrl(string $param): void |
420
|
|
|
{ |
421
|
|
|
$url = $this->getParam($param); |
422
|
|
|
if (empty($url) |
423
|
|
|
|| !GoogleLivresTemplate::isGoogleBookURL($url) |
424
|
|
|
) { |
425
|
|
|
return; |
426
|
|
|
} |
427
|
|
|
|
428
|
|
|
if (self::CONVERT_GOOGLEBOOK_TEMPLATE) { |
429
|
|
|
$template = GoogleLivresTemplate::createFromURL($url); |
430
|
|
|
if ($template) { |
|
|
|
|
431
|
|
|
$this->setParam($param, $template->serialize()); |
432
|
|
|
$this->log('{Google}'); |
433
|
|
|
$this->notCosmetic = true; |
434
|
|
|
|
435
|
|
|
return; |
436
|
|
|
} |
437
|
|
|
} |
438
|
|
|
|
439
|
|
|
$goo = GoogleLivresTemplate::simplifyGoogleUrl($url); |
440
|
|
|
if (!empty($goo) && $goo !== $url ) { |
441
|
|
|
$this->setParam($param, $goo); |
442
|
|
|
$this->log('Google'); |
443
|
|
|
$this->notCosmetic = true; |
444
|
|
|
} |
445
|
|
|
} |
446
|
|
|
|
447
|
|
|
/** |
448
|
|
|
* - {{lang|...}} dans titre => langue=... puis titre nettoyé |
449
|
|
|
* langue=L’utilisation de ce paramètre permet aussi aux synthétiseurs vocaux de reconnaître la langue du titre de |
450
|
|
|
* l’ouvrage. |
451
|
|
|
* Il est possible d'afficher plusieurs langues, en saisissant le nom séparé par des espaces ou des virgules. La première langue doit être celle du titre. |
452
|
|
|
* |
453
|
|
|
* @throws Exception |
454
|
|
|
*/ |
455
|
|
|
private function langInTitle() |
456
|
|
|
{ |
457
|
|
|
if (preg_match( |
458
|
|
|
'#^{{ ?(?:lang|langue) ?\| ?([a-z-]{2,5}) ?\| ?(?:texte=)?([^{}=]+)(?:\|dir=rtl)?}}$#i', |
459
|
|
|
$this->getParam('titre'), |
460
|
|
|
$matches |
461
|
|
|
) > 0 |
462
|
|
|
) { |
463
|
|
|
$lang = trim($matches[1]); |
464
|
|
|
$newtitre = str_replace($matches[0], trim($matches[2]), $this->getParam('titre')); |
465
|
|
|
$this->setParam('titre', $newtitre); |
466
|
|
|
$this->log('°titre'); |
467
|
|
|
if (self::WIKI_LANGUAGE !== $lang && empty($this->getParam('langue'))) { |
468
|
|
|
$this->setParam('langue', $lang); |
469
|
|
|
$this->log('+langue='.$lang); |
470
|
|
|
} |
471
|
|
|
} |
472
|
|
|
} |
473
|
|
|
|
474
|
|
|
private function processDates() |
475
|
|
|
{ |
476
|
|
|
// dewikification |
477
|
|
|
$params = ['date', 'année', 'mois', 'jour']; |
478
|
|
|
foreach ($params as $param) { |
479
|
|
|
if (WikiTextUtil::isWikify(' '.$this->getParam($param))) { |
480
|
|
|
$this->setParam($param, WikiTextUtil::unWikify($this->getParam($param))); |
|
|
|
|
481
|
|
|
} |
482
|
|
|
} |
483
|
|
|
|
484
|
|
|
try { |
485
|
|
|
$this->moveDate2Year(); |
486
|
|
|
} catch (Exception $e) { |
487
|
|
|
dump($e); |
488
|
|
|
} |
489
|
|
|
} |
490
|
|
|
|
491
|
|
|
/** |
492
|
|
|
* todo: move to AbstractWikiTemplate ? |
493
|
|
|
* Correction des parametres rejetés à l'hydratation données. |
494
|
|
|
* |
495
|
|
|
* @throws Exception |
496
|
|
|
*/ |
497
|
|
|
private function cleanAndPredictErrorParameters() |
498
|
|
|
{ |
499
|
|
|
if (empty($this->ouvrage->parametersErrorFromHydrate)) { |
