WeburgClient::getMovieIdByQuery() - Code Metrics - popstas/transmission-cli - Measure and Improve Code Quality continuously with Scrutinizer

WeburgClient::getMovieIdByQuery() A
last analyzed 2019-03-02 09:37 UTC

↳ Parent: WeburgClient

Complexity

Conditions	4
Paths	2

Size

Total Lines

Duplication

Lines	0
Ratio	0 %

Importance

Changes

Metric	Value
dl	0
loc	10
rs	9.9332
c	0
b	0
f	0
cc	4
nc	2
nop	1

<?php

namespace Popstas\Transmission\Console;

use GuzzleHttp\ClientInterface;
use GuzzleHttp\Cookie\CookieJar;
use GuzzleHttp\Exception\GuzzleException;
use Psr\Http\Message\ResponseInterface;
use Psr\Http\Message\StreamInterface;
use RuntimeException;

class WeburgClient
{
    /**
     * @var ClientInterface
     */
    private $httpClient;

    private $requestDelay;

    /**
     * WeburgClient constructor.
     * @param ClientInterface $httpClient
     * @param int $requestDelay
     */
    public function __construct(ClientInterface $httpClient, $requestDelay = 0)
    {
        $this->httpClient = $httpClient;
        $this->requestDelay = $requestDelay;
    }

    /**
     * @param $movieId
     * @return array
     * @throws GuzzleException
     */
    public function getMovieInfoById($movieId)
    {
        $movieUrl = $this->getMovieUrl($movieId);
        $body = $this->getUrlBody($movieUrl);
        $body = iconv('WINDOWS-1251', 'UTF-8', $body);
        $info = $this->getMovieInfo($body);
        return $info;
    }

    /**
     * @param $movieId
     * @return array urls of movie (not series)
     * @throws GuzzleException
     */
    public function getMovieTorrentUrlsById($movieId)
    {
        $torrentUrl = $this->getMovieTorrentUrl($movieId);
        $body = $this->getUrlBody($torrentUrl);
        $torrentsUrls = $this->getTorrentsUrls($body);
        return $torrentsUrls;
    }

    /**
     * @param $movieId
     * @param $hashes array hashes of torrents from movieInfo
     * @param int $daysMax torrents older last days will not matched
     * @param int $allowedMisses after x misses next checks will broken
     * @return array urls of matched torrent files
     * @throws GuzzleException
     */
    public function getSeriesTorrents($movieId, $hashes, $daysMax = 1, $allowedMisses = 0)
    {
        $torrentsUrls = [];
        $timestampFrom = strtotime('-' . $daysMax . 'days');

        $hashes = array_reverse($hashes);
        foreach ($hashes as $hash) {
            if ($allowedMisses < 0) {
                break;
            }

            $torrentUrl = $this->getMovieTorrentUrl($movieId, $hash);
            $body = $this->getUrlBody($torrentUrl);

            if ($this->checkTorrentDate($body, $timestampFrom) === false) {
                $allowedMisses--;
                continue;
            }

            $torrentsUrls = array_merge($torrentsUrls, $this->getTorrentsUrls($body));
        }

        return $torrentsUrls;
    }

    /**
     * @param string $moviesUrl
     * @return array
     * @throws GuzzleException
     */
    public function getMoviesIds($moviesUrl = 'https://weburg.net/movies/new/?clever_title=1&template=0&last=0')
    {
        if (!$moviesUrl) {
            $moviesUrl = 'https://weburg.net/movies/new/?clever_title=1&template=0&last=0';
        }
        $jsonRaw = $this->getUrlBody($moviesUrl, [
            'Content-Type'     => 'text/html; charset=utf-8',
            'User-Agent'       => 'Mozilla/5.0 (X11; Ubuntu; Linux x86_64; rv:27.0) Gecko/20100101 Firefox/27.0',
            'X-Requested-With' => 'XMLHttpRequest',
        ]);

        $moviesJson = @json_decode($jsonRaw);
        $html = $moviesJson ? $moviesJson->items : strval($jsonRaw);

        $moviesIds = $this->getInfoUrls($html);

        return $moviesIds;
    }

    /**
     * @param $q
     * @return bool|string
     * @throws GuzzleException
     */
    public function getMovieIdByQuery($q)
    {
        $results = $this->movieQuery($q);

        if ($results && $results[0] && $results[0]->object_id) {
            return $results[0]->object_id;
        }

        return false;
    }

    /**
     * @param $q
     * @return mixed
     * @throws GuzzleException
     */
    public function movieQuery($q)
    {
        $resultUrl = 'https://weburg.net/ajax/autocomplete/search/main?' . http_build_query(['q' => $q]);

        $jsonRaw = $this->getUrlBody($resultUrl, [
            'Content-Type' => 'text/html; charset=utf-8',
            'User-Agent'   => 'Mozilla/5.0 (X11; Ubuntu; Linux x86_64; rv:27.0) Gecko/20100101 Firefox/27.0',
            // 'X-Requested-With' => 'XMLHttpRequest',
        ]);

        $resultJson = json_decode($jsonRaw);

        return $resultJson;
    }

    public static function getMovieUrl($movieId)
    {
        return 'https://weburg.net/movies/info/' . $movieId;
    }

    /**
     * @param $url
     * @param array $headers
     * @return ResponseInterface
     * @throws RuntimeException
     * @throws GuzzleException
     */
    private function getUrl($url, $headers = [])
    {
        $jar = new CookieJar();

        $res = $this->httpClient->request('GET', $url, [
            'headers' => $headers,
            'cookies' => $jar,
        ]);

        if ($res->getStatusCode() != 200) {
            throw new RuntimeException('Error ' . $res->getStatusCode() . 'while get url ' . $url);
        }

        sleep($this->requestDelay);

        return $res;
    }

    /**
     * @param string $url
     * @param array $headers
     * @return StreamInterface
     * @throws RuntimeException
     * @throws GuzzleException
     */
    public function getUrlBody($url, $headers = [])
    {
        $res = $this->getUrl($url, $headers);
        $body = $res->getBody();
        return $body;
    }

    /**
     * @param $url
     * @param $torrentsDir
     * @return string path to downloaded file
     * @throws RuntimeException
     * @throws GuzzleException
     */
    public function downloadTorrent($url, $torrentsDir)
    {
        $res = $this->getUrl($url);
        $torrentBody = $res->getBody();

        $disposition = $res->getHeader('content-disposition');
        preg_match('/filename="(.*?)"/', $disposition[0], $res);
        $filename = $res[1];

        $filePath = $torrentsDir . '/' . $filename;
        file_put_contents($filePath, $torrentBody);

        return $filePath;
    }

    public function isTorrentPopular($movieInfo, $commentsMin, $imdbMin, $kinopoiskMin, $votesMin)
    {
        return $movieInfo['comments'] >= $commentsMin
            || $movieInfo['rating_imdb'] >= $imdbMin
            || $movieInfo['rating_kinopoisk'] >= $kinopoiskMin
            || $movieInfo['rating_votes'] >= $votesMin;
    }

    public function cleanMovieId($idOrUrl)
    {
        if (preg_match('/^\d+$/', $idOrUrl)) {
            return $idOrUrl;
        }
        preg_match('/^https?:\/\/weburg\.net\/(series|movies)\/info\/(\d+)$/', $idOrUrl, $res);
        $movieId = count($res) ? $res[2] : null;
        return $movieId;
    }

    private static function getMovieTorrentUrl($movieId, $hash = '')
    {
        return 'https://weburg.net/ajax/download/movie?'
            . ($hash ? 'hash=' . $hash . '&' : '')
            . 'obj_id=' . $movieId;
    }

    private static function getMovieInfo($body)
    {
        $info = [];

