A Quirk in File::Find?

Larson2042 has asked for the wisdom of the Perl Monks concerning the following question:

I'm relatively new to perl, so I don't know if my problem was a bug, quirk, feature, or whatnot, but I thought I would post it here to see what more experienced monks might think.

So the point of the program is to search through a directory and pull data from text files of a certain name and then write that data to an excel file. I also want to take part of the directory for use as a worksheet name, which is where I was having my problem. The relevant code, set up to produce the error:

use warnings;
use strict;
use File::Find;

my $directory = "\.\\v706_7-17";  # Set directory to search

find(\&wanted, $directory);

sub wanted {
    return unless -f;  # Operate only on files, not directories
    my $curfile = $_;
    if ($curfile =~ m|list\$\$|i) {  # If the file is a LIST$$
        print "$directory\n";  # For debug purposes
        print "$File::Find::dir\n";  # For debug purposes

        $File::Find::dir =~ m|$directory\/(.*)|;  # Variable way that 
+didn't work
        # $File::Find::dir =~ m|\.\\v706_7-17\/(.*)|;  # Hardcoded way
+ that worked

        my $name = $1;
        $name =~ s|\/| |;  # Replace "/" with " " for use in naming th
+e worksheet
        ... #and on to the rest of the code.
[download]

I'm on a windows system, which is why I set the search directory to .\v706_7-17 instead of ./v706_7-17. The debug prints gave the expected results:

.\v706_7-17
.\v706_7-17/DSP-23/6DCL_Ada_Sci
[download]

But only the hardcoded version of the regex would find a match. Then, I tried changing the backslash to a forwardslash on the original search directory:

my $directory = "\.\/v706_7-17";  # Set directory to search
[download]

And I was greeted with success. The variable regex found a match and everything worked as it should. The debug printouts also gave the expected results:

./v706_7-17
./v706_7-17/DSP-23/6DCL_Ada_Sci
[download]

So all in all, I found the whole thing a little strange that the debug printouts would match, but the regex wouldn't work as intended with the backslash.

Comment on A Quirk in File::Find? Select or Download Code

Replies are listed 'Best First'.
Re: A Quirk in File::Find? by ikegami (Patriarch) on Sep 23, 2009 at 21:15 UTC
`my $directory = "\.\\v706_7-17"; # Set directory to search m\|$directory\/(.)\| # Variable way that didn't work` [download] So `$directory` contains the string `.\v706_7-17`. As a regex pattern, it matches a string that contains any character followed by a vertical whitespace (\v) followed by "706_7-17". You forgot to convert your literal text into a regex pattern. You can use quotemeta to do this, most easily via \Q\E: `m\|\Q$directory\E\/(.)\|` [download] Note that you probably want a `^` at the start of that pattern.	[reply] [d/l] [select]
Re^2: A Quirk in File::Find? by Larson2042 (Novice) on Sep 23, 2009 at 21:42 UTC
Ah, I think I see what's going on. Basically, what's in `$directory` is being interpolated twice in the pattern match. The first interpolation is going from `$directory` to `.\v706_7-17` and then it's being interpolated again, so it's looking for a vertical tab and then 706_7-17. And the forwardslash worked because it isn't a metacharacter. Thanks. That quotemeta will be useful in the future. I'd never heard of it before.	[reply]
Re^3: A Quirk in File::Find? by ikegami (Patriarch) on Sep 23, 2009 at 22:11 UTC
and then it's being interpolated again, No. Then the string resulting from the interpolation is used as a regex pattern.	[reply]
Re: A Quirk in File::Find? by Corion (Patriarch) on Sep 23, 2009 at 21:16 UTC
The cause is that the backslash is a special character in regular expressions. This is why you need to use quotemeta on `$directory` or use `\Q...\E` when interpolating `$directory` into your regular expression. Also, Perl on Windows does not have any problems with using a forward slash instead of a backslash as the directory separator. `$File::Find::dir =~ m\|\Q$directory\E\/(.*)\|;`	[reply] [d/l] [select]