|
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\Domain\Transformers; |
|
11
|
|
|
|
|
12
|
|
|
use App\Domain\InfrastructurePorts\GoogleApiQuotaInterface; |
|
13
|
|
|
use App\Domain\InfrastructurePorts\GoogleBooksInterface; |
|
14
|
|
|
use App\Domain\Models\Wiki\OuvrageTemplate; |
|
15
|
|
|
use App\Domain\Publisher\GoogleBookMapper; |
|
16
|
|
|
use App\Domain\Publisher\GoogleBooksUtil; |
|
17
|
|
|
use App\Domain\Utils\NumberUtil; |
|
18
|
|
|
use App\Domain\Utils\WikiTextUtil; |
|
19
|
|
|
use App\Domain\WikiOptimizer\OptimizerFactory; |
|
20
|
|
|
use App\Domain\WikiTemplateFactory; |
|
21
|
|
|
use App\Infrastructure\Monitor\NullLogger; |
|
22
|
|
|
use DomainException; |
|
23
|
|
|
use Exception; |
|
24
|
|
|
use Psr\Log\LoggerInterface; |
|
25
|
|
|
use Scriptotek\GoogleBooks\Volume; |
|
26
|
|
|
use Throwable; |
|
27
|
|
|
|
|
28
|
|
|
/** |
|
29
|
|
|
* TODO REFAC : duplicate, extract methods in trait or in AbstractRefBotWorker + ExternBotWorker |
|
30
|
|
|
* -- |
|
31
|
|
|
* Transform <ref>https://books.google...</ref> to <ref>{{Ouvrage|...}}.</ref> |
|
32
|
|
|
* in an article wikitext. |
|
33
|
|
|
*/ |
|
34
|
|
|
class GoogleTransformer |
|
35
|
|
|
{ |
|
36
|
|
|
final public const SLEEP_GOOGLE_API_INTERVAL = 5; |
|
37
|
|
|
|
|
38
|
|
|
/** |
|
39
|
|
|
* @var OuvrageTemplate[] |
|
40
|
|
|
*/ |
|
41
|
|
|
protected array $cacheOuvrageTemplate = []; |
|
42
|
|
|
|
|
43
|
|
|
public function __construct( |
|
44
|
|
|
protected readonly GoogleApiQuotaInterface $quota, |
|
45
|
|
|
protected readonly GoogleBooksInterface $googleBooksAdapter, |
|
46
|
|
|
protected LoggerInterface $logger = new NullLogger() |
|
47
|
|
|
) |
|
48
|
|
|
{ |
|
49
|
|
|
} |
|
50
|
|
|
|
|
51
|
|
|
/** |
|
52
|
|
|
* Process page wikitext. Return wikitext with the <ref> or bullets list converted. |
|
53
|
|
|
*/ |
|
54
|
|
|
public function process(string $text): string |
|
55
|
|
|
{ |
|
56
|
|
|
if ($this->quota->isQuotaReached()) { |
|
|
|
|
|
|
57
|
|
|
$this->logger->debug('Quota Google atteint'); |
|
58
|
|
|
throw new DomainException('Quota Google atteint'); |
|
59
|
|
|
} |
|
60
|
|
|
|
|
61
|
|
|
$refsData = $this->extractAllGoogleRefs($text); |
|
62
|
|
|
$this->logger->debug('GoogleTransformer: refs found: ' . count($refsData)); |
|
63
|
|
|
if ($refsData !== []) { |
|
64
|
|
|
$text = $this->processRef($text, $refsData); |
|
65
|
|
|
} |
|
66
|
|
|
|
|
67
|
|
|
$links = $this->extractGoogleExternalBullets($text); |
|
68
|
|
|
$this->logger->debug('GoogleTransformer: links found: ' . count($links)); |
|
69
|
|
|
if ($links !== []) { |
|
70
|
|
|
$text = $this->processExternLinks($text, $links); |
|
71
|
|
|
} |
|
72
|
|
|
|
|
73
|
|
|
return $text; |
|
74
|
|
|
} |
|
75
|
|
|
|
|
76
|
|
|
/** |
|
77
|
|
|
* Extract all <ref>/{ref} with only GoogleBooks URL. |
|
78
|
|
|
* Todo : supprimer point final URL |
|
