| regular expressions [message #15713] |
Fri, 04 June 1999 00:00  |
Michael Werger
Messages: 34
Registered: May 1997
|
Member |
|
|
Dear IDL'ers
for a complex batch processing in IDL I need to do some regular
expression handling. Of course I can do this like:
function regexp_match,argument,pattern
defsysv,!true, 1 eq 1 ; defined here only for completeness
defsysv,!false, 1 eq 0 ; see above
command='perl -e ''print ("'+argument+'" =~ m/'+pattern+'/)'''
spawn,command,result
if (result[0] eq 1) then result = !true else result = !false
return,result
end
and then in some code:
if regexp_match(string,'\s*\d+') then print,'(spaces and) digits found!'
but this is rather slow, requires perl to be setup properly and
so on. Did anyone already wrote some routines like
regexp_replace and regexp_match (I think these names are speaking
for themselves? - like the tcl routines regsub and regexp?
Suggestions to improve the above routine are also welcome.
--
Michael Werger ------------o
ESA ESTEC & Praesepe B.V. |
Astrophysics Division mwerger@astro.estec.esa.nl|
| Postbus 299 http://astro.estec.esa.nl |
| 2200 AG Noordwijk +31 71 565 3783 (Voice)
o------------------- The Netherlands +31 71 565 4690 (FAX)
|
|
|
|
|
|
|
|
|
|
| Re: Regular Expressions [message #24188 is a reply to message #15713] |
Thu, 15 March 2001 16:09  |
John-David T. Smith
Messages: 384
Registered: January 2000
|
Senior Member |
|
|
Wayne Landsman wrote:
>
> The following is probably a simple question for anyone familiar with
> regular expressions, but I am still trying to learn the STREGEX
> function.
>
> Suppose I want to find the first occurence in a string of an 'l' ithat
> is not part of a double 'l'. For
> example, in the string
>
> IDL> st = 'The rolling hills and lake'
>
> I want to return the character position of the 'l' in lake (=21).
>
> The following expression almost works -- it will search for any 'l'
> which is both preceded and followed by anything that is not "l"
>
> IDL> print,stregex(st, '[^l]l[^l]' )
>
> but it won't work for the string 'The rolling hills and pool' because
> the final 'l' has no characters following it. Any suggestions?
IDL> print, stregex(st,'(^|[^l])l($|[^l])')
which means "a character that is not 'l', or the beginning of the
string, followed by an 'l', followed by a character that is not 'l', or
the end of the string". Aren't you glad Ken Thompson didn't decide
originally to develop regexps in english?
This will also work on
IDL> st = "let's all go the the movies"
JD
|
|
|
|
| Re: Regular Expressions [message #24195 is a reply to message #24188] |
Thu, 15 March 2001 19:24 |
Wayne Landsman
Messages: 117
Registered: January 1997
|
Senior Member |
|
|
JD Smith wrote:
> IDL> print, stregex(st,'(^|[^l])l($|[^l])')
>
> which means "a character that is not 'l', or the beginning of the
> string, followed by an 'l', followed by a character that is not 'l', or
> the end of the string". Aren't you glad Ken Thompson didn't decide
> originally to develop regexps in english?
>
> This will also work on
>
> IDL> st = "let's all go the the movies"
Thanks. But I now realize that my original formulation was not quite
correct, since the above expression (usually!) returns the position of the
character *before* the 'l', so to get the position of the first single 'l'
one has to add 1
IDL> l_position = stregex(st,'(^|[^l])l($|[^l])') + 1
Unfortunately, if 'l' is the first character, then you *don't* want to add
the 1. (The expression stregex(st,'(^|[^l])l($|[^l])') returns a value of
0 for both st ='long days' and st ='slow nights'. )
One solution is to forget about the beginning of string anchor and just
concatenate a blank to the beginning to the string
IDL> l_position = stregex(' ' + st,'[^l]l($|[^l])')
--Wayne
P.S. The real-life problem I am working on deals not with 'l' but with
apostrophes. I am trying to speed up the processing of FITS header
values, where is a string is delineated by non-repeating apostrophes, and a
possessive is indicated by a double apostrophe.
VALUE = 'This is Wayne''s FITS value' / Example string field
|
|
|
|
|
|
| Re: Regular Expressions [message #24236 is a reply to message #15713] |
Tue, 20 March 2001 15:23 |
Craig Markwardt
Messages: 1869
Registered: November 1996
|
Senior Member |
|
|
"Mark Hadfield" <m.hadfield@niwa.cri.nz> writes:
> "Wayne Landsman" <landsman@mpb.gsfc.nasa.gov> wrote in message
> news:3AB7BA3E.CA411E1B@mpb.gsfc.nasa.gov...
>> Of course, one should probably add an English comment to the use of STRGEX
>>
>> ; Find the substring beginning with an "=", followed by any number
>> ; of characters, followed by a quote, followed by any number of
>> ; characters (including double quotes) up to the last single quote.
>> ; Extract from this substring all characters between the first
>> ; and last single quotes.
>
> So, you're saying that STREGEX is a good thing because (like HISTOGRAM) it
> allows you to write code in which the executable statements are several
> times shorter than the comments required to explain them?
