|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
namespace MalScraper\Model\General; |
|
4
|
|
|
|
|
5
|
|
|
use MalScraper\Helper\Helper; |
|
6
|
|
|
use MalScraper\Model\MainModel; |
|
7
|
|
|
|
|
8
|
|
|
/** |
|
9
|
|
|
* PeopleModel class. |
|
10
|
|
|
*/ |
|
11
|
|
|
class PeopleModel extends MainModel |
|
12
|
|
|
{ |
|
13
|
|
|
/** |
|
14
|
|
|
* Id of the people. |
|
15
|
|
|
* |
|
16
|
|
|
* @var string|int |
|
17
|
|
|
*/ |
|
18
|
|
|
private $_id; |
|
19
|
|
|
|
|
20
|
|
|
/** |
|
21
|
|
|
* Biodata area. |
|
22
|
|
|
* |
|
23
|
|
|
* @var string |
|
24
|
|
|
*/ |
|
25
|
|
|
private $_biodata; |
|
26
|
|
|
|
|
27
|
|
|
/** |
|
28
|
|
|
* Default constructor. |
|
29
|
|
|
* |
|
30
|
|
|
* @param string|int $id |
|
31
|
|
|
* @param string $parserArea |
|
32
|
|
|
* |
|
33
|
|
|
* @return void |
|
34
|
|
|
*/ |
|
35
|
|
|
public function __construct($id, $parserArea = '#contentWrapper') |
|
36
|
|
|
{ |
|
37
|
|
|
$this->_id = $id; |
|
38
|
|
|
$this->_url = $this->_myAnimeListUrl.'/people/'.$id; |
|
39
|
|
|
$this->_parserArea = $parserArea; |
|
40
|
|
|
|
|
41
|
|
|
parent::errorCheck($this); |
|
42
|
|
|
|
|
43
|
|
|
if (!$this->_error) |
|
44
|
|
|
self::setBiodata(); |
|
|
|
|
|
|
45
|
|
|
} |
|
46
|
|
|
|
|
47
|
|
|
/** |
|
48
|
|
|
* Default call. |
|
49
|
|
|
* |
|
50
|
|
|
* @param string $method |
|
51
|
|
|
* @param array $arguments |
|
52
|
|
|
* |
|
53
|
|
|
* @return array|string|int |
|
54
|
|
|
*/ |
|
55
|
|
|
public function __call($method, $arguments) |
|
56
|
|
|
{ |
|
57
|
|
|
if ($this->_error) |
|
58
|
|
|
return $this->_error; |
|
59
|
|
|
return call_user_func_array([$this, $method], $arguments); |
|
60
|
|
|
} |
|
61
|
|
|
|
|
62
|
|
|
/** |
|
63
|
|
|
* Get people id. |
|
64
|
|
|
* |
|
65
|
|
|
* @return string |
|
66
|
|
|
*/ |
|
67
|
|
|
private function getId() |
|
68
|
|
|
{ |
|
69
|
|
|
return $this->_id; |
|
70
|
|
|
} |
|
71
|
|
|
|
|
72
|
|
|
/** |
|
73
|
|
|
* Get people name. |
|
74
|
|
|
* |
|
75
|
|
|
* @return string |
|
76
|
|
|
*/ |
|
77
|
|
|
private function getName() |
|
78
|
|
|
{ |
|
79
|
|
|
return $this->_parser->find('h1', 0)->plaintext; |
|
80
|
|
|
} |
|
81
|
|
|
|
|
82
|
|
|
/** |
|
83
|
|
|
* Get people image. |
|
84
|
|
|
* |
|
85
|
|
|
* @return string |
|
86
|
|
|
*/ |
|
87
|
|
|
private function getImage() |
|
88
|
|
|
{ |
|
89
|
|
|
$image = $this->_parser->find('#content table tr', 0)->find('td', 0)->find('img', 0); |
|
90
|
|
|
return $image ? $image->src : ''; |
|
|
|
|
|
|
91
|
|
|
} |
|
92
|
|
|
|
|
93
|
|
|
/** |
|
94
|
|
|
* Set people biodata. |
|
95
|
|
|
* |
|
96
|
|
|
* @return void |
|
97
|
|
|
*/ |
|
98
|
|
|
private function setBiodata() |
|
99
|
|
|
{ |
|
100
|
|
|
$html = $this->_parser->find('#content table tr', 0)->find('td', 0); |
|
101
|
|
|
$biodata = $html->innertext; |
|
102
|
|
|
$useless_biodata = ''; |
|
103
|
|
|
$useless_area = $html->find('div', 0); |
|
104
|
|
|
for ($i = 0; $i < 4; $i++) { |
|
105
|
|
|
$useless_biodata .= $useless_area->outertext; |
|
106
|
|
|
$useless_area = $useless_area->next_sibling(); |
|
107
|
|
|
} |
|
108
