Re: Specific fixed width question...

[CdeMills]

Sergei Steshenko-2 wrote:
> 
> 
>> From: John W. Eaton <address@hidden>
>> Subject: Re: Specific fixed width question...
>> To: "CdeMills" <address@hidden>
>> Cc: address@hidden
> [snip]
>> Is there a good way to write the above
>> using a regexp that
>> will only match the data in columns 4-6, 8-10, and 37-39?
>> 
>> jwe
> 
> 
> Well, in Perl (and, I guess, in PCRE used by Octave) '.' matches any
> character, so, say, the following
> 
> $s =~ m/^.{3}(.{4}).{5}(.{6})/;
> #              $1        $2
> 
> skips 3 character from the beginning, then matches 4 characters and puts
> them into $1, then skips 5 characters and then matches 6 characters and
> puts them into $2.
> 
> 

We have reached here a limitation of regexpes, that is, the inability to
specify what to return and what to not return. According to 
https://secure.wikimedia.org/wikipedia/en/wiki/Regexp#POSIX_Basic_Regular_Expressions
the construct \( ... \) and \1 up to \9 was not retained in POSIX regexpes. 

About the given problem, one idea would be to return the start of the regexp
'\d+' and to check if the start falls inside some allowed range:
regexp('WA 053 011 Clark County 123            45 ', '\d+') 
ans= 
        4    8   25   40
and skip the third reply, because it falls inside a range where numeric are
not relevant.

Regards

Pascal
-- 
View this message in context: 
http://octave.1599824.n4.nabble.com/Specific-fixed-width-question-tp2539259p2541930.html
Sent from the Octave - General mailing list archive at Nabble.com.