AdePipe::extractGenres() - Code Metrics - Inspection of "Improve age verification cookies handling" - NNTmux/newznab-tmux - Measure and Improve Code Quality continuously with Scrutinizer

Passed

Push — master ( 73c322...402528 )

by Darko

created 2025-12-22 11:29 UTC

AdePipe::extractGenres() B

↳ Parent: AdePipe

Complexity

Conditions	7
Paths	8

Size

Total Lines	31
Code Lines	17

Duplication

Lines	0
Ratio	0 %

Importance

Changes	1
Bugs	0	Features	0

Metric	Value
eloc	17
c	1
b	0
f	0
dl	0
loc	31
rs	8.8333
cc	7
nc	8
nop	0

<?php

namespace App\Services\AdultProcessing\Pipes;

use App\Services\AdultProcessing\AdultProcessingPassable;
use App\Services\AdultProcessing\AdultProcessingResult;

/**
 * Adult DVD Empire (ADE) provider pipe.
 *
 * Handles movie information extraction from adultdvdempire.com
 */
class AdePipe extends AbstractAdultProviderPipe
{
    protected int $priority = 40;

    private const BASE_URL = 'https://www.adultdvdempire.com';
    private const SEARCH_URL = '/dvd/search?q=';

    protected string $directUrl = '';
    protected string $title = '';
    protected string $response = '';

    public function getName(): string
    {
        return 'ade';
    }

    public function getDisplayName(): string
    {
        return 'Adult DVD Empire';
    }

    protected function getBaseUrl(): string
    {
        return self::BASE_URL;
    }

    protected function process(AdultProcessingPassable $passable): AdultProcessingResult
    {
        $movie = $passable->getCleanTitle();

        $searchResult = $this->search($movie);

        if ($searchResult === false) {

            $this->outputNotFound();
            return AdultProcessingResult::notFound($this->getName());
        }

        $this->title = $searchResult['title'];
        $this->directUrl = $searchResult['url'];

        // Fetch the movie details page
        $this->response = $this->fetchHtml($this->directUrl, $this->cookie);

        if ($this->response === false) {
            return AdultProcessingResult::failed('Failed to fetch movie details page', $this->getName());
        }

        $this->getHtmlParser()->loadHtml($this->response);

        $movieInfo = $this->getMovieInfo();

        if ($movieInfo === false) {
            return AdultProcessingResult::notFound($this->getName());
        }

        $this->outputMatch($this->title);

        return AdultProcessingResult::matched(
            $this->title,
            $this->getName(),
            $movieInfo
        );
    }

    protected function search(string $movie): array|false
    {
        if (empty($movie)) {
            return false;
        }

        // Initialize session with age verification cookies
        $this->initializeSession();

        $searchUrl = self::BASE_URL . self::SEARCH_URL . rawurlencode($movie);
        $response = $this->fetchHtml($searchUrl, $this->cookie);

        if ($response === false) {
            return false;
        }

        $this->getHtmlParser()->loadHtml($response);

        // Try multiple search result selectors
        $resultSelectors = [
            'a[class=fancybox-button]',
            'div.card a.boxcover-link',
            'a[href*="/item/"]',
        ];

        $bestMatch = null;
        $highestSimilarity = 0;

        foreach ($resultSelectors as $selector) {
            $res = $this->getHtmlParser()->find($selector);
            if (!empty($res)) {
                foreach ($res as $ret) {
                    $title = $ret->title ?? $ret->getAttribute('title') ?? trim($ret->plaintext);
                    $url = trim($ret->href ?? '');

                    if (empty($title) || empty($url)) {
                        continue;
                    }

                    $similarity = $this->calculateSimilarity($movie, $title);

                    if ($similarity > $highestSimilarity) {
                        $highestSimilarity = $similarity;
                        $bestMatch = [
                            'title' => trim($title),
                            'url' => str_starts_with($url, 'http') ? $url : self::BASE_URL . $url,
                        ];
                    }
                }

