|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
namespace MalScraper\Model; |
|
4
|
|
|
|
|
5
|
|
|
define('MAX_FILE_SIZE', 100000000); |
|
6
|
|
|
|
|
7
|
|
|
use HtmlDomParser; |
|
8
|
|
|
|
|
9
|
|
|
/** |
|
10
|
|
|
* MainModel class. |
|
11
|
|
|
* |
|
12
|
|
|
* Base model for all model class. |
|
13
|
|
|
*/ |
|
14
|
|
|
class MainModel |
|
15
|
|
|
{ |
|
16
|
|
|
/** |
|
17
|
|
|
* MyAnimeList main URL. |
|
18
|
|
|
* |
|
19
|
|
|
* @var string |
|
20
|
|
|
*/ |
|
21
|
|
|
protected $_myAnimeListUrl = 'https://myanimelist.net'; |
|
22
|
|
|
|
|
23
|
|
|
/** |
|
24
|
|
|
* Trimmed HtmlDomParser. |
|
25
|
|
|
* |
|
26
|
|
|
* @var \simplehtmldom_1_5\simple_html_dom |
|
27
|
|
|
*/ |
|
28
|
|
|
protected $_parser; |
|
29
|
|
|
|
|
30
|
|
|
/** |
|
31
|
|
|
* Area to be parsed. |
|
32
|
|
|
* |
|
33
|
|
|
* @var string |
|
34
|
|
|
*/ |
|
35
|
|
|
protected $_parserArea; |
|
36
|
|
|
|
|
37
|
|
|
/** |
|
38
|
|
|
* Complete MyAnimeList page URL. |
|
39
|
|
|
* |
|
40
|
|
|
* @var string |
|
41
|
|
|
*/ |
|
42
|
|
|
protected $_url; |
|
43
|
|
|
|
|
44
|
|
|
/** |
|
45
|
|
|
* Error response. |
|
46
|
|
|
* |
|
47
|
|
|
* @var string|int |
|
48
|
|
|
*/ |
|
49
|
|
|
protected $_error; |
|
50
|
|
|
|
|
51
|
|
|
/** |
|
52
|
|
|
* Get URL header. |
|
53
|
|
|
* |
|
54
|
|
|
* @param string $url URL of full MyAnimeList page |
|
55
|
|
|
* |
|
56
|
|
|
* @return int |
|
57
|
|
|
*/ |
|
58
|
|
|
public static function getHeader($url) |
|
59
|
|
|
{ |
|
60
|
|
|
$file_headers = @get_headers($url); |
|
61
|
|
|
if (empty($file_headers) || $file_headers[0] == 'HTTP/1.1 404 Not Found') { |
|
62
|
|
|
return 404; |
|
63
|
|
|
} |
|
64
|
|
|
|
|
65
|
|
|
if (empty($file_headers) || $file_headers[0] == 'HTTP/1.1 403 Forbidden') { |
|
66
|
|
|
return 403; |
|
67
|
|
|
} |
|
68
|
|
|
|
|
69
|
|
|
return 200; |
|
70
|
|
|
} |
|
71
|
|
|
|
|
72
|
|
|
/** |
|
73
|
|
|
* Get trimmed HtmlDomParser class. |
|
74
|
|
|
* |
|
75
|
|
|
* @param string $url URL of full MyAnimeList page |
|
76
|
|
|
* @param string $contentDiv Specific area to be parsed |
|
77
|
|
|
* |
|
78
|
|
|
* @return \simplehtmldom_1_5\simple_html_dom |
|
79
|
|
|
*/ |
|
80
|
|
|
public static function getParser($url, $contentDiv, $additionalSetting = false) |
|
81
|
|
|
{ |
|
82
|
|
|
$html = HtmlDomParser::file_get_html($url)->find($contentDiv, 0); |
|
83
|
|
|
$html = !$additionalSetting ? $html : $html->next_sibling(); |
|
84
|
|
|
$html = $html->outertext; |
|
85
|
|
|
$html = str_replace('"', '\"', $html); |
|
86
|
|
|
$html = html_entity_decode($html, ENT_QUOTES, 'UTF-8'); |
|
87
|
|
|
$html = HtmlDomParser::str_get_html($html); |
|
88
|
|
|
|
|
89
|
|
|
return $html; |
|
90
|
|
|
} |
|
91
|
|
|
|
|
92
|
|
|
/** |
|
93
|
|
|
* Header error check. |
|
94
|
|
|
* |
|
95
|
|
|
* @param MainModel $model Any model |
|
96
|
|
|
* |
|
97
|
|
|
* @return void |
|
98
|
|
|
