|
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 Exception; |
|
17
|
|
|
use Normalizer; |
|
18
|
|
|
|
|
19
|
|
|
class OuvrageComplete |
|
20
|
|
|
{ |
|
21
|
|
|
const ADD_PRESENTATION_EN_LIGNE = true; |
|
22
|
|
|
const WIKI_LANGUAGE = 'fr'; |
|
23
|
|
|
|
|
24
|
|
|
/** |
|
25
|
|
|
* @var OuvrageTemplate |
|
26
|
|
|
*/ |
|
27
|
|
|
private $origin; |
|
28
|
|
|
|
|
29
|
|
|
private $book; |
|
30
|
|
|
|
|
31
|
|
|
public $major = false; |
|
32
|
|
|
|
|
33
|
|
|
public $notCosmetic = false; |
|
34
|
|
|
|
|
35
|
|
|
private $log = []; |
|
36
|
|
|
|
|
37
|
|
|
private $sameBook; |
|
38
|
|
|
|
|
39
|
|
|
//todo: injection référence base ou mapping ? (Google |
|
40
|
|
|
public function __construct(OuvrageTemplate $origin, OuvrageTemplate $book) |
|
41
|
|
|
{ |
|
42
|
|
|
$this->origin = clone $origin; |
|
43
|
|
|
$this->book = $book; |
|
44
|
|
|
} |
|
45
|
|
|
|
|
46
|
|
|
public function getLog(): array |
|
47
|
|
|
{ |
|
48
|
|
|
return $this->log; |
|
49
|
|
|
} |
|
50
|
|
|
|
|
51
|
|
|
/** |
|
52
|
|
|
* @return OuvrageTemplate |
|
53
|
|
|
* @throws Exception |
|
54
|
|
|
*/ |
|
55
|
|
|
public function getResult() |
|
56
|
|
|
{ |
|
57
|
|
|
$this->complete(); |
|
58
|
|
|
|
|
59
|
|
|
return $this->origin; |
|
60
|
|
|
} |
|
61
|
|
|
|
|
62
|
|
|
/** |
|
63
|
|
|
* @return bool |
|
64
|
|
|
* @throws Exception |
|
65
|
|
|
*/ |
|
66
|
|
|
private function complete() |
|
67
|
|
|
{ |
|
68
|
|
|
// si livre suspect, on stoppe |
|
69
|
|
|
$this->sameBook = $this->predictSameBook(); |
|
70
|
|
|
if (!$this->sameBook) { |
|
71
|
|
|
dump('not same book'); |
|
72
|
|
|
|
|
73
|
|
|
return false; |
|
74
|
|
|
} |
|
75
|
|
|
|
|
76
|
|
|
$skipParam = [ |
|
77
|
|
|
'isbn invalide', |
|
78
|
|
|
'auteurs', |
|
79
|
|
|
'auteur1', |
|
80
|
|
|
'prénom1', |
|
81
|
|
|
'nom1', |
|
82
|
|
|
'auteur2', |
|
83
|
|
|
'prénom2', |
|
84
|
|
|
'nom2', |
|
85
|
|
|
'auteur3', |
|
86
|
|
|
'prénom3', |
|
87
|
|
|
'nom3', |
|
88
|
|
|
'auteur4', |
|
89
|
|
|
'prénom4', |
|
90
|
|
|
'nom4', |
|
91
|
|
|
'lire en ligne', |
|
92
|
|
|
'présentation en ligne', |
|
93
|
|
|
'date', |
|
94
|
|
|
'sous-titre', |
|
95
|
|
|
]; |
|
96
|
|
|
|
|
97
|
|
|
// completion automatique |
|
98
|
|
|
foreach ($this->book->toArray() as $param => $value) { |
|
99
|
|
|
if (empty($this->origin->getParam($param))) { |
|
100
|
|
|
if (in_array($param, $skipParam)) { |
|
101
|
|
|
continue; |
|
102
|
|
|
} |
|
103
|
|
|
// skip 'année' if 'date' not empty |
|
104
|
|
|
if ('année' === $param && !empty($this->origin->getParam('date'))) { |
|
105
|
|
|
continue; |
|
106
|
|
|
} |
|
107
|
|
|
|
|
108
|
|
|
$this->origin->setParam($param, $value); |
|
109
|
|
|
|
|
110
|
|
|
if ('langue' === $param && static::WIKI_LANGUAGE === $value) { |
|
111
|
|
|
//$this->log('fr'.$param); |
