RegEx - match !foo followed by foo

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

Replies are listed 'Best First'.
Re: RegEx - match !foo followed by foo by merlyn (Sage) on Mar 16, 2006 at 12:56 UTC
Not sure I completely understand, but does this do it for you? `if (/^(foobar:((?!=gonk\|\@).)+)/) { print "matched $1\n"; }` [download] That's your classic "inchworm" pattern, so it's not the fastest in the world. -- Randal L. Schwartz, Perl hacker Be sure to read my standard disclaimer if this is a reply. update: Yeah, just like in real life, he changes the spec after I write the implementation. {grin} And the spec is still broken: I'm having problems working out how to perform the following regex. I want to match foobar:Xgonk where X is between 1 and 5 characters, and can be anything except 'gonk' (or the beginning of 'gonk') or '@'. I then want to remove gonk or @. The last "or @" is spurious. There could never be an "@" there, since it has to have a trailing "gonk" where we stop.	[reply] [d/l]
Re^2: RegEx - match !foo followed by foo by Melly (Chaplain) on Mar 16, 2006 at 15:24 UTC
LOL - you think you've got it bad, this wasn't even my specification, and the guy doing that actual spec. had absolutely no idea what he was asking for (or at least was incapable of expressing it). It was like asking a 2-year old for the plot of "The Tale of Peter Rabbit" - they just say "rabbits" over and over again... Yeah, in my spec. the "or @" is spurious (and the whole @ is something of a red herring). In RealLife(tm) 'gonk' can either be 'gonk' or @ at all stages (the actual strings are '\|' or '\.br\') Anyway, Corion's final solution got me there... I'd get him to give you a few perl-lessons, he obviously knows way more than you ;) Tom Melly, tom@tomandlu.co.uk	[reply]
Re^3: RegEx - match !foo followed by foo by kwaping (Priest) on Mar 16, 2006 at 15:29 UTC
Extra ++ for a good sense of humor. :) --- It's all fine and dandy until someone has to look at the code.	[reply]
Re^2: RegEx - match !foo followed by foo by Melly (Chaplain) on Mar 16, 2006 at 13:38 UTC
Hi Merlyn Sorry, my explanation was ambiguous (and had a mistake) - please see my reply to Corion Many thanks (we are not worthy) Tom Melly, tom@tomandlu.co.uk	[reply]
Re: RegEx - match !foo followed by foo by Corion (Patriarch) on Mar 16, 2006 at 13:02 UTC
I read your problem differently than merlyn does. It seems to me that you want to match only if there is `gonk` at the end of the string, and you only care for up to the first five letters of the word between `foobar:` and `gonk`: `use strict; use Test::More tests => 4; sub ungonk { local $_ = $_[0]; if (/^(foobar:.{1,5}).*gonk$/) { return $1 } else { return undef }; }; is ungonk('foobar:hellogonk'), 'foobar:hello'; is ungonk('foobar:gonk'), undef; is ungonk('foobar:higonk'), 'foobar:hi'; is ungonk('foobar:helloworldgonk'), 'foobar:hello';` [download]	[reply] [d/l] [select]
Re^2: RegEx - match !foo followed by foo by Melly (Chaplain) on Mar 16, 2006 at 13:31 UTC
Thanks Corion, but I don't think either merlyn's or your solution works... also I made one mistake in my examples: foobar:helloworldgonk shouldn't match, because neither gonk nor @ can be found after no more than 5 preceding characters Merlyn's solution fails (afaik) because it doesn't require the non-gonk to be followed by gonk (or @). Your solution fails (afaik) because foobar:gonkgonk would return 'foobar:gonk' (and should return nothing). To put it another way, if 'gonk' was a single character (say 'Ł'), then I would do: `/foobar:[^Ł@]{1,5}[Ł@]/` [download] Hope that makes it clearer Tom Melly, tom@tomandlu.co.uk	[reply] [d/l]
