AebnPipe::extractFromJsonLd() - Code Metrics - NNTmux/newznab-tmux - Measure and Improve Code Quality continuously with Scrutinizer

AebnPipe::extractFromJsonLd() F
last analyzed 2025-12-27 16:35 UTC

↳ Parent: AebnPipe

Complexity

Conditions	19
Paths	1801

Size

Total Lines	71
Code Lines	38

Duplication

Lines	0
Ratio	0 %

Importance

Changes	1
Bugs	0	Features	0

Metric	Value
eloc	38
c	1
b	0
f	0
dl	0
loc	71
rs	0.3499
cc	19
nc	1801
nop	0

How to fix Long Method Complexity

<?php

namespace App\Services\AdultProcessing\Pipes;

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

/**
 * AEBN (Adult Entertainment Broadcast Network) provider pipe.
 *
 * Handles movie information extraction from straight.theater.aebn.net
 * with support for JSON-LD structured data and multiple fallback selectors.
 */
class AebnPipe extends AbstractAdultProviderPipe
{
    protected int $priority = 10;

    private const BASE_URL = 'https://straight.theater.aebn.net';
    private const SEARCH_URL = '/dispatcher/fts?theaterId=13992&genreId=101&locale=en&count=30&imageType=Large&targetSearchMode=basic&isAdvancedSearch=false&isFlushAdvancedSearchCriteria=false&sortType=Relevance&userQuery=title%3A+%2B';
    private const TRAILER_URL = '/dispatcher/previewPlayer?locale=en&theaterId=13992&genreId=101&movieId=';

    protected string $directUrl = '';
    protected string $title = '';
    protected array $res = [];
    protected string $response = '';
    protected ?array $jsonLdData = null;

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

    public function getDisplayName(): string
    {
        return 'Adult Entertainment Broadcast Network';
    }

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

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

        // Check cache first
        $cached = $this->getCachedSearch($movie);
        if ($cached !== null) {
            if ($cached === false) {
                return AdultProcessingResult::notFound($this->getName());
            }
            return AdultProcessingResult::matched(
                $cached['title'] ?? $movie,
                $this->getName(),
                $cached
            );
        }

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

        if ($searchResult === false) {
            $this->cacheSearchResult($movie, 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());
        }

        // Try to extract JSON-LD data first
        $this->jsonLdData = $this->extractJsonLd($this->response);

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

        $movieInfo = $this->getMovieInfo();

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

        // Cache the successful result
        $this->cacheSearchResult($movie, $movieInfo);

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

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

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

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

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

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

        $bestMatch = null;
        $highestSimilarity = 0;
        $i = 1;

        // Try multiple container selectors for search results
        $containerSelectors = [
            'div.movie',
            'div.search-result',
            'div[class*="movie"]',
            'article.movie',
        ];

        $movies = [];
        foreach ($containerSelectors as $containerSelector) {
            $movies = $this->getHtmlParser()->find($containerSelector);
            if (!empty($movies) && count($movies) > 0) {
                break;
            }
        }

        foreach ($movies as $mov) {
            // Try multiple selector patterns for the title link
            $selectors = [
                'a#FTSMovieSearch_link_title_detail_' . $i,
                'a.title-link',
                'a[href*="/movie/"]',
                'h3 a',
                'a[title]',
            ];

            $ret = null;
            foreach ($selectors as $selector) {
                $ret = $mov->findOne($selector);
                if ($ret && isset($ret->href)) {
                    break;
                }
            }

            if ($ret && isset($ret->href)) {
                $title = $ret->title ?? trim($ret->plaintext ?? '');

                if (!empty($title)) {
                    $similarity = $this->calculateSimilarity($movie, $title);

                    if ($similarity > $highestSimilarity) {
                        $highestSimilarity = $similarity;
                        $bestMatch = [
                            'title' => trim($title),
                            'url' => self::BASE_URL . html_entity_decode($ret->href),
                            'similarity' => $similarity,
                        ];
                    }
                }
            }
            $i++;
        }

        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;
        }

        // Try to get data from JSON-LD first (most reliable)
        if ($this->jsonLdData !== null) {
            $results = array_merge($results, $this->extractFromJsonLd());
        }