                // If we found results with this selector, don't try others
                if ($bestMatch !== null) {
                    break;
                }
            }
        }

        if ($bestMatch !== null && $highestSimilarity >= $this->minimumSimilarity) {

            return $bestMatch;
        }

        return false;
    }

    protected function getMovieInfo(): array|false
    {
        $results = [];

        if (!empty($this->directUrl)) {
            if (!empty($this->title)) {
                $results['title'] = $this->title;
            }
            $results['directurl'] = $this->directUrl;
        }

        // Get all the movie data
        $synopsis = $this->extractSynopsis();
        if (is_array($synopsis)) {

            $results = array_merge($results, $synopsis);
        }

        $productInfo = $this->extractProductInfo(true);
        if (is_array($productInfo)) {

            $results = array_merge($results, $productInfo);
        }

        $cast = $this->extractCast();
        if (is_array($cast)) {

            $results = array_merge($results, $cast);
        }

        $genres = $this->extractGenres();
        if (is_array($genres)) {

            $results = array_merge($results, $genres);
        }

        $covers = $this->extractCovers();
        if (is_array($covers)) {

            $results = array_merge($results, $covers);
        }

        $trailers = $this->extractTrailers();
        if (is_array($trailers)) {

            $results = array_merge($results, $trailers);
        }

        if (empty($results)) {
            return false;
        }

        return $results;
    }

    protected function extractTrailers(): array
    {
        $res = [];

        $trailersUrl = str_replace('/item/', '/item/trailers/', $this->directUrl);
        $trailersResponse = $this->fetchHtml($trailersUrl, $this->cookie);

        if ($trailersResponse !== false) {
            if (preg_match("/([\"|'])(?P<swf>[^\"']+.swf)([\"|'])/i", $trailersResponse, $hits)) {
                $res['trailers']['url'] = self::BASE_URL . trim(trim($hits['swf']), '"');

                if (preg_match('#(?:streamID:\s\")(?P<streamid>[0-9A-Z]+)(?:\")#', $trailersResponse, $hits)) {
                    $res['trailers']['streamid'] = trim($hits['streamid']);
                }

                if (preg_match('#(?:BaseStreamingUrl:\s\")(?P<baseurl>[\d]+\.[\d]+\.[\d]+\.[\d]+)(?:\")#', $trailersResponse, $hits)) {
                    $res['trailers']['baseurl'] = $hits['baseurl'];
                }
            }
        }

        return $res;
    }

    protected function extractCovers(): array
    {
        $res = [];

        // Try multiple selectors for better reliability
        $selectors = [
            'div#Boxcover img[itemprop=image]',
            'img[itemprop=image]',
            'div#Boxcover img',
            'div.boxcover img',
        ];

        foreach ($selectors as $selector) {
            $ret = $this->getHtmlParser()->findOne($selector);
            if ($ret && isset($ret->src)) {
                // Get high-resolution covers
                $res['boxcover'] = preg_replace('/[ms]\.jpg$/', 'h.jpg', $ret->src);
                $res['backcover'] = preg_replace('/[ms]\.jpg$/', 'bh.jpg', $ret->src);

                return $res;
            }
        }

        return $res;
    }

    protected function extractSynopsis(): array
    {
        $res = [];

        // Try multiple selectors in priority order
        $selectors = [
            'meta[property="og:description"]' => 'content',
            'meta[name="description"]' => 'content',
            'div[itemprop="description"]' => 'plaintext',
            'p.synopsis' => 'plaintext',
        ];

        foreach ($selectors as $selector => $property) {
            $meta = $this->getHtmlParser()->findOne($selector);
            if ($meta && isset($meta->$property) && $meta->$property !== false && !empty(trim($meta->$property))) {
                $res['synopsis'] = trim($meta->$property);

                return $res;
            }
        }

        return $res;
    }

    protected function extractCast(): array
    {
        $res = [];
        $cast = [];

        // Try multiple selector strategies
        $selectors = [
            'div[itemprop="actor"] span[itemprop="name"]',
            'div.performer-list a',
            'a[href*="/performer/"]',
        ];

        foreach ($selectors as $selector) {
            $elements = $this->getHtmlParser()->find($selector);
            if (!empty($elements)) {
                foreach ($elements as $a) {
                    if ($a->plaintext !== false && !empty(trim($a->plaintext))) {
                        $cast[] = trim($a->plaintext);
                    }
                }

                if (!empty($cast)) {
                    break;
                }
            }
        }