500
|
|
|
return; |
501
|
|
|
} |
502
|
|
|
$allParamsAndAlias = $this->ouvrage->getParamsAndAlias(); |
503
|
|
|
|
504
|
|
|
foreach ($this->ouvrage->parametersErrorFromHydrate as $name => $value) { |
505
|
|
|
if (!is_string($name)) { |
506
|
|
|
// example : 1 => "ouvrage collectif" from |ouvrage collectif| |
507
|
|
|
continue; |
508
|
|
|
} |
509
|
|
|
|
510
|
|
|
// delete error parameter if no value |
511
|
|
|
if (empty($value)) { |
512
|
|
|
unset($this->ouvrage->parametersErrorFromHydrate[$name]); |
513
|
|
|
|
514
|
|
|
continue; |
515
|
|
|
} |
516
|
|
|
|
517
|
|
|
$maxDistance = 1; |
518
|
|
|
if (mb_strlen($name) >= 4) { |
519
|
|
|
$maxDistance = 2; |
520
|
|
|
} |
521
|
|
|
if (mb_strlen($name) >= 8) { |
522
|
|
|
$maxDistance = 3; |
523
|
|
|
} |
524
|
|
|
|
525
|
|
|
$predName = TextUtil::predictCorrectParam($name, $allParamsAndAlias, $maxDistance); |
526
|
|
|
if ($predName && mb_strlen($name) >= 5) { |
527
|
|
|
if (empty($this->getParam($predName))) { |
528
|
|
|
$predName = $this->ouvrage->getAliasParam($predName); |
529
|
|
|
$this->setParam($predName, $value); |
530
|
|
|
$this->log(sprintf('%s⇒%s ?', $name, $predName)); |
531
|
|
|
$this->notCosmetic = true; |
532
|
|
|
unset($this->ouvrage->parametersErrorFromHydrate[$name]); |
533
|
|
|
} |
534
|
|
|
} |
535
|
|
|
} |
536
|
|
|
} |
537
|
|
|
|
538
|
|
|
/** |
539
|
|
|
* TODO : return "" instead of null ? |
540
|
|
|
* |
541
|
|
|
* @param $name |
542
|
|
|
* |
543
|
|
|
* @return string|null |
544
|
|
|
* @throws Exception |
545
|
|
|
*/ |
546
|
|
|
private function getParam(string $name): ?string |
547
|
|
|
{ |
548
|
|
|
return $this->ouvrage->getParam($name); |
549
|
|
|
} |
550
|
|
|
|
551
|
|
|
private function setParam($name, $value) |
552
|
|
|
{ |
553
|
|
|
// todo : overwrite setParam() ? |
554
|
|
|
if (!empty($value) || $this->ouvrage->getParam($name)) { |
555
|
|
|
$this->ouvrage->setParam($name, $value); |
556
|
|
|
} |
557
|
|
|
} |
558
|
|
|
|
559
|
|
|
private function log(string $string): void |
560
|
|
|
{ |
561
|
|
|
if (!empty($string)) { |
562
|
|
|
$this->log[] = trim($string); |
563
|
|
|
} |
564
|
|
|
} |
565
|
|
|
|
566
|
|
|
/** |
567
|
|
|
* Bool ? |
568
|
|
|
* déwikification du titre : consensus Bistro 27 août 2011 |
569
|
|
|
* idem 'titre chapitre'. |
570
|
|
|
* |
571
|
|
|
* @param string $param |
572
|
|
|
* |
573
|
|
|
* @throws Exception |
574
|
|
|
*/ |
575
|
|
|
private function deWikifyExternalLink(string $param): void |
576
|
|
|
{ |
577
|
|
|
if (empty($this->getParam($param))) { |
578
|
|
|
return; |
579
|
|
|
} |
580
|
|
|
if (preg_match('#^\[(http[^ \]]+) ([^]]+)]#i', $this->getParam($param), $matches) > 0) { |
581
|
|
|
$this->setParam($param, str_replace($matches[0], $matches[2], $this->getParam($param))); |
582
|
|
|
$this->log('±'.$param); |
583
|
|
|
|
584
|
|
|
if (in_array($param, ['titre', 'titre chapitre'])) { |
585
|
|
