|
| 1 | +<?php |
| 2 | +namespace hmerritt; |
| 3 | + |
| 4 | +/** |
| 5 | +* Class Imdb |
| 6 | +* |
| 7 | +* |
| 8 | +* @package hmerritt/imdb-api |
| 9 | +* @author Harry Merritt |
| 10 | +*/ |
| 11 | +class Imdb |
| 12 | +{ |
| 13 | + |
| 14 | + /** |
| 15 | + * Returns default options combined with any user options |
| 16 | + * |
| 17 | + * @param string $options |
| 18 | + * @return array $defaults |
| 19 | + */ |
| 20 | + private function populateOptions(array $options = []): array |
| 21 | + { |
| 22 | + // Default options |
| 23 | + $defaults = [ |
| 24 | + 'cache' => true, |
| 25 | + 'category' => 'all', |
| 26 | + 'curlHeaders' => ['Accept-Language: en-US,en;q=0.5'], |
| 27 | + 'techSpecs' => true, |
| 28 | + ]; |
| 29 | + |
| 30 | + // Merge any user options with the default ones |
| 31 | + foreach ($options as $key => $option) { |
| 32 | + $defaults[$key] = $option; |
| 33 | + } |
| 34 | + |
| 35 | + // Return final options array |
| 36 | + return $defaults; |
| 37 | + } |
| 38 | + |
| 39 | + /** |
| 40 | + * Gets film data from IMDB. Will first search if the |
| 41 | + * film name is passed instead of film-id |
| 42 | + * @param string $filmId |
| 43 | + * @param array $options |
| 44 | + * @return array $response |
| 45 | + */ |
| 46 | + public function film(string $filmId, array $options = []): array |
| 47 | + { |
| 48 | + // Combine user options with default ones |
| 49 | + $options = $this->populateOptions($options); |
| 50 | + |
| 51 | + // Initiate response object |
| 52 | + // -> handles what the api returns |
| 53 | + $response = new Response; |
| 54 | + |
| 55 | + // Initiate cache object |
| 56 | + // -> handles storing/retrieving cached results |
| 57 | + $cache = new Cache; |
| 58 | + |
| 59 | + // Initiate dom object |
| 60 | + // -> handles page scraping |
| 61 | + $dom = new Dom; |
| 62 | + |
| 63 | + // Initiate html-pieces object |
| 64 | + // -> handles finding specific content from the dom |
| 65 | + $htmlPieces = new HtmlPieces; |
| 66 | + |
| 67 | + // Check for 'tt' at start of $filmId |
| 68 | + if (substr($filmId, 0, 2) !== "tt") |
| 69 | + { |
| 70 | + // Search $filmId and use first result |
| 71 | + $search_film = $this->search($filmId, [ "category" => "tt" ]); |
| 72 | + if ($htmlPieces->count($search_film["titles"]) > 0) |
| 73 | + { |
| 74 | + // Use first film returned from search |
| 75 | + $filmId = $search_film["titles"][0]["id"]; |
| 76 | + } else |
| 77 | + { |
| 78 | + // No film found |
| 79 | + // -> return default (empty) response |
| 80 | + return $response->default('film'); |
| 81 | + } |
| 82 | + } |
| 83 | + |
| 84 | + // If caching is enabled |
| 85 | + if ($options["cache"]) { |
| 86 | + // Check cache for film |
| 87 | + if ($cache->has($filmId)) { |
| 88 | + return $cache->get($filmId)->film; |
| 89 | + } |
| 90 | + } |
| 91 | + |
| 92 | + // Load imdb film page and parse the dom |
| 93 | + $page = $dom->fetch("https://www.imdb.com/title/".$filmId, $options); |
| 94 | + |
| 95 | + // Add all film data to response $store |
| 96 | + $response->add("id", $filmId); |
| 97 | + $response->add("title", $htmlPieces->get($page, "title")); |
| 98 | + $response->add("year", $htmlPieces->get($page, "year")); |
| 99 | + $response->add("length", $htmlPieces->get($page, "length")); |
| 100 | + $response->add("plot", $htmlPieces->get($page, "plot")); |
| 101 | + $response->add("rating", $htmlPieces->get($page, "rating")); |
| 102 | + $response->add("rating_votes", $htmlPieces->get($page, "rating_votes")); |
| 103 | + $response->add("poster", $htmlPieces->get($page, "poster")); |
| 104 | + $response->add("trailer", $htmlPieces->get($page, "trailer")); |
| 105 | + $response->add("cast", $htmlPieces->get($page, "cast")); |
| 106 | + |
| 107 | + // If techSpecs is enabled in user $options |
| 108 | + // -> Make a second request to load the full techSpecs page |
| 109 | + if ($options["techSpecs"]) { |
| 110 | + $page_techSpecs = $dom->fetch("https://www.imdb.com/title/$filmId/technical", $options); |
| 111 | + $response->add("technical_specs", $htmlPieces->get($page_techSpecs, "technical_specs")); |
| 112 | + } |
| 113 | + else { |
| 114 | + $response->add("technical_specs", []); |
| 115 | + } |
| 116 | + |
| 117 | + // If caching is enabled |
| 118 | + if ($options["cache"]) { |
| 119 | + // Add result to the cache |
| 120 | + $cache->add($filmId, $response->return()); |
| 121 | + } |
| 122 | + |
| 123 | + // Return the response $store |
| 124 | + return $response->return(); |
| 125 | + } |
| 126 | + |
| 127 | + /** |
| 128 | + * Searches IMDB for films, people and companies |
| 129 | + * @param string $search |
| 130 | + * @param array $options |
| 131 | + * @return array $searchData |
| 132 | + */ |
| 133 | + public function search(string $search, array $options = []): array |
| 134 | + { |
| 135 | + // Combine user options with default ones |
| 136 | + $options = $this->populateOptions($options); |
| 137 | + |
| 138 | + // Initiate response object |
| 139 | + // -> handles what the api returns |
| 140 | + $response = new Response; |
| 141 | + |
| 142 | + // Initiate dom object |
| 143 | + // -> handles page scraping |
| 144 | + $dom = new Dom; |
| 145 | + |
| 146 | + // Initiate html-pieces object |
| 147 | + // -> handles finding specific content from the dom |
| 148 | + $htmlPieces = new HtmlPieces; |
| 149 | + |
| 150 | + // Load imdb search page and parse the dom |
| 151 | + $page = $dom->fetch("https://www.imdb.com/find?q=$search&s=".$options["category"], $options); |
| 152 | + |
| 153 | + // Add all search data to response $store |
| 154 | + $response->add("titles", $htmlPieces->get($page, "titles")); |
| 155 | + $response->add("names", $htmlPieces->get($page, "names")); |
| 156 | + $response->add("companies", $htmlPieces->get($page, "companies")); |
| 157 | + |
| 158 | + return $response->return(); |
| 159 | + } |
| 160 | + |
| 161 | +} |
0 commit comments