|
1
|
|
|
<?php |
|
2
|
|
|
/* |
|
3
|
|
|
* This file is part of dispositif/wikibot application (@github) |
|
4
|
|
|
* 2019-2023 © Philippe M./Irønie <[email protected]> |
|
5
|
|
|
* For the full copyright and MIT license information, view the license file. |
|
6
|
|
|
*/ |
|
7
|
|
|
|
|
8
|
|
|
declare(strict_types=1); |
|
9
|
|
|
|
|
10
|
|
|
namespace App\Application\OuvrageComplete; |
|
11
|
|
|
|
|
12
|
|
|
use App\Application\InfrastructurePorts\DbAdapterInterface; |
|
13
|
|
|
use App\Application\InfrastructurePorts\MemoryInterface; |
|
14
|
|
|
use App\Application\WikiBotConfig; |
|
15
|
|
|
use App\Domain\InfrastructurePorts\WikidataAdapterInterface; |
|
16
|
|
|
use App\Domain\Models\Wiki\OuvrageTemplate; |
|
17
|
|
|
use App\Domain\OptimizerFactory; |
|
18
|
|
|
use App\Domain\OuvrageComplete; |
|
19
|
|
|
use App\Domain\OuvrageFactory; |
|
20
|
|
|
use App\Domain\Publisher\Wikidata2Ouvrage; |
|
21
|
|
|
use App\Domain\SummaryLogTrait; |
|
22
|
|
|
use App\Domain\Utils\TemplateParser; |
|
23
|
|
|
use Exception; |
|
24
|
|
|
use Normalizer; |
|
25
|
|
|
use Psr\Log\LoggerInterface; |
|
26
|
|
|
use Psr\Log\NullLogger; |
|
27
|
|
|
use Throwable; |
|
28
|
|
|
|
|
29
|
|
|
/** |
|
30
|
|
|
* TODO Legacy class, to be refactored. To big, too many responsibilities. |
|
31
|
|
|
* TODO use DTO from DbAdapter. |
|
32
|
|
|
*/ |
|
33
|
|
|
class OuvrageCompleteWorker |
|
34
|
|
|
{ |
|
35
|
|
|
use SummaryLogTrait; |
|
36
|
|
|
|
|
37
|
|
|
/** |
|
38
|
|
|
* Exclusion requête BnF/Google/etc |
|
39
|
|
|
* Format EAN ou ISBN10 sans tiret. |
|
40
|
|
|
*/ |
|
41
|
|
|
public const ISBN_EAN_SKIP |
|
42
|
|
|
= [ |
|
43
|
|
|
'9782918758440', // Profils de lignes du réseau ferré français vol.2 |
|
44
|
|
|
'9782918758341', // Profils de lignes du réseau ferré français vol.1 |
|
45
|
|
|
'285608043X', // Dictionnaire encyclopédique d'électronique (langue erronée) |
|
46
|
|
|
'9782021401196', // sous-titre erroné |
|
47
|
|
|
]; |
|
48
|
|
|
|
|
49
|
|
|
/** |
|
50
|
|
|
* @var MemoryInterface |
|
51
|
|
|
*/ |
|
52
|
|
|
protected $memory; |
|
53
|
|
|
/** |
|
54
|
|
|
* @var DbAdapterInterface |
|
55
|
|
|
*/ |
|
56
|
|
|
private $queueAdapter; |
|
57
|
|
|
/** |
|
58
|
|
|
* @var string |
|
59
|
|
|
*/ |
|
60
|
|
|
private $raw = ''; |
|
61
|
|
|
private $page; // article title |
|
62
|
|
|
|
|
63
|
|
|
private $notCosmetic = false; |
|
64
|
|
|
private $major = false; |
|
65
|
|
|
/** |
|
66
|
|
|
* @var OuvrageTemplate |
|
67
|
|
|
*/ |
|
68
|
|
|
private $ouvrage; |
|
69
|
|
|
/** |
|
70
|
|
|
* @var LoggerInterface |
|
71
|
|
|
*/ |
|
72
|
|
|
private $logger; |
|
73
|
|
|
/** |
|
74
|
|
|
* @var WikidataAdapterInterface |
|
75
|
|
|
*/ |
|
76
|
|
|
private $wikidataAdapter; |
|
77
|
|
|
|
|
78
|
|
|
public function __construct( |
|
79
|
|
|
DbAdapterInterface $queueAdapter, |
|
80
|
|
|
WikidataAdapterInterface $wikidataAdapter, |
|
81
|
|
|
MemoryInterface $memory, |
|
82
|
|
|