|
if (empty($this->getParam('lire en ligne'))) { |
586
|
|
|
$this->setParam('lire en ligne', $matches[1]); |
587
|
|
|
$this->log('+lire en ligne'); |
588
|
|
|
|
589
|
|
|
return; |
590
|
|
|
} |
591
|
|
|
$this->log('autre lien externe: '.$matches[1]); |
592
|
|
|
} |
593
|
|
|
} |
594
|
|
|
} |
595
|
|
|
|
596
|
|
|
private function upperCaseFirstLetter($param) |
597
|
|
|
{ |
598
|
|
|
if (empty($this->getParam($param))) { |
599
|
|
|
return; |
600
|
|
|
} |
601
|
|
|
$newValue = TextUtil::mb_ucfirst(trim($this->getParam($param))); |
602
|
|
|
$this->setParam($param, $newValue); |
603
|
|
|
} |
604
|
|
|
|
605
|
|
|
/** |
606
|
|
|
* Typo internationale 'titre : sous-titre'. |
607
|
|
|
* Fix fantasy typo of subtitle with '. ' or ' - '. |
608
|
|
|
* International Standard Bibliographic Description : |
609
|
|
|
* https://fr.wikipedia.org/wiki/Wikip%C3%A9dia:Le_Bistro/13_janvier_2016#Modif_du_mod%C3%A8le:Ouvrage. |
610
|
|
|
* |
611
|
|
|
* @param $param |
612
|
|
|
* |
613
|
|
|
* @throws Exception |
614
|
|
|
*/ |
615
|
|
|
private function typoDeuxPoints($param) |
616
|
|
|
{ |
617
|
|
|
$origin = $this->getParam($param) ?? ''; |
618
|
|
|
if (empty($origin)) { |
619
|
|
|
return; |
620
|
|
|
} |
621
|
|
|
// FIXED bug [[fu:bar]] |
622
|
|
|
if (WikiTextUtil::isWikify($origin)) { |
623
|
|
|
return; |
624
|
|
|
} |
625
|
|
|
|
626
|
|
|
$origin = TextUtil::replaceNonBreakingSpaces($origin); |
627
|
|
|
|
628
|
|
|
$strTitle = $origin; |
629
|
|
|
|
630
|
|
|
// CORRECTING TYPO FANTASY OF SUBTITLE |
631
|
|
|
|
632
|
|
|
// Replace first '.' by ':' if no ': ' and no numbers around (as PHP 7.3) |
633
|
|
|
// exlude pattern "blabla... blabla" |
634
|
|
|
// TODO: statistics |
635
|
|
|
|
636
|
|
|
// Replace ' - ' or ' / ' (spaced!) by ' : ' if no ':' and no numbers after (as PHP 7.3 or 1939-1945) |
637
|
|
|
if (!mb_strpos(':', $strTitle) && preg_match('#.{6,} ?[-/] ?[^0-9)]{6,}#', $strTitle) > 0) { |
638
|
|
|
$strTitle = preg_replace('#(.{6,}) [-/] ([^0-9)]{6,})#', '$1 : $2', $strTitle); |
639
|
|
|
} |
640
|
|
|
|
641
|
|
|
// international typo style " : " (first occurrence) |
642
|
|
|
$strTitle = preg_replace('#[ ]*:[ ]*#', ' : ', $strTitle); |
643
|
|
|
|
644
|
|
|
if ($strTitle !== $origin) { |
645
|
|
|
$this->setParam($param, $strTitle); |
646
|
|
|
$this->log(sprintf(':%s', $param)); |
647
|
|
|
} |
648
|
|
|
} |
649
|
|
|
|
650
|
|
|
private function valideNumeroChapitre() |
651
|
|
|
{ |
652
|
|
|
$value = $this->getParam('numéro chapitre'); |
653
|
|
|
if (empty($value)) { |
654
|
|
|
return; |
655
|
|
|
} |
656
|
|
|
// "12" ou "VI", {{II}}, II:3 |
657
|
|
|
if (preg_match('#^[0-9IVXL\-.:{}]+$#i', $value) > 0) { |
658
|
|
|
return; |
659
|
|
|
} |
660
|
|
|
// déplace vers "titre chapitre" ? |
661
|
|
|
if (!$this->getParam('titre chapitre')) { |
662
|
|
|
$this->unsetParam('numéro chapitre'); |
663
|
|
|
$this->setParam('titre chapitre', $value); |
664
