Problems with regexps

[Kirk Strauser]

I'm writing a program to scan through a bunch of VB source.**An*example*of
one of the types of strings I'm trying to match is:

****response.write*Request.Cookies*("domain")("cna me")

but I do not want to match variable assignments like this (they are picked
up later by a different pattern):

****strXRSCust=*Request.Cookies*("domain")("cname" )

I'm differentiating between the two forms by looking for an equal sign
followed by zero or more spaces; if the '=' is there, then I don't want to
match.**Here's*where*it*gets*weird.**This*pattern* works*perfectly*(as*long
as there are 1 or more spaces after the '='), in that it will not match
the assignment example above:

****re.compile(r'(?<!=)\s+Request.Cookies\s*((\(\s *".*?"\s*\)\s*)+)')

I really want to use the pattern below to match for zero or more spaces
(not one or more).**Note*that*it's*identical*except*that*the*f irst*'\s+'
is replaced with a '\s*':

****re.compile(r'(?<!=)\s*Request.Cookies\s*((\(\s *".*?"\s*\)\s*)+)')

I don't know why, but the second pattern does match the assignment example
above, although I don't think it should.

It seems like there's a problem with the negative lookbehind assertion and
that the variable-length '\s' pattern immediately following it is throwing
it off.**Any*thoughts?
--
Kirk Strauser
The Day Companies

[Kirk Strauser]

OK, let me give a clearer demonstration. Given this program:

#!/usr/bin/env python

import re

sample = 'FOO= BAR'

if re.search(r'[^=]\s+BAR', sample):
print 'Match 1.'

if re.search(r'[^=]\s*BAR', sample):
print 'Match 2.'

When run, it produces "Match 2.". Why? I want to match:

1) Anything but '=', followed by
2) Zero or more spaces, followed by
3) 'BAR'

The first pattern correct does *not* match. Why doesn't the '= ' get struck
down by the '[^=]' pattern before a '\s+' but not before '\s*'?

Thanks,
--
Kirk Strauser
The Day Companies

[Dennis Reinhardt]

> sample = 'FOO= BAR'

if re.search(r'[^=]\s*BAR', sample):
print 'Match 2.'

The [^=] will match on the space preceding BAR. The \s* matches because
there are zero spaces remaining. Result: match.

Prediction: you can get the first expression to fail by putting in two
spaces, not one.

Caveat: untested.
--

Dennis Reinhardt

De*****@dair.com http://www.spamai.com?ng_py

[Ben Finney]

On Fri, 07 Nov 2003 23:00:34 GMT, Kirk Strauser wrote:

sample = 'FOO= BAR'

if re.search(r'[^=]\s+BAR', sample):
print 'Match 1.'

if re.search(r'[^=]\s*BAR', sample):
print 'Match 2.'

When run, it produces "Match 2.". Why? I want to match:

1) Anything but '=', followed by
2) Zero or more spaces, followed by
3) 'BAR'

To understand the answer, you must understand the question.

If you inspect the MatchObject returned from re.search(), you'll learn
more about what's going on:

import re
sample = 'FOO= BAR'
match = re.search( r'[^=]\s+BAR', sample )
match
The first match returns the None object.
match = re.search( r'[^=]\s*BAR', sample )
match <_sre.SRE_Match object at 0x4021b090>

The second match returns a Match object; let's see what it matched.
match.group(0) ' BAR'

The matching string contains the space ("Anything but '='"), followed by
nothing ("Zero or more spaces"), followed by BAR. Exactly what you
asked for.

--
\ "It is seldom that liberty of any kind is lost all at once." |
`\ -- David Hume |
_o__) |
Ben Finney <http://bignose.squidly.org/>

[Paul Foley]

On Fri, 07 Nov 2003 21:20:08 GMT, Kirk Strauser wrote:

I'm writing a program to scan through a bunch of VB source.**An*example*of
one of the types of strings I'm trying to match is: ****response.write*Request.Cookies*("domain")("cna me") but I do not want to match variable assignments like this (they are picked
up later by a different pattern): ****strXRSCust=*Request.Cookies*("domain")("cname" )

Try http://weitz.de/regex-coach/

--
Cogito ergo I'm right and you're wrong. -- Blair Houghton

(setq reply-to
(concatenate 'string "Paul Foley " "<mycroft" '(#\@) "actrix.gen.nz>"))

[Edward K. Ream]

I've just added script-based find-changed commands to Leo 4.1 beta 1. So
Leo can execute a script to find the next match, and optionally executes a
script to do the replacement. This gives Leo the full power of languages
like Icon or Snobol, using the plain Python language. There is a dead
simple way for the find script to communicate with the change script, and
for the find script to continue or not in the case of the Find All or Change
All commands.

BTW, scripts could search anywhere, not just in a Leo outline. For
example, the find script could pull in the found text from any file to any
Leo node (creating an outline node if you like). Etc.

I'm very excited about this new feature. At long last we are not limited to
non-extensible, wimpy tools like re. Fully interactive if you want; fully
batch if you want. Undo is easy, etc. The sky is the limit, and when you
are done, you will be able to understand the scripts you wrote. End of re.
Yeah!

Edward
--------------------------------------------------------------------
Edward K. Ream email: ed*******@charter.net
Leo: Literate Editor with Outlines
Leo: http://webpages.charter.net/edreamleo/front.html
--------------------------------------------------------------------

[Edward K. Ream]

Actually, I misspoke a little. Script-find is fully compatible with re,
because re can appear in any script. Anyway, there is nothing particularly
wrong with re for simple tasks. So it's not the end of re; it's the end of
re for complex tasks for which re just isn't suited.

Edward
--------------------------------------------------------------------
Edward K. Ream email: ed*******@charter.net
Leo: Literate Editor with Outlines
Leo: http://webpages.charter.net/edreamleo/front.html
--------------------------------------------------------------------