|
112
|
|
|
continue; |
|
113
|
|
|
} |
|
114
|
|
|
|
|
115
|
|
|
$this->log('++'.$param); |
|
116
|
|
|
$this->major = true; |
|
117
|
|
|
$this->notCosmetic = true; |
|
118
|
|
|
} |
|
119
|
|
|
} |
|
120
|
|
|
|
|
121
|
|
|
// $this->dateComplete(); |
|
122
|
|
|
$this->googleBookProcess(); |
|
123
|
|
|
$this->processSousTitre(); |
|
124
|
|
|
|
|
125
|
|
|
return true; |
|
126
|
|
|
} |
|
127
|
|
|
|
|
128
|
|
|
private function log(string $string): void |
|
129
|
|
|
{ |
|
130
|
|
|
if (!empty($string)) { |
|
131
|
|
|
$this->log[] = trim($string); |
|
132
|
|
|
} |
|
133
|
|
|
} |
|
134
|
|
|
|
|
135
|
|
|
/** |
|
136
|
|
|
* todo: test + refactor dirty logic/duplicate. |
|
137
|
|
|
* todo: bistro specs |
|
138
|
|
|
* Gestion doublon et accessibilité document Google Book. |
|
139
|
|
|
* |
|
140
|
|
|
* @throws Exception |
|
141
|
|
|
*/ |
|
142
|
|
|
private function googleBookProcess() |
|
143
|
|
|
{ |
|
144
|
|
|
$lire = $this->origin->getParam('lire en ligne') ?? false; |
|
145
|
|
|
if (!empty($lire) && GoogleLivresTemplate::isGoogleBookValue($lire)) { |
|
146
|
|
|
if (!empty($this->book->getParam('présentation en ligne'))) { |
|
147
|
|
|
// PARTIAL |
|
148
|
|
|
// on déplace sur présentation |
|
149
|
|
|
$this->origin->setParam('présentation en ligne', $lire); |
|
150
|
|
|
$this->origin->unsetParam('lire en ligne'); |
|
151
|
|
|
$this->log('Google partiel'); |
|
152
|
|
|
$this->notCosmetic = true; |
|
153
|
|
|
|
|
154
|
|
|
return; // ? |
|
155
|
|
|
} |
|
156
|
|
|
} |
|
157
|
|
|
// completion basique |
|
158
|
|
|
$booklire = $this->book->getParam('lire en ligne'); |
|
159
|
|
|
if (empty($lire) && !empty($booklire)) { |
|
160
|
|
|
$this->origin->setParam('lire en ligne', $booklire); |
|
161
|
|
|
$this->log('+lire en ligne'); |
|
162
|
|
|
$this->notCosmetic = true; |
|
163
|
|
|
$this->major = true; |
|
164
|
|
|
} |
|
165
|
|
|
unset($lire, $booklire); |
|
166
|
|
|
|
|
167
|
|
|
$presentation = $this->origin->getParam('présentation en ligne') ?? false; |
|
168
|
|
|
if (!empty($presentation) && GoogleLivresTemplate::isGoogleBookValue($presentation)) { |
|
169
|
|
|
if (!empty($this->book->getParam('lire en ligne'))) { |
|
170
|
|
|
// TOTAL |
|
171
|
|
|
// on déplace sur lire en ligne |
|
172
|
|
|
$this->origin->setParam('lire en ligne', $presentation); |
|
173
|
|
|
$this->origin->unsetParam('présentation en ligne'); |
|
174
|
|
|
$this->log('Google accessible'); |
|
175
|
|
|
$this->notCosmetic = true; |
|
176
|
|
|
} |
|
177
|
|
|
} |
|
178
|
|
|
|
|
179
|
|
|
// ajout de Google partial si présentation/lire sont vides |
|
180
|
|
|
$bookpresentation = $this->book->getParam('présentation en ligne'); |
|
181
|
|
|
if (self::ADD_PRESENTATION_EN_LIGNE |
|
182
|
|
|
&& empty($this->origin->getParam('présentation en ligne')) |
