How to read unbuffered input of lines ending with "\r" or "\n"?

Question

I'm writing a simple program which is intended to filter input from a pipeline. Most of the input is sent as output untouched. Part of it is modified or used to extract information. In its simplest form — doing nothing — the program filter is:

$|++;

while (<>) {
    print;
}

The main program sometimes outputs progress updates of its task, overwriting the same visual content through the use of lines ending with a carriage return. Piping such content to filter blocks all output:

$ perl -e '$|++; print ++$a, "\r" and sleep 1 while 1' | filter

Is there an easy way to read those lines in the same loop fashion, or should I go the sysread way? I'm looking for something similar to what would happen if it was possible to set the record separator to "\n or \r".

Answer 1

If the input isn't so large that efficiency is a concern, this version of filter is more robust

while (!eof(STDIN)) {
    $_ .= getc(STDIN);
    if (/[\r\n]/) {
        # manipulate $_, if desired
        print;
        $_ = "";
    }
}
print;