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; |
11
|
|
|
|
12
|
|
|
use App\Domain\Models\Wiki\OuvrageTemplate; |
13
|
|
|
use App\Domain\OuvrageComplete; |
14
|
|
|
use App\Domain\OuvrageFactory; |
15
|
|
|
use App\Domain\OuvrageOptimize; |
16
|
|
|
use App\Domain\Publisher\Wikidata2Ouvrage; |
17
|
|
|
use App\Domain\Utils\TemplateParser; |
18
|
|
|
use Normalizer; |
19
|
|
|
use Throwable; |
20
|
|
|
|
21
|
|
|
/** |
22
|
|
|
* Class CompleteProcess |
23
|
|
|
*/ |
24
|
|
|
class CompleteProcess |
25
|
|
|
{ |
26
|
|
|
/** |
27
|
|
|
* @var bool |
28
|
|
|
*/ |
29
|
|
|
public $verbose = false; |
30
|
|
|
|
31
|
|
|
/** |
32
|
|
|
* @var QueueInterface |
33
|
|
|
*/ |
34
|
|
|
private $queueAdapter; |
35
|
|
|
/** |
36
|
|
|
* @var string |
37
|
|
|
*/ |
38
|
|
|
private $raw = ''; |
39
|
|
|
private $log = []; |
40
|
|
|
private $notCosmetic = false; |
41
|
|
|
private $major = false; |
42
|
|
|
/** |
43
|
|
|
* @var OuvrageTemplate |
44
|
|
|
*/ |
45
|
|
|
private $ouvrage; |
46
|
|
|
|
47
|
|
|
public function __construct(QueueInterface $queueAdapter, ?bool $verbose = false) |
48
|
|
|
{ |
49
|
|
|
$this->queueAdapter = $queueAdapter; |
50
|
|
|
$this->verbose = (bool)$verbose; |
51
|
|
|
} |
52
|
|
|
|
53
|
|
|
public function run(?int $limit = 10000) |
54
|
|
|
{ |
55
|
|
|
$memory = new Memory(); |
56
|
|
|
while ($limit > 0) { |
57
|
|
|
$limit--; |
58
|
|
|
sleep(1); |
59
|
|
|
$this->raw = $this->getNewRaw(); |
60
|
|
|
|
61
|
|
|
echo sprintf( |
62
|
|
|
"-------------------------------\n%s [%s]\n\n%s\n", |
63
|
|
|
date("Y-m-d H:i:s"), |
64
|
|
|
Bot::getGitVersion() ?? '', |
65
|
|
|
$this->raw |
66
|
|
|
); |
67
|
|
|
if ($this->verbose) { |
68
|
|
|
$memory->echoMemory(true); |
69
|
|
|
} |
70
|
|
|
|
71
|
|
|
// initialise variables |
72
|
|
|
$this->log = []; |
73
|
|
|
$this->ouvrage = null; |
74
|
|
|
$this->notCosmetic = false; |
75
|
|
|
$this->major = false; |
76
|
|
|
|
77
|
|
|
|
78
|
|
|
try { |
79
|
|
|
$parse = TemplateParser::parseAllTemplateByName('ouvrage', $this->raw); |
80
|
|
|
$origin = $parse['ouvrage'][0]['model'] ?? null; |
81
|
|
|
} catch (Throwable $e) { |
82
|
|
|
echo sprintf("*** ERREUR impossible de transformer en modèle %s \n", $this->raw); |
83
|
|
|
continue; |
84
|
|
|
} |
85
|
|
|
|
86
|
|
|
if (!$origin instanceof OuvrageTemplate) { |
87
|
|
|
echo sprintf("*** ERREUR impossible de transformer en modèle %s \n", $this->raw); |
88
|
|
|
continue; |
89
|
|
|
} |
90
|
|
|
|
91
|
|
|
// Final optimizing (with online predictions) |
92
|
|
|
$optimizer = new OuvrageOptimize($origin, null); |
93
|
|
|
$optimizer->doTasks(); |
94
|
|
|
$this->ouvrage = $optimizer->getOuvrage(); |
95
|
|
|
$this->log = array_merge($this->log, $optimizer->getLog()); |
96
|
|
|
$this->notCosmetic = ($optimizer->notCosmetic || $this->notCosmetic); |
97
|
|
|
|
98
|
|
|
/** |
99
|
|
|
* RECHERCHE ONLINE |
100
|
|
|
*/ |
101
|
|
|
$isbn = $origin->getParam('isbn') ?? null; // avant mise en forme EAN>ISBN |
102
|
|
|
$isbn10 = $origin->getParam('isbn2') ?? $origin->getParam('isbn10') ?? null; |
103
|
|
|
if (!empty($isbn) |
104
|
|
|
&& empty($origin->getParam('isbn invalide')) |
105
|
|
|
&& empty($origin->getParam('isbn erroné')) |
106
|
|
|
) { |
107
|
|
|
$this->onlineIsbnSearch($isbn, $isbn10); |
108
|
|
|
} |
109
|
|
|
|
110
|
|
|
