|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
namespace Popstas\Transmission\Console; |
|
4
|
|
|
|
|
5
|
|
|
use GuzzleHttp\ClientInterface; |
|
6
|
|
|
use GuzzleHttp\Cookie\CookieJar; |
|
7
|
|
|
|
|
8
|
|
|
class WeburgClient |
|
9
|
|
|
{ |
|
10
|
|
|
/** |
|
11
|
|
|
* @var ClientInterface |
|
12
|
|
|
*/ |
|
13
|
|
|
private $httpClient; |
|
14
|
|
|
|
|
15
|
|
|
private $requestDelay; |
|
16
|
|
|
|
|
17
|
|
|
public function __construct(ClientInterface $httpClient, $requestDelay = 0) |
|
18
|
|
|
{ |
|
19
|
|
|
$this->httpClient = $httpClient; |
|
20
|
|
|
$this->requestDelay = $requestDelay; |
|
21
|
|
|
} |
|
22
|
|
|
|
|
23
|
|
|
public function getMovieInfoById($movieId) |
|
24
|
|
|
{ |
|
25
|
|
|
$movieUrl = $this->getMovieUrl($movieId); |
|
26
|
|
|
$body = $this->getUrlBody($movieUrl); |
|
27
|
|
|
$body = iconv('WINDOWS-1251', 'UTF-8', $body); |
|
28
|
|
|
$info = $this->getMovieInfo($body); |
|
29
|
|
|
return $info; |
|
30
|
|
|
} |
|
31
|
|
|
|
|
32
|
|
|
/** |
|
33
|
|
|
* @param $movieId |
|
34
|
|
|
* @return array urls of movie (not series) |
|
35
|
|
|
*/ |
|
36
|
|
|
public function getMovieTorrentUrlsById($movieId) |
|
37
|
|
|
{ |
|
38
|
|
|
$torrentUrl = $this->getMovieTorrentUrl($movieId); |
|
39
|
|
|
$body = $this->getUrlBody($torrentUrl); |
|
40
|
|
|
$torrentsUrls = $this->getTorrentsUrls($body); |
|
41
|
|
|
return $torrentsUrls; |
|
42
|
|
|
} |
|
43
|
|
|
|
|
44
|
|
|
/** |
|
45
|
|
|
* @param $movieId |
|
46
|
|
|
* @param $hashes array hashes of torrents from movieInfo |
|
47
|
|
|
* @param int $daysMax torrents older last days will not matched |
|
48
|
|
|
* @param int $allowedMisses after x misses next checks will broken |
|
49
|
|
|
* @return array urls of matched torrent files |
|
50
|
|
|
*/ |
|
51
|
|
|
public function getSeriesTorrents($movieId, $hashes, $daysMax = 1, $allowedMisses = 0) |
|
52
|
|
|
{ |
|
53
|
|
|
$torrentsUrls = []; |
|
54
|
|
|
$timestampFrom = strtotime('-' . $daysMax . 'days'); |
|
55
|
|
|
|
|
56
|
|
|
$hashes = array_reverse($hashes); |
|
57
|
|
|
foreach ($hashes as $hash) { |
|
58
|
|
|
if ($allowedMisses < 0) { |
|
59
|
|
|
break; |
|
60
|
|
|
} |
|
61
|
|
|
|
|
62
|
|
|
$torrentUrl = $this->getMovieTorrentUrl($movieId, $hash); |
|
63
|
|
|
$body = $this->getUrlBody($torrentUrl); |
|
64
|
|
|
|
|
65
|
|
|
if (!$this->checkTorrentDate($body, $timestampFrom)) { |
|
|
|
|
|
|
66
|
|
|
$allowedMisses--; |
|
67
|
|
|
continue; |
|
68
|
|
|
} |
|
69
|
|
|
|
|
70
|
|
|
$torrentsUrls = array_merge($torrentsUrls, $this->getTorrentsUrls($body)); |
|
71
|
|
|
} |
|
72
|
|
|
|
|
73
|
|
|
return $torrentsUrls; |
|
74
|
|
|
} |
|
75
|
|
|
|
|
76
|
|
|
public function getMoviesIds() |
|
77
|
|
|
{ |
|
78
|
|
|
$moviesUrl = 'http://weburg.net/movies/new/?clever_title=1&template=0&last=0'; |
|
79
|
|
|
|
|
80
|
|
|
$jsonRaw = $this->getUrlBody($moviesUrl, [ |
|
81
|
|
|
'Content-Type' => 'text/html; charset=utf-8', |
|
82
|
|
|
