2
\$\begingroup\$

I have been struggling with a regular expression involving path names. Immediately, this is a bit troublesome, owing to the embedded / in the pattern, but braces to the rescue.

First the convention I have imposed:

Every path in the set looks like:

/ifmxdev/files/file.0123

but that top directory can take a suffix, for example:

/ifmxdev_test/files/file.8765

The final suffix is exactly 4 digits

After much struggle and an hour of composing this plea, as well as my own thoughts, I came up with a truly ugly but working pattern:

$rawfile_pattern = qr{/ifmxdev[_0-9A-Za-z]*/files/file.\d{4}};
if (! $fname =~ $rawfile_pattern) {....

Now I'd just like some help in making that pattern more elegant, mainly compacting that bracketed section of the regex.

\$\endgroup\$
2
  • 4
    \$\begingroup\$ Just use \w for that character class: qr{/ifmxdev\w*/files/file\.\d{4}};. And don't forget to escape the period in the suffix. Finally, might want to add some anchors ^ and $, but that's up to you. \$\endgroup\$
    – Miller
    Jun 12, 2014 at 22:41
  • \$\begingroup\$ Thanks, Miller - this did it. And extra thanks for reminding me to escape that period. In the jumble of trying everything that got lost by the wayside. \$\endgroup\$
    – Jacob Salomon
    Jun 16, 2014 at 13:53

2 Answers 2

2
\$\begingroup\$

It probably doesn't matter, but the first element of that path can be ifmxdev______ or ifmxdev_x_x_x with that regex.

I suggest you use use the Unicode property alnum, which is [A-Za-z0-9] - \w without the underscore

qr{ \A /ifmxdev (?:_\p{alnum}+)? /files /file\.\d{4} \z }x
\$\endgroup\$
0
\$\begingroup\$

Here are some options:

As Miller mentioned in the comments, instead of [_0-9A-Za-z]* you could use

\w*

Or you could use a negated character class

[^/]*

which would be more efficient if you are doing this operation a lot of times. This is because * is greedy and [^/]* will stop before \w*.

You could also write /files/file as

(/files?){2}

I think that is less readable though. And if you did use it, you might want to use a non-capturing group, which also might make it harder to read:

(?:files?){2}.

Borodin also shows the free-format modifier /x. That causes whitespace to be ignored in your regular expression. This might help with readability.

Last, as Miller also mentions, you might want to add anchors: ^ and $. This would prevent something like "/ifmxdev/files/file.01231234" from matching.

qr{ ^ /ifmxdev[^/]* (?:/files?){2} \.\d{4} $ }x

Here is a nice FAQ from stack overflow.

And if you are interested in more information, I highly recommend Mastering Regular Expressions. I'm still working my way through it but it has taught me a lot.

PS. I would have posted more links, but I don't have enough rep points yet ;-)

\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.