redis_import_dict.php 2.0 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990
  1. <?php
  2. /*
  3. 将csv字典载入redis
  4. #例:
  5. # php redis_import_dict.php ../../dicttext/rich/rech.json
  6. */
  7. require_once "../config.php";
  8. require_once "../install/filelist.php";
  9. require_once "../redis/function.php";
  10. if (PHP_SAPI == "cli") {
  11. if ($argc >= 2) {
  12. $list = $argv[1];
  13. if(isset($argv[2])){
  14. $tableNum = (int)$argv[2];
  15. }
  16. else{
  17. $tableNum = -1;
  18. }
  19. $redis = redis_connect();
  20. if ($redis == false) {
  21. echo "no redis connect\n";
  22. exit;
  23. }
  24. $taskList = json_decode(file_get_contents(__DIR__."/".$list));
  25. $dir = dirname(__DIR__."/".$list);
  26. if($tableNum<0){
  27. foreach ($taskList as $key => $task) {
  28. # code...
  29. runTask($redis,$task,$dir);
  30. }
  31. }
  32. else{
  33. if($tableNum<count($taskList)){
  34. runTask($redis,$taskList[$tableNum],$dir);
  35. }
  36. else{
  37. echo "wrong task number task length is ".count($taskList);
  38. }
  39. }
  40. }
  41. }
  42. function runTask($redis,$task,$dir){
  43. $count=0;
  44. $redis->del($task->rediskey);
  45. foreach ($task->csv as $csv) {
  46. $csvfile = $dir."/".$csv;
  47. if (($fp = fopen($csvfile, "r")) !== false) {
  48. echo "单词表load {$csvfile}\n";
  49. $row=0;
  50. while (($data = fgetcsv($fp)) !== false) {
  51. $row++;
  52. $data1 = $data;
  53. if(count($data1)>7){
  54. if($data1[2]==".comp." && $data1[1]===$data1[7]){
  55. continue;
  56. }
  57. $old = $redis->hGet($task->rediskey,$data1[$task->keycol]);
  58. $new = array();
  59. if($old){
  60. $new = json_decode($old,true);
  61. array_push($new,$data1);
  62. }
  63. else{
  64. $new[] = $data1;
  65. }
  66. $redis->hSet($task->rediskey,$data1[$task->keycol],json_encode($new, JSON_UNESCAPED_UNICODE));
  67. }
  68. else{
  69. echo "列不足够:行:{$row} 列:".count($data1)." 数据:{$data} \n";
  70. }
  71. $count++;
  72. if($count%50000==0){
  73. sleep(1);
  74. echo $count."\n";
  75. }
  76. }
  77. fclose($fp);
  78. sleep(1);
  79. echo "task : {$task->rediskey}:".$redis->hLen($task->rediskey)."\n";
  80. } else {
  81. echo "can not open csv file. ";
  82. }
  83. }
  84. }
  85. ?>