Wouldn't that be APL?
Craig
--
------------------------------------------------------------ --------------
Craig B. Markwardt, Ph.D. EMAIL: craigmnet@cow.physics.wisc.edu
Astrophysics, IDL, Finance, Derivatives | Remove "net" for better response
------------------------------------------------------------ --------------
|
|
|
|
| Re: Regular Expressions [message #24237 is a reply to message #15713] |
Tue, 20 March 2001 15:28 |
John-David T. Smith
Messages: 384
Registered: January 2000
|
Senior Member |
|
|
Mark Hadfield wrote:
>>
>> ; Find the substring beginning with an "=", followed by any number
>> ; of characters, followed by a quote, followed by any number of
>> ; characters (including double quotes) up to the last single quote.
>> ; Extract from this substring all characters between the first
>> ; and last single quotes.
>
> So, you're saying that STREGEX is a good thing because (like HISTOGRAM) it
> allows you to write code in which the executable statements are several
> times shorter than the comments required to explain them?
I take that as a personal jab. Actually, the code I write is much more
comprehensible than the examples I post here -- I *do* have a reputation
to maintain though.
Somehow, I think the equivalent byte array version would be even
uglier. Anyone care to whip up a version for comparison, using the
detailed comments above?
JD
|
|
|
|
| Re: Regular Expressions [message #24239 is a reply to message #15713] |
Tue, 20 March 2001 14:48 |
Mark Hadfield
Messages: 783
Registered: May 1995
|
Senior Member |
|
|
"Wayne Landsman" <landsman@mpb.gsfc.nasa.gov> wrote in message
news:3AB7BA3E.CA411E1B@mpb.gsfc.nasa.gov...
> Of course, one should probably add an English comment to the use of STRGEX
>
> ; Find the substring beginning with an "=", followed by any number
> ; of characters, followed by a quote, followed by any number of
> ; characters (including double quotes) up to the last single quote.
> ; Extract from this substring all characters between the first
> ; and last single quotes.
So, you're saying that STREGEX is a good thing because (like HISTOGRAM) it
allows you to write code in which the executable statements are several
times shorter than the comments required to explain them?
---
Mark Hadfield
m.hadfield@niwa.cri.nz http://katipo.niwa.cri.nz/~hadfield
National Institute for Water and Atmospheric Research
|
|
|
|
| Re: Regular Expressions [message #24240 is a reply to message #15713] |
Tue, 20 March 2001 12:14 |
Wayne Landsman
Messages: 117
Registered: January 1997
|
Senior Member |
|
|
"Pavel A. Romashkin" wrote:
> Wouldn't it be easier to analyse a byte array with more human-readible
> functions, than those beautiful regular expressions you guys brought up?
>
It depends on what you mean by "easier". One nice thing about STREGEX is
that it works on vector strings. One can always convert the string
array to a byte array and analyze, but -- **if you are trying to avoid
loops** -- the indexing can be become extremely opaque, and exercise at
least as many brain cells as using STREGEX. For example, JD's solution
can also apply to a string array where one is trying to extract the
substrings beginning and ending with a singe quote:
IDL> st = ["value1 = 'Wayne''s dog' / First string ", $
"value2 = 'Sue''s dog and Ralph''s cat' / Second string ", $
"value3 = 'two pigeons'" ]
IDL> val = (stregex(st, /SUBEXPR,/EXTRACT,"= *'(.*)'([^']|$)"))[1,*]
IDL> print,val
Wayne''s dog
Sue''s dog and Ralph''s cat
two pigeons
Of course, one should probably add an English comment to the use of STRGEX
; Find the substring beginning with an "=", followed by any number of
characters,
; followed by a quote, followed by any number of characters (including
double
; quotes) up to the last single quote. Extract from this substring all
; characters between the first and last single quotes.
|
|
|
|
| Re: Regular Expressions [message #24250 is a reply to message #15713] |
Tue, 20 March 2001 04:39 |
Martin Schultz
Messages: 515
Registered: August 1997
|
Senior Member |
|
|
"Pavel A. Romashkin" wrote:
>
> Wouldn't it be easier to analyse a byte array with more human-readible
> functions, than those beautiful regular expressions you guys brought up?
>
> Cheers,
> Pavel
Oh no! Pavel! That would mean to take all the fun out of it! Just
imagine IDL got rid of all the quirks we spend so much time musing
upon in this group. Wouldn't that be boring (and David would be out of
bread and butter, too). With regular expressions, it's a similar
thing: they are brain sport! Somewhere I read that people who train
their brain regularily have a better chance of avoiding dementia later
on. So, where's the weekly regular expression contest?
Cheers,
Martin
PS: As far as I know, emacs is nothing else than a smart and well
balanced collection of regular expressions ;-)
--
[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[ [[[[[[[
[[ Dr. Martin Schultz Max-Planck-Institut fuer Meteorologie [[
[[ Bundesstr. 55, 20146 Hamburg [[
[[ phone: +49 40 41173-308 [[
[[ fax: +49 40 41173-298 [[
[[ martin.schultz@dkrz.de [[
[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[ [[[[[[[
|
|
|
|
|
|
comp.lang.idl-pvwave archive
Members
Search
Help
Login
Home