$this->sendCompleted(); |
111
|
|
|
unset($optimizer); |
112
|
|
|
unset($parse); |
113
|
|
|
unset($origin); |
114
|
|
|
} // END WHILE |
115
|
|
|
|
116
|
|
|
return true; |
117
|
|
|
} |
118
|
|
|
|
119
|
|
|
/** |
120
|
|
|
* Get raw string to complete from AMQP queue, SQL Select or file reading. |
121
|
|
|
* |
122
|
|
|
* @return string|null |
123
|
|
|
* @throws \Exception |
124
|
|
|
*/ |
125
|
|
|
private function getNewRaw(): ?string |
126
|
|
|
{ |
127
|
|
|
$raw = $this->queueAdapter->getNewRaw(); |
128
|
|
|
if (!$raw) { |
129
|
|
|
echo "STOP: no more queue to process \n"; |
130
|
|
|
throw new \Exception('no more queue to process'); |
131
|
|
|
} |
132
|
|
|
|
133
|
|
|
return $raw; |
134
|
|
|
} |
135
|
|
|
|
136
|
|
|
private function onlineIsbnSearch(string $isbn, ?string $isbn10 = null) |
137
|
|
|
{ |
138
|
|
|
online: |
139
|
|
|
if ($this->verbose) { |
140
|
|
|
echo "sleep 10...\n"; |
141
|
|
|
} |
142
|
|
|
sleep(10); |
143
|
|
|
|
144
|
|
|
try { |
145
|
|
|
if ($this->verbose) { |
146
|
|
|
dump('BIBLIO NAT FRANCE...'); |
147
|
|
|
} |
148
|
|
|
// BnF sait pas trouver un vieux livre (10) d'après ISBN-13... FACEPALM ! |
149
|
|
|
if ($isbn10) { |
150
|
|
|
$bnfOuvrage = OuvrageFactory::BnfFromIsbn($isbn10); |
151
|
|
|
sleep(2); |
152
|
|
|
} |
153
|
|
|
if (!$isbn10 || empty($bnfOuvrage) || empty($bnfOuvrage->getParam('titre'))) { |
154
|
|
|
$bnfOuvrage = OuvrageFactory::BnfFromIsbn($isbn); |
155
|
|
|
} |
156
|
|
|
if (isset($bnfOuvrage) and $bnfOuvrage instanceof OuvrageTemplate) { |
157
|
|
|
$this->completeOuvrage($bnfOuvrage); |
158
|
|
|
|
159
|
|
|
// Wikidata requests from $infos (ISBN/ISNI) |
160
|
|
|
if (!empty($bnfOuvrage->getInfos())) { |
161
|
|
|
if ($this->verbose) { |
162
|
|
|
dump('WIKIDATA...'); |
163
|
|
|
} |
164
|
|
|
$wdComplete = new Wikidata2Ouvrage(clone $bnfOuvrage); |
165
|
|
|
$this->completeOuvrage($wdComplete->getOuvrage()); |
166
|
|
|
} |
167
|
|
|
} |
168
|
|
|
} catch (Throwable $e) { |
169
|
|
|
echo sprintf( |
170
|
|
|
"*** ERREUR BnF Isbn Search %s %s %s \n", |
171
|
|
|
$e->getMessage(), |
172
|
|
|
$e->getFile(), |
173
|
|
|
$e->getLine() |
174
|
|
|
); |
175
|
|
|
} |
176
|
|
|
|
177
|
|
|
if (!isset($bnfOuvrage) || !$this->skipGoogle($bnfOuvrage)) { |
178
|
|
|
try { |
179
|
|
|
if ($this->verbose) { |
180
|
|
|
dump('GOOGLE...'); |
181
|
|
|
} |
182
|
|
|
$googleOuvrage = OuvrageFactory::GoogleFromIsbn($isbn); |
183
|
|
|
$this->completeOuvrage($googleOuvrage); |
184
|
|
|
} catch (Throwable $e) { |
185
|
|
|
echo "*** ERREUR GOOGLE Isbn Search ***".$e->getMessage()."\n"; |
186
|
|
|
throw $e; |
187
|
|
|
} |
188
|
|
|
} |
189
|
|
|
|
190
|
|
|
if (!isset($bnfOuvrage) && !isset($googleOuvrage)) { |
191
|
|
|
try { |
192
|
|
|
if ($this->verbose) { |
193
|
|
|
dump('OpenLibrary...'); |
194
|
|
|
} |
195
|
|
|
$openLibraryOuvrage = OuvrageFactory::OpenLibraryFromIsbn($isbn); |
196
|
|
|
if (!empty($openLibraryOuvrage)) { |
197
|
|
|
$this->completeOuvrage($openLibraryOuvrage); |
198
|
|
|
} |
199
|
|
|
} catch (Throwable $e) { |
200
|
|
|
echo '**** ERREUR OpenLibrary Isbn Search'; |
201
|
|
|
} |
202
|
|
|
} |
203
|
|
|
} |
204
|
|
|
|
205
|
|
|
// private function onlineQuerySearch(string $query) |
206
|
|
|
// { |
207
|
|
|
// echo "sleep 40..."; |
208
|
|
|
// sleep(20); |
209
|
|
|
