Parse.php 6.5 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290
  1. <?php
  2. namespace Spider\Lib;
  3. use Dever;
  4. use Spider\Lib\Doc\Dom;
  5. class Parse
  6. {
  7. private $url = '';
  8. private $host = '';
  9. private $log;
  10. private $doc = array();
  11. private $data = array();
  12. public function __construct($url, $project, $rule, $param, $col, $set, $push)
  13. {
  14. $doc = Doc::getInstance($url, $rule);
  15. $doc->log(new Log($project));
  16. $data = $doc->get($param);
  17. if ($data) {
  18. if (!is_array($data) && !is_object($data)) {
  19. $state = Dever::json_decode($data);
  20. if ($state) {
  21. $data = $state;
  22. }
  23. }
  24. if ($data) {
  25. if (is_array($data)) {
  26. $domain = parse_url($url);
  27. $host = $domain['scheme'] . '://' . $domain['host'] . '/';
  28. foreach ($data as $k => $v) {
  29. if (is_string($v) && !strstr($v, 'http')) {
  30. $v = $host . ltrim($v, '/');
  31. }
  32. $this->data[$k] = $this->load($doc, $k, $v, $col, $set, $push, $project, $v);
  33. }
  34. } else {
  35. $this->data = $this->load($doc, 0, $data, $col, $set, $push, $project, $url);
  36. }
  37. }
  38. }
  39. $doc->saveLog();
  40. }
  41. public function get()
  42. {
  43. return $this->data;
  44. }
  45. private function load($doc, $index, $data, $col, $set, $push, $project, $source)
  46. {
  47. if (!$col) {
  48. if (Dever::input('test') == 1) {
  49. $doc->outLog();
  50. echo 'error';die;
  51. }
  52. return false;
  53. }
  54. $result = $table = array();
  55. if (isset($col[1])) {
  56. $data = $doc->init($data);
  57. $this->getCol($doc, $col[1], $data, $result, $table);
  58. }
  59. if (isset($col[2])) {
  60. $data = $doc->getCur();
  61. $this->getCol($doc, $col[2], $data, $result, $table);
  62. }
  63. if ($set) {
  64. foreach ($set as $v) {
  65. $value = $this->set($index, $v, $project);
  66. $result[$v['key']] = $value;
  67. if (Dever::input('test') == 1) {
  68. $table[$v['name']] = $value;
  69. }
  70. }
  71. }
  72. if ($push) {
  73. $result['test'] = Dever::input('test');
  74. $this->push($push, $result, $project);
  75. }
  76. if (Dever::input('test') == 1) {
  77. $doc->outLog();
  78. echo Dever::table($table);die;
  79. }
  80. $this->update($result, $project, $source);
  81. return $result;
  82. }
  83. private function getCol($doc, $col, $data, &$result, &$table)
  84. {
  85. foreach ($col as $v) {
  86. $callback = false;
  87. if (strpos($v['key'], '.') !== false) {
  88. $temp = explode('.', $v['key']);
  89. $v['key'] = $temp[1];
  90. $callback = $temp[0];
  91. }
  92. $value = $doc->rule($data, $col, $v);
  93. if ($value == 'error') {
  94. break;
  95. }
  96. if ($callback) {
  97. if (function_exists($callback)) {
  98. $value = $callback($value);
  99. } else {
  100. $value = Dever::{$callback}($value);
  101. }
  102. }
  103. if ($v['local'] == 1) {
  104. $this->res = $v['res_key'];
  105. $value = $this->local($value, $v['type']);
  106. }
  107. if ($v['collect_filter_link'] == 1) {
  108. $value = $this->filter($value);
  109. }
  110. if ($value) {
  111. $result[$v['key']] = $value;
  112. if (Dever::input('test') == 1) {
  113. $table[$v['name']] = $value;
  114. }
  115. }
  116. }
  117. }
  118. private function push($push, $data, $project)
  119. {
  120. $push = Dever::split($push);
  121. $data['project_id'] = $project;
  122. foreach ($push as $k => $v) {
  123. if (strstr($v, 'http')) {
  124. Dever::curl($v, $data, 'post');
  125. } else {
  126. Dever::load($v, $data);
  127. }
  128. }
  129. }
  130. private function set($index, $data, $project)
  131. {
  132. if ($data['type'] == 1) {
  133. return $data['value'];
  134. } elseif ($data['type'] == 2) {
  135. $old = 0;
  136. $info = Dever::db('spider/data')->getOne(array('pid' => $project));
  137. if($info) {
  138. $value = json_decode($info['value'], true);
