如果时间戳用户名对是一个uniq标识符，则可以运行join来只提取不成对的行。
如果你有一个例子

"2023-08-01 05:54:18","Lolla"
"2023-08-01 17:44:27","Dave"
"2023-08-01 17:42:23","John"

然后

"2023-08-01 05:54:18","Lolla"
"2023-08-01 17:44:27","Dave"
"2023-08-01 17:42:23","John"
"2023-08-01 17:42:28","Pam"
"2023-08-01 19:38:15","John"

您可以使用Miller并运行

mlr --csv -N join --ur --np -j 1,2 -f first.csv second.csv

得到

2023-08-01 17:42:28,Pam
2023-08-01 19:38:15,John

• -j将1和2个字段设置为键
• --ur从正确的文件发出未配对的记录
• --np不发出配对记录

因为这些行都有时间戳，所以它们可能是唯一的，所以你可以保存上一次迭代的最后一行，在当前文件中搜索它，然后获取它之后的内容：

$file = '/some/watched/file.txt';
$tail = './saved_last_line.txt';

$last_line = @file_get_contents($tail);

copy($file, $tail) && ($fp = fopen($tail,'r+')) || exit(1);

if ( $last_line !== false ) {
    do {
        $line = fgets($fp);
        $found = ($line === $last_line);
    } while ( !$found && $line !== false );
    if ( !$found )
        rewind($fp);
}

$new_content = stream_get_contents($fp);

if ( $new_content !== "" ) {
    $pos = strrpos($new_content, "\n", -2);
    $last_line = ( false === $pos ? $new_content : substr($new_content, $pos+1) );
}

ftruncate($fp, 0);
rewind($fp);
fwrite($fp, $last_line);
fclose($fp);

# THEN DO YOUR STRUFF
echo $new_content;