|
183
|
|
|
&& empty($this->origin->getParam('lire en ligne')) |
|
184
|
|
|
&& !empty($bookpresentation) |
|
185
|
|
|
) { |
|
186
|
|
|
$this->origin->setParam('présentation en ligne', $bookpresentation); |
|
187
|
|
|
$this->log('+présentation en ligne'); |
|
188
|
|
|
$this->notCosmetic = true; |
|
189
|
|
|
$this->major = true; |
|
190
|
|
|
} |
|
191
|
|
|
} |
|
192
|
|
|
|
|
193
|
|
|
/** |
|
194
|
|
|
* @return bool |
|
195
|
|
|
* @throws Exception |
|
196
|
|
|
*/ |
|
197
|
|
|
private function predictSameBook() |
|
198
|
|
|
{ |
|
199
|
|
|
if ($this->hasSameISBN() && ($this->hasSameBookTitles() || $this->hasSameAuthors())) { |
|
200
|
|
|
return true; |
|
201
|
|
|
} |
|
202
|
|
|
if ($this->hasSameBookTitles() && $this->hasSameAuthors()) { |
|
203
|
|
|
return true; |
|
204
|
|
|
} |
|
205
|
|
|
|
|
206
|
|
|
return false; |
|
207
|
|
|
} |
|
208
|
|
|
|
|
209
|
|
|
/** |
|
210
|
|
|
* @return bool |
|
211
|
|
|
* @throws Exception |
|
212
|
|
|
*/ |
|
213
|
|
|
private function hasSameAuthors(): bool |
|
214
|
|
|
{ |
|
215
|
|
|
if ($this->authorsFromBook($this->origin) === $this->authorsFromBook($this->book)) { |
|
216
|
|
|
return true; |
|
217
|
|
|
} |
|
218
|
|
|
|
|
219
|
|
|
// if there is only 2 char of difference (i.e. typo error) |
|
220
|
|
|
if (levenshtein($this->authorsFromBook($this->origin), $this->authorsFromBook($this->book)) <= 2) { |
|
221
|
|
|
$this->log('typo auteurs?'); |
|
222
|
|
|
|
|
223
|
|
|
return true; |
|
224
|
|
|
} |
|
225
|
|
|
|
|
226
|
|
|
// Si auteur manquant sur wikipedia |
|
227
|
|
|
if (empty($this->authorsFromBook($this->origin))) { |
|
228
|
|
|
return true; |
|
229
|
|
|
} |
|
230
|
|
|
|
|
231
|
|
|
return false; |
|
232
|
|
|
} |
|
233
|
|
|
|
|
234
|
|
|
/** |
|
235
|
|
|
* @param OuvrageTemplate $ouv |
|
236
|
|
|
* |
|
237
|
|
|
* @return string |
|
238
|
|
|
* @throws Exception |
|
239
|
|
|
*/ |
|
240
|
|
|
private function authorsFromBook(OuvrageTemplate $ouv) |
|
241
|
|
|
{ |
|
242
|
|
|
$text = ''; |
|
243
|
|
|
$paramAuteurs = [ |
|
244
|
|
|
'auteurs', |
|
245
|
|
|
'auteur1', |
|
246
|
|
|
'prénom1', |
|
247
|
|
|
'nom1', |
|
248
|
|
|
'auteur2', |
|
249
|
|
|
'prénom2', |
|
250
|
|
|
'nom2', |
|
251
|
|
|
'auteur3', |
|
252
|
|
|
'prénom3', |
|
253
|
|
|
'nom3', |
|
254
|
|
|
'auteur4', |
|
255
|
|
|
'prénom4', |
|
256
|
|
|
'nom4', |
|
257
|
|
|
]; |
|
258
|
|
|
foreach ($paramAuteurs as $param) { |
|
259
|
|
|
$value = $ouv->getParam($param); |
|
260
|
|
|
// retire wikilien sur auteur |
|
261
|
|
|
if (!empty($value)) { |
|
262
|
|
|
$text .= WikiTextUtil::unWikify($value); |
|
263
|
|
|
} |
|
264
|
|
|
} |
|
265
|
|
|
|
|
266
|
|
|
return $this->stripAll($text); |
|
267
|
|
|
} |
|
268
|
|
|
|
|
269
|
|
|
/** |
|
270
|
|
|
* @return bool |
|
271
|
|
|
* @throws Exception |
