| Re: stregex - lookaround operators? [message #56348] |
Thu, 18 October 2007 09:53  |
rtowler
Messages: 28
Registered: June 2006
|
Junior Member |
|
|
Thanks Chris,
I ended up pretty much doing the same thing you did but I used
STRSPLIT or STRMID depending on my mood.
rid= STREGEX(dg.dgram, '[0-9]*</rid>', /EXTRACT)
rid = LONG((STRSPLIT(rid, '<'))[0])
xcvrs = STREGEX(dg.dgram, '[ a-zA-Z0-9,\-]*</value>', /EXTRACT)
xcvrs = STRMID(xcvrs, 0, STRLEN(xcvrs) - 8)
Luckily my input strings will always only contain one pair of tags for
each type so I don't have to resort to loops for processing. Still, I
think this is rather ugly and will request that the regex engine in
IDL be updated to include the lookaround operators.
-Rick
On Oct 18, 12:56 pm, Spon wrote:
> On Oct 17, 5:35 pm, wrote:
>
>> Does stregex support the lookahead and lookbehind operators? I'm
>> guessing it doesn't, which is a real bummer. Specifically I am trying
>> to use the lookahead operator to extract data from an XML like string:
>> '[^>]*(?=</rid>)' which should extract all characters before the </
>> rid> tag up to ">". But I get an error:
>
>> STREGEX: Error processing regular expression: [^>]*(?=</rid>)
>> repetition-operator operand invalid
>
>> I think the expression is valid. Has anyone used the lookaround
>> operators in IDL?
>
>> -r
>
> This is the best I could do. Not elegant, probably not efficient, but
> it throws up a few questions of its own:
>
> *******
>
> FUNCTION STREGEXTEST, StrExpr
>
> IF N_PARAMS() EQ 0 THEN $
> StrExpr = ['<rid> Stuff in here </rid>', $
> '<rid> All of this stuff in here </rid>', $
> '<html> Not this stuff </html>', 'frog', $
> '>.<', '<grid> Not this either </grid>', $
> '</html> <rid> This stuff too </rid>', $
> '<rid> Even this</rid> <foo>!</foo>', $
> 'The acid test </rid>' ]
>
> ; StrExprTemp = StrExpr
>
> RegStr = '>[^>]*(</rid>)'
>
> Streg = STREGEX (StrExpr, RegStr, $
> LENGTH = Length)
> ExtrStreg = STREGEX (StrExpr, RegStr, /EXTR)
>
> PRINT, Streg
> PRINT, ExtrStreg
>
> Index = WHERE (Streg GE 0, Count)
> IF Count EQ 0 THEN RETURN, ''
>
> Streg = Streg [Index]
> Length = Length [Index]
> PosExp = StrExpr [Index]
>
> Streg += 1 ; Manually move your location
> Length -= 7 ; and length pointers to
> ; exclude the Lookarounds
>
> ; ; This doesn't work for string arrays,
> ; ; even if Streg and Length have the
> ; ; same dimensions:
> ; Strings = STRMID (PosExp, Streg, Length)
>
> NS = N_ELEMENTS (Streg)
> Strings = STRARR (NS)
> FOR i = 0L, NS - 1 DO $
> Strings [i] = STRMID (PosExp [i], $
> Streg [i], Length [i] )
>
> HELP, Strings
> PRINT, Strings
> RETURN, Strings
> END
>
> ********
>
> As you can see, I had foolishly assumed that if you call STRMID with
> three arrays of equal dimensions, IDL would somehow know to
> 'vectorise' the calculation, but I had to pump it through a FOR loop
> instead. Can someone show me a way around this?
>
> The other commented-out line is this:
> ; StrExprTemp = StrExpr
> Which I was using because I thought STREGEX (or possibly STRMID) was
> changing my input function (I've written functions of my own that
> ended up doing this, and watched in dismay as my nice 4-dimensional
> array was decimated to a tiny subset, when all I was doing was trying
> to was extract the subset from the data, leaving the original intact.
> It's easy not to do a second time, but quite frustrating the first
> time you do it!)
> I seemed to be getting a large array of blank strings after using
> STREGEX and them STRMID on the input array (the resulting array was of
> expected dimensions, just empty, contrary to my expectations.) I can't
> seem to replicate this now, so chances are it was some silly error on
> my part, but I thought I'd ask if anyone else had come across this?