  139. if (isset($value[$data['key']])) {
  140. $old = $value[$data['key']];
  141. }
  142. }
  143. return $data['value'] + $index + $old;
  144. } elseif ($data['type'] == 3) {
  145. $eval = '$value = ' . $data['value'] . ';';
  146. eval($eval);
  147. return $value;
  148. } elseif ($data['type'] == 4) {
  149. $temp = Dever::split($data['value']);
  150. return mt_rand($temp[0], $temp[1]);
  151. } elseif ($data['type'] == 5) {
  152. $temp = Dever::split($data['value']);
  153. $temp[0] = Dever::maketime($temp[0]);
  154. $temp[1] = Dever::maketime($temp[1]);
  155. return mt_rand($temp[0], $temp[1]);
  156. }
  157. }
  158. private function filter($content)
  159. {
  160. $rule = '<(a).+href="(.*?)"(.*?)>(.*?)<\/a>';
  161. $content = preg_replace_callback('/' . $rule . '/i', array($this, 'filter_replace'), $content);
  162. return $content;
  163. }
  164. private function filter_replace($result)
  165. {
  166. if (isset($result[4]) && $result[4]) {
  167. return $result[4];
  168. }
  169. }
  170. private function local($content, $type = 1)
  171. {
  172. if ($type == 1) {
  173. $doc = Dom::init($content);
  174. $pic = Dom::find($doc, '$("img").each().attr("src")');
  175. if ($pic) {
  176. $content = $this->local_replace($pic, $content);
  177. }
  178. $video = Dom::find($doc, '$("video").each().attr("src")');
  179. if ($video) {
  180. $content = $this->local_replace($video, $content);
  181. }
  182. $audio = Dom::find($doc, '$("audio").each().attr("src")');
  183. if ($audio) {
  184. $content = $this->local_replace($audio, $content);
  185. }
  186. /*
  187. $rule = '<(img|video|audio).+src=\"?(.+\.(jpg|gif|bmp|bnp|png))\"?.+>';
  188. $content = preg_replace_callback('/' . $rule . '/i', array($this, 'local_rule_replace'), $content);
  189. */
  190. } else {
  191. $content = $this->copy($content);
  192. }
  193. return $content;
  194. }
  195. private function local_replace($file, $content)
  196. {
  197. if (is_string($file) && strstr($file, '[')) {
  198. $file = Dever::json_decode($file);
  199. }
  200. if (is_array($file)) {
  201. foreach ($file as $k => $v) {
  202. $content = $this->local_replace($v, $content);
  203. }
  204. } else {
  205. $result = $this->copy($file);
  206. if ($result) {
  207. $content = str_replace($file, $result, $content);
  208. }
  209. }
  210. return $content;
  211. }
  212. private function local_rule_replace($result)
  213. {
  214. if (isset($result[2]) && $result[2]) {
  215. $file = $this->copy($result[2]);
  216. if ($file) {
  217. $result[0] = str_replace($result[2], $file, $result[0]);
  218. return $result[0];
  219. }
  220. }
  221. }
  222. private function copy($file)
  223. {
  224. if (strstr($file, ',')) {
  225. $temp = explode(',', $file);
  226. $file = array();
  227. foreach($temp as $k => $v) {
  228. $f = $this->copy($v);
  229. if ($f) {
  230. $file[] = $f;
  231. }
  232. }
  233. $file = implode(',', $file);
  234. return $file;
  235. }
  236. $data = Dever::load('upload/save.copy?file=' . $file . '&key=' . $this->res . '&state=1');
  237. if (isset($data['status']) && $data['status'] == -1) {
  238. return '';
  239. } elseif (isset($data['url'])) {
  240. return $data['url'];
  241. } else {
  242. return '';
  243. }
  244. }
  245. private function update($data, $project, $source)
  246. {
  247. if (!$data) {
  248. return;
  249. }
  250. $param['pid'] = $project;
  251. $param['source'] = $source;
  252. $info = Dever::db('spider/data')->one($param);
  253. $param['value'] = json_encode($data, JSON_UNESCAPED_UNICODE);
  254. if ($info) {
  255. $update = $param;
  256. $id = $update['where_id'] = $info['id'];
  257. Dever::db('spider/data')->update($update);
  258. } else {
  259. $update = $param;
  260. $id = Dever::db('spider/data')->insert($update);
  261. }
  262. echo $id;
  263. echo "\r\n";
  264. }
  265. }