php简单去除大型文本重复

php简单去除大型文本重复

<?php
error_reporting(0);
@ini_set('memory_limit','-1');
set_time_limit(0);
echo" 去除文本重复工具"."\r\n\r\n";
echo"\n"."输入要整理的文件:"."\n";
$dic=trim(fgets(STDIN));
/*while (!feof($dic)){
        $file[]=stream_get_line($fp,65535,"\r\n");
}
 */
 
$file=file($dic);
$array=preg_replace('/($\s*$)|(^\s*^)/m','',$file); //消除空行
$new_array=array_values(array_unique($array));//消除重复行
$new_filename="new_". basename($dic);
 
if(file_put_contents("$new_filename",join("\r\n",$new_array))){
 
        $num=count($file);
        $new_num=count($new_array);
        $counts=$num-$new_num;
        $files=dirname(__FILE__).DIRECTORY_SEPARATOR.$new_filename;
 
echo<<<INFO
 +----------------------------------------------+
 | [+] 去除重复完毕!   www.2cto.com            |
 | [+] 整理后的文件为:$files     |
 | [+] 原始字典数量:$num 行                     |
 | [+] 整理后为:$new_num 行                         |
 | [+] 共替换了$counts 行                           |
 +----------------------------------------------+
INFO;
 
}
 
else{ 
        echo"------------------------------------------"."\r\n";
        echo"[*] 错误!"."\r\n\r\n";
        echo"[*] 找不到文件!请检查输入路径是否存在!"."\r\n";
        echo"------------------------------------------"."\r\n";
exit();
}
 
 
?>