Sed 或 Perl：一个带有正则表达式指令的文件，每行一条指令，在另一个文件上执行

Question

I'm setting up a regex learning environment purely in bash/tmux with a pane for the file containing a regex, a pane for a text-file-for-processing, and a pane for the bash shell. I'm at the start of " The Bastards Book of Ruby "-regex chapter.

The 'Bastard's Book' shows an example of a 'negative-lookahead' regex (perfect, lets learn), where perl is recommended over sed. As I'm going for a CLI approach-> Bash command: $ perl -p file_with_regex.pl test.txt (This prints the lines from test.txt with the intended substitutions)

Question: How would I add a second regex (on a new line) of the regex.pl file, and have perl execute both the first and (next) this second instruction for processing the text file?

    # regex.pl
    s/^(?!Mr)/Ms./g
    s/Ms./Mrs./g

(Adding the second regex results in "Execution of regex.pl aborted due to compilation errors.")

The overall aim here is to progress in Ruby, while testing Regular Expressions as concisely as possible. Picking up a bare minimum of sed/perl while doing so would be a plus, as a proper dive into perl would take time from Ruby (and when it's time for the perl dive, I'll have had some time with the basics). The more I look at this the more it seems necessary to just do it in Ruby, if there isn't a perl switch that would enable a command-line-with-files approach.

Answer 1

The basic answer is that you need a semicolon after each line.

Paraphrased from perlrun , -p reads all lines of input, runs the commands you specified, and then prints out the value in $_ (the implicit variable you're running your substitute commands on in this script).

So, removing the magic, -p transformed your code into:

LINE:
while (<>) {
    # regex.pl
    s/^(?!Mr)/Ms./g
    s/Ms./Mrs./g
} continue {
    print or die "-p destination: $!\n";
}

Perl requires a semicolon between statements (but a terminal semicolon at the end of a block is optional) hence the error.

I personally would recommend writing the whole script above into the file instead of using -p because it is far less magical, but you're welcome to do it either way.

If you were going to write the whole script, I would recommend something more like the following:

use strict;
use warnings;

while ( my $line = <ARGV> ) {

    $line =~ s/^(?!Mr)/Ms./g;
    print "After first subst: $line";

    $line =~ s/Ms./Mrs./g;
    print "After second subst: $line";
}

use strict and use warnings are the boilerplate you want at the top of any perl script (to catch typos and other common mistakes) and explicitly calling the variable $line gives you a better understanding of how the script is working ( $_ is very magical for beginners and the source of many errors IMO, but great when you know what's what).

If you're wondering about <> vs. <ARGV> they are the same thing and mean "Read through all the lines of files provided as command-line arguments to this script or standard input if no files are provided"."