?LoggerInterface $logger = null |
|
83
|
|
|
) |
|
84
|
|
|
{ |
|
85
|
|
|
$this->queueAdapter = $queueAdapter; |
|
86
|
|
|
$this->logger = $logger ?? new NullLogger(); |
|
87
|
|
|
$this->wikidataAdapter = $wikidataAdapter; |
|
88
|
|
|
$this->memory = $memory; |
|
89
|
|
|
} |
|
90
|
|
|
|
|
91
|
|
|
public function run(?int $limit = 10000): bool |
|
92
|
|
|
{ |
|
93
|
|
|
while ($limit > 0) { |
|
94
|
|
|
$limit--; |
|
95
|
|
|
sleep(1); |
|
96
|
|
|
$row = $this->getNewRow2Complete(); |
|
97
|
|
|
$this->raw = $row['raw']; |
|
98
|
|
|
$this->page = $row['page']; |
|
99
|
|
|
// Note : $row['id'] défini |
|
100
|
|
|
|
|
101
|
|
|
echo sprintf( |
|
102
|
|
|
"-------------------------------\n%s [%s]\n%s\n%s\n", |
|
103
|
|
|
date("Y-m-d H:i:s"), |
|
104
|
|
|
WikiBotConfig::VERSION ?? '', |
|
105
|
|
|
$this->page, |
|
106
|
|
|
$this->raw |
|
107
|
|
|
); |
|
108
|
|
|
|
|
109
|
|
|
$this->logger->debug($this->memory->getMemory(true)); |
|
110
|
|
|
|
|
111
|
|
|
// initialise variables |
|
112
|
|
|
$this->resetSummaryLog(); |
|
113
|
|
|
$this->ouvrage = null; |
|
114
|
|
|
$this->notCosmetic = false; |
|
115
|
|
|
$this->major = false; |
|
116
|
|
|
|
|
117
|
|
|
|
|
118
|
|
|
try { |
|
119
|
|
|
$parse = TemplateParser::parseAllTemplateByName('ouvrage', $this->raw); |
|
120
|
|
|
$origin = $parse['ouvrage'][0]['model'] ?? null; |
|
121
|
|
|
} catch (Throwable $e) { |
|
122
|
|
|
$this->logger->warning( |
|
123
|
|
|
sprintf( |
|
124
|
|
|
"*** ERREUR 432 impossible de transformer en modèle => skip %s : %s \n", |
|
125
|
|
|
$row['id'], |
|
126
|
|
|
$this->raw |
|
127
|
|
|
) |
|
128
|
|
|
); |
|
129
|
|
|
$this->queueAdapter->skipRow((int) $row['id']); |
|
130
|
|
|
sleep(10); |
|
131
|
|
|
continue; |
|
132
|
|
|
} |
|
133
|
|
|
|
|
134
|
|
|
if (!$origin instanceof OuvrageTemplate) { |
|
135
|
|
|
$this->logger->warning( |
|
136
|
|
|
sprintf( |
|
137
|
|
|
"*** ERREUR 433 impossible de transformer en modèle => skip %s : %s \n", |
|
138
|
|
|
$row['id'], |
|
139
|
|
|
$this->raw |
|
140
|
|
|
) |
|
141
|
|
|
); |
|
142
|
|
|
$this->queueAdapter->skipRow((int) $row['id']); |
|
143
|
|
|
sleep(10); |
|
144
|
|
|
continue; |
|
145
|
|
|
} |
|
146
|
|
|
|
|
147
|
|
|
// Final optimizing (with online predictions) |
|
148
|
|
|
$optimizer = OptimizerFactory::fromTemplate($origin, $this->page, $this->logger); |
|
149
|
|
|
$optimizer->doTasks(); |
|
150
|
|
|
$this->ouvrage = $optimizer->getOptiTemplate(); |
|
151
|
|
|
$this->summaryLog = array_merge($this->getSummaryLog(), $optimizer->getSummaryLog()); |
|
152
|
|
|
$this->notCosmetic = ($optimizer->notCosmetic || $this->notCosmetic); |
|
153
|
|
|
|
|
154
|
|
|
/** |
|
155
|
|
|
* RECHERCHE ONLINE |
|
156
|
|
|
*/ |
|
157
|
|
|
$isbn = $origin->getParam('isbn') ?? null; // avant mise en forme EAN>ISBN |
|
158
|
|
|
$isbn10 = $origin->getParam('isbn2') ?? $origin->getParam('isbn10') ?? null; |
|
159
|
|
|
if (!empty($isbn) |
|
160
|
|
|