*/ |
|
99
|
|
|
public static function errorCheck($model) |
|
100
|
|
|
{ |
|
101
|
|
|
$className = self::getCleanClassName($model); |
|
102
|
|
|
|
|
103
|
|
|
if (strpos($className, 'Search') !== false) { |
|
104
|
|
|
if (strlen($model->_query) < 3) { |
|
105
|
|
|
$model->_error = 400; |
|
106
|
|
|
} |
|
107
|
|
|
} |
|
108
|
|
|
|
|
109
|
|
|
if (!$model->_error) { |
|
110
|
|
|
$header = self::getHeader($model->_url); |
|
111
|
|
|
if ($header == 200) { |
|
112
|
|
|
if ($className != 'UserListModel') { |
|
113
|
|
|
$additionalSetting = ($className == 'CharacterPeoplePictureModel'); |
|
114
|
|
|
$model->_parser = self::getParser(self::getAbsoluteUrl($model), $model->_parserArea, $additionalSetting); |
|
115
|
|
|
} |
|
116
|
|
|
} else { |
|
117
|
|
|
$model->_error = $header; |
|
118
|
|
|
} |
|
119
|
|
|
} |
|
120
|
|
|
} |
|
121
|
|
|
|
|
122
|
|
|
/** |
|
123
|
|
|
* Get the correct url. |
|
124
|
|
|
* |
|
125
|
|
|
* @param MainModel $model Any model |
|
126
|
|
|
* |
|
127
|
|
|
* @return string |
|
128
|
|
|
*/ |
|
129
|
|
|
public static function getAbsoluteUrl($model) |
|
130
|
|
|
{ |
|
131
|
|
|
$className = self::getCleanClassName($model); |
|
132
|
|
|
$additionalUrl = ''; |
|
133
|
|
|
|
|
134
|
|
|
switch ($className) { |
|
135
|
|
|
case 'CharacterStaffModel': |
|
136
|
|
|
$area = 'li a[href$=characters]'; |
|
137
|
|
|
break; |
|
138
|
|
|
case 'StatModel': |
|
139
|
|
|
$area = 'li a[href$=stats]'; |
|
140
|
|
|
$additionalUrl = '?m=all&show=1'; |
|
141
|
|
|
break; |
|
142
|
|
|
case 'PictureModel': |
|
143
|
|
|
$area = 'li a[href$=pics]'; |
|
144
|
|
|
break; |
|
145
|
|
|
case 'CharacterPeoplePictureModel': |
|
146
|
|
|
$area = 'li a[href$=pictures]'; |
|
147
|
|
|
break; |
|
148
|
|
|
default: |
|
149
|
|
|
return $model->_url; |
|
150
|
|
|
} |
|
151
|
|
|
|
|
152
|
|
|
$html = HtmlDomParser::file_get_html($model->_url)->find($area, 0)->href; |
|
153
|
|
|
|
|
154
|
|
|
if ($model->getType() == 'manga') { |
|
155
|
|
|
return 'https://myanimelist.net'.$html.$additionalUrl; |
|
156
|
|
|
} |
|
157
|
|
|
|
|
158
|
|
|
return $html.$additionalUrl; |
|
159
|
|
|
} |
|
160
|
|
|
|
|
161
|
|
|
/** |
|
162
|
|
|
* Get clean class name. |
|
163
|
|
|
* |
|
164
|
|
|
* @param MainModel $model Any model |
|
165
|
|
|
* |
|
166
|
|
|
* @return string |
|
167
|
|
|
*/ |
|
168
|
|
|
public static function getCleanClassName($model) |
|
169
|
|
|
{ |
|
170
|
|
|
$className = get_class($model); |
|
171
|
|
|
$className = explode('\\', $className); |
|
172
|
|
|
|
|
173
|
|
|
return $className[count($className) - 1]; |
|
174
|
|
|
} |
|
175
|
|
|
|
|
176
|
|
|
/** |
|
177
|
|
|
* Get default type. |
|
178
|
|
|
* |
|
179
|
|
|
* @return string |
|
180
|
|
|
*/ |
|
181
|
|
|
public function getType() |
|
182
|
|
|
{ |
|
183
|
|
|
return 'anime'; |
|
184
|
|
|
} |
|
185
|
|
|
} |
|
186
|
|
|
|
rl404 /
MAL-Scraper