        $res['cast'] = array_values(array_unique($cast));

        return $res;
    }

    protected function extractGenres(): array
    {
        $res = [];
        $genres = [];

        // Try multiple selector strategies
        $selectors = [
            'a[Label="Category"]',
            'div.categories a',
            'a[href*="/category/"]',
            'span[itemprop="genre"]',
        ];

        foreach ($selectors as $selector) {
            $elements = $this->getHtmlParser()->find($selector);
            if (!empty($elements)) {
                foreach ($elements as $a) {
                    if ($a->plaintext !== false && !empty(trim($a->plaintext))) {
                        $genres[] = trim($a->plaintext);
                    }
                }

                if (!empty($genres)) {
                    break;
                }
            }
        }

        $res['genres'] = array_values(array_unique($genres));

        return $res;
    }

    protected function extractProductInfo(bool $extras = false): array
    {
        $res = [];
        $dofeature = null;

        $tmpResponse = str_ireplace('Section ProductInfo', 'spdinfo', $this->response);
        $tmpHtml = new \voku\helper\HtmlDomParser();
filter:
    dependency_paths: ["lib/*"]
        $tmpHtml->loadHtml($tmpResponse);

        if ($ret = $tmpHtml->findOne('div[class=spdinfo]')) {
            $productinfo = [];
            $extrasData = [];

            $tmpResponse = trim($ret->outertext);
            $ret = $tmpHtml->loadHtml($tmpResponse);

            foreach ($ret->find('text') as $strong) {
                if (trim($strong->innertext) === 'Features') {
                    $dofeature = true;
                }
                if ($dofeature !== true) {
                    if (trim($strong->innertext) !== '&nbsp;') {
                        $productinfo[] = trim($strong->innertext);
                    }
                } else {
                    if ($extras === true) {
                        $extrasData[] = trim($strong->innertext);
                    }
                }
            }

            array_shift($productinfo);
            array_shift($productinfo);
            $res['productinfo'] = array_chunk($productinfo, 2, false);

            if (!empty($extrasData)) {
                $res['extras'] = $extrasData;
            }
        }

        return $res;
    }

    /**
     * Initialize session by visiting the site to establish cookies.
     * ADE uses JavaScript-based age verification with cookies.
     */
    protected function initializeSession(): void
    {
        try {
            $client = $this->getHttpClient();

            // Visit the homepage first to establish a session
            $client->get(self::BASE_URL, [
                'headers' => $this->getDefaultHeaders(),
                'allow_redirects' => true,
            ]);

            usleep(300000); // 300ms delay

        } catch (\Exception $e) {
            \Illuminate\Support\Facades\Log::debug('ADE session initialization: ' . $e->getMessage());
        }
    }
}