|
|
|
$biodata = str_replace($useless_biodata, '', $biodata); |
|
109
|
|
|
$this->_biodata = preg_replace("/([\s])+/", ' ', $biodata); |
|
110
|
|
|
} |
|
111
|
|
|
|
|
112
|
|
|
/** |
|
113
|
|
|
* Get people biodata. |
|
114
|
|
|
* |
|
115
|
|
|
* @param string $type Biodata type |
|
116
|
|
|
* |
|
117
|
|
|
* @return string|array |
|
118
|
|
|
*/ |
|
119
|
|
|
private function getBiodata($type) |
|
120
|
|
|
{ |
|
121
|
|
|
switch ($type) { |
|
122
|
|
|
case 'given_name': |
|
123
|
|
|
preg_match("/(Given name:<\/span>)[^<]*/", $this->_biodata, $biodata); |
|
124
|
|
|
break; |
|
125
|
|
|
case 'family_name': |
|
126
|
|
|
preg_match("/(Family name:<\/span>)[^<]*/", $this->_biodata, $biodata); |
|
127
|
|
|
break; |
|
128
|
|
|
case 'alternative_name': |
|
129
|
|
|
preg_match("/(Alternate names:<\/span>)[^<]*/", $this->_biodata, $biodata); |
|
130
|
|
|
break; |
|
131
|
|
|
case 'birthday': |
|
132
|
|
|
preg_match("/(Birthday:<\/span>)([^<])*/", $this->_biodata, $biodata); |
|
133
|
|
|
break; |
|
134
|
|
|
case 'website': |
|
135
|
|
|
preg_match("/(Website:<\/span> <a)([^<])*/", $this->_biodata, $biodata); |
|
136
|
|
|
break; |
|
137
|
|
|
case 'favorite': |
|
138
|
|
|
preg_match("/(Member Favorites:<\/span>)([^<])*/", $this->_biodata, $biodata); |
|
139
|
|
|
break; |
|
140
|
|
|
default: |
|
141
|
|
|
return null; |
|
142
|
|
|
} |
|
143
|
|
|
|
|
144
|
|
|
if ($biodata) { |
|
145
|
|
|
if ($type != 'website') { |
|
146
|
|
|
$biodata = strip_tags($biodata[0]); |
|
147
|
|
|
$biodata = explode(': ', $biodata); |
|
148
|
|
|
$biodata = trim($biodata[1]); |
|
149
|
|
|
} |
|
150
|
|
|
|
|
151
|
|
|
if ($type == 'given_name' || $type == 'family_name' || $type == 'birthday') |
|
152
|
|
|
return $biodata; |
|
153
|
|
|
|
|
154
|
|
|
if ($type == 'alternative_name') |
|
155
|
|
|
return explode(', ', $biodata); |
|
156
|
|
|
|
|
157
|
|
|
if ($type == 'favorite') |
|
158
|
|
|
return str_replace(',', '', $biodata); |
|
159
|
|
|
|
|
160
|
|
|
if ($type == 'website') { |
|
161
|
|
|
preg_match('/".+"/', $biodata[0], $biodata); |
|
162
|
|
|
if ($biodata[0] != '"http://"') |
|
163
|
|
|
return str_replace('"', '', $biodata[0]); |
|
164
|
|
|
} |
|
165
|
|
|
} |
|
166
|
|
|
return null; |
|
167
|
|
|
} |
|
168
|
|
|
|
|
169
|
|
|
/** |
|
170
|
|
|
* Get people more information. |
|
171
|
|
|
* |
|
172
|
|
|
* @return string |
|
173
|
|
|
*/ |
|
174
|
|
|
private function getMore() |
|
175
|
|
|
{ |
|
176
|
|
|
$more = $this->_parser->find('#content table tr', 0)->find('td', 0); |
|
177
|
|
|
$more = $more->find('div[class^=people-informantion-more]', 0)->plaintext; |
|
178
|
|
|
return preg_replace('/\n[^\S\n]*/', "\n", $more); |
|
179
|
|
|
} |
|
180
|
|
|
|
|
181
|
|
|
/** |
|
182
|
|
|
* Get people voice actor list. |
|
183
|
|
|
* |
|
184
|
|
|
* @return array |
|
185
|
|
|
*/ |
|
186
|
|
|
private function getVa() |
|
187
|
|
|
{ |
|
188
|
|
|
$va = []; |
|
189
|
|
|
$va_index = 0; |
|
190
|
|
|
$html = $this->_parser->find('#content table tr', 0)->find('td', 0)->next_sibling(); |
|
191
|
|
|
