Path: utzoo!attcan!utgpu!jarvis.csri.toronto.edu!rutgers!ucsd!nosc!humu!uhccux!julian
From: julian@uhccux.uhcc.hawaii.edu (Julian Cowley)
Newsgroups: comp.emacs
Subject: Re: Lisp Programming: Substring Searches, etc.
Message-ID: <4307@uhccux.uhcc.hawaii.edu>
Date: 10 Jul 89 01:42:59 GMT
References: <397@lxn.eds.com> <65891@yale-celray.yale.UUCP> <SPENCER.89Jul9162408@spline.eecs.umich.edu>
Organization: University of Hawaii at Manoa
Lines: 36

In article <SPENCER.89Jul9162408@spline.eecs.umich.edu> spencer@eecs.umich.edu (Spencer W. Thomas) writes:
>In article <65891@yale-celray.yale.UUCP> Duchier-Denys@cs.yale.edu (Denys Duchier) writes:
>   (defun extract-file-extension (filename)
>     (and (string-match ".+\\.\\(.*\\)$" filename)
>	  (substring filename (match-beginning 1) (match-end 1))))
>
>Actually, probably better would be
>	 (string-match ".+\\.\\([^.]*\\)$" filename)
>This pattern will match the last ".", while the given pattern will
>match the first one.

This has kind of regexp has come up before, I think.  Two things
need pointing out:

1) The above regexps incorrectly return the names of hidden
files.  E.g.,
	(extract-file-extension "~/.letter")
returns
	"letter".
The correct result should be nil.

2) They also don't work with directories.  E.g.,
	(extract-file-extension "/usr/lib/sccs.help/cmds")
returns
	"help/cmds".
Again, the correct result should be nil.

So, here is an improved regexp that correctly handles the above
cases:

	(string-match "[^/]\\.\\([^/.]*\\)$" filename)

julian@uhccux.uhcc.hawaii.edu	| "Hawaii has always been a very pivotal role 
uunet!ucsd!nosc!uhccux!julian	|  in the Pacific.  It is in the Pacific.
julian@uhccux.bitnet		|  It is part of the United States that is an
University of Hawaii at Manoa	|  island that is right here." - Quayle