        // Get all the movie data (HTML fallback)
        $synopsis = $this->extractSynopsis();
        if (is_array($synopsis) && !empty($synopsis)) {
            $results = array_merge($results, $synopsis);
        }

        $productInfo = $this->extractProductInfo(true);
        if (is_array($productInfo) && !empty($productInfo)) {
            $results = array_merge($results, $productInfo);
        }

        $cast = $this->extractCast();
        if (is_array($cast) && !empty($cast)) {
            $results = array_merge($results, $cast);
        }

        $genres = $this->extractGenres();
        if (is_array($genres) && !empty($genres)) {
            $results = array_merge($results, $genres);
        }

        $covers = $this->extractCovers();
        if (is_array($covers) && !empty($covers)) {
            $results = array_merge($results, $covers);
        }

        $trailers = $this->extractTrailers();
        if (is_array($trailers) && !empty($trailers)) {
            $results = array_merge($results, $trailers);
        }

        if (empty($results) || (empty($results['title'] ?? '') && empty($results['boxcover'] ?? ''))) {
            return false;
        }

        return $results;
    }

    /**
     * Extract data from JSON-LD structured data.
     */
    protected function extractFromJsonLd(): array
    {
        $results = [];

        if ($this->jsonLdData === null) {
            return $results;
        }

        // Title
        if (!empty($this->jsonLdData['name'])) {
            $results['title'] = $this->jsonLdData['name'];
        }

        // Synopsis/Description
        if (!empty($this->jsonLdData['description'])) {
            $results['synopsis'] = $this->jsonLdData['description'];
        }

        // Image/Cover
        if (!empty($this->jsonLdData['image'])) {
            $image = is_array($this->jsonLdData['image']) ? ($this->jsonLdData['image'][0] ?? '') : $this->jsonLdData['image'];
            if (!empty($image)) {
                $results['boxcover'] = $image;
                $results['backcover'] = str_ireplace(['xlf.jpg', 'front'], ['xlb.jpg', 'back'], $image);
            }
        }

        // Duration
        if (!empty($this->jsonLdData['duration'])) {
            $results['duration'] = $this->jsonLdData['duration'];
        }

        // Director
        if (!empty($this->jsonLdData['director'])) {
            $director = $this->jsonLdData['director'];
            if (is_array($director)) {
                $results['director'] = $director['name'] ?? ($director[0]['name'] ?? '');
            } else {
                $results['director'] = $director;
            }
        }

        // Actors
        if (!empty($this->jsonLdData['actor'])) {
            $actors = $this->jsonLdData['actor'];
            $cast = [];
            if (is_array($actors)) {
                foreach ($actors as $actor) {
                    if (is_array($actor) && !empty($actor['name'])) {
                        $cast[] = $actor['name'];
                    } elseif (is_string($actor)) {
                        $cast[] = $actor;
                    }
                }
            }
            if (!empty($cast)) {
                $results['cast'] = $cast;
            }
        }

        // Genre
        if (!empty($this->jsonLdData['genre'])) {
            $genres = $this->jsonLdData['genre'];
            if (is_array($genres)) {
                $results['genres'] = $genres;
            } else {
                $results['genres'] = [$genres];
            }
        }

        return $results;
    }

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

        // Try multiple selectors
        $selectors = [
            'a[itemprop=trailer]',
            'a[href*="previewPlayer"]',
            'a.trailer-link',
        ];

        foreach ($selectors as $selector) {
            $ret = $this->getHtmlParser()->find($selector, 0);
            if ($ret && isset($ret->href) && preg_match('/movieId=(?<movieid>\d+)/', trim($ret->href), $hits)) {
                $movieid = $hits['movieid'];
                $res['trailers']['url'] = self::BASE_URL . self::TRAILER_URL . $movieid;
                return $res;
            }
        }

        return $res;
    }

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

        // Try multiple selectors
        $selectors = [
            'img[itemprop=thumbnailUrl]',
            'img[itemprop=image]',
            'div#md-boxCover img',
            'div.boxcover img',
            'img.boxcover',
            'img[src*="boxcover"]',
            'meta[property="og:image"]',
        ];

        foreach ($selectors as $selector) {
            $ret = $this->getHtmlParser()->findOne($selector);
            if ($ret) {
                $coverUrl = $ret->src ?? $ret->content ?? null;

                if (!empty($coverUrl)) {
                    // Ensure URL has protocol
                    if (str_starts_with($coverUrl, '//')) {
                        $coverUrl = 'https:' . $coverUrl;
                    }