$va_area = $html->find('.normal_header', 0)->next_sibling(); |
|
192
|
|
|
if ($va_area->tag == 'table') { |
|
193
|
|
|
if ($va_area->find('tr')) { |
|
194
|
|
|
foreach ($va_area->find('tr') as $each_va) { |
|
195
|
|
|
// anime image |
|
196
|
|
|
$anime_image = $each_va->find('td', 0)->find('img', 0)->getAttribute('data-src'); |
|
197
|
|
|
$va[$va_index]['anime']['image'] = Helper::imageUrlCleaner($anime_image); |
|
198
|
|
|
|
|
199
|
|
|
$anime_area = $each_va->find('td', 1); |
|
200
|
|
|
|
|
201
|
|
|
// anime id |
|
202
|
|
|
$anime_id = $anime_area->find('a', 0)->href; |
|
203
|
|
|
$parsed_anime_id = explode('/', $anime_id); |
|
204
|
|
|
$anime_id = $parsed_anime_id[4]; |
|
205
|
|
|
$va[$va_index]['anime']['id'] = $anime_id; |
|
206
|
|
|
|
|
207
|
|
|
// anime title |
|
208
|
|
|
$anime_title = $anime_area->find('a', 0)->plaintext; |
|
209
|
|
|
$va[$va_index]['anime']['title'] = $anime_title; |
|
210
|
|
|
|
|
211
|
|
|
// character image |
|
212
|
|
|
$character_image = $each_va->find('td', 3)->find('img', 0)->getAttribute('data-src'); |
|
213
|
|
|
$va[$va_index]['character']['image'] = Helper::imageUrlCleaner($character_image); |
|
214
|
|
|
|
|
215
|
|
|
$character_area = $each_va->find('td', 2); |
|
216
|
|
|
|
|
217
|
|
|
// character id |
|
218
|
|
|
$character_id = $character_area->find('a', 0)->href; |
|
219
|
|
|
$parsed_character_id = explode('/', $character_id); |
|
220
|
|
|
$character_id = $parsed_character_id[4]; |
|
221
|
|
|
$va[$va_index]['character']['id'] = $character_id; |
|
222
|
|
|
|
|
223
|
|
|
// character name |
|
224
|
|
|
$character_name = $character_area->find('a', 0)->plaintext; |
|
225
|
|
|
$va[$va_index]['character']['name'] = $character_name; |
|
226
|
|
|
|
|
227
|
|
|
// character role |
|
228
|
|
|
$character_role = $character_area->find('div', 0)->plaintext; |
|
229
|
|
|
$va[$va_index]['character']['role'] = $character_role; |
|
230
|
|
|
|
|
231
|
|
|
$va_index++; |
|
232
|
|
|
} |
|
233
|
|
|
} |
|
234
|
|
|
} |
|
235
|
|
|
return $va; |
|
236
|
|
|
} |
|
237
|
|
|
|
|
238
|
|
|
/** |
|
239
|
|
|
* Get people staff list. |
|
240
|
|
|
* |
|
241
|
|
|
* @return array |
|
242
|
|
|
*/ |
|
243
|
|
|
private function getStaff() |
|
244
|
|
|
{ |
|
245
|
|
|
$staff = []; |
|
246
|
|
|
$staff_index = 0; |
|
247
|
|
|
$html = $this->_parser->find('#content table tr', 0)->find('td', 0)->next_sibling(); |
|
248
|
|
|
$staff_area = $html->find('.normal_header', 1)->next_sibling(); |
|
249
|
|
|
if ($staff_area->tag == 'table') { |
|
250
|
|
|
foreach ($staff_area->find('tr') as $each_staff) { |
|
251
|
|
|
$anime_image = $each_staff->find('td', 0)->find('img', 0)->getAttribute('data-src'); |
|
252
|
|
|
$staff[$staff_index]['image'] = Helper::imageUrlCleaner($anime_image); |
|
253
|
|
|
|
|
254
|
|
|
$each_staff = $each_staff->find('td', 1); |
|
255
|
|
|
|
|
256
|
|
|
// anime id |
|
257
|
|
|
$anime_id = $each_staff->find('a', 0)->href; |
|
258
|
|
|
$parsed_anime_id = explode('/', $anime_id); |
|
259
|
|
|
$anime_id = $parsed_anime_id[4]; |
|
260
|
|
|
$staff[$staff_index]['id'] = $anime_id; |
|
261
|
|
|
|
|
262
