| 123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990 |
- <?php
- /*
- 将csv字典载入redis
- #例:
- # php redis_import_dict.php ../../dicttext/rich/rech.json
- */
- require_once __DIR__."/../config.php";
- require_once __DIR__."/../install/filelist.php";
- require_once __DIR__."/../redis/function.php";
- if (PHP_SAPI == "cli") {
- if ($argc >= 2) {
- $list = $argv[1];
- if(isset($argv[2])){
- $tableNum = (int)$argv[2];
- }
- else{
- $tableNum = -1;
- }
- $redis = redis_connect();
- if ($redis == false) {
- echo "no redis connect\n";
- exit;
- }
-
- $taskList = json_decode(file_get_contents(__DIR__."/".$list));
- $dir = dirname(__DIR__."/".$list);
- if($tableNum<0){
- foreach ($taskList as $key => $task) {
- # code...
- runTask($redis,$task,$dir);
- }
- }
- else{
- if($tableNum<count($taskList)){
- runTask($redis,$taskList[$tableNum],$dir);
- }
- else{
- echo "wrong task number task length is ".count($taskList);
- }
- }
- }
- }
- function runTask($redis,$task,$dir){
- $count=0;
- $redis->del($task->rediskey);
- foreach ($task->csv as $csv) {
- $csvfile = $dir."/".$csv;
- if (($fp = fopen($csvfile, "r")) !== false) {
- echo "单词表load {$csvfile}\n";
- $row=0;
- while (($data = fgetcsv($fp)) !== false) {
- $row++;
- $data1 = $data;
- if(count($data1)>7){
- if($data1[2]==".comp." && $data1[1]===$data1[7]){
- continue;
- }
- $old = $redis->hGet($task->rediskey,$data1[$task->keycol]);
- $new = array();
- if($old){
- $new = json_decode($old,true);
- array_push($new,$data1);
- }
- else{
- $new[] = $data1;
- }
- $redis->hSet($task->rediskey,$data1[$task->keycol],json_encode($new, JSON_UNESCAPED_UNICODE));
- }
- else{
- echo "列不足够:行:{$row} 列:".count($data1)." 数据:{$data} \n";
- }
- $count++;
- if($count%50000==0){
- sleep(1);
- echo $count."\n";
- }
- }
- fclose($fp);
- sleep(1);
- echo "task : {$task->rediskey}:".$redis->hLen($task->rediskey)."\n";
-
- } else {
- echo "can not open csv file. ";
- }
- }
- }
- ?>
|