From airboss at nodewarrior.org  Wed Apr 26 20:46:35 2006
From: airboss at nodewarrior.org (Dan Debertin)
Date: Wed, 26 Apr 2006 15:46:35 -0500
Subject: [cl-ppcre-devel] Detecting "partial" matches
Message-ID: <21059.1146084395@nodewarrior.org>

Hi,

I'm using CL-PPCRE to develop a character-at-a-time lexer. This is
causing me some perplexity, though, with regexes like the common
notation for hexadecimal literals:

"^0(?:x[0-9A-Fa-f]+)?$"

This should match both the string "0", between positions 0 and 1, as
just a bare literal zero, and should also match things like "0xa6"
between positions 0 and 3, but should not match simply "0x". But I want
the longest match possible, so (for example) I'd like to know that while
"0x" didn't match, parts of the regex *did* match and might produce a
"real" match depending on what comes after "x".

So, in succession, if the input is "0xa6 ", my scanner gets called thus:

1. Input: "0". a) A match. b) But it *could* possibly match more, depending on
   what comes next.
2. Input: "0x". a) Not a match. b) But, once again, the possibility
   exists that more input could still produce a longer match than "0".
3. Input: "0xa". a) A match. b) Because of the "+" attached to the
   character class, a longer match is still possible.
4. Input: "0xa6". a) A match. b) As above.
5. Input: "0xa6 ". a) Not a match. b) Will *never* match no matter how
   much more input you add to it.

CL-PPCRE just tells me a), and I also want to know b). Is there any way
to get this information (if it even exists) out of the scanner? 

TIA,

-Dan
--
Dan Debertin             | 
airboss at nodewarrior.org  | 
www.nodewarrior.org      | 





From edi at agharta.de  Wed Apr 26 21:26:12 2006
From: edi at agharta.de (Edi Weitz)
Date: Wed, 26 Apr 2006 23:26:12 +0200
Subject: [cl-ppcre-devel] Detecting "partial" matches
In-Reply-To: <21059.1146084395@nodewarrior.org> (Dan Debertin's message of
	"Wed, 26 Apr 2006 15:46:35 -0500")
References: <21059.1146084395@nodewarrior.org>
Message-ID: <uzmi8hvuz.fsf@agharta.de>

On Wed, 26 Apr 2006 15:46:35 -0500, Dan Debertin <airboss at nodewarrior.org> wrote:

> CL-PPCRE just tells me a), and I also want to know b). Is there any
> way to get this information (if it even exists) out of the scanner?

If I understand you correctly, this information isn't available.  The
regex engine doesn't make any plans about what might have happened if
it had scanned another target string instead of the current one, so to
say... :)

You might want to look at filters, though:

  <http://weitz.de/cl-ppcre/#filters>

But generally, your problem to me looks as if regular expressions
aren't the right way to tackle it.

Cheers,
Edi.


From daniel.caune at ubisoft.com  Thu Apr 27 23:03:25 2006
From: daniel.caune at ubisoft.com (Daniel Caune)
Date: Thu, 27 Apr 2006 19:03:25 -0400
Subject: [cl-ppcre-devel] Matching problem while using \s (?)
Message-ID: <1E293D3FF63A3740B10AD5AAD88535D2021A2E4A@UBIMAIL1.ubisoft.org>

Hi,

I'm facing a strange result while trying to scan a string:

CL-USER> (defconstant +iso-8601-regex+
"([0-9]+)-([0-9]+)-([0-9]+)[\sT]*([0-9]+):([0-9]+):([0-9]+)[+-]([0-9]+)"
)
+ISO-8601-REGEX+

CL-USER> (cl-ppcre:scan-to-strings +iso-8601-regex+ "2006-04-13
00:00:00+00")
NIL

CL-USER> (cl-ppcre:scan-to-strings +iso-8601-regex+
"2006-04-13T00:00:00+00")
"2006-04-13T00:00:00+00"
#("2006" "04" "13" "00" "00" "00" "00")


For some reasons, \s seems not to match a whitespace character, unless
the space within "2006-04-13 00:00:00+00" is not the space cl-ppcre
expects (encoding on Linux?! Hmm... a space is more likely to be encoded
with 0x20, isn't it?!).

I tried to match both "2006-04-13 00:00:00+00" and
"2006-04-13T00:00:00+00" with the Regex Coach (on Windows), and it works
perfectly with that same regex!

Any idea?  What can I check?

Regards,


--
Daniel CAUNE
Ubisoft Online Technology
(514) 4090 2040 ext. 5418



From emailmac at gmail.com  Fri Apr 28 02:00:05 2006
From: emailmac at gmail.com (Mac Chan)
Date: Thu, 27 Apr 2006 19:00:05 -0700
Subject: [cl-ppcre-devel] Matching problem while using \s (?)
In-Reply-To: <1E293D3FF63A3740B10AD5AAD88535D2021A2E4A@UBIMAIL1.ubisoft.org>
References: <1E293D3FF63A3740B10AD5AAD88535D2021A2E4A@UBIMAIL1.ubisoft.org>
Message-ID: <4877ae640604271900x583dd13es7e97d767dba7bfa@mail.gmail.com>

you need to double escape \\s, since the lisp reader will consume one
and cl-ppcre:scan-to-strings will only see s.


On 4/27/06, Daniel Caune <daniel.caune at ubisoft.com> wrote:
> Hi,
>
> I'm facing a strange result while trying to scan a string:
>
> CL-USER> (defconstant +iso-8601-regex+
> "([0-9]+)-([0-9]+)-([0-9]+)[\sT]*([0-9]+):([0-9]+):([0-9]+)[+-]([0-9]+)"
> )
> +ISO-8601-REGEX+
>
> CL-USER> (cl-ppcre:scan-to-strings +iso-8601-regex+ "2006-04-13
> 00:00:00+00")
> NIL
>
> CL-USER> (cl-ppcre:scan-to-strings +iso-8601-regex+
> "2006-04-13T00:00:00+00")
> "2006-04-13T00:00:00+00"
> #("2006" "04" "13" "00" "00" "00" "00")
>
>
> For some reasons, \s seems not to match a whitespace character, unless
> the space within "2006-04-13 00:00:00+00" is not the space cl-ppcre
> expects (encoding on Linux?! Hmm... a space is more likely to be encoded
> with 0x20, isn't it?!).
>
> I tried to match both "2006-04-13 00:00:00+00" and
> "2006-04-13T00:00:00+00" with the Regex Coach (on Windows), and it works
> perfectly with that same regex!
>
> Any idea?  What can I check?
>
> Regards,
>
>
> --
> Daniel CAUNE
> Ubisoft Online Technology
> (514) 4090 2040 ext. 5418
>
> _______________________________________________
> cl-ppcre-devel site list
> cl-ppcre-devel at common-lisp.net
> http://common-lisp.net/mailman/listinfo/cl-ppcre-devel
>