        $checks = [
            'title'            => '<title>(.*?)( — Weburg)?( — Вебург)?<\/title>',
            'comments'         => 'Комментариев:&nbsp;(\d+)',
            'rating_kinopoisk' => 'external-ratings-link_type_kinopoisk.*?>(\d+\.\d+)',
            'rating_imdb'      => 'external-ratings-link_type_imdb.*?>(\d+\.\d+)',
            'rating_votes'     => 'count-votes" value="([0-9]+)"',
        ];

        foreach ($checks as $name => $regexp) {
            preg_match('/' . $regexp . '/mis', $body, $res);
            $info[$name] = count($res) ? $res[1] : null;
        }

        preg_match_all('/js-search-button.*?hash="(.*?)"/mis', $body, $res);
        $info['hashes'] = count($res) ? $res[1] : null;

        return $info;
    }

    /**
     * @param $body
     * @param $fromTimestamp
     * @return bool|string date if matched, false if not
     */
    private static function checkTorrentDate($body, $fromTimestamp)
    {
        preg_match('/(\d{2})\.(\d{2})\.(\d{4})/mis', $body, $res);
        if (empty($res)) {
            return false;
        }

        $torrentTimestamp = mktime(0, 0, 0, $res[2], $res[1], $res[3]);
        return $torrentTimestamp >= $fromTimestamp ? $res[0] : false;
    }

    private static function getInfoUrls($body)
    {
        preg_match_all('/\/movies\/info\/([0-9]+)/', $body, $res);
        $moviesIds = array_unique($res[1]);
        return $moviesIds;
    }

    /**
     * @param $body
     * @return array
     */
    private static function getTorrentsUrls($body)
    {
        preg_match_all('/(http:\/\/.*?gettorrent.*?)"/', $body, $res);
        $torrentsUrls = $res[1];
        $torrentsUrls = array_unique($torrentsUrls);
        return $torrentsUrls;
    }
}