|
|
|
} |
665
|
|
|
$this->log('≠numéro chapitre'); |
666
|
|
|
} |
667
|
|
|
|
668
|
|
|
private function unsetParam($name) |
669
|
|
|
{ |
670
|
|
|
$this->ouvrage->unsetParam($name); |
671
|
|
|
} |
672
|
|
|
|
673
|
|
|
/** |
674
|
|
|
* TODO move+refac |
675
|
|
|
* TODO PlumeTemplate CommentaireBiblioTemplate ExtraitTemplate |
676
|
|
|
* Probleme {{commentaire biblio}} <> {{commentaire biblio SRL}} |
677
|
|
|
* Generate supplementary templates from obsoletes params. |
678
|
|
|
* |
679
|
|
|
* @throws Exception |
680
|
|
|
*/ |
681
|
|
|
protected function externalTemplates() |
682
|
|
|
{ |
683
|
|
|
// "plume=bla" => {{plume}} |
684
|
|
|
// =oui selon doc, mais testé OK avec "non" |
685
|
|
|
// todo detect duplication ouvrage/plume dans externalTemplate ? |
686
|
|
|
if (!empty($this->getParam('plume'))) { |
687
|
|
|
$plumeValue = $this->getParam('plume'); |
688
|
|
|
$this->ouvrage->externalTemplates[] = (object)[ |
689
|
|
|
'template' => 'plume', |
690
|
|
|
'1' => $plumeValue, |
691
|
|
|
'raw' => '{{nobr|. {{plume}}}}', |
692
|
|
|
]; |
693
|
|
|
$this->unsetParam('plume'); |
694
|
|
|
$this->log('{plume}'); |
695
|
|
|
} |
696
|
|
|
|
697
|
|
|
// "extrait=bla" => {{citation bloc|bla}} |
698
|
|
|
if (!empty($this->getParam('extrait'))) { |
699
|
|
|
$extrait = $this->getParam('extrait'); |
700
|
|
|
// todo bug {{citation bloc}} si "=" ou "|" dans texte de citation |
701
|
|
|
// Legacy : use {{début citation}} ... {{fin citation}} |
702
|
|
|
if (preg_match('#[=|]#', $extrait) > 0) { |
703
|
|
|
$this->ouvrage->externalTemplates[] = (object)[ |
704
|
|
|
'template' => 'début citation', |
705
|
|
|
'1' => '', |
706
|
|
|
'raw' => '{{début citation}}'.$extrait.'{{fin citation}}', |
707
|
|
|
]; |
708
|
|
|
$this->log('{début citation}'); |
709
|
|
|
$this->notCosmetic = true; |
710
|
|
|
} else { |
711
|
|
|
// StdClass |
712
|
|
|
$this->ouvrage->externalTemplates[] = (object)[ |
713
|
|
|
'template' => 'citation bloc', |
714
|
|
|
'1' => $extrait, |
715
|
|
|
'raw' => '{{extrait|'.$extrait.'}}', |
716
|
|
|
]; |
717
|
|
|
$this->log('{extrait}'); |
718
|
|
|
$this->notCosmetic = true; |
719
|
|
|
} |
720
|
|
|
|
721
|
|
|
$this->unsetParam('extrait'); |
722
|
|
|
$this->notCosmetic = true; |
723
|
|
|
} |
724
|
|
|
|
725
|
|
|
// "commentaire=bla" => {{Commentaire biblio|1=bla}} |
726
|
|
|
if (!empty($this->getParam('commentaire'))) { |
727
|
|
|
$commentaire = $this->getParam('commentaire'); |
728
|
|
|
$this->ouvrage->externalTemplates[] = (object)[ |
729
|
|
|
'template' => 'commentaire biblio', |
730
|
|
|
'1' => $commentaire, |
731
|
|
|
'raw' => '{{commentaire biblio|'.$commentaire.'}}', |
732
|
|
|
]; |
733
|
|
|
$this->unsetParam('commentaire'); |
734
|
|
|
$this->log('{commentaire}'); |
735
|
|
|
$this->notCosmetic = true; |
736
|
|
|
} |
737
|
|
|
} |
738
|
|
|
|
739
|
|
|
// ---------------------- |
740
|
|
|
// ---------------------- |
741
|