>
> Thanks,
> Chris
|
|
|
|
| Re: stregex - lookaround operators? [message #56358 is a reply to message #56348] |
Thu, 18 October 2007 05:56  |
Spon
Messages: 178
Registered: September 2007
|
Senior Member |
|
|
On Oct 17, 5:35 pm, rtow...@gmail.com wrote:
> Does stregex support the lookahead and lookbehind operators? I'm
> guessing it doesn't, which is a real bummer. Specifically I am trying
> to use the lookahead operator to extract data from an XML like string:
> '[^>]*(?=</rid>)' which should extract all characters before the </
> rid> tag up to ">". But I get an error:
>
> STREGEX: Error processing regular expression: [^>]*(?=</rid>)
> repetition-operator operand invalid
>
> I think the expression is valid. Has anyone used the lookaround
> operators in IDL?
>
> -r
This is the best I could do. Not elegant, probably not efficient, but
it throws up a few questions of its own:
*******
FUNCTION STREGEXTEST, StrExpr
IF N_PARAMS() EQ 0 THEN $
StrExpr = ['<rid> Stuff in here </rid>', $
'<rid> All of this stuff in here </rid>', $
'<html> Not this stuff </html>', 'frog', $
'>.<', '<grid> Not this either </grid>', $
'</html> <rid> This stuff too </rid>', $
'<rid> Even this</rid> <foo>!</foo>', $
'The acid test </rid>' ]
; StrExprTemp = StrExpr
RegStr = '>[^>]*(</rid>)'
Streg = STREGEX (StrExpr, RegStr, $
LENGTH = Length)
ExtrStreg = STREGEX (StrExpr, RegStr, /EXTR)
PRINT, Streg
PRINT, ExtrStreg
Index = WHERE (Streg GE 0, Count)
IF Count EQ 0 THEN RETURN, ''
Streg = Streg [Index]
Length = Length [Index]
PosExp = StrExpr [Index]
Streg += 1 ; Manually move your location
Length -= 7 ; and length pointers to
; exclude the Lookarounds
; ; This doesn't work for string arrays,
; ; even if Streg and Length have the
; ; same dimensions:
; Strings = STRMID (PosExp, Streg, Length)
NS = N_ELEMENTS (Streg)
Strings = STRARR (NS)
FOR i = 0L, NS - 1 DO $
Strings [i] = STRMID (PosExp [i], $
Streg [i], Length [i] )
HELP, Strings
PRINT, Strings
RETURN, Strings
END
********
As you can see, I had foolishly assumed that if you call STRMID with
three arrays of equal dimensions, IDL would somehow know to
'vectorise' the calculation, but I had to pump it through a FOR loop
instead. Can someone show me a way around this?
The other commented-out line is this:
; StrExprTemp = StrExpr
Which I was using because I thought STREGEX (or possibly STRMID) was
changing my input function (I've written functions of my own that
ended up doing this, and watched in dismay as my nice 4-dimensional
array was decimated to a tiny subset, when all I was doing was trying
to was extract the subset from the data, leaving the original intact.
It's easy not to do a second time, but quite frustrating the first
time you do it!)
I seemed to be getting a large array of blank strings after using
STREGEX and them STRMID on the input array (the resulting array was of
expected dimensions, just empty, contrary to my expectations.) I can't
seem to replicate this now, so chances are it was some silly error on
my part, but I thought I'd ask if anyone else had come across this?
Thanks,
Chris
|
|
|
|
| Re: stregex - lookaround operators? [message #56685 is a reply to message #56348] |
Mon, 12 November 2007 03:59 |
Spon
Messages: 178
Registered: September 2007
|
Senior Member |
|
|
On Oct 18, 4:53 pm, rtow...@gmail.com wrote:
> Thanks Chris,
>
> I ended up pretty much doing the same thing you did but I used
> STRSPLIT or STRMID depending on my mood.
>
> rid= STREGEX(dg.dgram, '[0-9]*</rid>', /EXTRACT)
> rid = LONG((STRSPLIT(rid, '<'))[0])
>
> xcvrs = STREGEX(dg.dgram, '[ a-zA-Z0-9,\-]*</value>', /EXTRACT)
> xcvrs = STRMID(xcvrs, 0, STRLEN(xcvrs) - 8)
Ok, how about this?
regex='>([^>]*)<\rid>'
test=stregex(strexpr,regex,/extr,/sub)
print,transpose((test(1,*))[where(test(1,*)ne'')])
So long as you only get one sub-expression per string, it should work.
Spon
|
|
|
|
comp.lang.idl-pvwave archive
Members
Search
Help
Login
Home