|
272
|
|
|
*/ |
|
273
|
|
|
private function hasSameISBN(): bool |
|
274
|
|
|
{ |
|
275
|
|
|
if (empty($this->origin->getParam('isbn')) || empty($this->book->getParam('isbn'))) { |
|
276
|
|
|
return false; |
|
277
|
|
|
} |
|
278
|
|
|
// TODO replace with calcul isbn13 |
|
279
|
|
|
$isbn1 = IsbnFacade::isbn2ean($this->origin->getParam('isbn')); |
|
|
|
|
|
|
280
|
|
|
$isbn2 = IsbnFacade::isbn2ean($this->book->getParam('isbn')); |
|
281
|
|
|
if ($isbn1 === $isbn2) { |
|
282
|
|
|
return true; |
|
283
|
|
|
} |
|
284
|
|
|
|
|
285
|
|
|
return false; |
|
286
|
|
|
} |
|
287
|
|
|
|
|
288
|
|
|
/** |
|
289
|
|
|
* Add or extract subtitle like in second book. |
|
290
|
|
|
* |
|
291
|
|
|
* @throws Exception |
|
292
|
|
|
*/ |
|
293
|
|
|
private function processSousTitre() |
|
294
|
|
|
{ |
|
295
|
|
|
if (empty($this->book->getParam('sous-titre'))) { |
|
296
|
|
|
return; |
|
297
|
|
|
} |
|
298
|
|
|
|
|
299
|
|
|
// Skip pour éviter conflit entre 'sous-titre' et 'collection' ou 'titre volume' |
|
300
|
|
|
if (!empty($this->origin->getParam('titre volume')) |
|
301
|
|
|
|| !empty($this->origin->getParam('titre chapitre')) |
|
302
|
|
|
|| !empty($this->origin->getParam('titre tome')) |
|
303
|
|
|
|| !empty($this->origin->getParam('collection')) |
|
304
|
|
|
|| !empty($this->origin->getParam('nature ouvrage')) |
|
305
|
|
|
) { |
|
306
|
|
|
return; |
|
307
|
|
|
} |
|
308
|
|
|
|
|
309
|
|
|
// simple : titres identiques mais sous-titre manquant |
|
310
|
|
|
if ($this->stripAll($this->origin->getParam('titre')) === $this->stripAll($this->book->getParam('titre'))) { |
|
|
|
|
|
|
311
|
|
|
// même titre mais sous-titre manquant |
|
312
|
|
|
if (empty($this->origin->getParam('sous-titre'))) { |
|
313
|
|
|
$this->origin->setParam('sous-titre', $this->book->getParam('sous-titre')); |
|
|
|
|
|
|
314
|
|
|
$this->log('++sous-titre'); |
|
315
|
|
|
$this->major = true; |
|
316
|
|
|
$this->notCosmetic = true; |
|
317
|
|
|
|
|
318
|
|
|
return; |
|
319
|
|
|
} |
|
320
|
|
|
} |
|
321
|
|
|
|
|
322
|
|
|
// compliqué : sous-titre inclus dans titre original => on copie titre/sous-titre de book |
|
323
|
|
|
if ($this->charsFromBigTitle($this->origin) === $this->charsFromBigTitle($this->book)) { |
|
324
|
|
|
if (empty($this->origin->getParam('sous-titre'))) { |
|
325
|
|
|
$this->origin->setParam('titre', $this->book->getParam('titre')); |
|
326
|
|
|
$this->origin->setParam('sous-titre', $this->book->getParam('sous-titre')); |
|
327
|
|
|
$this->log('>titre>sous-titre'); |
|
328
|
|
|
} |
|
329
|
|
|
} |
|
330
|
|
|
} |
|
331
|
|
|
|
|
332
|
|
|
/** |
|
333
|
|
|
* @return bool |
|
334
|
|
|
* @throws Exception |
|
335
|
|
|
*/ |
|
336
|
|
|
private function hasSameBookTitles(): bool |
|
337
|
|
|
{ |
|
338
|
|
|
