Parse.php 5.2 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228
  1. <?php
  2. namespace Spider\Lib;
  3. use Dever;
  4. class Parse
  5. {
  6. private $url = '';
  7. private $host = '';
  8. private $log;
  9. private $doc = array();
  10. private $data = array();
  11. public function __construct($url, $project, $rule, $col, $set, $push)
  12. {
  13. $doc = Doc::getInstance($url, $rule);
  14. $doc->log(new Log($project));
  15. $data = $doc->get();
  16. if ($data) {
  17. if (!is_array($data) && !is_object($data)) {
  18. $state = Dever::json_decode($data);
  19. if ($state) {
  20. $data = $state;
  21. }
  22. }
  23. if ($data) {
  24. if (is_array($data)) {
  25. $domain = parse_url($url);
  26. $host = $domain['scheme'] . '://' . $domain['host'];
  27. foreach ($data as $k => $v) {
  28. if (is_string($v) && !strstr($v, 'http')) {
  29. $v = $host . $v;
  30. }
  31. $this->data[$k] = $this->load($doc, $k, $v, $col, $set, $push, $project);
  32. }
  33. } else {
  34. $this->data = $this->load($doc, 0, $data, $col, $set, $push, $project);
  35. }
  36. }
  37. }
  38. $doc->saveLog();
  39. }
  40. public function get()
  41. {
  42. return $this->data;
  43. }
  44. private function load($doc, $index, $data, $col, $set, $push, $project)
  45. {
  46. $result = $table = array();
  47. $data = $doc->init($data);
  48. if (!$col) {
  49. if (Dever::input('test') == 1) {
  50. $doc->outLog();
  51. echo 'error';die;
  52. }
  53. return false;
  54. }
  55. foreach ($col as $v) {
  56. $callback = false;
  57. if (strpos($v['key'], '.') !== false) {
  58. $temp = explode('.', $v['key']);
  59. $v['key'] = $temp[1];
  60. $callback = $temp[0];
  61. }
  62. $value = $doc->rule($data, $col, $v);
  63. if ($value == 'error') {
  64. break;
  65. }
  66. if ($callback) {
  67. if (function_exists($callback)) {
  68. $value = $callback($value);
  69. } else {
  70. $value = Dever::{$callback}($value);
  71. }
  72. }
  73. if ($v['local'] == 1) {
  74. $this->res = $v['res_key'];
  75. $value = $this->local($value, $v['type']);
  76. }
  77. if ($v['collect_filter_link'] == 1) {
  78. $value = $this->filter($value);
  79. }
  80. $result[$v['key']] = $value;
  81. if (Dever::input('test') == 1) {
  82. $table[$v['name']] = $value;
  83. }
  84. }
  85. if ($set) {
  86. foreach ($set as $v) {
  87. $value = $this->set($index, $v, $project);
  88. $result[$v['key']] = $value;
  89. if (Dever::input('test') == 1) {
  90. $table[$v['name']] = $value;
  91. }
  92. }
  93. }
  94. if ($push) {
  95. $result['test'] = Dever::input('test');
  96. $this->push($push, $result);
  97. }
  98. if (Dever::input('test') == 1) {
  99. $doc->outLog();
  100. echo Dever::table($table);die;
  101. }
  102. $this->update($result, $project);
  103. return $result;
  104. }
  105. private function push($push, $data)
  106. {
  107. $push = explode("\n", str_replace("\r", '', $push));
  108. foreach ($push as $k => $v) {
  109. if (strstr($v, 'http')) {
  110. Dever::curl($v, $data, 'post');
  111. } else {
  112. Dever::load($v, $data);
  113. }
  114. }
  115. }
  116. private function set($index, $data, $project)
  117. {
  118. if ($data['type'] == 1) {
  119. return $data['value'];
  120. } elseif ($data['type'] == 2) {
  121. $old = 0;
  122. $info = Dever::db('spider/data')->getOne(array('pid' => $project));
  123. if($info) {
  124. $value = json_decode($info['value'], true);
  125. if (isset($value[$data['key']])) {
  126. $old = $value[$data['key']];
  127. }
  128. }
  129. return $data['value'] + $index + $old;
  130. } elseif ($data['type'] == 3) {
  131. $eval = '$value = ' . $data['value'] . ';';
  132. eval($eval);
  133. return $value;
  134. } elseif ($data['type'] == 4) {
  135. $temp = explode("\n", str_replace("\r", '', $data['value']));
  136. return mt_rand($temp[0], $temp[1]);
  137. } elseif ($data['type'] == 5) {
  138. $temp = explode("\n", str_replace("\r", '', $data['value']));
  139. $temp[0] = Dever::maketime($temp[0]);
  140. $temp[1] = Dever::maketime($temp[1]);
  141. return mt_rand($temp[0], $temp[1]);
  142. }
  143. }
  144. private function filter($content)
  145. {
  146. $rule = '<(a).+href="(.*?)"(.*?)>(.*?)<\/a>';
  147. $content = preg_replace_callback('/' . $rule . '/i', array($this, 'filter_replace'), $content);
  148. return $content;
  149. }
  150. private function filter_replace($result)
  151. {
  152. if (isset($result[4]) && $result[4]) {
  153. return $result[4];
  154. }
  155. }
  156. private function local($content, $type = 1)
  157. {
  158. if ($type == 1) {
  159. $rule = '<(img|video|audio).+src=\"?(.+\.(jpg|gif|bmp|bnp|png))\"?.+>';
  160. $content = preg_replace_callback('/' . $rule . '/i', array($this, 'local_replace'), $content);
  161. } else {
  162. $content = $this->copy($content);
  163. }
  164. return $content;
  165. }
  166. private function local_replace($result)
  167. {
  168. if (isset($result[2]) && $result[2]) {
  169. $file = $this->copy($result[2]);
  170. if ($file) {
  171. $result[0] = str_replace($result[2], $file, $result[0]);
  172. return $result[0];
  173. }
  174. }
  175. }
  176. private function copy($file)
  177. {
  178. $data = Dever::load('upload/save.copy?file=' . $file . '&key=' . $this->res . '&state=1');
  179. if (isset($data['url'])) {
  180. return $data['url'];
  181. } else {
  182. return $file;
  183. }
  184. }
  185. private function update($data, $project)
  186. {
  187. $param['option_pid'] = $project;
  188. $param['option_value'] = json_encode($data, JSON_UNESCAPED_UNICODE);
  189. $info = Dever::db('spider/data')->one($param);
  190. if ($info) {
  191. $update = array();
  192. foreach ($param as $i => $j) {
  193. $i = str_replace('option_', 'set_', $i);
  194. $update[$i] = $j;
  195. }
  196. $id = $update['where_id'] = $info['id'];
  197. Dever::db('spider/data')->update($update);
  198. } else {
  199. $update = array();
  200. foreach ($param as $i => $j) {
  201. $i = str_replace('option_', 'add_', $i);
  202. $update[$i] = $j;
  203. }
  204. $id = Dever::db('spider/data')->insert($update);
  205. }
  206. }
  207. }