1			<?php
2
3			namespace App\Services\AdultProcessing\Pipes;
4
5			use App\Services\AdultProcessing\AdultProcessingPassable;
6			use App\Services\AdultProcessing\AdultProcessingResult;
7
8			/**
9			* Adult DVD Empire (ADE) provider pipe.
10			*
11			* Handles movie information extraction from adultdvdempire.com
12			*/
13			class AdePipe extends AbstractAdultProviderPipe
14			{
15			protected int $priority = 40;
16
17			private const BASE_URL = 'https://www.adultdvdempire.com';
18			private const SEARCH_URL = '/dvd/search?q=';
19
20			protected string $directUrl = '';
21			protected string $title = '';
22			protected string $response = '';
23
24			public function getName(): string
25			{
26			return 'ade';
27			}
28
29			public function getDisplayName(): string
30			{
31			return 'Adult DVD Empire';
32			}
33
34			protected function getBaseUrl(): string
35			{
36			return self::BASE_URL;
37			}
38
39			protected function process(AdultProcessingPassable $passable): AdultProcessingResult
40			{
41			$movie = $passable->getCleanTitle();
42
43			$searchResult = $this->search($movie);
44
45			if ($searchResult === false) {
			0 ignored issues – show introduced 2025-12-16 14:06 UTC by Report Bug Copy Issue Report The condition `$searchResult === false` is always `true`. Loading history...
46			$this->outputNotFound();
47			return AdultProcessingResult::notFound($this->getName());
48			}
49
50			$this->title = $searchResult['title'];
51			$this->directUrl = $searchResult['url'];
52
53			// Fetch the movie details page
54			$this->response = $this->fetchHtml($this->directUrl, $this->cookie);
55
56			if ($this->response === false) {
57			return AdultProcessingResult::failed('Failed to fetch movie details page', $this->getName());
58			}
59
60			$this->getHtmlParser()->loadHtml($this->response);
61
62			$movieInfo = $this->getMovieInfo();
63
64			if ($movieInfo === false) {
65			return AdultProcessingResult::notFound($this->getName());
66			}
67
68			$this->outputMatch($this->title);
69
70			return AdultProcessingResult::matched(
71			$this->title,
72			$this->getName(),
73			$movieInfo
74			);
75			}
76
77			protected function search(string $movie): array\|false
78			{
79			if (empty($movie)) {
80			return false;
81			}
82
83			// Initialize session with age verification cookies
84			$this->initializeSession();
85
86			$searchUrl = self::BASE_URL . self::SEARCH_URL . rawurlencode($movie);
87			$response = $this->fetchHtml($searchUrl, $this->cookie);
88
89			if ($response === false) {
90			return false;
91			}
92
93			$this->getHtmlParser()->loadHtml($response);
94
95			// Try multiple search result selectors
96			$resultSelectors = [
97			'a[class=fancybox-button]',
98			'div.card a.boxcover-link',
99			'a[href*="/item/"]',
100			];
101
102			$bestMatch = null;
103			$highestSimilarity = 0;
104
105			foreach ($resultSelectors as $selector) {
106			$res = $this->getHtmlParser()->find($selector);
107			if (!empty($res)) {
108			foreach ($res as $ret) {
109			$title = $ret->title ?? $ret->getAttribute('title') ?? trim($ret->plaintext);
110			$url = trim($ret->href ?? '');
111
112			if (empty($title) \|\| empty($url)) {
113			continue;
114			}
115
116			$similarity = $this->calculateSimilarity($movie, $title);
117
118			if ($similarity > $highestSimilarity) {
119			$highestSimilarity = $similarity;
120			$bestMatch = [
121			'title' => trim($title),
122			'url' => str_starts_with($url, 'http') ? $url : self::BASE_URL . $url,
123			];
124			}
125			}
126
127			// If we found results with this selector, don't try others
128			if ($bestMatch !== null) {
129			break;
130			}
131			}
132			}
133
134			if ($bestMatch !== null && $highestSimilarity >= $this->minimumSimilarity) {
			0 ignored issues – show introduced 2025-12-16 14:06 UTC by Report Bug Copy Issue Report The condition `$bestMatch !== null` is always `false`. Loading history...
135			return $bestMatch;
136			}
137
138			return false;
139			}
140
141			protected function getMovieInfo(): array\|false
142			{
143			$results = [];
144
145			if (!empty($this->directUrl)) {
146			if (!empty($this->title)) {
147			$results['title'] = $this->title;
