redis_import_dict.php 2.1 KB

1234567891011121314151617181920212223242526272829303132333435363738394041424344454647484950515253545556575859606162636465666768697071727374757677787980818283848586878889909192
  1. <?php
  2. /*
  3. 将csv字典载入redis
  4. #例:
  5. # php redis_import_dict.php ../../dicttext/rich/rich.json
  6. */
  7. require_once __DIR__."/../config.php";
  8. require_once __DIR__."/../install/filelist.php";
  9. require_once __DIR__."/../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. fwrite(STDERR,"no redis connect\n") ;
  22. exit;
  23. }
  24. $taskList = json_decode(file_get_contents(__DIR__."/".$list));
  25. $dir = dirname(__DIR__."/".$list);
  26. if($tableNum<0){
  27. //全部都导入
  28. foreach ($taskList as $key => $task) {
  29. # code...
  30. runTask($redis,$task,$dir);
  31. }
  32. }
  33. else{
  34. //只导入指定的
  35. if($tableNum<count($taskList)){
  36. runTask($redis,$taskList[$tableNum],$dir);
  37. }
  38. else{
  39. fwrite(STDERR, "wrong task number task length is ".count($taskList));
  40. }
  41. }
  42. }
  43. }
  44. function runTask($redis,$task,$dir){
  45. $count=0;
  46. $redis->del($task->rediskey);
  47. foreach ($task->csv as $csv) {
  48. $csvfile = $dir."/".$csv;
  49. if (($fp = fopen($csvfile, "r")) !== false) {
  50. fwrite(STDOUT, "单词表load {$csvfile}\n");
  51. $row=0;
  52. while (($data = fgetcsv($fp)) !== false) {
  53. $row++;
  54. $data1 = $data;
  55. if(count($data1)>7){
  56. if($data1[2]==".comp." && $data1[1]===$data1[7]){
  57. continue;
  58. }
  59. $old = $redis->hGet($task->rediskey,$data1[$task->keycol]);
  60. $new = array();
  61. if($old){
  62. $new = json_decode($old,true);
  63. array_push($new,$data1);
  64. }
  65. else{
  66. $new[] = $data1;
  67. }
  68. $redis->hSet($task->rediskey,$data1[$task->keycol],json_encode($new, JSON_UNESCAPED_UNICODE));
  69. }
  70. else{
  71. //echo "列不足够:行:{$row} 列:".count($data1)." 数据:{$data} \n";
  72. }
  73. $count++;
  74. if($count%50000==0){
  75. sleep(1);
  76. fwrite(STDOUT, $count."\n");
  77. }
  78. }
  79. fclose($fp);
  80. sleep(1);
  81. fwrite(STDOUT, "task : {$task->rediskey}:".$redis->hLen($task->rediskey)."\n");
  82. } else {
  83. fwrite(STDERR, "can not open csv file. ".PHP_EOL);
  84. }
  85. }
  86. }
  87. ?>