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\Models\Wiki; |
11
|
|
|
|
12
|
|
|
use App\Domain\Publisher\GoogleBooksTrait; |
13
|
|
|
use App\Domain\Utils\ArrayProcessTrait; |
14
|
|
|
use DomainException; |
15
|
|
|
use Exception; |
16
|
|
|
|
17
|
|
|
/** |
18
|
|
|
* https://fr.wikipedia.org/wiki/Mod%C3%A8le:Google_Livres |
19
|
|
|
* Le premier paramètre (ou id) est obligatoire. |
20
|
|
|
* Le deuxième (ou titre) est requis si on ne veut pas fabriquer le lien brut (inclusion {{ouvrage}} 'Lire en ligne') |
21
|
|
|
* Class GoogleLivresTemplate. |
22
|
|
|
*/ |
23
|
|
|
class GoogleLivresTemplate extends AbstractWikiTemplate |
24
|
|
|
{ |
25
|
|
|
use ArrayProcessTrait, GoogleBooksTrait; |
26
|
|
|
|
27
|
|
|
/** |
28
|
|
|
* todo utile ici ? |
29
|
|
|
*/ |
30
|
|
|
const DEFAULT_GOOGLEBOOKS_URL = 'https://books.google.com/books'; |
31
|
|
|
|
32
|
|
|
/** |
33
|
|
|
* todo URL avec fin |
34
|
|
|
*/ |
35
|
|
|
const GOOGLEBOOKS_START_URL_PATTERN = 'https?://(?:books|play)\.google\.[a-z\.]{2,6}/(?:books)?(?:books/[^\?]+\.html)?(?:/reader)?\?(?:[a-zA-Z=&]+&)?id='; |
36
|
|
|
|
37
|
|
|
const TRACKING_PARAMETERS |
38
|
|
|
= [ |
39
|
|
|
'xtor', |
40
|
|
|
'ved', |
41
|
|
|
'ots', |
42
|
|
|
'sig', |
43
|
|
|
'source', |
44
|
|
|
'utm_source', |
45
|
|
|
'utm_medium', |
46
|
|
|
'utm_campaign', |
47
|
|
|
'utm_term', |
48
|
|
|
'utm_content', |
49
|
|
|
]; |
50
|
|
|
|
51
|
|
|
const WIKITEMPLATE_NAME = 'Google Livres'; |
52
|
|
|
|
53
|
|
|
const REQUIRED_PARAMETERS = ['id']; |
54
|
|
|
|
55
|
|
|
const MINIMUM_PARAMETERS = ['id' => '']; |
56
|
|
|
|
57
|
|
|
const PARAM_ALIAS |
58
|
|
|
= [ |
59
|
|
|
'1' => 'id', |
60
|
|
|
'2' => 'titre', |
61
|
|
|
'surligné' => 'surligne', |
62
|
|
|
'BuchID' => 'id', |
63
|
|
|
]; |
64
|
|
|
|
65
|
|
|
/** |
66
|
|
|
* @var array |
67
|
|
|
*/ |
68
|
|
|
protected $parametersByOrder |
69
|
|
|
= ['id', 'titre', 'couv', 'page', 'romain', 'page autre', 'surligne']; |
70
|
|
|
|
71
|
|
|
/** |
72
|
|
|
* Create {Google Book} from URL. |
73
|
|
|
* See also https://fr.wikipedia.org/wiki/Utilisateur:Jack_ma/GB |
74
|
|
|
* https://stackoverflow.com/questions/11584551/need-information-on-query-parameters-for-google-books-e-g-difference-between-d. |
75
|
|
|
* |
76
|
|
|
* @param string $url |
77
|
|
|
* |
78
|
|
|
* @return GoogleLivresTemplate|null |
79
|
|
|
* @throws Exception |
80
|
|
|
*/ |
81
|
|
|
public static function createFromURL(string $url): ?self |
82
|
|
|
{ |
83
|
|
|
if (!self::isGoogleBookURL($url)) { |
84
|
|
|
throw new DomainException('not a Google Book URL'); |
85
|
|
|
} |
86
|
|
|
$gooDat = self::parseGoogleBookQuery($url); |
87
|
|
|
|
88
|
|
|
if (empty($gooDat['id'])) { |
89
|
|
|
throw new DomainException("no GoogleBook 'id' in URL"); |
90
|
|
|
} |
91
|
|
|
if (!preg_match('#[0-9A-Za-z_\-]{12}#', $gooDat['id'])) { |
92
|
|
|
throw new DomainException("GoogleBook 'id' malformed [0-9A-Za-z_\-]{12}"); |
93
|
|
|
} |
94
|
|
|
|
95
|
|
|
$data = self::mapGooData($gooDat); |
96
|
|
|
|
97
|
|
|
$templ = new self(); |
98
|
|
|
$templ->hydrate($data); |
