On Thu, Nov 10, 2011 at 7:17 AM, Skip Cave <s...@caveconsulting.com> wrote:
> OK, Now I am working on real data.
>
> This works:
>
> getSkipsTags =: ('RESULT[0]';crlf,'CONFIDENCE'
> )& getTagContents
> > getSkipsTags shell zipth, filepth
> = u s air one ninety one
> = yes
>
> So, I have captured the strings I want from the log.gz file. If I put the
> result in a noun like this:
>
> bb =: > getSkipsTags shell zipth, filepth
>
> Now I can operate on the noun to replace the equals sign with a blank, and
> then remove all the spaces by this:
>
> deb "1 bb rplc"1 '= '
> u s air one ninety one
> yes
Kip has sorted your initial problem. I just have a couple of comments
on the code.
Firstly I'd be hesitant to include the CR,LF,TAB as part of the endtag
unless the file format is really specified like that.
For single character substitutions I think that charsub (also in the
strings script) is more performant. Here are the in-script docs for
it:
NB. =========================================================
NB.*charsub v character substitution
NB. characterpairs charsub string
NB. For example:
NB. '-_$ ' charsub '$123 -456 -789'
NB. 123 _456 _789
NB. Use <rplc> for arbitrary string replacement.
NB.
NB. thanks to Dan Bron/Jforum 25 April 2006
Lastly when dealing with lists (e.g. strings) of different lengths I
usually use boxing to prevent the issues you have discussed around J
adding padding to regularize the array.
bb =: getSkipsTags shell zipth, filepth
deb each '= ' charsub each bb
or
cc=: ([: deb '= '&charsub) each bb
If you need to write the strings out at some stage (without extra
padding), you can use something like:
'mystrings.txt' fwrites~ ; ,&LF each cc
----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm
