Re: Guile regular expressions are too greedy

[Ludovic Courtès]

Hello,

Chris Dennis <address@hidden> writes:

> Hello Guile People
>
> Is there a way to make Guile regular expressions less greedy?  I
> understand that POSIX doesn't define non-greedy modifiers.
>
> Specifically, I'm trying to parse font names such as
>
>     Arial 12
>     Arial Bold Italic 14
>     Nimbus Sans L Bold Italic Condensed 11
>
> so that I can construct CSS styles from them.
>
> I've tried the following, but the first (.*) gobbles up everything
> before the size because the other elements are optional:
>
> (define s (string-match "(.*)( +(bold|semi-bold|regular|light))?(
> +(italic|oblique))?( +(condensed))? +([0-9]+)" "nimbus sans l bold
> italic condensed 11"))

Here's a possible solution:

--8<---------------cut here---------------start------------->8---
scheme@(guile-user)> (define rx (make-regexp "^([^ ]+)( 
+(bold|semi-bold|regular|light))?( +(italic|oblique))?( +(condensed))? 
+([0-9]+)" regexp/extended))
scheme@(guile-user)> (define m (regexp-exec rx "nimbus bold italic condensed 
11"))
scheme@(guile-user)> (match:substring m 1)
"nimbus"
scheme@(guile-user)> (match:substring m 2)
" bold"
scheme@(guile-user)> (match:substring m 3)
"bold"
scheme@(guile-user)> (match:substring m 4)
" italic"
scheme@(guile-user)> (match:substring m 5)
"italic"
--8<---------------cut here---------------end--------------->8---

Note that I slightly modified the string that's matched because "sans"
and "l" are not meant to be matched.

Hope this helps,
Ludo'.