79
|
|
|
* |
|
80
|
|
|
* @param string $text Page wikitext |
|
81
|
|
|
* |
|
82
|
|
|
* @return array [0 => ['<ref>http...</ref>', 'http://'], 1 => ...] |
|
83
|
|
|
*/ |
|
84
|
|
|
public function extractAllGoogleRefs(string $text): array |
|
85
|
|
|
{ |
|
86
|
|
|
// <ref>...</ref> or {{ref|...}} |
|
87
|
|
|
// GoogleLivresTemplate::GOOGLEBOOK_URL_PATTERN |
|
88
|
|
|
if (preg_match_all( |
|
89
|
|
|
'#(?:<ref[^>]*>|{{ref\|) ?(' . GoogleBooksUtil::GOOGLEBOOKS_START_URL_PATTERN . '[^>\]} \n]+) ?(?:</ref>|}})#i', |
|
90
|
|
|
$text, |
|
91
|
|
|
$matches, |
|
92
|
|
|
PREG_SET_ORDER |
|
93
|
|
|
) |
|
94
|
|
|
) { |
|
95
|
|
|
return $matches; |
|
96
|
|
|
} |
|
97
|
|
|
|
|
98
|
|
|
return []; |
|
99
|
|
|
} |
|
100
|
|
|
|
|
101
|
|
|
protected function processRef(string $text, array $refsData): string |
|
102
|
|
|
{ |
|
103
|
|
|
foreach ($refsData as $ref) { |
|
104
|
|
|
$this->logger->info('Process GoogleBooks ref: ' . $ref[1]); |
|
105
|
|
|
if ($this->quota->isQuotaReached()) { |
|
106
|
|
|
$this->logger->debug('Quota Google atteint'); |
|
107
|
|
|
throw new DomainException('Quota Google atteint'); |
|
108
|
|
|
} |
|
109
|
|
|
try { |
|
110
|
|
|
$citation = $this->convertGBurl2OuvrageCitation(WikiTextUtil::stripFinalPoint($ref[1])); |
|
111
|
|
|
sleep(2); |
|
112
|
|
|
} catch (Throwable $e) { |
|
113
|
|
|
$this->logger->debug("Exception " . $e->getMessage()); |
|
114
|
|
|
continue; |
|
115
|
|
|
} |
|
116
|
|
|
|
|
117
|
|
|
// ajout point final pour référence |
|
118
|
|
|
$citation .= '.'; |
|
119
|
|
|
|
|
120
|
|
|
$newRef = str_replace($ref[1], $citation, (string)$ref[0]); |
|
121
|
|
|
$this->logger->info($newRef); |
|
122
|
|
|
|
|
123
|
|
|
$text = str_replace($ref[0], $newRef, $text); |
|
124
|
|
|
|
|
125
|
|
|
$this->logger->notice("sleep " . self::SLEEP_GOOGLE_API_INTERVAL); |
|
126
|
|
|
sleep(self::SLEEP_GOOGLE_API_INTERVAL); |
|
127
|
|
|
} |
|
128
|
|
|
|
|
129
|
|
|
return $text; |
|
130
|
|
|
} |
|
131
|
|
|
|
|
132
|
|
|
/** |
|
133
|
|
|
* TODO : extract. TODO private ? |
|
134
|
|
|
* Convert GoogleBooks URL to wiki-template {ouvrage} citation. |
|
135
|
|
|
* Need GoogleBooksAdapter injection. |
|
136
|
|
|
* @throws Throwable |
|
137
|
|
|
*/ |
|
138
|
|
|
public function convertGBurl2OuvrageCitation(string $url): string |
|
139
|
|
|
{ |
|
140
|
|
|
if (!GoogleBooksUtil::isGoogleBookURL($url)) { |
|
141
|
|
|
throw new DomainException('Pas de URL Google Books'); |
|
142
|
|
|
} |
|
143
|
|
|
|
|
144
|
|
|
$gooDat = GoogleBooksUtil::parseGoogleBookQuery($url); |
|
145
|
|
|
if (empty($gooDat['isbn']) && empty($gooDat['id'])) { |
|
146
|
|
|
throw new DomainException('Pas de ISBN ou ID Google Books'); |
|
147
|
|
|
} |
|
148
|
|
|
|
|
149
|
|
|
try { |
|
150
|
|
|
$identifiant = $gooDat['id'] ?? $gooDat['isbn']; |
|
151
|
|
|
$isISBN = !empty($gooDat['isbn']); |
|
152
|
|
|