|
|
|
// anime title |
|
263
|
|
|
$anime_title = $each_staff->find('a', 0)->plaintext; |
|
264
|
|
|
$staff[$staff_index]['title'] = $anime_title; |
|
265
|
|
|
|
|
266
|
|
|
// role |
|
267
|
|
|
$role = $each_staff->find('small', 0)->plaintext; |
|
268
|
|
|
$staff[$staff_index]['role'] = $role; |
|
269
|
|
|
|
|
270
|
|
|
$staff_index++; |
|
271
|
|
|
} |
|
272
|
|
|
} |
|
273
|
|
|
return $staff; |
|
274
|
|
|
} |
|
275
|
|
|
|
|
276
|
|
|
/** |
|
277
|
|
|
* Get people published manga list. |
|
278
|
|
|
* |
|
279
|
|
|
* @return array |
|
280
|
|
|
*/ |
|
281
|
|
|
private function getManga() |
|
282
|
|
|
{ |
|
283
|
|
|
$published_manga = []; |
|
284
|
|
|
$manga_index = 0; |
|
285
|
|
|
$html = $this->_parser->find('#content table tr', 0)->find('td', 0)->next_sibling(); |
|
286
|
|
|
$manga_area = $html->find('.normal_header', 2)->next_sibling(); |
|
287
|
|
|
if ($manga_area->tag == 'table') { |
|
288
|
|
|
foreach ($manga_area->find('tr') as $each_manga) { |
|
289
|
|
|
$manga_image = $each_manga->find('td', 0)->find('img', 0)->getAttribute('data-src'); |
|
290
|
|
|
$published_manga[$manga_index]['image'] = Helper::imageUrlCleaner($manga_image); |
|
291
|
|
|
|
|
292
|
|
|
$each_manga = $each_manga->find('td', 1); |
|
293
|
|
|
|
|
294
|
|
|
// manga id |
|
295
|
|
|
$manga_id = $each_manga->find('a', 0)->href; |
|
296
|
|
|
$parsed_manga_id = explode('/', $manga_id); |
|
297
|
|
|
$manga_id = $parsed_manga_id[4]; |
|
298
|
|
|
$published_manga[$manga_index]['id'] = $manga_id; |
|
299
|
|
|
|
|
300
|
|
|
// manga title |
|
301
|
|
|
$manga_title = $each_manga->find('a', 0)->plaintext; |
|
302
|
|
|
$published_manga[$manga_index]['title'] = $manga_title; |
|
303
|
|
|
|
|
304
|
|
|
// role |
|
305
|
|
|
$role = $each_manga->find('small', 0)->plaintext; |
|
306
|
|
|
$published_manga[$manga_index]['role'] = $role; |
|
307
|
|
|
|
|
308
|
|
|
$manga_index++; |
|
309
|
|
|
} |
|
310
|
|
|
} |
|
311
|
|
|
return $published_manga; |
|
312
|
|
|
} |
|
313
|
|
|
|
|
314
|
|
|
/** |
|
315
|
|
|
* Get people all information. |
|
316
|
|
|
* |
|
317
|
|
|
* @return array |
|
318
|
|
|
*/ |
|
319
|
|
|
private function getAllInfo() |
|
320
|
|
|
{ |
|
321
|
|
|
$data = [ |
|
322
|
|
|
'id' => self::getId(), |
|
|
|
|
|
|
323
|
|
|
'name' => self::getName(), |
|
|
|
|
|
|
324
|
|
|
'image' => self::getImage(), |
|
|
|
|
|
|
325
|
|
|
'given_name' => self::getBiodata('given_name'), |
|
|
|
|
|
|
326
|
|
|
'family_name' => self::getBiodata('family_name'), |
|
327
|
|
|
'alternative_name' => self::getBiodata('alternative_name'), |
|
328
|
|
|
'birthday' => self::getBiodata('birthday'), |
|
329
|
|
|
'website' => self::getBiodata('website'), |
|
330
|
|
|
'favorite' => self::getBiodata('favorite'), |
|
331
|
|
|
'more' => self::getMore(), |
|
|
|
|
|
|
332
|
|
|
'va' => self::getVa(), |
|
|
|
|
|
|
333
|
|
|
'staff' => self::getStaff(), |
|
|
|
|
|
|
334
|
|
|
'published_manga' => self::getManga(), |
|
|
|
|
|
|
335
|
|
|
]; |
|
336
|
|
|
|
|
337
|
|
|
return $data; |
|
338
|
|
|
} |
|
339
|
|
|
} |
rl404 /
MAL-Scraper
Loading history...