                    // Get high-resolution versions
                    $res['boxcover'] = str_ireplace(['160w.jpg', '120w.jpg', '_small', '_thumb'], ['xlf.jpg', 'xlf.jpg', '', ''], $coverUrl);
                    $res['backcover'] = str_ireplace(['160w.jpg', '120w.jpg', 'xlf.jpg', 'front'], ['xlb.jpg', 'xlb.jpg', 'xlb.jpg', 'back'], $res['boxcover']);

                    return $res;
                }
            }
        }

        return $res;
    }

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

        // Try multiple selectors
        $selectors = [
            'div.md-detailsCategories a[itemprop=genre]',
            'a[itemprop=genre]',
            'span[itemprop=genre]',
            'div.categories a',
            'a[href*="/category/"]',
        ];

        foreach ($selectors as $selector) {
            $elements = $this->getHtmlParser()->find($selector);
            if (!empty($elements)) {
                foreach ($elements as $genre) {
                    $text = trim($genre->plaintext ?? '');
                    if (!empty($text)) {
                        $genres[] = $text;
                    }
                }
                if (!empty($genres)) {
                    break;
                }
            }
        }

        if (!empty($genres)) {
            $res['genres'] = array_unique($genres);
        }

        return $res;
    }

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

        // Try multiple selectors
        $selectors = [
            'div.starsFull span[itemprop=name]',
            'span[itemprop=name]',
            'a[href*="/stars/"]',
            'div.cast a',
            'div.performers a',
        ];

        foreach ($selectors as $selector) {
            $elements = $this->getHtmlParser()->find($selector);
            if (!empty($elements)) {
                foreach ($elements as $star) {
                    $text = trim($star->plaintext ?? '');
                    if (!empty($text) && strlen($text) > 2) {
                        $cast[] = $text;
                    }
                }
                if (!empty($cast)) {
                    break;
                }
            }
        }

        // Fallback: try detailsLink div
        if (empty($cast)) {
            $ret = $this->getHtmlParser()->findOne('div.detailsLink');
            if ($ret && !($ret instanceof \voku\helper\SimpleHtmlDomNodeBlank)) {
                foreach ($ret->find('span') as $star) {
                    $text = $star->plaintext ?? '';
                    if (str_contains($text, '/More/') && str_contains($text, '/Stars/')) {
                        $cast[] = trim($text);
                    }
                }
            }
        }

        if (!empty($cast)) {
            $res['cast'] = array_unique($cast);
        }

        return $res;
    }

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

        // Try multiple selectors
        $selectors = [
            'div#md-detailsLeft',
            'div.movie-details',
            'div.product-info',
        ];

        foreach ($selectors as $selector) {
            $ret = $this->getHtmlParser()->find($selector, 0);
            if ($ret) {
                $productinfo = [];
                foreach ($ret->find('div') as $div) {
                    foreach ($div->find('span') as $span) {
                        $text = rawurldecode($span->plaintext ?? '');
                        $text = preg_replace('/&nbsp;/', '', $text);
                        $text = trim($text);
                        if (!empty($text)) {
                            $productinfo[] = $text;
                        }
                    }
                }

                if (!empty($productinfo)) {
                    if (false !== $key = array_search('Running Time:', $productinfo, false)) {
                        unset($productinfo[$key + 2]);
                    }

                    if (false !== $key = array_search('Director:', $productinfo, false)) {
                        $res['director'] = $productinfo[$key + 1] ?? '';
                        unset($productinfo[$key], $productinfo[$key + 1]);
                    }