$ouvrage = $this->generateOuvrageFromGoogleData($identifiant, $isISBN); |
|
153
|
|
|
} catch (Throwable $e) { |
|
154
|
|
|
// ID n'existe pas sur Google Books |
|
155
|
|
|
if (strpos($e->getMessage(), '"message": "The volume ID could n')) { |
|
156
|
|
|
return sprintf( |
|
157
|
|
|
'{{lien brisé |url= %s |titre=%s |brisé le=%s}}', |
|
158
|
|
|
$url, |
|
159
|
|
|
'Ouvrage inexistant sur Google Books', |
|
160
|
|
|
date('d-m-Y') |
|
161
|
|
|
); |
|
162
|
|
|
} |
|
163
|
|
|
throw $e; |
|
164
|
|
|
} |
|
165
|
|
|
|
|
166
|
|
|
$cleanUrl = GoogleBooksUtil::simplifyGoogleUrl($url); |
|
167
|
|
|
$ouvrage->unsetParam('présentation en ligne'); |
|
168
|
|
|
$ouvrage->setParam('lire en ligne', $cleanUrl); |
|
169
|
|
|
$ouvrage->userSeparator = ' |'; |
|
170
|
|
|
|
|
171
|
|
|
// Si titre absent |
|
172
|
|
|
if (!$ouvrage->hasParamValue('titre')) { |
|
173
|
|
|
throw new DomainException("Ouvrage sans titre (data Google?)"); |
|
174
|
|
|
} |
|
175
|
|
|
|
|
176
|
|
|
// Google page => 'passage' |
|
177
|
|
|
if (!empty($gooDat['pg'])) { |
|
178
|
|
|
// Exclusion de page=1, page=2 (vue par défaut sur Google Book) |
|
179
|
|
|
if (preg_match('#(?:PA|PT)(\d+)$#', (string)$gooDat['pg'], $matches) && (int)$matches[1] >= 3) { |
|
180
|
|
|
$page = $matches[1]; |
|
181
|
|
|
} |
|
182
|
|
|
// conversion chiffres Romain pour PR |
|
183
|
|
|
// Exclusion de page=1, page=2 (vue par défaut sur Google Book) |
|
184
|
|
|
if (preg_match('#PR(\d+)$#', (string)$gooDat['pg'], $matches) && (int)$matches[1] >= 3) { |
|
185
|
|
|
$page = NumberUtil::arab2roman((int)$matches[1], true); |
|
186
|
|
|
} |
|
187
|
|
|
|
|
188
|
|
|
if (!empty($page)) { |
|
189
|
|
|
$ouvrage->setParam('passage', $page); |
|
190
|
|
|
// ajout commentaire '<!-- utile? -->' ? |
|
191
|
|
|
} |
|
192
|
|
|
} |
|
193
|
|
|
|
|
194
|
|
|
$optimizer = OptimizerFactory::fromTemplate($ouvrage, null, $this->logger); |
|
195
|
|
|
$optimizer->doTasks(); |
|
196
|
|
|
$ouvrage2 = $optimizer->getOptiTemplate(); |
|
197
|
|
|
|
|
198
|
|
|
return $ouvrage2->serialize(); |
|
199
|
|
|
} |
|
200
|
|
|
|
|
201
|
|
|
/** |
|
202
|
|
|
* todo: move (injection) to other class. |
|
203
|
|
|
* Generate wiki-template {ouvrage} from GoogleBook ID. |
|
204
|
|
|
* |
|
205
|
|
|
* @param string $id GoogleBooks ID |
|
206
|
|
|
* |
|
207
|
|
|
* @throws Exception |
|
208
|
|
|
*/ |
|
209
|
|
|
protected function generateOuvrageFromGoogleData(string $id, ?bool $isISBN = false): OuvrageTemplate |
|
210
|
|
|
{ |
|
211
|
|
|
// return cached OuvrageTemplate |
|
212
|
|
|
if (!$isISBN && isset($this->cacheOuvrageTemplate[$id])) { |
|
213
|
|
|
return clone $this->cacheOuvrageTemplate[$id]; |
|
214
|
|
|
} |
|
215
|
|
|
|
|
216
|
|
|
// Get Google data by ID ZvhBAAAAcAAJ |
|
217
|
|
|
$volume = $isISBN === true |
|
218
|
|
|
? $this->googleBooksAdapter->getDataByIsbn($id) |
|
|
|
|
|