148			}
149			$results['directurl'] = $this->directUrl;
150			}
151
152			// Get all the movie data
153			$synopsis = $this->extractSynopsis();
154			if (is_array($synopsis)) {
			0 ignored issues – show introduced 2025-12-16 14:06 UTC by Report Bug Copy Issue Report The condition `is_array($synopsis)` is always `true`. Loading history...
155			$results = array_merge($results, $synopsis);
156			}
157
158			$productInfo = $this->extractProductInfo(true);
159			if (is_array($productInfo)) {
			0 ignored issues – show introduced 2025-12-16 14:06 UTC by Report Bug Copy Issue Report The condition `is_array($productInfo)` is always `true`. Loading history...
160			$results = array_merge($results, $productInfo);
161			}
162
163			$cast = $this->extractCast();
164			if (is_array($cast)) {
			0 ignored issues – show introduced 2025-12-16 14:06 UTC by Report Bug Copy Issue Report The condition `is_array($cast)` is always `true`. Loading history...
165			$results = array_merge($results, $cast);
166			}
167
168			$genres = $this->extractGenres();
169			if (is_array($genres)) {
			0 ignored issues – show introduced 2025-12-16 14:06 UTC by Report Bug Copy Issue Report The condition `is_array($genres)` is always `true`. Loading history...
170			$results = array_merge($results, $genres);
171			}
172
173			$covers = $this->extractCovers();
174			if (is_array($covers)) {
			0 ignored issues – show introduced 2025-12-16 14:06 UTC by Report Bug Copy Issue Report The condition `is_array($covers)` is always `true`. Loading history...
175			$results = array_merge($results, $covers);
176			}
177
178			$trailers = $this->extractTrailers();
179			if (is_array($trailers)) {
			0 ignored issues – show introduced 2025-12-16 14:06 UTC by Report Bug Copy Issue Report The condition `is_array($trailers)` is always `true`. Loading history...
180			$results = array_merge($results, $trailers);
181			}
182
183			if (empty($results)) {
184			return false;
185			}
186
187			return $results;
188			}
189
190			protected function extractTrailers(): array
191			{
192			$res = [];
193
194			$trailersUrl = str_replace('/item/', '/item/trailers/', $this->directUrl);
195			$trailersResponse = $this->fetchHtml($trailersUrl, $this->cookie);
196
197			if ($trailersResponse !== false) {
198			if (preg_match("/([\"\|'])(?P<swf>[^\"']+.swf)([\"\|'])/i", $trailersResponse, $hits)) {
199			$res['trailers']['url'] = self::BASE_URL . trim(trim($hits['swf']), '"');
200
201			if (preg_match('#(?:streamID:\s\")(?P<streamid>[0-9A-Z]+)(?:\")#', $trailersResponse, $hits)) {
202			$res['trailers']['streamid'] = trim($hits['streamid']);
203			}
204
205			if (preg_match('#(?:BaseStreamingUrl:\s\")(?P<baseurl>[\d]+\.[\d]+\.[\d]+\.[\d]+)(?:\")#', $trailersResponse, $hits)) {
206			$res['trailers']['baseurl'] = $hits['baseurl'];
207			}
208			}
209			}
210
211			return $res;
212			}
213
214			protected function extractCovers(): array
215			{
216			$res = [];
217
218			// Try multiple selectors for better reliability
219			$selectors = [
220			'div#Boxcover img[itemprop=image]',
221			'img[itemprop=image]',
222			'div#Boxcover img',
223			'div.boxcover img',
224			];
225
226			foreach ($selectors as $selector) {
227			$ret = $this->getHtmlParser()->findOne($selector);
228			if ($ret && isset($ret->src)) {
229			// Get high-resolution covers
230			$res['boxcover'] = preg_replace('/[ms]\.jpg$/', 'h.jpg', $ret->src);
231			$res['backcover'] = preg_replace('/[ms]\.jpg$/', 'bh.jpg', $ret->src);
232
233			return $res;
234			}
235			}
236
237			return $res;
238			}
239
240			protected function extractSynopsis(): array
241			{
242			$res = [];
243
244			// Try multiple selectors in priority order
245			$selectors = [
246			'meta[property="og:description"]' => 'content',
247			'meta[name="description"]' => 'content',
248			'div[itemprop="description"]' => 'plaintext',
249			'p.synopsis' => 'plaintext',
250			];
251
252			foreach ($selectors as $selector => $property) {
253			$meta = $this->getHtmlParser()->findOne($selector);
254			if ($meta && isset($meta->$property) && $meta->$property !== false && !empty(trim($meta->$property))) {
255			$res['synopsis'] = trim($meta->$property);