|
|
// ---------------------- |
742
|
|
|
|
743
|
|
|
/** |
744
|
|
|
* Date->année (nécessaire pour OuvrageComplete). |
745
|
|
|
* |
746
|
|
|
* @throws Exception |
747
|
|
|
*/ |
748
|
|
|
private function moveDate2Year() |
749
|
|
|
{ |
750
|
|
|
$date = $this->getParam('date') ?? false; |
751
|
|
|
if ($date) { |
752
|
|
|
if (preg_match('#^-?[12][0-9][0-9][0-9]$#', $date)) { |
753
|
|
|
$this->setParam('année', $date); |
754
|
|
|
$this->unsetParam('date'); |
755
|
|
|
//$this->log('>année'); |
756
|
|
|
} |
757
|
|
|
} |
758
|
|
|
} |
759
|
|
|
|
760
|
|
|
private function predictFormatByPattern() |
761
|
|
|
{ |
762
|
|
|
if (($value = $this->getParam('format'))) { |
763
|
|
|
// predict if 'format électronique' |
764
|
|
|
// format electronique lié au champ 'lire en ligne' |
765
|
|
|
// 2015 https://fr.wikipedia.org/wiki/Discussion_mod%C3%A8le:Ouvrage#format,_format_livre,_format_%C3%A9lectronique |
766
|
|
|
// if (preg_match('#(pdf|epub|html|kindle|audio|\{\{aud|jpg)#i', $value) > 0) { |
767
|
|
|
// |
768
|
|
|
// $this->setParam('format électronique', $value); |
769
|
|
|
// $this->unsetParam('format'); |
770
|
|
|
// $this->log('format:électronique?'); |
771
|
|
|
// |
772
|
|
|
// return; |
773
|
|
|
// } |
774
|
|
|
if (preg_match( |
775
|
|
|
'#(ill\.|couv\.|in-[0-9]|in-fol|poche|broché|relié|{{unité|{{Dunité|[0-9]{2} ?cm|\|cm}}|vol\.|A4)#i', |
776
|
|
|
$value |
777
|
|
|
) > 0 |
778
|
|
|
) { |
779
|
|
|
$this->setParam('format livre', $value); |
780
|
|
|
$this->unsetParam('format'); |
781
|
|
|
$this->log('format:livre?'); |
782
|
|
|
$this->notCosmetic = true; |
783
|
|
|
} |
784
|
|
|
// Certainement 'format électronique'... |
785
|
|
|
} |
786
|
|
|
} |
787
|
|
|
|
788
|
|
|
/** |
789
|
|
|
* @return bool |
790
|
|
|
* @throws Exception |
791
|
|
|
*/ |
792
|
|
|
public function checkMajorEdit(): bool |
793
|
|
|
{ |
794
|
|
|
// Correction paramètre |
795
|
|
|
if ($this->ouvrage->parametersErrorFromHydrate !== $this->original->parametersErrorFromHydrate) { |
796
|
|
|
return true; |
797
|
|
|
} |
798
|
|
|
// Complétion langue ? |
799
|
|
|
if (!empty($this->getParam('langue')) && empty($this->original->getParam('langue')) |
800
|
|
|
&& self::WIKI_LANGUAGE !== $this->getParam('langue') |
801
|
|
|
) { |
802
|
|
|
return true; |
803
|
|
|
} |
804
|
|
|
// TODO replace conditions ci-dessous par event flagMajor() |
805
|
|
|
// Retire le param/value 'langue' (pas major si conversion nom langue) |
806
|
|
|
$datOuvrage = $this->ouvrage->toArray(); |
807
|
|
|
$datOriginal = $this->original->toArray(); |
808
|
|
|
unset($datOriginal['langue'], $datOuvrage['langue']); |
809
|
|
|
|
810
|
|
|
// Modification données |
811
|
|
|
if ($datOriginal !== $datOuvrage) { |
812
|
|
|
return true; |
813
|
|
|
} |
814
|
|
|
|
815
|
|
|
return false; |
816
|
|
|
} |
817
|
|
|
|
818
|
|
|
/** |
819
|
|
|
* @return array |
820
|
|
|
*/ |
821
|
|
|
public function getLog(): array |