'User-Agent' => 'Mozilla/5.0 (X11; Ubuntu; Linux x86_64; rv:27.0) Gecko/20100101 Firefox/27.0', |
|
83
|
|
|
'X-Requested-With' => 'XMLHttpRequest', |
|
84
|
|
|
]); |
|
85
|
|
|
|
|
86
|
|
|
$moviesJson = json_decode($jsonRaw); |
|
87
|
|
|
|
|
88
|
|
|
$moviesIds = $this->getInfoUrls($moviesJson->items); |
|
89
|
|
|
|
|
90
|
|
|
return $moviesIds; |
|
91
|
|
|
} |
|
92
|
|
|
|
|
93
|
|
|
public function getMovieUrl($movieId) |
|
94
|
|
|
{ |
|
95
|
|
|
return 'http://weburg.net/movies/info/' . $movieId; |
|
96
|
|
|
} |
|
97
|
|
|
|
|
98
|
|
|
/** |
|
99
|
|
|
* @param $url |
|
100
|
|
|
* @param array $headers |
|
101
|
|
|
* @return \Psr\Http\Message\ResponseInterface |
|
102
|
|
|
* @throws \RuntimeException |
|
103
|
|
|
*/ |
|
104
|
|
|
private function getUrl($url, $headers = []) |
|
105
|
|
|
{ |
|
106
|
|
|
$jar = new CookieJar(); |
|
107
|
|
|
|
|
108
|
|
|
$res = $this->httpClient->request('GET', $url, [ |
|
109
|
|
|
'headers' => $headers, |
|
110
|
|
|
'cookies' => $jar, |
|
111
|
|
|
]); |
|
112
|
|
|
|
|
113
|
|
|
if ($res->getStatusCode() != 200) { |
|
114
|
|
|
throw new \RuntimeException('Error ' . $res->getStatusCode() . 'while get url ' . $url); |
|
115
|
|
|
} |
|
116
|
|
|
|
|
117
|
|
|
sleep($this->requestDelay); |
|
118
|
|
|
|
|
119
|
|
|
return $res; |
|
120
|
|
|
} |
|
121
|
|
|
|
|
122
|
|
|
/** |
|
123
|
|
|
* @param $url |
|
124
|
|
|
* @param array $headers |
|
125
|
|
|
* @return \Psr\Http\Message\StreamInterface |
|
126
|
|
|
* @throws \RuntimeException |
|
127
|
|
|
*/ |
|
128
|
|
|
public function getUrlBody($url, $headers = []) |
|
129
|
|
|
{ |
|
130
|
|
|
$res = $this->getUrl($url, $headers); |
|
131
|
|
|
$body = $res->getBody(); |
|
132
|
|
|
return $body; |
|
133
|
|
|
} |
|
134
|
|
|
|
|
135
|
|
|
/** |
|
136
|
|
|
* @param $url |
|
137
|
|
|
* @param $torrentsDir |
|
138
|
|
|
* @throws \RuntimeException |
|
139
|
|
|
*/ |
|
140
|
|
|
public function downloadTorrent($url, $torrentsDir) |
|
141
|
|
|
{ |
|
142
|
|
|
$res = $this->getUrl($url); |
|
143
|
|
|
$torrentBody = $res->getBody(); |
|
144
|
|
|
|
|
145
|
|
|
$disposition = $res->getHeader('content-disposition'); |
|
146
|
|
|
preg_match('/filename="(.*?)"/', $disposition[0], $res); |
|
147
|
|
|
$filename = $res[1]; |
|
148
|
|
|
|
|
149
|
|
|
$filePath = $torrentsDir . '/' . $filename; |
|
150
|
|
|
file_put_contents($filePath, $torrentBody); |
|
151
|
|
|
} |
|
152
|
|
|
|
|
153
|
|
|
public function isTorrentPopular($movieInfo, $commentsMin, $imdbMin, $kinopoiskMin, $votesMin) |
|
154
|
|
|
{ |
|
155
|
|
|
return $movieInfo['comments'] >= $commentsMin |
|
156
|
|
|
|| $movieInfo['rating_imdb'] >= $imdbMin |
|
157
|
|
|
|| $movieInfo['rating_kinopoisk'] >= $kinopoiskMin |
|
158
|
|
|
|| $movieInfo['rating_votes'] >= $votesMin; |
|
159
|
|
|
} |
|
160
|
|
|
|
|
161
|
|
|
public function cleanMovieId($idOrUrl) |
|
162
|
|
|
{ |
|
163
|
|
|
if (preg_match('/^\d+$/', $idOrUrl)) { |
|
164
|
|
|
