backend.php 4.0 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196
  1. <?php
  2. class backend{
  3. public function __construct($scraper){
  4. $this->scraper = $scraper;
  5. }
  6. /*
  7. Proxy stuff
  8. */
  9. public function get_ip(){
  10. $pool = constant("config::PROXY_" . strtoupper($this->scraper));
  11. if($pool === false){
  12. // we don't want a proxy, fuck off!
  13. return 'raw_ip::::';
  14. }
  15. // indent
  16. $proxy_index_raw = apcu_inc("p." . $this->scraper);
  17. $proxylist = file_get_contents("data/proxies/" . $pool . ".txt");
  18. $proxylist = explode("\n", $proxylist);
  19. // ignore empty or commented lines
  20. $proxylist = array_filter($proxylist, function($entry){
  21. $entry = ltrim($entry);
  22. return strlen($entry) > 0 && substr($entry, 0, 1) != "#";
  23. });
  24. $proxylist = array_values($proxylist);
  25. return $proxylist[$proxy_index_raw % count($proxylist)];
  26. }
  27. // this function is also called directly on nextpage
  28. public function assign_proxy(&$curlproc, $ip){
  29. // parse proxy line
  30. [
  31. $type,
  32. $address,
  33. $port,
  34. $username,
  35. $password
  36. ] = explode(":", $ip, 5);
  37. switch($type){
  38. case "raw_ip":
  39. return;
  40. break;
  41. case "http":
  42. case "https":
  43. curl_setopt($curlproc, CURLOPT_PROXYTYPE, CURLPROXY_HTTP);
  44. curl_setopt($curlproc, CURLOPT_PROXY, $type . "://" . $address . ":" . $port);
  45. break;
  46. case "socks4":
  47. curl_setopt($curlproc, CURLOPT_PROXYTYPE, CURLPROXY_SOCKS4);
  48. curl_setopt($curlproc, CURLOPT_PROXY, $address . ":" . $port);
  49. break;
  50. case "socks5":
  51. curl_setopt($curlproc, CURLOPT_PROXYTYPE, CURLPROXY_SOCKS5);
  52. curl_setopt($curlproc, CURLOPT_PROXY, $address . ":" . $port);
  53. break;
  54. case "socks4a":
  55. curl_setopt($curlproc, CURLOPT_PROXYTYPE, CURLPROXY_SOCKS4A);
  56. curl_setopt($curlproc, CURLOPT_PROXY, $address . ":" . $port);
  57. break;
  58. case "socks5_hostname":
  59. curl_setopt($curlproc, CURLOPT_PROXYTYPE, CURLPROXY_SOCKS5_HOSTNAME);
  60. curl_setopt($curlproc, CURLOPT_PROXY, $address . ":" . $port);
  61. break;
  62. }
  63. if($username != ""){
  64. curl_setopt($curlproc, CURLOPT_PROXYUSERPWD, $username . ":" . $password);
  65. }
  66. }
  67. /*
  68. Next page stuff
  69. */
  70. public function store($payload, $page, $proxy){
  71. $page = $page[0];
  72. $password = random_bytes(256); // 2048 bit
  73. $salt = random_bytes(16);
  74. $key = hash_pbkdf2("sha512", $password, $salt, 20000, 32, true);
  75. $iv =
  76. random_bytes(
  77. openssl_cipher_iv_length("aes-256-gcm")
  78. );
  79. $tag = "";
  80. $out = openssl_encrypt($payload, "aes-256-gcm", $key, OPENSSL_RAW_DATA, $iv, $tag, "", 16);
  81. $requestid = apcu_inc("requestid");
  82. apcu_store(
  83. $page . "." .
  84. $this->scraper .
  85. $requestid,
  86. gzdeflate($proxy . "," . $salt.$iv.$out.$tag),
  87. 900 // cache information for 15 minutes blaze it
  88. );
  89. return
  90. $this->scraper . $requestid . "." .
  91. rtrim(strtr(base64_encode($password), '+/', '-_'), '=');
  92. }
  93. public function get($npt, $page){
  94. $page = $page[0];
  95. $explode = explode(".", $npt, 2);
  96. if(count($explode) !== 2){
  97. throw new Exception("Malformed nextPageToken!");
  98. }
  99. $apcu = $page . "." . $explode[0];
  100. $key = $explode[1];
  101. $payload = apcu_fetch($apcu);
  102. if($payload === false){
  103. throw new Exception("The nextPageToken is invalid or has expired!");
  104. }
  105. $key =
  106. base64_decode(
  107. str_pad(
  108. strtr($key, '-_', '+/'),
  109. strlen($key) % 4,
  110. '=',
  111. STR_PAD_RIGHT
  112. )
  113. );
  114. $payload = gzinflate($payload);
  115. // get proxy
  116. [
  117. $proxy,
  118. $payload
  119. ] = explode(",", $payload, 2);
  120. $key =
  121. hash_pbkdf2(
  122. "sha512",
  123. $key,
  124. substr($payload, 0, 16), // salt
  125. 20000,
  126. 32,
  127. true
  128. );
  129. $ivlen = openssl_cipher_iv_length("aes-256-gcm");
  130. $payload =
  131. openssl_decrypt(
  132. substr(
  133. $payload,
  134. 16 + $ivlen,
  135. -16
  136. ),
  137. "aes-256-gcm",
  138. $key,
  139. OPENSSL_RAW_DATA,
  140. substr($payload, 16, $ivlen),
  141. substr($payload, -16)
  142. );
  143. if($payload === false){
  144. throw new Exception("The nextPageToken is invalid or has expired!");
  145. }
  146. // remove the key after using
  147. apcu_delete($apcu);
  148. return [$payload, $proxy];
  149. }
  150. }