822
|
|
|
{ |
823
|
|
|
return $this->log; |
824
|
|
|
} |
825
|
|
|
|
826
|
|
|
/** |
827
|
|
|
* @return OuvrageTemplate |
828
|
|
|
*/ |
829
|
|
|
public function getOuvrage(): OuvrageTemplate |
830
|
|
|
{ |
831
|
|
|
return $this->ouvrage; |
832
|
|
|
} |
833
|
|
|
|
834
|
|
|
/** |
835
|
|
|
* todo : vérif lien rouge |
836
|
|
|
* todo 'lien éditeur' affiché 1x par page |
837
|
|
|
* opti : Suppression lien éditeur si c'est l'article de l'éditeur. |
838
|
|
|
* |
839
|
|
|
* @throws Exception |
840
|
|
|
*/ |
841
|
|
|
private function processEditeur() |
842
|
|
|
{ |
843
|
|
|
$editeur = $this->getParam('éditeur'); |
844
|
|
|
if (empty($editeur)) { |
845
|
|
|
return; |
846
|
|
|
} |
847
|
|
|
|
848
|
|
|
// FIX bug "GEO Art ([[Prisma Media]]) ; [[Le Monde]]" |
849
|
|
|
if (preg_match('#\[.*\[.*\[#', $editeur) > 0) { |
850
|
|
|
return; |
851
|
|
|
} |
852
|
|
|
// FIX bug "[[Fu|Bar]] bla" => [[Fu|Bar bla]] |
853
|
|
|
if (preg_match('#(.+\[\[|\]\].+)#', $editeur) > 0) { |
854
|
|
|
return; |
855
|
|
|
} |
856
|
|
|
|
857
|
|
|
// [[éditeur]] |
858
|
|
|
if (preg_match('#\[\[([^|]+)]]#', $editeur, $matches) > 0) { |
859
|
|
|
$editeurUrl = $matches[1]; |
860
|
|
|
} |
861
|
|
|
// [[bla|éditeur]] |
862
|
|
|
if (preg_match('#\[\[([^]|]+)\|.+]]#', $editeur, $matches) > 0) { |
863
|
|
|
$editeurUrl = $matches[1]; |
864
|
|
|
} |
865
|
|
|
|
866
|
|
|
// abréviations communes |
867
|
|
|
// ['éd. de ', 'éd. du ', 'éd.', 'ed.', 'Éd. de ', 'Éd.', 'édit.', 'Édit.', '(éd.)', '(ed.)', 'Ltd.'] |
868
|
|
|
|
869
|
|
|
$editeurStr = WikiTextUtil::unWikify($editeur); |
870
|
|
|
// On garde minuscule sur éditeur, pour nuance Éditeur/éditeur permettant de supprimer "éditeur" |
871
|
|
|
// ex: "éditions Milan" => "Milan" |
872
|
|
|
|
873
|
|
|
// $editeurStr = TextUtil::mb_ucfirst($editeurStr); |
874
|
|
|
|
875
|
|
|
// Déconseillé : 'lien éditeur' (obsolete 2019) |
876
|
|
|
if (!empty($this->getParam('lien éditeur'))) { |
877
|
|
|
if (empty($editeurUrl)) { |
878
|
|
|
$editeurUrl = $this->getParam('lien éditeur'); |
879
|
|
|
} |
880
|
|
|
$this->unsetParam('lien éditeur'); |
881
|
|
|
} |
882
|
|
|
|
883
|
|
|
// if (isset($editeurUrl)) { |
884
|
|
|
// $editeurUrl = TextUtil::mb_ucfirst($editeurUrl); |
885
|
|
|
// } |
886
|
|
|
$newEditeur = $editeurStr; |
887
|
|
|
if (isset($editeurUrl) && $editeurUrl !== $editeurStr) { |
888
|
|
|
$newEditeur = '[['.$editeurUrl.'|'.$editeurStr.']]'; |
889
|
|
|
} |
890
|
|
|
if (isset($editeurUrl) && $editeurUrl === $editeurStr) { |
891
|
|
|
$newEditeur = '[['.$editeurStr.']]'; |
892
|
|
|
} |
893
|
|
|
|
894
|
|
|
if ($newEditeur !== $editeur) { |
895
|
|
|
$this->setParam('éditeur', $newEditeur); |
896
|
|
|
$this->log('±éditeur'); |
897
|
|
|
$this->notCosmetic = true; |
898
|
|
|
} |
899
|
|
|
} |
900
|
|
|
} |
901
|
|
|
|
This check looks for private methods that have been defined, but are not used inside the class.