File name regex

dev2dev has asked for the wisdom of the Perl Monks concerning the following question:

Replies are listed 'Best First'.
Re: File name regex by ikegami (Patriarch) on Jan 20, 2006 at 06:52 UTC
It's not possible to combine those regexps if those are trully meant to be captures (`(...)`) and not groupings (`(?:...)`). At least, it wouldn't be possible without change the variables in which the data is captured. Let's assume that you only used the parens for grouping (and that you forgot `$` at the end of each regexp), then you'd be starting with the following: `if ($file =~ /^\d{8}(?:_\d+)\.ilf$/i \|\| $file =~ /^\d{8}(?:\s\d+)\.sht$/i ) { ... }` [download] The possibilities for joining are still quite limited since there is a mixture of common and not common elements. The following illustrates this: `vvvvvvvvv vvv vvv vvvv common /^\d{8}(?:_ \d+)\.ilf$/ix /^\d{8}(?:\s\d+)\.sht$/ix ^^ ^^^ not common` [download] The best that can be done, as far as I can tell, is to join the common beginning and the common ending. The following illustrates this: `(?:_ \d+)\.ilf /^\d{8} $/ix (?:\s\d+)\.sht` [download] We get the following regexp: `if ($file =~ /^\d{8}(?:(?:_\d+)\.ilf\|(?:\s\d+)\.sht)$/i) { ... }` [download] This is definitely less readable.	[reply] [d/l] [select]
Re: File name regex by duff (Parson) on Jan 20, 2006 at 06:59 UTC
This should do for you: `if ($file =~ /^\d{8}((_\d+)\.ilf\|(\s\d+)\.sht)/i) { ... }` [download] You don't need parens around `\d`. Update: What ikegami said about captures vs. grouping. I assumed you were only using the parens for grouping. duff	[reply] [d/l] [select]
Re: File name regex by BrowserUk (Patriarch) on Jan 20, 2006 at 07:53 UTC
I believe this will do what you ask, where if the option '_nnn' is present then the extension should be '.ilf', or if it is ' nnn' then the extension should be '.sht'; whilst capturing the entire optional part to $1 and the last digit of that optional part to $2: Update: tightened slightly. Update2: Tightening removed; unnecessary. `m[ ^ \d{8} ( (?(?=.* \. ilf) _ \| \s \| (?!) ) ) (\d)+ )* \. (?:ilf\| +sht) ]x` [download] Whether you would call that a 'shortcut' is debatable. Examine what is said, not who speaks -- Silence betokens consent -- Love the truth but pardon error. Lingua non convalesco, consenesco et abolesco. -- Rule 1 has a caveat! -- Who broke the cabal? "Science is about questioning the status quo. Questioning authority". In the absence of evidence, opinion is indistinguishable from prejudice.	[reply] [d/l]
Re: File name regex by grinder (Bishop) on Jan 20, 2006 at 08:56 UTC
(Warning: more Regexp::Assemble pimping ahead). Assuming you do not really care about those captures, if I "unroll" the patterns, it looks like you're interested matching against the following patterns: `^\d{8}\.ilf ^\d{8}_\d+\.ilf ^\d{8}\.sht ^\d{8}\s\d+\.sht` [download] Using the `assemble` script from the above module, or something like `my $re = Regexp::Assemble->new(flags=>'i')->add(@list)` it produces the following pattern: `^\d{8}(?:\.(?:ilf\|sht)\|\s\d+\.sht\|_\d+\.ilf)` ... which looks roughly like that the other people came up with manually. If you do need the captures, the pattern becomes `^\d{8}(?:(\s(\d+))\.sht\|(_(\d+))\.ilf\|\.(?:ilf\|sht))` in which case you can perform the match and get the captures with something like `my @result = map {defined} ($file =~ /$re/);` • another intruder with the mooring in the heart of the Perl	[reply] [d/l] [select]
Re: File name regex by parv (Parson) on Jan 20, 2006 at 07:02 UTC
never mind what i wrote earlier; i was quite wrong as i failed to note what ikegami noted.	[reply]
Re^2: File name regex by duff (Parson) on Jan 20, 2006 at 07:05 UTC
That would work if he can live with the false positives. i.e., you pattern will match files named `12345678_1.sht` and `12345678 1.ilf` duff	[reply] [d/l] [select]
Re^2: File name regex by Fletch (Bishop) on Jan 20, 2006 at 14:23 UTC
Erm, and now that you've deleted the previous contents we'll never know what was wrong. It's better form to add an update (possibly bracketing the incorrect parts with <strike> tags) stating that somethings wrong. Now the existing reply to your node makes no sense because it has no context.	[reply]