Re^2: Reverse regexp a regexp?

This is possible by converting a regex into some sort of finite automaton and doing a traversal on it.

Perl would be my prefered choice in this though. The use of such would be great in creating examples in how regexes get processed.

This problem is a bit underspecified. For example, if you give me any sample of, say, 10 URLs

http://www.youtube.com/user/test1 http://www.youtube.com/user/test2 http://www.youtube.com/user/test3 http://www.youtube.com/user/test4 http://www.youtube.com/user/test5:
regex created: http://www.youtube.com/user/(.*)

http://be.apple.itunes.com/blabla/blabla/id/1543285432 http://us.apple.itunes.com/blabla/blabla/id/154328545 http://uk.apple.itunes.com/blabla/blabla/id/5532424342 http://be.apple.itunes.com/blabla/blabla/id/15432 http://be.apple.itunes.com/blabla/id/4324
regex created: http://(\d)+\.apple.itunes.com/blabla/blabla/id/(.*)

Comment on Re^2: Reverse regexp a regexp?

Replies are listed 'Best First'.
Re^3: Reverse regexp a regexp? by Anonymous Monk on Feb 12, 2010 at 20:20 UTC
Reread what Roboticus said. Your first example could just as easily have been `http://www.youtube.com/user/test[1-5]` and your second could have been `http://(u[sk]\|be)\.apple.itunes.com/blabla/blabla/id/(\d)5(\d+).` How do you make the choices? Once you can answer those questions, you'll be able to make some progress towards what you want.*	[reply] [d/l] [select]
Re^4: Reverse regexp a regexp? (by an algorithm?) by freakingwildchild (Scribe) on Feb 13, 2010 at 22:56 UTC
How do you make the choices? Once you can answer those questions, you'll be able to make some progress towards what you want. I'd say ... an algorithm of minimum requirements could do the trick here. for example: 10 url's of the same domain of a 95% "similarity" would be matched as "clean url". 100 url's of the same domain would give more weight The string detected has to be atleast 4 characters difference before it adds weight/validity Any others could be selected manually or kept in a cache till they offer a difference Some url's will be easier than some others (especially script url's); but most sites got a general structure, to be SEO friendly towards search engines; which would make most URL's a heaven to work with...	[reply]


There's more than one way to do things
	PerlMonks