|
219
|
|
|
: $this->googleBooksAdapter->getDataByGoogleId($id); |
|
|
|
|
|
|
220
|
|
|
if (!$volume instanceof Volume) { |
|
221
|
|
|
throw new DomainException('googleBooks Volume not found for that GB-id/isbn'); |
|
222
|
|
|
} |
|
223
|
|
|
|
|
224
|
|
|
$mapper = new GoogleBookMapper(); // todo inject |
|
225
|
|
|
$mapper->mapLanguageData(true); |
|
226
|
|
|
$data = $mapper->process($volume); |
|
227
|
|
|
|
|
228
|
|
|
// Generate wiki-template {ouvrage} |
|
229
|
|
|
$ouvrage = WikiTemplateFactory::create('ouvrage'); |
|
230
|
|
|
$ouvrage->hydrate($data); |
|
231
|
|
|
$ouvrage->setParam('consulté le', date('d-m-Y')); |
|
232
|
|
|
|
|
233
|
|
|
// cache |
|
234
|
|
|
$this->cacheOuvrageTemplate[$id] = clone $ouvrage; |
|
235
|
|
|
|
|
236
|
|
|
return $ouvrage; |
|
|
|
|
|
|
237
|
|
|
} |
|
238
|
|
|
|
|
239
|
|
|
/** |
|
240
|
|
|
* todo move |
|
241
|
|
|
*/ |
|
242
|
|
|
public function extractGoogleExternalBullets(string $text): array |
|
243
|
|
|
{ |
|
244
|
|
|
// match "* https://books.google.fr/..." |
|
245
|
|
|
if (preg_match_all( |
|
246
|
|
|
'#^\* *(' . GoogleBooksUtil::GOOGLEBOOKS_START_URL_PATTERN . '[^ <{\]}\n\r]+) *$#im', |
|
247
|
|
|
$text, |
|
248
|
|
|
$matches, |
|
249
|
|
|
PREG_SET_ORDER |
|
250
|
|
|
) |
|
251
|
|
|
) { |
|
252
|
|
|
return $matches; |
|
253
|
|
|
} |
|
254
|
|
|
|
|
255
|
|
|
return []; |
|
256
|
|
|
} |
|
257
|
|
|
|
|
258
|
|
|
/** |
|
259
|
|
|
* TODO Duplication du dessus... |
|
260
|
|
|
* @return string|string[] |
|
261
|
|
|
* @throws Throwable |
|
262
|
|
|
*/ |
|
263
|
|
|
protected function processExternLinks(string $text, array $links): string|array |
|
264
|
|
|
{ |
|
265
|
|
|
foreach ($links as $pattern) { |
|
266
|
|
|
$this->logger->info('Process links: ' . $pattern[1]); |
|
267
|
|
|
if ($this->quota->isQuotaReached()) { |
|
268
|
|
|
$this->logger->debug('Quota Google atteint'); |
|
269
|
|
|
throw new DomainException('Quota Google atteint'); |
|
270
|
|
|
} |
|
271
|
|
|
try { |
|
272
|
|
|
$citation = $this->convertGBurl2OuvrageCitation(WikiTextUtil::stripFinalPoint($pattern[1])); |
|
273
|
|
|
} catch (Exception $e) { |
|
274
|
|
|
$this->logger->debug("Exception " . $e->getMessage()); |
|
275
|
|
|
continue; |
|
276
|
|
|
} |
|
277
|
|
|
|
|
278
|
|
|
// todo : ajout point final pour référence ??? |
|
279
|
|
|
$citation .= '.'; |
|
280
|
|
|
|
|
281
|
|
|
$newRef = str_replace($pattern[1], $citation, (string)$pattern[0]); |
|
282
|
|
|
$this->logger->info($newRef); |
|
283
|
|
|
|
|
284
|
|
|
$text = str_replace($pattern[0], $newRef, $text); |
|
285
|
|
|
|
|
286
|
|
|
$this->logger->info("Sleep " . self::SLEEP_GOOGLE_API_INTERVAL); |
|
287
|
|
|
sleep(self::SLEEP_GOOGLE_API_INTERVAL); |
|
288
|
|
|
} |
|
289
|
|
|
|
|
290
|
|
|
return $text; |
|
291
|
|
|
} |
|
292
|
|
|
|
|
293
|
|
|
} |
|
294
|
|
|
|
Dispositif /
Wikibot
Loading history...