                    $res['productinfo'] = array_chunk(array_values($productinfo), 2, false);
                    break;
                }
            }
        }

        return $res;
    }

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

        // Try multiple selectors in order of preference
        $selectors = [
            'span[itemprop=about]',
            'span[itemprop=description]',
            'div[itemprop=description]',
            'div.movieDetailDescription',
            'div.synopsis',
            'p.description',
            'meta[name="description"]',
            'meta[property="og:description"]',
        ];

        foreach ($selectors as $selector) {
            $ret = $this->getHtmlParser()->findOne($selector);
            if ($ret) {
                $text = $ret->plaintext ?? $ret->content ?? '';
                if (!empty(trim($text))) {
                    $text = trim($text);
                    $text = preg_replace('/^Description:\s*/i', '', $text);
                    $res['synopsis'] = $text;
                    return $res;
                }
            }
        }

        return $res;
    }
}



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			* AEBN (Adult Entertainment Broadcast Network) provider pipe.
10			*
11			* Handles movie information extraction from straight.theater.aebn.net
12			* with support for JSON-LD structured data and multiple fallback selectors.
13			*/
14			class AebnPipe extends AbstractAdultProviderPipe
15			{
16			protected int $priority = 10;
17
18			private const BASE_URL = 'https://straight.theater.aebn.net';
19			private const SEARCH_URL = '/dispatcher/fts?theaterId=13992&genreId=101&locale=en&count=30&imageType=Large&targetSearchMode=basic&isAdvancedSearch=false&isFlushAdvancedSearchCriteria=false&sortType=Relevance&userQuery=title%3A+%2B';
20			private const TRAILER_URL = '/dispatcher/previewPlayer?locale=en&theaterId=13992&genreId=101&movieId=';
21
22			protected string $directUrl = '';
23			protected string $title = '';
24			protected array $res = [];
25			protected string $response = '';
26			protected ?array $jsonLdData = null;
27
28			public function getName(): string
29			{
30			return 'aebn';
31			}
32
33			public function getDisplayName(): string
34			{
35			return 'Adult Entertainment Broadcast Network';
36			}
37
38			protected function getBaseUrl(): string
39			{
40			return self::BASE_URL;
41			}
42
43			protected function process(AdultProcessingPassable $passable): AdultProcessingResult
44			{
45			$movie = $passable->getCleanTitle();
46
47			// Check cache first
48			$cached = $this->getCachedSearch($movie);
49			if ($cached !== null) {
50			if ($cached === false) {
51			return AdultProcessingResult::notFound($this->getName());
52			}
53			return AdultProcessingResult::matched(
54			$cached['title'] ?? $movie,
55			$this->getName(),
56			$cached
57			);
58			}
59
60			$searchResult = $this->search($movie);
61
62			if ($searchResult === false) {
63			$this->cacheSearchResult($movie, false);
64			$this->outputNotFound();
65			return AdultProcessingResult::notFound($this->getName());
66			}
67
68			$this->title = $searchResult['title'];
69			$this->directUrl = $searchResult['url'];
70
71			// Fetch the movie details page
72			$this->response = $this->fetchHtml($this->directUrl, $this->cookie);
73
74			if ($this->response === false) {
			0 ignored issues – show introduced 2025-12-16 14:06 UTC by Report Bug Copy Issue Report The condition `$this->response === false` is always `false`. Loading history...
75			return AdultProcessingResult::failed('Failed to fetch movie details page', $this->getName());
76			}
77
78			// Try to extract JSON-LD data first
79			$this->jsonLdData = $this->extractJsonLd($this->response);
80
81			$this->getHtmlParser()->loadHtml($this->response);
82
83			$movieInfo = $this->getMovieInfo();
84
85			if ($movieInfo === false) {
86			$this->cacheSearchResult($movie, false);
87			return AdultProcessingResult::notFound($this->getName());