return $idOrUrl; |
|
165
|
|
|
} |
|
166
|
|
|
preg_match('/^http:\/\/weburg\.net\/(series|movies)\/info\/(\d+)$/', $idOrUrl, $res); |
|
167
|
|
|
$movieId = count($res) ? $res[2] : null; |
|
168
|
|
|
return $movieId; |
|
169
|
|
|
} |
|
170
|
|
|
|
|
171
|
|
|
private function getMovieTorrentUrl($movieId, $hash = '') |
|
172
|
|
|
{ |
|
173
|
|
|
return 'http://weburg.net/ajax/download/movie?' |
|
174
|
|
|
. ($hash ? 'hash=' . $hash . '&' : '') |
|
175
|
|
|
. 'obj_id=' . $movieId; |
|
176
|
|
|
} |
|
177
|
|
|
|
|
178
|
|
|
private function getMovieInfo($body) |
|
179
|
|
|
{ |
|
180
|
|
|
$info = []; |
|
181
|
|
|
|
|
182
|
|
|
$checks = [ |
|
183
|
|
|
'title' => '<title>(.*?) — Weburg<\/title>', |
|
184
|
|
|
'comments' => 'Комментариев: (\d+)', |
|
185
|
|
|
'rating_kinopoisk' => 'external-ratings-link_type_kinopoisk.*?>(\d+\.\d+)', |
|
186
|
|
|
'rating_imdb' => 'external-ratings-link_type_imdb.*?>(\d+\.\d+)', |
|
187
|
|
|
'rating_votes' => 'count-votes" value="([0-9]+)"', |
|
188
|
|
|
]; |
|
189
|
|
|
|
|
190
|
|
|
foreach ($checks as $name => $regexp) { |
|
191
|
|
|
preg_match('/' . $regexp . '/mis', $body, $res); |
|
192
|
|
|
$info[$name] = count($res) ? $res[1] : null; |
|
193
|
|
|
} |
|
194
|
|
|
|
|
195
|
|
|
preg_match_all('/js-search-button.*?hash="(.*?)"/mis', $body, $res); |
|
196
|
|
|
$info['hashes'] = count($res) ? $res[1] : null; |
|
197
|
|
|
|
|
198
|
|
|
return $info; |
|
199
|
|
|
} |
|
200
|
|
|
|
|
201
|
|
|
/** |
|
202
|
|
|
* @param $body |
|
203
|
|
|
* @param $fromTimestamp |
|
204
|
|
|
* @return bool|string date if matched, false if not |
|
205
|
|
|
*/ |
|
206
|
|
|
private function checkTorrentDate($body, $fromTimestamp) |
|
207
|
|
|
{ |
|
208
|
|
|
preg_match('/(\d{2})\.(\d{2})\.(\d{4})/mis', $body, $res); |
|
209
|
|
|
if (empty($res)) { |
|
210
|
|
|
return false; |
|
211
|
|
|
} |
|
212
|
|
|
|
|
213
|
|
|
$torrentTimestamp = mktime(0, 0, 0, $res[2], $res[1], $res[3]); |
|
214
|
|
|
return $torrentTimestamp >= $fromTimestamp ? $res[0] : false; |
|
215
|
|
|
} |
|
216
|
|
|
|
|
217
|
|
|
private function getInfoUrls($body) |
|
218
|
|
|
{ |
|
219
|
|
|
preg_match_all('/\/movies\/info\/([0-9]+)/', $body, $res); |
|
220
|
|
|
$moviesIds = array_unique($res[1]); |
|
221
|
|
|
return $moviesIds; |
|
222
|
|
|
} |
|
223
|
|
|
|
|
224
|
|
|
/** |
|
225
|
|
|
* @param $body |
|
226
|
|
|
* @return array |
|
227
|
|
|
*/ |
|
228
|
|
|
private function getTorrentsUrls($body) |
|
229
|
|
|
{ |
|
230
|
|
|
preg_match_all('/(http:\/\/.*?gettorrent.*?)"/', $body, $res); |
|
231
|
|
|
$torrentsUrls = $res[1]; |
|
232
|
|
|
$torrentsUrls = array_unique($torrentsUrls); |
|
233
|
|
|
return $torrentsUrls; |
|
234
|
|
|
} |
|
235
|
|
|
} |
|
236
|
|
|
|
In PHP, under loose comparison (like
==, or!=, orswitchconditions), values of different types might be equal.For
stringvalues, the empty string''is a special case, in particular the following results might be unexpected: