|
| 1 | +<?php |
| 2 | + |
| 3 | +namespace AllegroScraper; |
| 4 | + |
| 5 | +/** |
| 6 | + * Extractor (scraper, crawler, parser) of products from Allegro. |
| 7 | + * It receives the search output of Allegro.pl, as well as detailed information about the products. |
| 8 | + * |
| 9 | + * @see https://www.parser.best/en/allegro-scraper-extractor.php |
| 10 | + * @author Telegram: @JWprogrammer |
| 11 | + * @copyright Telegram: @JWprogrammer |
| 12 | + */ |
| 13 | + |
| 14 | +class AllegroScraper |
| 15 | +{ |
| 16 | + private string $db_host; |
| 17 | + private string $db_name; |
| 18 | + private string $db_user; |
| 19 | + private string $db_password; |
| 20 | + |
| 21 | + private bool $remote_access_exists; |
| 22 | + private string $remote_uri; |
| 23 | + private string $remote_api_key; |
| 24 | + |
| 25 | + public function __construct(string $db_host, string $db_name, string $db_user, string $db_password, string $remote_uri = null, string $remote_api_key = null) |
| 26 | + { |
| 27 | + $this->db_host = $db_host; |
| 28 | + $this->db_name = $db_name; |
| 29 | + $this->db_user = $db_user; |
| 30 | + $this->db_password = $db_password; |
| 31 | + |
| 32 | + if (!empty($remote_uri) && !empty($remote_api_key)){ |
| 33 | + $this->remote_access_exists = true; |
| 34 | + $this->remote_uri = $remote_uri; |
| 35 | + $this->remote_api_key = $remote_api_key; |
| 36 | + } |
| 37 | + else { |
| 38 | + $this->remote_access_exists = false; |
| 39 | + } |
| 40 | + } |
| 41 | + |
| 42 | + /** |
| 43 | + * Get search results from Allegro |
| 44 | + * @param array $parameters |
| 45 | + * @param bool $return_as_object |
| 46 | + * @return array|object |
| 47 | + */ |
| 48 | + public function search(array $parameters, bool $return_as_object = true) |
| 49 | + { |
| 50 | + if ($this->isOnThisServer()){ |
| 51 | + return $this->localRequest(__FUNCTION__, $return_as_object, $parameters['language'] ?? '', $parameters['page'] ?? 1, $parameters['category'] ?? null, $parameters['query'] ?? null, $parameters['sort'] ?? null, $parameters['status'] ?? null, $parameters['make'] ?? null, $parameters['model'] ?? null, $parameters['type'] ?? null, $parameters['price_from'] ?? null, $parameters['price_to'] ?? null, $parameters['filters'] ?? [], $parameters['seller_id'] ?? '', $parameters['seller_name'] ?? '', $parameters['show_lokalnie'] ?? false, $parameters['show_filters'] ?? false, $parameters['filters_excluded'] ?? []); |
| 52 | + } |
| 53 | + else { |
| 54 | + return $this->remoteRequest(__FUNCTION__, $return_as_object, $parameters); |
| 55 | + } |
| 56 | + } |
| 57 | + |
| 58 | + /** |
| 59 | + * Get detailed product information from Allegro |
| 60 | + * @param array $parameters |
| 61 | + * @param bool $return_as_object |
| 62 | + * @return array|object |
| 63 | + */ |
| 64 | + public function details(array $parameters, bool $return_as_object = true) |
| 65 | + { |
| 66 | + if ($this->isOnThisServer()){ |
| 67 | + return $this->localRequest(__FUNCTION__, $return_as_object, $parameters['language'] ?? '', $parameters['product_id'] ?? null); |
| 68 | + } |
| 69 | + else { |
| 70 | + return $this->remoteRequest(__FUNCTION__, $return_as_object, $parameters); |
| 71 | + } |
| 72 | + } |
| 73 | + |
| 74 | + /** |
| 75 | + * Get a list of categories from Allegro |
| 76 | + * @param array $parameters |
| 77 | + * @param bool $return_as_object |
| 78 | + * @return array|object |
| 79 | + */ |