88			}
89
90			// Cache the successful result
91			$this->cacheSearchResult($movie, $movieInfo);
92
93			$this->outputMatch($this->title);
94
95			return AdultProcessingResult::matched(
96			$this->title,
97			$this->getName(),
98			$movieInfo
99			);
100			}
101
102			protected function search(string $movie): array\|false
103			{
104			if (empty($movie)) {
105			return false;
106			}
107
108			$searchUrl = self::BASE_URL . self::SEARCH_URL . urlencode($movie);
109			$response = $this->fetchHtml($searchUrl, $this->cookie);
110
111			if ($response === false) {
112			return false;
113			}
114
115			$this->getHtmlParser()->loadHtml($response);
116
117			$bestMatch = null;
118			$highestSimilarity = 0;
119			$i = 1;
120
121			// Try multiple container selectors for search results
122			$containerSelectors = [
123			'div.movie',
124			'div.search-result',
125			'div[class*="movie"]',
126			'article.movie',
127			];
128
129			$movies = [];
130			foreach ($containerSelectors as $containerSelector) {
131			$movies = $this->getHtmlParser()->find($containerSelector);
132			if (!empty($movies) && count($movies) > 0) {
133			break;
134			}
135			}
136
137			foreach ($movies as $mov) {
138			// Try multiple selector patterns for the title link
139			$selectors = [
140			'a#FTSMovieSearch_link_title_detail_' . $i,
141			'a.title-link',
142			'a[href*="/movie/"]',
143			'h3 a',
144			'a[title]',
145			];
146
147			$ret = null;
148			foreach ($selectors as $selector) {
149			$ret = $mov->findOne($selector);
150			if ($ret && isset($ret->href)) {
151			break;
152			}
153			}
154
155			if ($ret && isset($ret->href)) {
156			$title = $ret->title ?? trim($ret->plaintext ?? '');
157
158			if (!empty($title)) {
159			$similarity = $this->calculateSimilarity($movie, $title);
160
161			if ($similarity > $highestSimilarity) {
162			$highestSimilarity = $similarity;
163			$bestMatch = [
164			'title' => trim($title),
165			'url' => self::BASE_URL . html_entity_decode($ret->href),
166			'similarity' => $similarity,
167			];
168			}
169			}
170			}
171			$i++;
172			}
173
174			if ($bestMatch !== null && $highestSimilarity >= $this->minimumSimilarity) {
175			return $bestMatch;
176			}
177
178			return false;
179			}
180
181			protected function getMovieInfo(): array\|false
182			{
183			$results = [];
184
185			if (!empty($this->directUrl)) {
186			if (!empty($this->title)) {
187			$results['title'] = $this->title;
188			}
189			$results['directurl'] = $this->directUrl;
190			}
191
192			// Try to get data from JSON-LD first (most reliable)
193			if ($this->jsonLdData !== null) {
194			$results = array_merge($results, $this->extractFromJsonLd());
195			}
196
197			// Get all the movie data (HTML fallback)
198			$synopsis = $this->extractSynopsis();
199			if (is_array($synopsis) && !empty($synopsis)) {
200			$results = array_merge($results, $synopsis);
201			}
202
203			$productInfo = $this->extractProductInfo(true);
204			if (is_array($productInfo) && !empty($productInfo)) {
205			$results = array_merge($results, $productInfo);
206			}
207
208			$cast = $this->extractCast();
209			if (is_array($cast) && !empty($cast)) {
210			$results = array_merge($results, $cast);
211			}
212
213			$genres = $this->extractGenres();
214			if (is_array($genres) && !empty($genres)) {
215			$results = array_merge($results, $genres);
216			}
217
218			$covers = $this->extractCovers();
219			if (is_array($covers) && !empty($covers)) {
220			$results = array_merge($results, $covers);
221			}
222
223			$trailers = $this->extractTrailers();
224			if (is_array($trailers) && !empty($trailers)) {
225			$results = array_merge($results, $trailers);
226			}
227
228			if (empty($results) \|\| (empty($results['title'] ?? '') && empty($results['boxcover'] ?? ''))) {
229			return false;
230			}
231
232			return $results;
233			}
234
235			/**
236			* Extract data from JSON-LD structured data.
237			*/
238			protected function extractFromJsonLd(): array
239			{