1			<?php
2
3			namespace Popstas\Transmission\Console;
4
5			use GuzzleHttp\ClientInterface;
6			use GuzzleHttp\Cookie\CookieJar;
7			use GuzzleHttp\Exception\GuzzleException;
8			use Psr\Http\Message\ResponseInterface;
9			use Psr\Http\Message\StreamInterface;
10			use RuntimeException;
11
12			class WeburgClient
13			{
14			/**
15			* @var ClientInterface
16			*/
17			private $httpClient;
18
19			private $requestDelay;
20
21			/**
22			* WeburgClient constructor.
23			* @param ClientInterface $httpClient
24			* @param int $requestDelay
25			*/
26			public function __construct(ClientInterface $httpClient, $requestDelay = 0)
27			{
28			$this->httpClient = $httpClient;
29			$this->requestDelay = $requestDelay;
30			}
31
32			/**
33			* @param $movieId
34			* @return array
35			* @throws GuzzleException
36			*/
37			public function getMovieInfoById($movieId)
38			{
39			$movieUrl = $this->getMovieUrl($movieId);
40			$body = $this->getUrlBody($movieUrl);
41			$body = iconv('WINDOWS-1251', 'UTF-8', $body);
42			$info = $this->getMovieInfo($body);
43			return $info;
44			}
45
46			/**
47			* @param $movieId
48			* @return array urls of movie (not series)
49			* @throws GuzzleException
50			*/
51			public function getMovieTorrentUrlsById($movieId)
52			{
53			$torrentUrl = $this->getMovieTorrentUrl($movieId);
54			$body = $this->getUrlBody($torrentUrl);
55			$torrentsUrls = $this->getTorrentsUrls($body);
56			return $torrentsUrls;
57			}
58
59			/**
60			* @param $movieId
61			* @param $hashes array hashes of torrents from movieInfo
62			* @param int $daysMax torrents older last days will not matched
63			* @param int $allowedMisses after x misses next checks will broken
64			* @return array urls of matched torrent files
65			* @throws GuzzleException
66			*/
67			public function getSeriesTorrents($movieId, $hashes, $daysMax = 1, $allowedMisses = 0)
68			{
69			$torrentsUrls = [];
70			$timestampFrom = strtotime('-' . $daysMax . 'days');
71
72			$hashes = array_reverse($hashes);
73			foreach ($hashes as $hash) {
74			if ($allowedMisses < 0) {
75			break;
76			}
77
78			$torrentUrl = $this->getMovieTorrentUrl($movieId, $hash);
79			$body = $this->getUrlBody($torrentUrl);
80
81			if ($this->checkTorrentDate($body, $timestampFrom) === false) {
82			$allowedMisses--;
83			continue;
84			}
85
86			$torrentsUrls = array_merge($torrentsUrls, $this->getTorrentsUrls($body));
87			}
88
89			return $torrentsUrls;
90			}
91
92			/**
93			* @param string $moviesUrl
94			* @return array
95			* @throws GuzzleException
96			*/
97			public function getMoviesIds($moviesUrl = 'https://weburg.net/movies/new/?clever_title=1&template=0&last=0')
98			{
99			if (!$moviesUrl) {
100			$moviesUrl = 'https://weburg.net/movies/new/?clever_title=1&template=0&last=0';
101			}
102			$jsonRaw = $this->getUrlBody($moviesUrl, [
103			'Content-Type' => 'text/html; charset=utf-8',
104			'User-Agent' => 'Mozilla/5.0 (X11; Ubuntu; Linux x86_64; rv:27.0) Gecko/20100101 Firefox/27.0',
105			'X-Requested-With' => 'XMLHttpRequest',
106			]);
107
108			$moviesJson = @json_decode($jsonRaw);
109			$html = $moviesJson ? $moviesJson->items : strval($jsonRaw);
110
111			$moviesIds = $this->getInfoUrls($html);
112
113			return $moviesIds;
114			}
115
116			/**
117			* @param $q
118			* @return bool\|string
119			* @throws GuzzleException
120			*/
121			public function getMovieIdByQuery($q)
122			{
123			$results = $this->movieQuery($q);
124
125			if ($results && $results[0] && $results[0]->object_id) {
126			return $results[0]->object_id;
127			}
128
129			return false;
130			}
131
132			/**
133			* @param $q
134			* @return mixed
135			* @throws GuzzleException
136			*/
137			public function movieQuery($q)
138			{
139			$resultUrl = 'https://weburg.net/ajax/autocomplete/search/main?' . http_build_query(['q' => $q]);
140
141			$jsonRaw = $this->getUrlBody($resultUrl, [
142			'Content-Type' => 'text/html; charset=utf-8',
143			'User-Agent' => 'Mozilla/5.0 (X11; Ubuntu; Linux x86_64; rv:27.0) Gecko/20100101 Firefox/27.0',
144			// 'X-Requested-With' => 'XMLHttpRequest',
145			]);
146
147			$resultJson = json_decode($jsonRaw);
148
149			return $resultJson;
150			}
151
152			public static function getMovieUrl($movieId)
153			{
154			return 'https://weburg.net/movies/info/' . $movieId;
155			}
156
157			/**
158			* @param $url
159			* @param array $headers
