redis_import_dict.php 2.1 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293
  1. <?php
  2. /*
  3. 将csv字典载入redis
  4. #例:
  5. # php redis_import_dict.php ../../dicttext/rich/rich.json
  6. # php redis_import_dict.php ../../dicttext/system/system.json
  7. */
  8. require_once __DIR__."/../config.php";
  9. require_once __DIR__."/../install/filelist.php";
  10. require_once __DIR__."/../redis/function.php";
  11. if (PHP_SAPI == "cli") {
  12. if ($argc >= 2) {
  13. $list = $argv[1];
  14. if(isset($argv[2])){
  15. $tableNum = (int)$argv[2];
  16. }
  17. else{
  18. $tableNum = -1;
  19. }
  20. $redis = redis_connect();
  21. if ($redis == false) {
  22. fwrite(STDERR,"no redis connect\n") ;
  23. exit;
  24. }
  25. $taskList = json_decode(file_get_contents(__DIR__."/".$list));
  26. $dir = dirname(__DIR__."/".$list);
  27. if($tableNum<0){
  28. //全部都导入
  29. foreach ($taskList as $key => $task) {
  30. # code...
  31. runTask($redis,$task,$dir);
  32. }
  33. }
  34. else{
  35. //只导入指定的
  36. if($tableNum<count($taskList)){
  37. runTask($redis,$taskList[$tableNum],$dir);
  38. }
  39. else{
  40. fwrite(STDERR, "wrong task number task length is ".count($taskList));
  41. }
  42. }
  43. }
  44. }
  45. function runTask($redis,$task,$dir){
  46. $count=0;
  47. $redis->del($task->rediskey);
  48. foreach ($task->csv as $csv) {
  49. $csvfile = $dir."/".$csv;
  50. if (($fp = fopen($csvfile, "r")) !== false) {
  51. fwrite(STDOUT, "单词表load {$csvfile}\n");
  52. $row=0;
  53. while (($data = fgetcsv($fp)) !== false) {
  54. $row++;
  55. $data1 = $data;
  56. if(count($data1)>7){
  57. if($data1[2]==".comp." && $data1[1]===$data1[7]){
  58. continue;
  59. }
  60. $old = $redis->hGet($task->rediskey,$data1[$task->keycol]);
  61. $new = array();
  62. if($old){
  63. $new = json_decode($old,true);
  64. array_push($new,$data1);
  65. }
  66. else{
  67. $new[] = $data1;
  68. }
  69. $redis->hSet($task->rediskey,$data1[$task->keycol],json_encode($new, JSON_UNESCAPED_UNICODE));
  70. }
  71. else{
  72. //echo "列不足够:行:{$row} 列:".count($data1)." 数据:{$data} \n";
  73. }
  74. $count++;
  75. if($count%50000==0){
  76. sleep(1);
  77. fwrite(STDOUT, $count."\n");
  78. }
  79. }
  80. fclose($fp);
  81. sleep(1);
  82. fwrite(STDOUT, "task : {$task->rediskey}:".$redis->hLen($task->rediskey)."\n");
  83. } else {
  84. fwrite(STDERR, "can not open csv file. ".PHP_EOL);
  85. }
  86. }
  87. }
  88. ?>