240			$results = [];
241
242			if ($this->jsonLdData === null) {
243			return $results;
244			}
245
246			// Title
247			if (!empty($this->jsonLdData['name'])) {
248			$results['title'] = $this->jsonLdData['name'];
249			}
250
251			// Synopsis/Description
252			if (!empty($this->jsonLdData['description'])) {
253			$results['synopsis'] = $this->jsonLdData['description'];
254			}
255
256			// Image/Cover
257			if (!empty($this->jsonLdData['image'])) {
258			$image = is_array($this->jsonLdData['image']) ? ($this->jsonLdData['image'][0] ?? '') : $this->jsonLdData['image'];
259			if (!empty($image)) {
260			$results['boxcover'] = $image;
261			$results['backcover'] = str_ireplace(['xlf.jpg', 'front'], ['xlb.jpg', 'back'], $image);
262			}
263			}
264
265			// Duration
266			if (!empty($this->jsonLdData['duration'])) {
267			$results['duration'] = $this->jsonLdData['duration'];
268			}
269
270			// Director
271			if (!empty($this->jsonLdData['director'])) {
272			$director = $this->jsonLdData['director'];
273			if (is_array($director)) {
274			$results['director'] = $director['name'] ?? ($director[0]['name'] ?? '');
275			} else {
276			$results['director'] = $director;
277			}
278			}
279
280			// Actors
281			if (!empty($this->jsonLdData['actor'])) {
282			$actors = $this->jsonLdData['actor'];
283			$cast = [];
284			if (is_array($actors)) {
285			foreach ($actors as $actor) {
286			if (is_array($actor) && !empty($actor['name'])) {
287			$cast[] = $actor['name'];
288			} elseif (is_string($actor)) {
289			$cast[] = $actor;
290			}
291			}
292			}
293			if (!empty($cast)) {
294			$results['cast'] = $cast;
295			}
296			}
297
298			// Genre
299			if (!empty($this->jsonLdData['genre'])) {
300			$genres = $this->jsonLdData['genre'];
301			if (is_array($genres)) {
302			$results['genres'] = $genres;
303			} else {
304			$results['genres'] = [$genres];
305			}
306			}
307
308			return $results;
309			}
310
311			protected function extractTrailers(): array
312			{
313			$res = [];
314
315			// Try multiple selectors
316			$selectors = [
317			'a[itemprop=trailer]',
318			'a[href*="previewPlayer"]',
319			'a.trailer-link',
320			];
321
322			foreach ($selectors as $selector) {
323			$ret = $this->getHtmlParser()->find($selector, 0);
324			if ($ret && isset($ret->href) && preg_match('/movieId=(?<movieid>\d+)/', trim($ret->href), $hits)) {
325			$movieid = $hits['movieid'];
326			$res['trailers']['url'] = self::BASE_URL . self::TRAILER_URL . $movieid;
327			return $res;
328			}
329			}
330
331			return $res;
332			}
333
334			protected function extractCovers(): array
335			{
336			$res = [];
337
338			// Try multiple selectors
339			$selectors = [
340			'img[itemprop=thumbnailUrl]',
341			'img[itemprop=image]',
342			'div#md-boxCover img',
343			'div.boxcover img',
344			'img.boxcover',
345			'img[src*="boxcover"]',
346			'meta[property="og:image"]',
347			];
348
349			foreach ($selectors as $selector) {
350			$ret = $this->getHtmlParser()->findOne($selector);
351			if ($ret) {
352			$coverUrl = $ret->src ?? $ret->content ?? null;
353
354			if (!empty($coverUrl)) {
355			// Ensure URL has protocol
356			if (str_starts_with($coverUrl, '//')) {
357			$coverUrl = 'https:' . $coverUrl;
358			}
359
360			// Get high-resolution versions
361			$res['boxcover'] = str_ireplace(['160w.jpg', '120w.jpg', '_small', '_thumb'], ['xlf.jpg', 'xlf.jpg', '', ''], $coverUrl);
362			$res['backcover'] = str_ireplace(['160w.jpg', '120w.jpg', 'xlf.jpg', 'front'], ['xlb.jpg', 'xlb.jpg', 'xlb.jpg', 'back'], $res['boxcover']);
363
364			return $res;
365			}
366			}
367			}
368
369			return $res;
370			}
371
372			protected function extractGenres(): array
373			{
374			$res = [];
375			$genres = [];
376
377			// Try multiple selectors
378			$selectors = [
379			'div.md-detailsCategories a[itemprop=genre]',
380			'a[itemprop=genre]',
381			'span[itemprop=genre]',
382			'div.categories a',
383			'a[href*="/category/"]',
384			];
385
386			foreach ($selectors as $selector) {