99
|
|
|
|
100
|
|
|
return $templ; |
101
|
|
|
} |
102
|
|
|
|
103
|
|
|
/** |
104
|
|
|
* Mapping Google URL data to {Google Livres} data. |
105
|
|
|
* |
106
|
|
|
* @param array $gooData |
107
|
|
|
* |
108
|
|
|
* @return array |
109
|
|
|
*/ |
110
|
|
|
private static function mapGooData(array $gooData): array |
111
|
|
|
{ |
112
|
|
|
$data = []; |
113
|
|
|
$data['id'] = $gooData['id']; |
114
|
|
|
|
115
|
|
|
// show cover ? |
116
|
|
|
if (isset($gooData['printsec']) && 'frontcover' === $gooData['printsec']) { |
117
|
|
|
$data['couv'] = '1'; |
118
|
|
|
} |
119
|
|
|
|
120
|
|
|
// page number |
121
|
|
|
if (!empty($gooData['pg'])) { |
122
|
|
|
$data['page autre'] = $gooData['pg']; |
123
|
|
|
|
124
|
|
|
// pg=PAx => "page=x" |
125
|
|
|
if (preg_match('/^PA([0-9]+)$/', $gooData['pg'], $matches) > 0) { |
126
|
|
|
$data['page'] = $matches[1]; |
127
|
|
|
unset($data['page autre']); |
128
|
|
|
} |
129
|
|
|
// pg=PRx => "page=x|romain=1" |
130
|
|
|
if (preg_match('/^PR([0-9]+)$/', $gooData['pg'], $matches) > 0) { |
131
|
|
|
$data['page'] = $matches[1]; |
132
|
|
|
$data['romain'] = '1'; |
133
|
|
|
unset($data['page autre']); |
134
|
|
|
} |
135
|
|
|
} |
136
|
|
|
// q : keywords search / dq : quoted phrase search |
137
|
|
|
// affichage Google : dq ignoré si q existe |
138
|
|
|
if (!empty($gooData['dq']) || !empty($gooData['q'])) { |
139
|
|
|
$data['surligne'] = $gooData['q'] ?? $gooData['dq']; // q prévaut |
140
|
|
|
$data['surligne'] = self::googleUrlEncode($data['surligne']); |
141
|
|
|
} |
142
|
|
|
|
143
|
|
|
return $data; |
144
|
|
|
} |
145
|
|
|
|
146
|
|
|
/** |
147
|
|
|
* Instead of url_encode(). No UTF-8 encoding. |
148
|
|
|
* |
149
|
|
|
* @param string $str |
150
|
|
|
* |
151
|
|
|
* @return string |
152
|
|
|
*/ |
153
|
|
|
public static function googleUrlEncode(string $str): string |
154
|
|
|
{ |
155
|
|
|
return str_replace(' ', '+', trim(urldecode($str))); |
156
|
|
|
} |
157
|
|
|
|
158
|
|
|
/** |
159
|
|
|
* Check if Google URL or wiki {Google Books} template. |
160
|
|
|
* |
161
|
|
|
* @param string $text |
162
|
|
|
* |
163
|
|
|
* @return bool |
164
|
|
|
*/ |
165
|
|
|
public static function isGoogleBookValue(string $text): bool |
166
|
|
|
{ |
167
|
|
|
if (true === self::isGoogleBookURL($text)) { |
168
|
|
|
return true; |
169
|
|
|
} |
170
|
|
|
if (preg_match('#^{{[ \n]*Google (Livres|Books)[^}]+}}$#i', $text) > 0) { |
171
|
|
|
return true; |
172
|
|
|
} |
173
|
|
|
|
174
|
|
|
return false; |
175
|
|
|
} |
176
|
|
|
|
177
|
|
|
/** |
178
|
|
|
* Serialize the wiki-template. |
179
|
|
|
* Improvement : force param order : id/titre/... |
180
|
|
|
* |
181
|
|
|
* @param bool|null $cleanOrder |
182
|
|
|
* |
183
|
|
|
* @return string |
184
|
|
|
*/ |
185
|
|
|
public function serialize(?bool $cleanOrder = true): string |
186
|
|
|
{ |
187
|
|
|
$text = parent::serialize(); |
188
|
|
|
|
189
|
|
|
// Documentation suggère non affichage de ces 2 paramètres |
190
|
|
|
return str_replace(['id=', 'titre='], '', $text); |
191
|
|
|
} |
192
|
|
|
} |
193
|
|
|
|