[COFF] Re: Grep has gone overboard

[Ralph Corderoy <ralph@inputplus.co.uk>]

Hi Bakul,

> Forgot to add, you can solve this another way  (taking the abc
> example):
>   grep ‘^...$’ | grep ‘[^a]*a[^a]*’ | grep ‘[^b]*b[^b]*’ | grep ‘[^c]*c[^c]*’

I know you know this and the error is just from typing an email, but in
case other listeners are following along, all the greps need start and
end-of-line anchors.

Without them,

    grep '[^a]*a[^a]*'

still matches ‘aa’ because there are zero non-a before the first a, then
the first a matches, and there are zero non-a after it.  Here's ‘aa’
with indexes between the characters.

     a a  
    0 1 2

    [^a]*  matches  0-0
    a      matches  0-1
    [^a]*  matches  1-1

And it's easier to invert the test into

    grep -v 'a.*a'

I'll stop waffling on about regexps now.  Though I do have a SNOBOL4
question for TUHS which I should get around to writing.  :-)

-- 
Cheers, Ralph.