Re: regexp confusion

On Sun, 07 Dec 2003 19:47:59 -0500, Jack Tanner <[email protected]> wrote:

> I can't tell if this is a bug or not. I'm using ACL 6.2 with all
> recent patches applied.
>
> CL-USER(19): (match-regexp "^[0-9 ,]*$" "4, 7")
> T
> "4, 7"
> CL-USER(21): (match-regexp "^[0-9\\b,]*$" "4, 7")
> NIL
>
> The only difference is that in the regexp, I substitute a space for
> \\b, and all of a sudden the string matches. Isn't \\b supposed to
> be a catch-all for all kinds of whitespace?

  CL-USER(10): (compile-regexp "[0-9 ,]")
  #(EXCL::REGULAR-EXPRESSION "[0-9 ,]" 1
    ((:GROUP-START 0)
     (:OR
      #*0000000000000000000000000000000010000000000010001111111111000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000)
     (:GROUP-END 0))
    0)
  CL-USER(11): (compile-regexp "[0-9\\b,]")
  #(EXCL::REGULAR-EXPRESSION "[0-9\\b,]" 1
    ((:GROUP-START 0)
     (:OR
      #*0000000000000000000000000000000000000000000010001111111111000000000000000000000000000000000010000010000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000)
     (:GROUP-END 0))
    0)

This looks like the first [...] consists of #\Space, the comma, and
the ten digits, while the second [...] doesn't have the #\Space but
#\b and backslash instead. Which seems to imply that \\b isn't special
in [...] character classes.

Shameless plug: Use CL-PPCRE[1] instead... :)

  CL-USER(21): (cl-ppcre:scan-to-strings "^[0-9 ,]*$" "4, 7")
  "4, 7"
  #()
  CL-USER(22): (cl-ppcre:scan-to-strings "^[0-9\\s,]*$" "4, 7")
  "4, 7"
  #()

Cheers,
Edi.

[1] <http://weitz.de/cl-ppcre/>