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 DomainException; |
13
|
|
|
use Exception; |
14
|
|
|
|
15
|
|
|
/** |
16
|
|
|
* https://fr.wikipedia.org/wiki/Mod%C3%A8le:Google_Livres |
17
|
|
|
* Le premier paramètre (ou id) est obligatoire. L |
18
|
|
|
* Le deuxième (ou titre) est requis si on ne veut pas fabriquer le lien brut (inclusion {{ouvrage}} 'Lire en ligne') |
19
|
|
|
* Class GoogleLivresTemplate. |
20
|
|
|
*/ |
21
|
|
|
class GoogleLivresTemplate extends AbstractWikiTemplate |
22
|
|
|
{ |
23
|
|
|
const DEFAULT_GOOGLEBOOK_URL = 'https://books.google.com/books'; |
24
|
|
|
|
25
|
|
|
const ALLOW_USER_ORDER = false; |
26
|
|
|
|
27
|
|
|
const MODEL_NAME = 'Google Livres'; |
28
|
|
|
|
29
|
|
|
const REQUIRED_PARAMETERS = ['id' => '']; |
30
|
|
|
|
31
|
|
|
const PARAM_ALIAS |
32
|
|
|
= [ |
33
|
|
|
'1' => 'id', |
34
|
|
|
'2' => 'titre', |
35
|
|
|
'surligné' => 'surligne', |
36
|
|
|
'BuchID' => 'id', |
37
|
|
|
]; |
38
|
|
|
|
39
|
|
|
protected $parametersByOrder |
40
|
|
|
= ['id', 'titre', 'couv', 'page', 'romain', 'page autre', 'surligne']; |
41
|
|
|
|
42
|
|
|
/** |
43
|
|
|
* Serialize the wiki-template. |
44
|
|
|
* Improvement : force param order : id/titre/... |
45
|
|
|
* |
46
|
|
|
* @param bool|null $cleanOrder |
47
|
|
|
* |
48
|
|
|
* @return string |
49
|
|
|
*/ |
50
|
|
|
public function serialize(?bool $cleanOrder = true): string |
51
|
|
|
{ |
52
|
|
|
$text = parent::serialize(); |
53
|
|
|
|
54
|
|
|
// Documentation suggère non affichage de ces 2 paramètres |
55
|
|
|
return str_replace(['id=', 'titre='], '', $text); |
56
|
|
|
} |
57
|
|
|
|
58
|
|
|
/** |
59
|
|
|
* Create {Google Book} from URL. |
60
|
|
|
* See also https://fr.wikipedia.org/wiki/Utilisateur:Jack_ma/GB |
61
|
|
|
* https://stackoverflow.com/questions/11584551/need-information-on-query-parameters-for-google-books-e-g-difference-between-d. |
62
|
|
|
* |
63
|
|
|
* @param string $url |
64
|
|
|
* |
65
|
|
|
* @return GoogleLivresTemplate|null |
66
|
|
|
* @throws Exception |
67
|
|
|
*/ |
68
|
|
|
public static function createFromURL(string $url): ?self |
69
|
|
|
{ |
70
|
|
|
if (!self::isGoogleBookURL($url)) { |
71
|
|
|
throw new DomainException('not a Google Book URL'); |
72
|
|
|
} |
73
|
|
|
$gooDat = self::parseGoogleBookQuery($url); |
74
|
|
|
|
75
|
|
|
if (empty($gooDat['id'])) { |
76
|
|
|
throw new DomainException("no GoogleBook 'id' in URL"); |
77
|
|
|
} |
78
|
|
|
|
79
|
|
|
$data = self::mapGooData($gooDat); |
80
|
|
|
|
81
|
|
|
$templ = new self(); |
82
|
|
|
$templ->hydrate($data); |
83
|
|
|
|
84
|
|
|
return $templ; |
85
|
|
|
} |
86
|
|
|
|
87
|
|
|
/** |
88
|
|
|
* Mapping Google URL data to {Google Livres} data. |
89
|
|
|
* |
90
|
|
|
* @param array $gooData |
91
|
|
|
* |
92
|