256
257			return $res;
258			}
259			}
260
261			return $res;
262			}
263
264			protected function extractCast(): array
265			{
266			$res = [];
267			$cast = [];
268
269			// Try multiple selector strategies
270			$selectors = [
271			'div[itemprop="actor"] span[itemprop="name"]',
272			'div.performer-list a',
273			'a[href*="/performer/"]',
274			];
275
276			foreach ($selectors as $selector) {
277			$elements = $this->getHtmlParser()->find($selector);
278			if (!empty($elements)) {
279			foreach ($elements as $a) {
280			if ($a->plaintext !== false && !empty(trim($a->plaintext))) {
281			$cast[] = trim($a->plaintext);
282			}
283			}
284
285			if (!empty($cast)) {
286			break;
287			}
288			}
289			}
290
291			$res['cast'] = array_values(array_unique($cast));
292
293			return $res;
294			}
295
296			protected function extractGenres(): array
297			{
298			$res = [];
299			$genres = [];
300
301			// Try multiple selector strategies
302			$selectors = [
303			'a[Label="Category"]',
304			'div.categories a',
305			'a[href*="/category/"]',
306			'span[itemprop="genre"]',
307			];
308
309			foreach ($selectors as $selector) {
310			$elements = $this->getHtmlParser()->find($selector);
311			if (!empty($elements)) {
312			foreach ($elements as $a) {
313			if ($a->plaintext !== false && !empty(trim($a->plaintext))) {
314			$genres[] = trim($a->plaintext);
315			}
316			}
317
318			if (!empty($genres)) {
319			break;
320			}
321			}
322			}
323
324			$res['genres'] = array_values(array_unique($genres));
325
326			return $res;
327			}
328
329			protected function extractProductInfo(bool $extras = false): array
330			{
331			$res = [];
332			$dofeature = null;
333
334			$tmpResponse = str_ireplace('Section ProductInfo', 'spdinfo', $this->response);
335			$tmpHtml = new \voku\helper\HtmlDomParser();
			0 ignored issues – show Bug introduced 2025-12-16 14:06 UTC by Report Bug Copy Issue Report The type `voku\helper\HtmlDomParser` was not found. Maybe you did not declare it correctly or list all dependencies? The issue could also be caused by a filter entry in the build configuration. If the path has been excluded in your configuration, e.g. `excluded_paths: ["lib/"]`, you can move it to the dependency path list as follows: filter: dependency_paths: ["lib/"] For further information see https://scrutinizer-ci.com/docs/tools/php/php-scrutinizer/#list-dependency-paths Loading history...
336			$tmpHtml->loadHtml($tmpResponse);
337
338			if ($ret = $tmpHtml->findOne('div[class=spdinfo]')) {
339			$productinfo = [];
340			$extrasData = [];
341
342			$tmpResponse = trim($ret->outertext);
343			$ret = $tmpHtml->loadHtml($tmpResponse);
344
345			foreach ($ret->find('text') as $strong) {
346			if (trim($strong->innertext) === 'Features') {
347			$dofeature = true;
348			}
349			if ($dofeature !== true) {
350			if (trim($strong->innertext) !== ' ') {
351			$productinfo[] = trim($strong->innertext);
352			}
353			} else {
354			if ($extras === true) {
355			$extrasData[] = trim($strong->innertext);
356			}
357			}
358			}
359
360			array_shift($productinfo);
361			array_shift($productinfo);
362			$res['productinfo'] = array_chunk($productinfo, 2, false);
363
364			if (!empty($extrasData)) {
365			$res['extras'] = $extrasData;
366			}
367			}
368
369			return $res;
370			}
371
372			/**
373			* Initialize session by visiting the site to establish cookies.
374			* ADE uses JavaScript-based age verification with cookies.
375			*/
376			protected function initializeSession(): void
377			{
378			try {
379			$client = $this->getHttpClient();
380
381			// Visit the homepage first to establish a session
382			$client->get(self::BASE_URL, [
383			'headers' => $this->getDefaultHeaders(),
384			'allow_redirects' => true,
385			]);
386
387			usleep(300000); // 300ms delay
388
389			} catch (\Exception $e) {
390			\Illuminate\Support\Facades\Log::debug('ADE session initialization: ' . $e->getMessage());
391			}
392			}
393			}
394
395

NNTmux / newznab-tmux

Push — master ( 73c322...402528 )

AdePipe::extractGenres() B

Complexity

Size

Duplication

Importance

Duplication Side-by-Side

Filter issues like