// onlineQuerySearch: |
210
|
|
|
// |
211
|
|
|
// try { |
212
|
|
|
// dump('GOOGLE SEARCH...'); |
213
|
|
|
// // $googleOuvrage = OuvrageFactory::GoogleFromIsbn($isbn); |
214
|
|
|
// $adapter = new GoogleBooksAdapter(); |
215
|
|
|
// $data = $adapter->search('blabla'); |
216
|
|
|
// dump($data); |
217
|
|
|
// //die; |
218
|
|
|
// // return $import->getOuvrage(); |
219
|
|
|
// // $this->completeOuvrage($googleOuvrage); |
220
|
|
|
// } catch (Throwable $e) { |
221
|
|
|
// echo "*** ERREUR GOOGLE QuerySearch *** ".$e->getMessage()."\n"; |
222
|
|
|
// echo "sleep 30min"; |
223
|
|
|
// sleep(60 * 30); |
224
|
|
|
// echo "Wake up\n"; |
225
|
|
|
// goto onlineQuerySearch; |
226
|
|
|
// } |
227
|
|
|
// } |
228
|
|
|
|
229
|
|
|
private function completeOuvrage(OuvrageTemplate $onlineOuvrage) |
230
|
|
|
{ |
231
|
|
|
if ($this->verbose) { |
232
|
|
|
dump($onlineOuvrage->serialize(true)); |
233
|
|
|
} |
234
|
|
|
$optimizer = new OuvrageOptimize($onlineOuvrage); |
235
|
|
|
$onlineOptimized = ($optimizer)->doTasks()->getOuvrage(); |
236
|
|
|
|
237
|
|
|
$completer = new OuvrageComplete($this->ouvrage, $onlineOptimized); |
238
|
|
|
$this->ouvrage = $completer->getResult(); |
239
|
|
|
if ($this->verbose) { |
240
|
|
|
dump($completer->getLog()); |
241
|
|
|
} |
242
|
|
|
if ($completer->major) { |
243
|
|
|
$this->major = true; |
244
|
|
|
} |
245
|
|
|
$this->notCosmetic = ($completer->notCosmetic || $this->notCosmetic); |
246
|
|
|
$this->log = array_merge($this->log, $completer->getLog()); |
247
|
|
|
unset($optimizer); |
248
|
|
|
unset($completer); |
249
|
|
|
} |
250
|
|
|
|
251
|
|
|
private function sendCompleted() |
252
|
|
|
{ |
253
|
|
|
$isbn13 = $this->ouvrage->getParam('isbn') ?? null; |
254
|
|
|
|
255
|
|
|
$finalData = [ |
256
|
|
|
// 'page' => |
257
|
|
|
'raw' => $this->raw, |
258
|
|
|
'opti' => $this->serializeFinalOpti(), |
259
|
|
|
'optidate' => date("Y-m-d H:i:s"), |
260
|
|
|
'modifs' => mb_substr(implode(',', $this->log), 0, 250), |
261
|
|
|
'notcosmetic' => ($this->notCosmetic) ? 1 : 0, |
262
|
|
|
'major' => ($this->major) ? 1 : 0, |
263
|
|
|
'isbn' => substr($isbn13, 0, 20), |
264
|
|
|
'version' => Bot::getGitVersion() ?? null, |
265
|
|
|
]; |
266
|
|
|
if ($this->verbose) { |
267
|
|
|
dump($finalData); |
268
|
|
|
} |
269
|
|
|
// Json ? |
270
|
|
|
$result = $this->queueAdapter->sendCompletedData($finalData); |
271
|
|
|
|
272
|
|
|
dump($result); // bool |
273
|
|
|
} |
274
|
|
|
|
275
|
|
|
/** |
276
|
|
|
* Final serialization of the completed OuvrageTemplate. |
277
|
|
|
* |
278
|
|
|
* @return string |
279
|
|
|
*/ |
280
|
|
|
private function serializeFinalOpti(): string |
281
|
|
|
{ |
282
|
|
|
$finalOpti = $this->ouvrage->serialize(true); |
283
|
|
|
$finalOpti = Normalizer::normalize($finalOpti); |
284
|
|
|
|
285
|
|
|
return $finalOpti; |
286
|
|
|
} |
287
|
|
|
|
288
|
|
|
private function skipGoogle($bnfOuvrage): bool |
289
|
|
|
{ |
290
|
|
|
if ($bnfOuvrage instanceOf OuvrageTemplate |
291
|
|
|
&& !empty($bnfOuvrage->getParam('titre')) |
292
|
|
|
&& (!empty($this->ouvrage->getParam('lire en ligne')) |
293
|
|
|
|| !empty($this->ouvrage->getParam('présentation en ligne'))) |
294
|
|
|
) { |
295
|
|
|
return true; |
296
|
|
|
} |
297
|
|
|
|
298
|
|
|
return false; |
299
|
|
|
} |
300
|
|
|
} |
301
|
|
|
|