160			* @return ResponseInterface
161			* @throws RuntimeException
162			* @throws GuzzleException
163			*/
164			private function getUrl($url, $headers = [])
165			{
166			$jar = new CookieJar();
167
168			$res = $this->httpClient->request('GET', $url, [
169			'headers' => $headers,
170			'cookies' => $jar,
171			]);
172
173			if ($res->getStatusCode() != 200) {
174			throw new RuntimeException('Error ' . $res->getStatusCode() . 'while get url ' . $url);
175			}
176
177			sleep($this->requestDelay);
178
179			return $res;
180			}
181
182			/**
183			* @param string $url
184			* @param array $headers
185			* @return StreamInterface
186			* @throws RuntimeException
187			* @throws GuzzleException
188			*/
189			public function getUrlBody($url, $headers = [])
190			{
191			$res = $this->getUrl($url, $headers);
192			$body = $res->getBody();
193			return $body;
194			}
195
196			/**
197			* @param $url
198			* @param $torrentsDir
199			* @return string path to downloaded file
200			* @throws RuntimeException
201			* @throws GuzzleException
202			*/
203			public function downloadTorrent($url, $torrentsDir)
204			{
205			$res = $this->getUrl($url);
206			$torrentBody = $res->getBody();
207
208			$disposition = $res->getHeader('content-disposition');
209			preg_match('/filename="(.*?)"/', $disposition[0], $res);
210			$filename = $res[1];
211
212			$filePath = $torrentsDir . '/' . $filename;
213			file_put_contents($filePath, $torrentBody);
214
215			return $filePath;
216			}
217
218			public function isTorrentPopular($movieInfo, $commentsMin, $imdbMin, $kinopoiskMin, $votesMin)
219			{
220			return $movieInfo['comments'] >= $commentsMin
221			\|\| $movieInfo['rating_imdb'] >= $imdbMin
222			\|\| $movieInfo['rating_kinopoisk'] >= $kinopoiskMin
223			\|\| $movieInfo['rating_votes'] >= $votesMin;
224			}
225
226			public function cleanMovieId($idOrUrl)
227			{
228			if (preg_match('/^\d+$/', $idOrUrl)) {
229			return $idOrUrl;
230			}
231			preg_match('/^https?:\/\/weburg\.net\/(series\|movies)\/info\/(\d+)$/', $idOrUrl, $res);
232			$movieId = count($res) ? $res[2] : null;
233			return $movieId;
234			}
235
236			private static function getMovieTorrentUrl($movieId, $hash = '')
237			{
238			return 'https://weburg.net/ajax/download/movie?'
239			. ($hash ? 'hash=' . $hash . '&' : '')
240			. 'obj_id=' . $movieId;
241			}
242
243			private static function getMovieInfo($body)
244			{
245			$info = [];
246
247			$checks = [
248			'title' => '<title>(.*?)( — Weburg)?( — Вебург)?<\/title>',
249			'comments' => 'Комментариев: (\d+)',
250			'rating_kinopoisk' => 'external-ratings-link_type_kinopoisk.*?>(\d+\.\d+)',
251			'rating_imdb' => 'external-ratings-link_type_imdb.*?>(\d+\.\d+)',
252			'rating_votes' => 'count-votes" value="([0-9]+)"',
253			];
254
255			foreach ($checks as $name => $regexp) {
256			preg_match('/' . $regexp . '/mis', $body, $res);
257			$info[$name] = count($res) ? $res[1] : null;
258			}
259
260			preg_match_all('/js-search-button.?hash="(.?)"/mis', $body, $res);
261			$info['hashes'] = count($res) ? $res[1] : null;
262
263			return $info;
264			}
265
266			/**
267			* @param $body
268			* @param $fromTimestamp
269			* @return bool\|string date if matched, false if not
270			*/
271			private static function checkTorrentDate($body, $fromTimestamp)
272			{
273			preg_match('/(\d{2})\.(\d{2})\.(\d{4})/mis', $body, $res);
274			if (empty($res)) {
275			return false;
276			}
277
278			$torrentTimestamp = mktime(0, 0, 0, $res[2], $res[1], $res[3]);
279			return $torrentTimestamp >= $fromTimestamp ? $res[0] : false;
280			}
281
282			private static function getInfoUrls($body)
283			{
284			preg_match_all('/\/movies\/info\/([0-9]+)/', $body, $res);
285			$moviesIds = array_unique($res[1]);
286			return $moviesIds;
287			}
288
289			/**
290			* @param $body
291			* @return array
292			*/
293			private static function getTorrentsUrls($body)
294			{
295			preg_match_all('/(http:\/\/.?gettorrent.?)"/', $body, $res);
296			$torrentsUrls = $res[1];
297			$torrentsUrls = array_unique($torrentsUrls);
298			return $torrentsUrls;
299			}
300			}
301

popstas / transmission-cli

WeburgClient::getMovieIdByQuery() A last analyzed 2019-03-02 09:37 UTC

Complexity

Size

Duplication

Importance

Duplication Side-by-Side

Filter issues like

WeburgClient::getMovieIdByQuery() A
last analyzed 2019-03-02 09:37 UTC