|
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\Examples; |
|
11
|
|
|
|
|
12
|
|
|
use App\Application\Bot; |
|
13
|
|
|
use App\Application\Memory; |
|
14
|
|
|
use App\Application\QueueInterface; |
|
15
|
|
|
use App\Domain\Models\Wiki\OuvrageTemplate; |
|
16
|
|
|
use App\Domain\OuvrageComplete; |
|
17
|
|
|
use App\Domain\OuvrageFactory; |
|
18
|
|
|
use App\Domain\OuvrageOptimize; |
|
19
|
|
|
use App\Domain\Utils\TemplateParser; |
|
20
|
|
|
use App\Infrastructure\DbAdapter; |
|
21
|
|
|
use App\Infrastructure\GoogleBooksAdapter; |
|
22
|
|
|
use Throwable; |
|
23
|
|
|
|
|
24
|
|
|
include __DIR__.'/../myBootstrap.php'; |
|
25
|
|
|
|
|
26
|
|
|
$dirty = new CompleteProcess(new DbAdapter()); |
|
27
|
|
|
$dirty->run(); |
|
28
|
|
|
|
|
29
|
|
|
/** |
|
30
|
|
|
* Class CompleteProcess |
|
31
|
|
|
*/ |
|
32
|
|
|
class CompleteProcess |
|
33
|
|
|
{ |
|
34
|
|
|
/** |
|
35
|
|
|
* @var QueueInterface |
|
36
|
|
|
*/ |
|
37
|
|
|
private $queueAdapter; |
|
38
|
|
|
/** |
|
39
|
|
|
* @var string |
|
40
|
|
|
*/ |
|
41
|
|
|
private $raw = ''; |
|
42
|
|
|
private $log = []; |
|
43
|
|
|
private $notCosmetic = false; |
|
44
|
|
|
private $major = false; |
|
45
|
|
|
/** |
|
46
|
|
|
* @var OuvrageTemplate |
|
47
|
|
|
*/ |
|
48
|
|
|
private $ouvrage; |
|
49
|
|
|
private $continue = true; |
|
|
|
|
|
|
50
|
|
|
|
|
51
|
|
|
public function __construct(QueueInterface $queueAdapter) |
|
52
|
|
|
{ |
|
53
|
|
|
$this->queueAdapter = $queueAdapter; |
|
54
|
|
|
} |
|
55
|
|
|
|
|
56
|
|
|
public function run(?int $limit = 10000) |
|
57
|
|
|
{ |
|
58
|
|
|
$memory = new Memory(); |
|
59
|
|
|
while ($limit > 0) { |
|
60
|
|
|
$limit--; |
|
61
|
|
|
sleep(1); |
|
62
|
|
|
$this->raw = $this->getNewRaw(); |
|
63
|
|
|
|
|
64
|
|
|
echo sprintf( |
|
65
|
|
|
"-------------------------------\n%s [%s]\n\n%s\n", |
|
66
|
|
|
date("Y-m-d H:i:s"), |
|
67
|
|
|
Bot::getGitVersion() ?? '', |
|
68
|
|
|
$this->raw |
|
69
|
|
|
); |
|
70
|
|
|
$memory->echoMemory(true); |
|
71
|
|
|
|
|
72
|
|
|
// initialise variables |
|
73
|
|
|
$this->log = []; |
|
74
|
|
|
$this->ouvrage = null; |
|
75
|
|
|
$this->notCosmetic = false; |
|
76
|
|
|
$this->major = false; |
|
77
|
|
|
|
|
78
|
|
|
|
|
79
|
|
|
try { |
|
80
|
|
|
$parse = TemplateParser::parseAllTemplateByName('ouvrage', $this->raw); |
|
81
|
|
|
$origin = $parse['ouvrage'][0]['model'] ?? null; |
|
82
|
|
|
} catch (Throwable $e) { |
|
83
|
|
|
echo sprintf("*** ERREUR impossible de transformer en modèle %s \n", $this->raw); |
|
84
|
|
|
continue; |
|
85
|
|
|
} |
|
86
|
|
|
|
|
87
|
|
|
if (!$origin instanceof OuvrageTemplate) { |
|
88
|
|
|
echo sprintf("*** ERREUR impossible de transformer en modèle %s \n", $this->raw); |
|
89
|
|
|
continue; |
|
90
|
|
|
} |
|
91
|
|
|
|
|
92
|
|
|
// Final optimizing (with online predictions) |
|
93
|
|
|
$optimizer = new OuvrageOptimize($origin, null, true); |
|
94
|
|
|
$optimizer->doTasks(); |
|
95
|
|
|
$this->ouvrage = $optimizer->getOuvrage(); |
|
96
|
|
|
$this->log = array_merge($this->log, $optimizer->getLog()); |
|
97
|
|
|
$this->notCosmetic = ($optimizer->notCosmetic || $this->notCosmetic); |
|
98
|
|
|
|
|
99
|
|
|
/** |
|
100
|
|
|
* RECHERCHE ONLINE |
|
101
|
|
|
*/ |
|
102
|
|
|
$isbn = $origin->getParam('isbn') ?? null; // avant mise en forme EAN>ISBN |
|
103
|
|
|
if (!empty($isbn)) { |
|
104
|
|
|
$this->onlineIsbnSearch($isbn); |
|
105
|
|
|
} |
|
106
|
|
|
|
|
107
|
|
|
$this->sendCompleted(); |
|
108
|
|
|
unset($optimizer); |
|
109
|
|
|
unset($parse); |
|
110
|
|
|
unset($origin); |
|
111
|
|
|
} // END WHILE |
|
112
|
|
|
|
|
113
|
|
|
return true; |
|
114
|
|
|
} |
|
115
|
|
|
|
|
116
|
|
|
/** |
|
117
|
|
|
* Get raw string to complete from AMQP queue, SQL Select or file reading. |
|
118
|
|
|
* |
|
119
|
|
|
* @return string|null |
|
120
|
|
|
*/ |
|
121
|
|
|
private function getNewRaw(): ?string |
|
122
|
|
|
{ |
|
123
|
|
|
$raw = $this->queueAdapter->getNewRaw(); |
|
124
|
|
|
if (!$raw) { |
|
125
|
|
|
echo "STOP: no more queue to process \n"; |
|
126
|
|
|
exit; |
|
|
|
|
|
|
127
|
|
|
} |
|
128
|
|
|
|
|
129
|
|
|
return $raw; |
|
130
|
|
|
} |
|
131
|
|
|
|
|
132
|
|
|
private function onlineIsbnSearch(string $isbn) |
|
133
|
|
|
{ |
|
134
|
|
|
online: |
|
135
|
|
|
echo "sleep 40...\n"; |
|
136
|
|
|
sleep(40); |
|
137
|
|
|
|
|
138
|
|
|
try { |
|
139
|
|
|
dump('BIBLIO NAT FRANCE...'); |
|
140
|
|
|
$bnfOuvrage = OuvrageFactory::BnfFromIsbn($isbn); |
|
141
|
|
|
$this->completeOuvrage($bnfOuvrage); |
|
142
|
|
|
} catch (Throwable $e) { |
|
143
|
|
|
echo "*** ERREUR BnF Isbn Search".$e->getMessage()."\n"; |
|
144
|
|
|
echo "sleep 5min\n"; |
|
145
|
|
|
sleep(60 * 5); |
|
146
|
|
|
goto online; |
|
147
|
|
|
} |
|
148
|
|
|
|
|
149
|
|
|
try { |
|
150
|
|
|
if (!$this->skipGoogle()) { |
|
151
|
|
|
dump('GOOGLE...'); |
|
152
|
|
|
$googleOuvrage = OuvrageFactory::GoogleFromIsbn($isbn); |
|
153
|
|
|
$this->completeOuvrage($googleOuvrage); |
|
154
|
|
|
} |
|
155
|
|
|
} catch (Throwable $e) { |
|
156
|
|
|
echo "*** ERREUR GOOGLE Isbn Search ***".$e->getMessage()."\n"; |
|
157
|
|
|
if (strpos($e->getMessage(), 'Daily Limit Exceeded') !== true) { |
|
158
|
|
|
echo "sleep 1h\n"; |
|
159
|
|
|
sleep(60 * 60); |
|
160
|
|
|
echo "Try goto\n"; |
|
161
|
|
|
goto online; |
|
162
|
|
|
} |
|
163
|
|
|
} |
|
164
|
|
|
|
|
165
|
|
|
|
|
166
|
|
|
try { |
|
167
|
|
|
dump('OpenLibrary...'); |
|
168
|
|
|
$openLibraryOuvrage = OuvrageFactory::OpenLibraryFromIsbn($isbn); |
|
169
|
|
|
if (!empty($openLibraryOuvrage)) { |
|
170
|
|
|
$this->completeOuvrage($openLibraryOuvrage); |
|
171
|
|
|
} |
|
172
|
|
|
} catch (Throwable $e) { |
|
173
|
|
|
echo '**** ERREUR OpenLibrary Isbn Search'; |
|
174
|
|
|
} |
|
175
|
|
|
} |
|
176
|
|
|
|
|
177
|
|
|
private function onlineQuerySearch(string $query) |
|
|
|
|
|
|
178
|
|
|
{ |
|
179
|
|
|
echo "sleep 40..."; |
|
180
|
|
|
sleep(40); |
|
181
|
|
|
onlineQuerySearch: |
|
182
|
|
|
|
|
183
|
|
|
try { |
|
184
|
|
|
dump('GOOGLE SEARCH...'); |
|
185
|
|
|
// $googleOuvrage = OuvrageFactory::GoogleFromIsbn($isbn); |
|
186
|
|
|
$adapter = new GoogleBooksAdapter(); |
|
187
|
|
|
$data = $adapter->search('blabla'); |
|
188
|
|
|
dump($data); |
|
189
|
|
|
die; |
|
|
|
|
|
|
190
|
|
|
// return $import->getOuvrage(); |
|
191
|
|
|
// $this->completeOuvrage($googleOuvrage); |
|
192
|
|
|
} catch (Throwable $e) { |
|
193
|
|
|
echo "*** ERREUR GOOGLE QuerySearch *** ".$e->getMessage()."\n"; |
|
194
|
|
|
sleep(60 * 30); |
|
195
|
|
|
goto onlineQuerySearch; |
|
196
|
|
|
} |
|
197
|
|
|
} |
|
198
|
|
|
|
|
199
|
|
|
private function completeOuvrage(OuvrageTemplate $onlineOuvrage) |
|
200
|
|
|
{ |
|
201
|
|
|
dump($onlineOuvrage->serialize(true)); |
|
202
|
|
|
$optimizer = new OuvrageOptimize($onlineOuvrage); |
|
203
|
|
|
$onlineOptimized = ($optimizer)->doTasks()->getOuvrage(); |
|
204
|
|
|
|
|
205
|
|
|
$completer = new OuvrageComplete($this->ouvrage, $onlineOptimized); |
|
206
|
|
|
$this->ouvrage = $completer->getResult(); |
|
207
|
|
|
dump($completer->getLog()); |
|
208
|
|
|
if ($completer->major) { |
|
209
|
|
|
$this->major = true; |
|
210
|
|
|
} |
|
211
|
|
|
$this->notCosmetic = ($completer->notCosmetic || $this->notCosmetic); |
|
212
|
|
|
$this->log = array_merge($this->log, $completer->getLog()); |
|
213
|
|
|
unset($optimizer); |
|
214
|
|
|
unset($completer); |
|
215
|
|
|
} |
|
216
|
|
|
|
|
217
|
|
|
private function sendCompleted() |
|
218
|
|
|
{ |
|
219
|
|
|
$isbn13 = $this->ouvrage->getParam('isbn') ?? null; |
|
220
|
|
|
|
|
221
|
|
|
$finalData = [ |
|
222
|
|
|
// 'page' => |
|
223
|
|
|
'raw' => $this->raw, |
|
224
|
|
|
'opti' => $this->serializeFinalOpti(), |
|
225
|
|
|
'optidate' => date("Y-m-d H:i:s"), |
|
226
|
|
|
'modifs' => mb_substr(implode(',', $this->log), 0, 250), |
|
227
|
|
|
'notcosmetic' => ($this->notCosmetic) ? 1 : 0, |
|
228
|
|
|
'major' => ($this->major) ? 1 : 0, |
|
229
|
|
|
'isbn' => substr($isbn13, 0, 20), |
|
230
|
|
|
'version' => Bot::getGitVersion() ?? null, |
|
231
|
|
|
]; |
|
232
|
|
|
dump($finalData); |
|
233
|
|
|
// Json ? |
|
234
|
|
|
$result = $this->queueAdapter->sendCompletedData($finalData); |
|
235
|
|
|
dump($result); // bool |
|
236
|
|
|
} |
|
237
|
|
|
|
|
238
|
|
|
/** |
|
239
|
|
|
* Final serialization of the completed OuvrageTemplate. |
|
240
|
|
|
* Delete 'langue=fr' |
|
241
|
|
|
* |
|
242
|
|
|
* @return string |
|
243
|
|
|
* @throws \Exception |
|
244
|
|
|
*/ |
|
245
|
|
|
private function serializeFinalOpti(): string |
|
246
|
|
|
{ |
|
247
|
|
|
$ouvrage = $this->ouvrage; |
|
248
|
|
|
$langue = $ouvrage->getParam('langue'); |
|
249
|
|
|
if ($langue && 'fr' === $langue) { |
|
250
|
|
|
$ouvrage->unsetParam('langue'); |
|
251
|
|
|
} |
|
252
|
|
|
|
|
253
|
|
|
return $ouvrage->serialize(true); |
|
254
|
|
|
} |
|
255
|
|
|
|
|
256
|
|
|
private function skipGoogle(): bool |
|
257
|
|
|
{ |
|
258
|
|
|
return false; |
|
259
|
|
|
} |
|
260
|
|
|
} |
|
261
|
|
|
|