带有 -pl 命令行选项的 Perl 单行程序

Question

I'm running

perl -ple '$_=length' datafile

The datafile contains the following:

algorithm
student government
Fiumichino

The result is that it prints

9
18
10

What do the -p and -l options do? Also, what is $_ ?

Answer 1

$_ is the default input and pattern-searching variable. -p is a command line switch that puts an implicit while(<>) loop around your program, with a print statement at the end. The -l switch sets $/ and $\\ to "\\n" (newline) and chomp s your input, or in layman's terms, it handles newlines for you.

The diamond operator <> is "magic" in that it automatically chooses your input channel. If your script has arguments, it will interpret it as a file name, and open that file and read it. If not, it checks STDIN. In your case, it opens the file "datafile".

What that oneliner does is read each line of datafile and sets $_ to the length of $_ (since length uses $_ if no argument is supplied), then prints that number.

You can deparse the one-liner and see what the code looks like:

$ perl -MO=Deparse -ple '$_=length' datafile
BEGIN { $/ = "\n"; $\ = "\n"; }
LINE: while (defined($_ = <ARGV>)) {
    chomp $_;
    $_ = length $_;
}
continue {
    die "-p destination: $!\n" unless print $_;
}
-e syntax OK

Answer 2

-p causes the Perl interpreter to loop over its input and run the code (supplied via -e in this case) once per line, instead of just once total. At the bottom of each loop iteration it prints the line out.

The actual line is read into the $_ variable, and the contents of that variable are what is printed at the bottom of the loop body, so by changing the contents of $_ inside the loop, you can change what is output.

The length operator with no argument counts the length of $_ . So $_=length replaces the line with its length before it gets printed.

-l just causes the Perl interpreter to remove newlines from input (so they won't be counted in the length values here) and automatically put them back on output (otherwise the output would be jammed together as 91810).