|
|
* @return array |
93
|
|
|
*/ |
94
|
|
|
private static function mapGooData(array $gooData): array |
95
|
|
|
{ |
96
|
|
|
$data = []; |
97
|
|
|
$data['id'] = $gooData['id']; |
98
|
|
|
|
99
|
|
|
// show cover ? |
100
|
|
|
if (isset($gooData['printsec']) && 'frontcover' === $gooData['printsec']) { |
101
|
|
|
$data['couv'] = '1'; |
102
|
|
|
} |
103
|
|
|
|
104
|
|
|
// page number |
105
|
|
|
if (!empty($gooData['pg'])) { |
106
|
|
|
$data['page autre'] = $gooData['pg']; |
107
|
|
|
|
108
|
|
|
// pg=PAx => "page=x" |
109
|
|
|
if (preg_match('/^PA([0-9]+)$/', $gooData['pg'], $matches) > 0) { |
110
|
|
|
$data['page'] = $matches[1]; |
111
|
|
|
unset($data['page autre']); |
112
|
|
|
} |
113
|
|
|
// pg=PRx => "page=x|romain=1" |
114
|
|
|
if (preg_match('/^PR([0-9]+)$/', $gooData['pg'], $matches) > 0) { |
115
|
|
|
$data['page'] = $matches[1]; |
116
|
|
|
$data['romain'] = '1'; |
117
|
|
|
unset($data['page autre']); |
118
|
|
|
} |
119
|
|
|
} |
120
|
|
|
|
121
|
|
|
if (!empty($gooData['dq']) || !empty($gooData['q'])) { |
122
|
|
|
$data['surligne'] = $gooData['dq'] ?? $gooData['q']; |
123
|
|
|
$data['surligne'] = self::googleUrlEncode($data['surligne']); |
124
|
|
|
} |
125
|
|
|
|
126
|
|
|
return $data; |
127
|
|
|
} |
128
|
|
|
|
129
|
|
|
/** |
130
|
|
|
* Instead of url_encode(). No UTF-8 encoding. |
131
|
|
|
*/ |
132
|
|
|
private static function googleUrlEncode(string $str): string |
133
|
|
|
{ |
134
|
|
|
return str_replace(' ', '+', trim(urldecode($str))); |
135
|
|
|
} |
136
|
|
|
|
137
|
|
|
/** |
138
|
|
|
* Clean the google book URL from optional&tracking data. |
139
|
|
|
* |
140
|
|
|
* @param string $url |
141
|
|
|
* |
142
|
|
|
* @return string URL |
143
|
|
|
*/ |
144
|
|
|
public static function simplifyGoogleUrl(string $url): string |
145
|
|
|
{ |
146
|
|
|
if (!self::isGoogleBookURL($url)) { |
147
|
|
|
throw new DomainException('not a Google Book URL'); |
148
|
|
|
} |
149
|
|
|
|
150
|
|
|
$gooDat = self::parseGoogleBookQuery($url); |
151
|
|
|
if (empty($gooDat['id'])) { |
152
|
|
|
throw new DomainException("no GoogleBook 'id' in URL"); |
153
|
|
|
} |
154
|
|
|
|
155
|
|
|
$dat = []; |
156
|
|
|
// keep only a few parameters (+'q' ?) |
157
|
|
|
$keeps = ['id', 'pg', 'printsec', 'dq']; |
158
|
|
|
foreach ($keeps as $keep) { |
159
|
|
|
if (!empty($gooDat[$keep])) { |
160
|
|
|
$dat[$keep] = $gooDat[$keep]; |
161
|
|
|
} |
162
|
|
|
} |
163
|
|
|
|
164
|
|
|
$googleURL = self::DEFAULT_GOOGLEBOOK_URL; |
165
|
|
|
|
166
|
|
|
// domain .com .fr |
167
|
|
|
$gooDomain = self::parseGoogleDomain($url); |
168
|
|
|
if ($gooDomain) { |
169
|
|
|
$googleURL = str_replace('.com', $gooDomain, $googleURL); |
170
|
|
|
} |
171
|
|
|
|
172
|
|
|
return $googleURL.'?'.http_build_query($dat); |