387			$elements = $this->getHtmlParser()->find($selector);
388			if (!empty($elements)) {
389			foreach ($elements as $genre) {
390			$text = trim($genre->plaintext ?? '');
391			if (!empty($text)) {
392			$genres[] = $text;
393			}
394			}
395			if (!empty($genres)) {
396			break;
397			}
398			}
399			}
400
401			if (!empty($genres)) {
402			$res['genres'] = array_unique($genres);
403			}
404
405			return $res;
406			}
407
408			protected function extractCast(): array
409			{
410			$res = [];
411			$cast = [];
412
413			// Try multiple selectors
414			$selectors = [
415			'div.starsFull span[itemprop=name]',
416			'span[itemprop=name]',
417			'a[href*="/stars/"]',
418			'div.cast a',
419			'div.performers a',
420			];
421
422			foreach ($selectors as $selector) {
423			$elements = $this->getHtmlParser()->find($selector);
424			if (!empty($elements)) {
425			foreach ($elements as $star) {
426			$text = trim($star->plaintext ?? '');
427			if (!empty($text) && strlen($text) > 2) {
428			$cast[] = $text;
429			}
430			}
431			if (!empty($cast)) {
432			break;
433			}
434			}
435			}
436
437			// Fallback: try detailsLink div
438			if (empty($cast)) {
439			$ret = $this->getHtmlParser()->findOne('div.detailsLink');
440			if ($ret && !($ret instanceof \voku\helper\SimpleHtmlDomNodeBlank)) {
441			foreach ($ret->find('span') as $star) {
442			$text = $star->plaintext ?? '';
443			if (str_contains($text, '/More/') && str_contains($text, '/Stars/')) {
444			$cast[] = trim($text);
445			}
446			}
447			}
448			}
449
450			if (!empty($cast)) {
451			$res['cast'] = array_unique($cast);
452			}
453
454			return $res;
455			}
456
457			protected function extractProductInfo(bool $extras = false): array
458			{
459			$res = [];
460
461			// Try multiple selectors
462			$selectors = [
463			'div#md-detailsLeft',
464			'div.movie-details',
465			'div.product-info',
466			];
467
468			foreach ($selectors as $selector) {
469			$ret = $this->getHtmlParser()->find($selector, 0);
470			if ($ret) {
471			$productinfo = [];
472			foreach ($ret->find('div') as $div) {
473			foreach ($div->find('span') as $span) {
474			$text = rawurldecode($span->plaintext ?? '');
475			$text = preg_replace('/ /', '', $text);
476			$text = trim($text);
477			if (!empty($text)) {
478			$productinfo[] = $text;
479			}
480			}
481			}
482
483			if (!empty($productinfo)) {
484			if (false !== $key = array_search('Running Time:', $productinfo, false)) {
485			unset($productinfo[$key + 2]);
486			}
487
488			if (false !== $key = array_search('Director:', $productinfo, false)) {
489			$res['director'] = $productinfo[$key + 1] ?? '';
490			unset($productinfo[$key], $productinfo[$key + 1]);
491			}
492
493			$res['productinfo'] = array_chunk(array_values($productinfo), 2, false);
494			break;
495			}
496			}
497			}
498
499			return $res;
500			}
501
502			protected function extractSynopsis(): array
503			{
504			$res = [];
505
506			// Try multiple selectors in order of preference
507			$selectors = [
508			'span[itemprop=about]',
509			'span[itemprop=description]',
510			'div[itemprop=description]',
511			'div.movieDetailDescription',
512			'div.synopsis',
513			'p.description',
514			'meta[name="description"]',
515			'meta[property="og:description"]',
516			];
517
518			foreach ($selectors as $selector) {
519			$ret = $this->getHtmlParser()->findOne($selector);
520			if ($ret) {
521			$text = $ret->plaintext ?? $ret->content ?? '';
522			if (!empty(trim($text))) {
523			$text = trim($text);
524			$text = preg_replace('/^Description:\s*/i', '', $text);
525			$res['synopsis'] = $text;
526			return $res;
527			}
528			}
529			}
530
531			return $res;
532			}
533			}
534
535

NNTmux / newznab-tmux

AebnPipe::extractFromJsonLd() F last analyzed 2025-12-27 16:35 UTC

Complexity

Size

Duplication

Importance

How to fix Long Method Complexity

Long Method

Duplication Side-by-Side

Filter issues like

AebnPipe::extractFromJsonLd() F
last analyzed 2025-12-27 16:35 UTC