&& !$origin->hasParamValue('isbn invalide') |
|
161
|
|
|
&& !$origin->hasParamValue('isbn erroné') |
|
162
|
|
|
) { |
|
163
|
|
|
$this->onlineIsbnSearch($isbn, $isbn10); |
|
164
|
|
|
} |
|
165
|
|
|
|
|
166
|
|
|
$this->sendCompleted(); |
|
167
|
|
|
unset($optimizer); |
|
168
|
|
|
unset($parse); |
|
169
|
|
|
unset($origin); |
|
170
|
|
|
} // END WHILE |
|
171
|
|
|
|
|
172
|
|
|
return true; |
|
173
|
|
|
} |
|
174
|
|
|
|
|
175
|
|
|
/** |
|
176
|
|
|
* Get array (title+raw strings) to complete from AMQP queue, SQL Select or file reading. |
|
177
|
|
|
* |
|
178
|
|
|
* @return array |
|
179
|
|
|
* @throws Exception |
|
180
|
|
|
*/ |
|
181
|
|
|
private function getNewRow2Complete(): array |
|
182
|
|
|
{ |
|
183
|
|
|
$row = $this->queueAdapter->getNewRaw(); |
|
184
|
|
|
if (empty($row) || empty($row['raw'])) { |
|
185
|
|
|
echo "STOP: no more queue to process \n"; |
|
186
|
|
|
throw new Exception('no more queue to process'); |
|
187
|
|
|
} |
|
188
|
|
|
|
|
189
|
|
|
return $row; |
|
190
|
|
|
} |
|
191
|
|
|
|
|
192
|
|
|
/** |
|
193
|
|
|
* @param string $isbn |
|
194
|
|
|
* @param string|null $isbn10 |
|
195
|
|
|
* |
|
196
|
|
|
* @return bool |
|
197
|
|
|
*/ |
|
198
|
|
|
private function isIsbnSkipped(string $isbn, ?string $isbn10 = null): bool |
|
199
|
|
|
{ |
|
200
|
|
|
return in_array(str_replace('-', '', $isbn), self::ISBN_EAN_SKIP) |
|
201
|
|
|
|| ($isbn10 !== null |
|
202
|
|
|
&& in_array(str_replace('-', '', $isbn10), self::ISBN_EAN_SKIP)); |
|
203
|
|
|
} |
|
204
|
|
|
|
|
205
|
|
|
private function onlineIsbnSearch(string $isbn, ?string $isbn10 = null) |
|
206
|
|
|
{ |
|
207
|
|
|
if ($this->isIsbnSkipped($isbn, $isbn10)) { |
|
208
|
|
|
echo "*** SKIP THAT ISBN ***\n"; |
|
209
|
|
|
|
|
210
|
|
|
// Vérifier logique return |
|
211
|
|
|
return; |
|
212
|
|
|
} |
|
213
|
|
|
|
|
214
|
|
|
online: |
|
215
|
|
|
$this->logger->info("sleep 10...\n"); |
|
216
|
|
|
sleep(10); |
|
217
|
|
|
|
|
218
|
|
|
try { |
|
219
|
|
|
$this->logger->debug('BIBLIO NAT FRANCE...'); |
|
220
|
|
|
// BnF sait pas trouver un vieux livre (10) d'après ISBN-13... FACEPALM ! |
|
221
|
|
|
$bnfOuvrage = null; |
|
222
|
|
|
if ($isbn10) { |
|
223
|
|
|
$bnfOuvrage = OuvrageFactory::BnfFromIsbn($isbn10); |
|
224
|
|
|
sleep(2); |
|
225
|
|
|
} |
|
226
|
|
|
if (!$isbn10 || null === $bnfOuvrage || empty($bnfOuvrage->getParam('titre'))) { |
|
227
|
|
|
$bnfOuvrage = OuvrageFactory::BnfFromIsbn($isbn); |
|
228
|
|
|
} |
|
229
|
|
|
if ($bnfOuvrage instanceof OuvrageTemplate) { |
|
|
|
|
|
|
230
|
|
|
$this->completeOuvrage($bnfOuvrage); |
|
231
|
|
|
|
|
232
|
|
|
// Wikidata requests from $infos (ISBN/ISNI) |
|
233
|
|
|
if (!empty($bnfOuvrage->getInfos())) { |
|
234
|
|
|
$this->logger->info('WIKIDATA...'); |
|
235
|
|
|
|
|
236
|
|
|
// TODO move to factory |
|
237
|
|
|
$wdComplete = new Wikidata2Ouvrage($this->wikidataAdapter, clone $bnfOuvrage, $this->page); |
|
238
|
|
|
$this->completeOuvrage($wdComplete->getOuvrage()); |
