Core.php 1.9 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172
  1. <?php
  2. namespace Spider\Lib\Doc;
  3. use Dever;
  4. use Spider\Lib\Doc as Doc;
  5. class Core
  6. {
  7. public static function match($pattern, $string)
  8. {
  9. $temp = explode('||', $pattern);
  10. $index = isset($temp[1]) ? $temp[1] : 1;
  11. preg_match_all('/' . $temp[0] . '/i', $string, $match);
  12. $result = '';
  13. if (isset($match[$index][0])) {
  14. $result = $match[$index][0];
  15. }
  16. return $result;
  17. }
  18. public static function rule($doc, $data, $col, $rule, $key)
  19. {
  20. $rule = explode("\n", str_replace("\r", '', $rule));
  21. if (isset($rule[0]) && $rule[0]) {
  22. if (isset($col[$rule[0]]) && $rule[0] != $key) {
  23. if (isset($rule[1]) && $rule[1]) {
  24. $result = Doc::getInstance($doc->getUrl($data, $col, $col[$rule[0]]), $rule[1])->get();
  25. array_shift($rule);
  26. } else {
  27. $result = self::rule($doc, $data, $col, $col[$rule[0]]['collect_rule'], $key);
  28. }
  29. } else {
  30. if (is_string($data)) {
  31. $data = json_decode($data, true);
  32. }
  33. if (is_object($data)) {
  34. $result = Dom::find($data, $rule[0]);
  35. } elseif (is_array($data)) {
  36. $temp = explode('.', $rule[0]);
  37. $result = $data;
  38. foreach ($temp as $k => $v) {
  39. if (isset($result[$v])) {
  40. $result = $result[$v];
  41. }
  42. }
  43. }
  44. }
  45. }
  46. /*
  47. if (isset($rule[0]) && $rule[0]) {
  48. if (isset($col[$rule[0]]) && $rule[0] != $key) {
  49. # 此处开task
  50. if (isset($rule[1]) && $rule[1]) {
  51. $result = Doc::getInstance($doc->getUrl($data, $col, $col[$rule[0]]), $rule[1])->get();
  52. array_shift($rule);
  53. } else {
  54. $result = self::rule($doc, $data, $col, $col[$rule[0]]['collect_rule'], $key);
  55. }
  56. } elseif (isset($data[$rule[0]])) {
  57. $result = $data[$rule[0]];
  58. } else {
  59. $result = Dom::find($data, $rule[0]);
  60. }
  61. }
  62. */
  63. if (isset($rule[1]) && $rule[1]) $result = self::match($rule[1], $result);
  64. return $result;
  65. }
  66. }