Re^3: RegEx - match !foo followed by foo by Corion (Patriarch) on Mar 16, 2006 at 13:36 UTC
There are the edge cases of `foobar:gonkogonk` and `foobar:ogonkgonk`, which I've added to my below test cases. You need to decide if `gonkogonk` and `ogonkgonk` should be rejected or accepted. My solution rejects the first but accepts the second case. `use strict; use Test::More tests => 6; sub ungonk { local $_ = $_[0]; if (/^(foobar:(?!gonk).{1,5})gonk$/) { return $1 } else { return undef }; }; is ungonk('foobar:hellogonk'), 'foobar:hello'; is ungonk('foobar:gonk'), undef; is ungonk('foobar:higonk'), 'foobar:hi'; is ungonk('foobar:helloworldgonk'), undef; is ungonk('foobar:gonkgonk'), undef; is ungonk('foobar:gonkogonk'), undef; is ungonk('foobar:ogonkgonk'), 'ogonk';` [download] Update: After reading your specification again, you don't want `gonk` to be found within the first five characters, but it must appear at the end. I think the below program does that, and rejects `:ogonkgonk` and `:gonkogonk`. `use strict; use Test::More tests => 7; sub ungonk { local $_ = $_[0]; if (/^(foobar:(?:(?!gonk).){1,5})gonk$/) { return $1 } else { return undef }; }; is ungonk('foobar:hellogonk'), 'foobar:hello'; is ungonk('foobar:gonk'), undef; is ungonk('foobar:higonk'), 'foobar:hi'; is ungonk('foobar:helloworldgonk'), undef; is ungonk('foobar:gonkgonk'), undef; is ungonk('foobar:gonkogonk'), undef; is ungonk('foobar:ogonkgonk'), undef;` [download]	[reply] [d/l] [select]
Re^4: RegEx - match !foo followed by foo by Melly (Chaplain) on Mar 16, 2006 at 13:45 UTC
Re^5: RegEx - match !foo followed by foo by Corion (Patriarch) on Mar 16, 2006 at 13:49 UTC
Some notes below your chosen depth have not been shown here
Re: RegEx - match !foo followed by foo by johngg (Canon) on Mar 16, 2006 at 14:14 UTC
I think a good approach is to break the problem down into sections and make smaller compiled regular expression building blocks then put them together for the final solution. Hopefully, I've understood the requirement but the following seems to do the trick. #!/usr/local/bin/perl -w # use strict; # Make some building blocks. What we want at the start; what we don't +want # after "foobar:" using a negative look-ahead assertion; any five # characters; what we want at the end (which we will discard later). # our $rxFooBar = qr{foobar:}; our $rxNotGonkish = qr{(?!(?:g(?:o(?:n(?:k)))*\|@))}; our $rxFive = qr{.{1,5}}; our $rxGonk = qr{(?:gonk\|@)}; # Put the building blocks together using "()" to capture everything ex +cept # the "gonk" or "@" at the end; # our $rxPutItAllTogether = qr{($rxFooBar$rxNotGonkish$rxFive)$rxGonk}; while(<DATA>) { chomp; print "$_\n"; print /$rxPutItAllTogether/ ? "$1\n\n" : "Failed\n\n"; } __END__ foobar:hellogonk foobar:gonk foobar:higonk foobar:helloworldgonk foobar:ggonk foobar:gogonk foobar:gongonk foobar:gonkgonk foobar:gonkygonk foobar:@gonk foobar:its@ foobar:toomany@ [download] Which produces the following output. `foobar:hellogonk foobar:hello foobar:gonk Failed foobar:higonk foobar:hi foobar:helloworldgonk Failed foobar:ggonk Failed foobar:gogonk Failed foobar:gongonk Failed foobar:gonkgonk Failed foobar:gonkygonk Failed foobar:@gonk Failed foobar:its@ foobar:its foobar:toomany@ Failed` [download] Cheers, JohnGG	[reply] [d/l] [select]
Re: RegEx - match !foo followed by foo by holli (Abbot) on Mar 16, 2006 at 13:09 UTC
Shouldn't `foobar:helloworldgonk` give you `foobar:helloword`, instead of `foobar:hello`? holli, /regexed monk/	[reply] [d/l] [select]
Re^2: RegEx - match !foo followed by foo by Melly (Chaplain) on Mar 16, 2006 at 13:50 UTC
Sorry Holli, that one was a mistake - it should return nothing (now corrected) Tom Melly, tom@tomandlu.co.uk	[reply]