| 80 | + public function categories(array $parameters, bool $return_as_object = true) |
| 81 | + { |
| 82 | + if ($this->isOnThisServer()){ |
| 83 | + return $this->localRequest(__FUNCTION__, $return_as_object, $parameters['language'] ?? '', $parameters['category'] ?? null); |
| 84 | + } |
| 85 | + else { |
| 86 | + return $this->remoteRequest(__FUNCTION__, $return_as_object, $parameters); |
| 87 | + } |
| 88 | + } |
| 89 | + |
| 90 | + private function remoteRequest(string $method, bool $return_as_object, array $parameters) |
| 91 | + { |
| 92 | + $parameters['api_key'] = $this->remote_api_key; |
| 93 | + $parameters['method'] = $method; |
| 94 | + $curl = curl_init(); |
| 95 | + curl_setopt_array($curl, [ |
| 96 | + CURLOPT_URL => $this->remote_uri . '?' . http_build_query($parameters), |
| 97 | + CURLOPT_RETURNTRANSFER => true, |
| 98 | + CURLOPT_ENCODING => '', |
| 99 | + CURLOPT_MAXREDIRS => 5, |
| 100 | + CURLOPT_TIMEOUT => 360, |
| 101 | + CURLOPT_FOLLOWLOCATION => true, |
| 102 | + CURLOPT_HTTP_VERSION => CURL_HTTP_VERSION_1_1, |
| 103 | + CURLOPT_USERAGENT => $_SERVER['HTTP_HOST'], |
| 104 | + CURLOPT_CUSTOMREQUEST => 'GET', |
| 105 | + CURLOPT_SSL_VERIFYHOST => false, |
| 106 | + CURLOPT_SSL_VERIFYPEER => false |
| 107 | + ]); |
| 108 | + $response = curl_exec($curl); |
| 109 | + $error = curl_error($curl); |
| 110 | + curl_close($curl); |
| 111 | + if ($error){ |
| 112 | + throw new \RuntimeException($error); |
| 113 | + } |
| 114 | + $result = json_decode($response, !$return_as_object); |
| 115 | + if ($result == null){ |
| 116 | + throw new \RuntimeException('Error while request with response: "'. $response . '"'); |
| 117 | + } |
| 118 | + return $result; |
| 119 | + } |
| 120 | + |
| 121 | + private function localRequest(string $method, bool $return_as_object, $language, ...$parameters) |
| 122 | + { |
| 123 | + $timezone_old = ini_set('date.timezone', 'Europe/Kiev'); |
| 124 | + $array = (new \AllegroScraper\AllegroV4($this->db_host, $this->db_name, $this->db_user, $this->db_password, $language))->{$method}(...$parameters); |
| 125 | + ini_set('date.timezone', $timezone_old ?? 'Europe/Kiev'); |
| 126 | + if ($return_as_object){ |
| 127 | + return self::responseToObject($array); |
| 128 | + } |
| 129 | + else return $array; |
| 130 | + } |
| 131 | + |
| 132 | + private function isOnThisServer() |
| 133 | + { |
| 134 | + if (class_exists('\AllegroScraper\AllegroV4')) { |
| 135 | + return true; |
| 136 | + } |
| 137 | + elseif ($this->remote_access_exists){ |
| 138 | + return false; |
| 139 | + } |
| 140 | + else { |
| 141 | + throw new \RuntimeException('The Allegro scraper is not installed on your server. Please contact: https://t.me/JWprogrammer'); |
| 142 | + } |
| 143 | + } |
| 144 | + |
| 145 | + private static function responseToObject(array &$array) |
| 146 | + { |
| 147 | + $result_obj = new \stdClass(); |
| 148 | + $result_arr = []; |
| 149 | + $has_str_keys = false; |
| 150 | + foreach ($array as $k => &$v) { |
| 151 | + if (!$has_str_keys) { |
| 152 | + $has_str_keys = is_string($k); |
| 153 | + } |
| 154 | + $object = is_array($v) ? self::responseToObject($v) : $v; |
| 155 | + if (!$has_str_keys) { |
| 156 | + $result_arr[$k] = $object; |
| 157 | + } |
| 158 | + $result_obj->{$k} = $object; |
| 159 | + $v = null; |
| 160 | + } |
| 161 | + return ($has_str_keys) ? $result_obj : $result_arr; |
| 162 | + } |
| 163 | +} |
0 commit comments