if ($this->charsFromBigTitle($this->origin) === $this->charsFromBigTitle($this->book)) { |
|
339
|
|
|
return true; |
|
340
|
|
|
} |
|
341
|
|
|
|
|
342
|
|
|
// if there is only 2 chars of difference (i.e. typo error) |
|
343
|
|
|
if (levenshtein($this->charsFromBigTitle($this->origin), $this->charsFromBigTitle($this->book)) <= 2) { |
|
344
|
|
|
// $this->log('typo titre?'); // TODO Normalize:: text from external API |
|
345
|
|
|
|
|
346
|
|
|
return true; |
|
347
|
|
|
} |
|
348
|
|
|
|
|
349
|
|
|
// si l'un des ouvrages ne comporte pas le sous-titre |
|
350
|
|
|
if ($this->stripAll($this->origin->getParam('titre')) === $this->stripAll($this->book->getParam('titre'))) { |
|
|
|
|
|
|
351
|
|
|
return true; |
|
352
|
|
|
} |
|
353
|
|
|
|
|
354
|
|
|
// sous-titre inclus dans le titre |
|
355
|
|
|
// "Loiret : un département à l'élégance naturelle" <=> "Loiret" |
|
356
|
|
|
if ($this->stripAll($this->mainBookTitle($this->origin->getParam('titre'))) === $this->stripAll( |
|
|
|
|
|
|
357
|
|
|
$this->mainBookTitle($this->origin->getParam('titre')) |
|
358
|
|
|
) |
|
359
|
|
|
) { |
|
360
|
|
|
return true; |
|
361
|
|
|
} |
|
362
|
|
|
// titre manquant sur wiki |
|
363
|
|
|
if (empty($this->charsFromBigTitle($this->origin))) { |
|
364
|
|
|
return true; |
|
365
|
|
|
} |
|
366
|
|
|
|
|
367
|
|
|
return false; |
|
368
|
|
|
} |
|
369
|
|
|
|
|
370
|
|
|
/** |
|
371
|
|
|
* Give string before ":" (or same string if no ":"). |
|
372
|
|
|
* |
|
373
|
|
|
* @param string $str |
|
374
|
|
|
* |
|
375
|
|
|
* @return string |
|
376
|
|
|
*/ |
|
377
|
|
|
private function mainBookTitle(string $str) |
|
378
|
|
|
{ |
|
379
|
|
|
if (($pos = mb_strpos($str, ':'))) { |
|
380
|
|
|
$str = trim(mb_substr($str, 0, $pos)); |
|
381
|
|
|
} |
|
382
|
|
|
|
|
383
|
|
|
return $str; |
|
384
|
|
|
} |
|
385
|
|
|
|
|
386
|
|
|
/** |
|
387
|
|
|
* @param OuvrageTemplate $ouvrage |
|
388
|
|
|
* |
|
389
|
|
|
* @return string |
|
390
|
|
|
* @throws Exception |
|
391
|
|
|
*/ |
|
392
|
|
|
private function charsFromBigTitle(OuvrageTemplate $ouvrage): string |
|
393
|
|
|
{ |
|
394
|
|
|
$text = $ouvrage->getParam('titre').$ouvrage->getParam('sous-titre'); |
|
395
|
|
|
|
|
396
|
|
|
return $this->stripAll(Normalizer::normalize($text)); |
|
397
|
|
|
} |
|
398
|
|
|
|
|
399
|
|
|
/** |
|
400
|
|
|
* @param string $text |
|
401
|
|
|
* |
|
402
|
|
|
* @return string |
|
403
|
|
|
*/ |
|
404
|
|
|
private function stripAll(string $text): string |
|
405
|
|
|
{ |
|
406
|
|
|
$text = str_replace([' and ', ' et ', '&'], '', $text); |
|
407
|
|
|
$text = str_replace(' ', '', $text); |
|
408
|
|
|
$text = mb_strtolower(TextUtil::stripPunctuation(TextUtil::stripAccents($text))); |
|
409
|
|
|
|
|
410
|
|
|
return $text; |
|
411
|
|
|
} |
|
412
|
|
|
} |
|
413
|
|
|
|
Dispositif /
Wikibot
Loading history...