案例场景:玩家数据打包为一个blob存在mysql中,需要查看玩家特定时间的数据
策略:从dump sql文件中提取玩家数据块
实现:为了简单和跨平台,用php编写
具体步骤如图,1为简单版,2为优化版。
备注:关键点在于步长一定要大于玩家数据大小,这样如果搜索到起始位置,至多再有一次read就可以得到玩家数据.
另外重要的一点是:sql文件是mysqldump的输出文件,blob都是转义过的,不能拿来直接当作二进制数据做内存解析,只能再mysql导入到db中才能使用
源代码:
<?php
function write_data($uid, $data)
{
$fp = fopen("$uid"."_dump.rar", 'wb');
$real_data = "replace into kakag.t_player_data_1_1 values($uid,'".$data."');";
fwrite($fp, $real_data);
fclose($fp);
//echo md5($real_data)."\n";
}
function get_real_data($uid, $data, &$has_start_flag)
{
$start_flag = "($uid,'";
$start_pos = strpos($data, $start_flag);
if($start_pos != false)
{
$has_start_flag = true;
//first state
$check_flag = "'),(";
$end_pos = strpos($data, $check_flag, $start_pos);
if($end_pos != false)
{
$new_start_pos = $start_pos + strlen($start_flag);
$real_data = substr($data, $new_start_pos, $end_pos - $new_start_pos );
write_data($uid, $real_data);
return true;
}
//second state
$check_flag = "');";
$end_pos = strpos($data, $check_flag, $start_pos);
if($end_pos != false)
{
$new_start_pos = $start_pos + strlen($start_flag);
$real_data = substr($data, $new_start_pos, $end_pos - $new_start_pos );
write_data($uid, $real_data);
return true;
}
}
else
{
return false;
}
}
function read_data($uid, $file_name)
{
$fp = fopen($file_name, 'r');
$once_size = 1024 * 500;
while(1)
{
$data = fread($fp, $once_size);
if($data != "")
{
$has_start_flag = false;
$ret = get_real_data($uid, $data, $has_start_flag);
if($ret == true)
{
echo "load user data success\n";
break;
}
else
{
if($has_start_flag == true)
{
$data_extra = fread($fp, $once_size);
if($data_extra != "")
{
$new_data = $data.$data_extra;
$has_start_flag = false;
$ret = get_real_data($uid, $new_data, $has_start_flag);
if($ret == true)
{
echo "load user data success\n";
break;
}
else
{
break;
}
}
else
{
break;
}
}
else
{
continue;
}
}
}
else
{
break;
}
}
fclose($fp);
}
read_data($argv[1], $argv[2]);
?>