knowledge/tools/es_add_comments.php

<?php
// nohup php /data1/www/zhishiku.kuailelunwen.com/tools/es_add_comments.php &
include_once(dirname(dirname(__FILE__))."/vendor/autoload.php");
include_once(dirname(dirname(__FILE__))."/tools/es.php");

function parseDbCnf($dbflag) {
    $configs = parse_ini_file(dirname(dirname(__FILE__)).'/config/database.ini', true);
    $config = $configs[strtolower($dbflag)];

    $host_port = $config['master'];

    $hps = explode(',', $host_port);
    $random = rand(0, count($hps)-1);
    $hp = $hps[$random];

    list($host, $port) = explode(':', $hp);

    $cnf = array();
    $cnf['host'] = trim($host);
    $cnf['port'] = trim($port);
    $cnf['user'] = trim($config['user']);
    $cnf['pwd'] = trim($config['passwd']);
    $cnf['db'] = trim($config['db']);
    return $cnf;
}

$mysqlconfig = parseDbCnf('simplyphp');
$servername = $mysqlconfig['host'];
$username = $mysqlconfig['user'];
$password = $mysqlconfig['pwd'];
$dbname = $mysqlconfig['db'];

$obj = new \ES('weibo');
$conn = new mysqli($servername, $username, $password, $dbname);
if ($conn->connect_error) {
    die("连接失败: " . $conn->connect_error);
}
$conn->set_charset("utf8mb4");

$limit = 5000;
$min_id = 0;

$log_path = '/datacenter/zhishiku/es_comment.log';
$log_path_success = '/datacenter/zhishiku/es_success_comment.log';
$log_path_err = '/datacenter/zhishiku/es_error_comment.log';

for($page=0;;$page++){
    $sql = "SELECT * FROM spider_weibo_comments where is_search=1 and id>".$min_id." order by id asc limit ".$limit;
    $result = $conn->query($sql);
    if ($result->num_rows > 0) {
        while($row = $result->fetch_assoc()) {
            $min_id = $row['id'];

            if(str_replace(" ", "", $row["content"])=="") continue;

            $data['id'] = $row['id']+0;

            $data['uid'] = $row['uid']+0;
            $data['weibo_id'] = $row['weibo_id']+0;
            $data['content'] = $row['content'];

            if (!empty($row['comment_time'])) {
                $data['comment_time'] = date('c', strtotime($row['comment_time']));
            } else {
                $data['comment_time'] = null;
            }

            $params['index'] = 'comments';
            $params['type'] = 'doc';
            $params['id'] = $data['id'];
            $params['body'] = $data;

            try {
                $resc = $obj->add_doc($params);
                if($resc['result'] != 'created') {
                    error_log('error:'.json_encode($data)."\n", 3, $log_path_err);
                }else{
                    error_log($resc['_id']."|{$min_id}\n", 3, $log_path_success);
                }
            } catch (\Exception $th) {
                error_log('excption:'.$th->getMessage().'|'.json_encode($data).'|'.$min_id."\n", 3, $log_path_err);
            }
        }
        error_log((($page*$limit)+$result->num_rows)."|".$page."\n", 3, $log_path);
        if($result->num_rows<$limit) break;
    } else {
        error_log((($page*$limit)+$result->num_rows)."|".$page."\n", 3, $log_path);
        break;
    }
}
$conn->close();
?>