|
239
|
|
|
} |
|
240
|
|
|
} |
|
241
|
|
|
} catch (Throwable $e) { |
|
242
|
|
|
if (strpos($e->getMessage(), 'Could not resolve host') !== false) { |
|
243
|
|
|
throw $e; |
|
244
|
|
|
} |
|
245
|
|
|
$this->logger->error( |
|
246
|
|
|
sprintf( |
|
247
|
|
|
"*** ERREUR BnF Isbn Search %s %s %s \n", |
|
248
|
|
|
$e->getMessage(), |
|
249
|
|
|
$e->getFile(), |
|
250
|
|
|
$e->getLine() |
|
251
|
|
|
) |
|
252
|
|
|
); |
|
253
|
|
|
} |
|
254
|
|
|
|
|
255
|
|
|
if (!isset($bnfOuvrage) || !$this->skipGoogle($bnfOuvrage)) { |
|
256
|
|
|
try { |
|
257
|
|
|
$this->logger->info('GOOGLE...'); |
|
258
|
|
|
|
|
259
|
|
|
$googleOuvrage = OuvrageFactory::GoogleFromIsbn($isbn); |
|
260
|
|
|
$this->completeOuvrage($googleOuvrage); |
|
261
|
|
|
} catch (Throwable $e) { |
|
262
|
|
|
$this->logger->warning("*** ERREUR GOOGLE Isbn Search ***".$e->getMessage()); |
|
263
|
|
|
if (strpos($e->getMessage(), 'Could not resolve host: www.googleapis.com') === false) { |
|
264
|
|
|
throw $e; |
|
265
|
|
|
} |
|
266
|
|
|
unset($e); |
|
267
|
|
|
} |
|
268
|
|
|
} |
|
269
|
|
|
|
|
270
|
|
|
if (!isset($bnfOuvrage) && !isset($googleOuvrage)) { |
|
271
|
|
|
try { |
|
272
|
|
|
$this->logger->info('OpenLibrary...'); |
|
273
|
|
|
$openLibraryOuvrage = OuvrageFactory::OpenLibraryFromIsbn($isbn); |
|
274
|
|
|
if (!empty($openLibraryOuvrage)) { |
|
275
|
|
|
$this->completeOuvrage($openLibraryOuvrage); |
|
276
|
|
|
} |
|
277
|
|
|
} catch (Throwable $e) { |
|
278
|
|
|
$this->logger->warning('**** ERREUR OpenLibrary Isbn Search'); |
|
279
|
|
|
} |
|
280
|
|
|
} |
|
281
|
|
|
} |
|
282
|
|
|
|
|
283
|
|
|
// private function onlineQuerySearch(string $query) |
|
284
|
|
|
// { |
|
285
|
|
|
// echo "sleep 40..."; |
|
286
|
|
|
// sleep(20); |
|
287
|
|
|
// onlineQuerySearch: |
|
288
|
|
|
// |
|
289
|
|
|
// try { |
|
290
|
|
|
// dump('GOOGLE SEARCH...'); |
|
291
|
|
|
// // $googleOuvrage = OuvrageFactory::GoogleFromIsbn($isbn); |
|
292
|
|
|
// $adapter = new GoogleBooksAdapter(); |
|
293
|
|
|
// $data = $adapter->search('blabla'); |
|
294
|
|
|
// dump($data); |
|
295
|
|
|
// //die; |
|
296
|
|
|
// // return $import->getOuvrage(); |
|
297
|
|
|
// // $this->completeOuvrage($googleOuvrage); |
|
298
|
|
|
// } catch (Throwable $e) { |
|
299
|
|
|
// echo "*** ERREUR GOOGLE QuerySearch *** ".$e->getMessage()."\n"; |
|
300
|
|
|
// echo "sleep 30min"; |
|
301
|
|
|
// sleep(60 * 30); |
|
302
|
|
|
// echo "Wake up\n"; |
|
303
|
|
|
// goto onlineQuerySearch; |
|
304
|
|
|
// } |
|
305
|
|
|
// } |
|
306
|
|
|
|
|
307
|
|
|
private function completeOuvrage(OuvrageTemplate $onlineOuvrage) |
|
308
|
|
|
{ |
|
309
|
|
|
$this->logger->info($onlineOuvrage->serialize(true)); |
|
310
|
|
|
$optimizer = OptimizerFactory::fromTemplate($onlineOuvrage, $this->page, $this->logger); |
|
311
|
|
|
$onlineOptimized = ($optimizer)->doTasks()->getOptiTemplate(); |
|
312
|
|
|
|
|
313
|
|
|
$completer = new OuvrageComplete($this->ouvrage, $onlineOptimized, $this->logger); |
|
314
|
|
|
$this->ouvrage = $completer->getResult(); |
|
315
|
|
|
|
|
316
|
|
|
// todo move that optimizing in OuvrageComplete ? |
|
317
|
|
|
$optimizer = OptimizerFactory::fromTemplate($this->ouvrage, $this->page, $this->logger); |
|
318
|
|
|
$this->ouvrage = $optimizer->doTasks()->getOptiTemplate(); |
|
319
|
|
|
|
|
320
|
|
|
$this->logger->info('Summary', $completer->getSummaryLog()); |
|
321
|
|
|
|
|
322
|
|
|
if ($completer->major) { |
|
323
|
|
|
$this->major = true; |
|
324
|
|
|
} |
|
325
|
|
|
$this->notCosmetic = ($completer->notCosmetic || $this->notCosmetic); |
|
326
|
|
|
$this->summaryLog = array_merge($this->getSummaryLog(), $completer->getSummaryLog()); |
|
327
|
|
|
unset($optimizer); |
|
328
|
|
|
unset($completer); |
|
329
|
|
|
} |
|
330
|
|
|
|
|
331
|
|
|
private function sendCompleted() |
|
332
|
|
|
{ |
|
333
|
|
|
$isbn = $this->ouvrage->getParam('isbn'); |
|
334
|
|
|
$finalData = [ |
|
335
|
|
|
// 'page' => |
|
336
|
|
|
'raw' => $this->raw, |
|
337
|
|
|
'opti' => $this->serializeFinalOpti(), |
|
338
|
|
|
'optidate' => date("Y-m-d H:i:s"), |
|
339
|
|
|
'modifs' => mb_substr(implode(',', $this->getSummaryLog()), 0, 250), |
|
340
|
|
|
'notcosmetic' => ($this->notCosmetic) ? 1 : 0, |
|
341
|
|
|
'major' => ($this->major) ? 1 : 0, |
|
342
|
|
|
'isbn' => substr($isbn,0,19), |
|
|
|
|
|
|
343
|
|
|
'version' => WikiBotConfig::VERSION ?? null, |
|
344
|
|
|
]; |
|
345
|
|
|
$this->logger->info('finalData', $finalData); |
|
346
|
|
|
// Json ? |
|
347
|
|
|
$result = $this->queueAdapter->sendCompletedData($finalData); |
|
348
|
|
|
|
|
349
|
|
|
$this->logger->debug($result ? 'OK DB' : 'erreur sendCompletedData()'); |
|
350
|
|
|
} |
|
351
|
|
|
|
|
352
|
|
|
/** |
|
353
|
|
|
* Final serialization of the completed OuvrageTemplate. |
|
354
|
|
|
*/ |
|
355
|
|
|
private function serializeFinalOpti(): string |
|
356
|
|
|
{ |
|
357
|
|
|
// // Améliore style compact : plus espacé |
|
358
|
|
|
// if ('|' === $this->ouvrage->userSeparator) { |
|
359
|
|
|
// $this->ouvrage->userSeparator = ' |'; |
|
360
|
|
|
// } |
|
361
|
|
|
$finalOpti = $this->ouvrage->serialize(true); |
|
362
|
|
|
$finalOpti = Normalizer::normalize($finalOpti); |
|
363
|
|
|
if (empty($finalOpti) || !is_string($finalOpti)) { |
|
364
|
|
|
throw new Exception('normalized $finalOpti serialize in OuvrageComplete is not a string'); |
|
365
|
|
|
} |
|
366
|
|
|
|
|
367
|
|
|
return $finalOpti; |
|
368
|
|
|
} |
|
369
|
|
|
|
|
370
|
|
|
private function skipGoogle($bnfOuvrage): bool |
|
371
|
|
|
{ |
|
372
|
|
|
return $bnfOuvrage instanceof OuvrageTemplate |
|
373
|
|
|
&& $bnfOuvrage->hasParamValue('titre') |
|
374
|
|
|
&& ($this->ouvrage->hasParamValue('lire en ligne') |
|
375
|
|
|
|| $this->ouvrage->hasParamValue('présentation en ligne')); |
|
376
|
|
|
} |
|
377
|
|
|
} |
|
378
|
|
|
|
Dispositif /
Wikibot
Loading history...