173
|
|
|
} |
174
|
|
|
|
175
|
|
|
/** |
176
|
|
|
* Parse URL argument from ?query and #fragment. |
177
|
|
|
* |
178
|
|
|
* @param string $url |
179
|
|
|
* |
180
|
|
|
* @return array |
181
|
|
|
*/ |
182
|
|
|
private static function parseGoogleBookQuery(string $url): array |
183
|
|
|
{ |
184
|
|
|
// Note : Also datas in URL after the '#' !!! (URL fragment) |
185
|
|
|
$queryData = parse_url($url, PHP_URL_QUERY); // after ? |
186
|
|
|
$fragmentData = parse_url($url, PHP_URL_FRAGMENT); // after # |
187
|
|
|
// queryData precedence over fragmentData |
188
|
|
|
parse_str(implode('&', [$fragmentData, $queryData]), $val); |
189
|
|
|
|
190
|
|
|
return self::arrayKeysToLower($val); |
191
|
|
|
} |
192
|
|
|
|
193
|
|
|
/** |
194
|
|
|
* Set all array keys to lower case. |
195
|
|
|
* TODO: move to ArrayProcessTrait ? |
196
|
|
|
* |
197
|
|
|
* @param array $array |
198
|
|
|
* |
199
|
|
|
* @return array |
200
|
|
|
*/ |
201
|
|
|
private static function arrayKeysToLower(array $array): array |
202
|
|
|
{ |
203
|
|
|
$res = []; |
204
|
|
|
foreach ($array as $key => $val) { |
205
|
|
|
$res[strtolower($key)] = $val; |
206
|
|
|
} |
207
|
|
|
|
208
|
|
|
return $res; |
209
|
|
|
} |
210
|
|
|
|
211
|
|
|
/** |
212
|
|
|
* return '.fr' or '.com'. |
213
|
|
|
* |
214
|
|
|
* @param string $url |
215
|
|
|
* |
216
|
|
|
* @return string|null |
217
|
|
|
*/ |
218
|
|
|
private static function parseGoogleDomain(string $url): ?string |
219
|
|
|
{ |
220
|
|
|
$host = parse_url($url, PHP_URL_HOST); |
221
|
|
|
if (!empty($host) && preg_match('#\.[a-z]{2,3}$#', $host, $matches) > 0) { |
222
|
|
|
return $matches[0]; // .fr |
223
|
|
|
} |
224
|
|
|
|
225
|
|
|
return null; |
226
|
|
|
} |
227
|
|
|
|
228
|
|
|
/** |
229
|
|
|
* Check google URL pattern. |
230
|
|
|
* |
231
|
|
|
* @param string $text |
232
|
|
|
* |
233
|
|
|
* @return bool |
234
|
|
|
*/ |
235
|
|
|
public static function isGoogleBookURL(string $text): bool |
236
|
|
|
{ |
237
|
|
|
if (preg_match('#^https?://(books|play)\.google\.[a-z]{2,3}/(books)?(/reader)?\?id=#i', $text) > 0) { |
238
|
|
|
return true; |
239
|
|
|
} |
240
|
|
|
|
241
|
|
|
return false; |
242
|
|
|
} |
243
|
|
|
|
244
|
|
|
/** |
245
|
|
|
* Check if Google URL or wiki {Google Books} template. |
246
|
|
|
* |
247
|
|
|
* @param string $text |
248
|
|
|
* |
249
|
|
|
* @return bool |
250
|
|
|
*/ |
251
|
|
|
public static function isGoogleBookValue(string $text): bool |
252
|
|
|
{ |
253
|
|
|
if (true === self::isGoogleBookURL($text)) { |
254
|
|
|
return true; |
255
|
|
|
} |
256
|
|
|
if (preg_match('#^{{[ \n]*Google (Livres|Books)[^}]+}}$#i', $text) > 0) { |
257
|
|
|
return true; |
258
|
|
|
} |
259
|
|
|
|
260
|
|
|
return false; |
261
|
|
|
} |
262
|
|
|
} |
263
|
|
|
|