Re: [bug-gawk] FIELDWIDTHS extension request

[Arnold Robbins]

Hi.

OK. I've added this feature to gawk's master. It'll be in the next
release. Andrew Schorr was helpful in shaking it out.  The doc
on FIELDWIDTHS has also been considerably revised and there are now
a number of additional tests.

The only thing that's allowed is '*' as the last field width
designator. None of the other flights of fancy discussed in
the comp.lang.awk thread will be implemented.

Thanks,

Arnold

> Subject: Re: [bug-gawk] FIELDWIDTHS extension request
> To: address@hidden, address@hidden
> From: Ed Morton <address@hidden>
> Message-ID: <address@hidden>
> Date: Fri, 19 May 2017 07:45:00 -0500
>
> People are really using it this way. I personally have to do it a couple of 
> times a year with different situations (parsing log files, schedules, etc. in 
> various formats) and then there's, for example, the following questions and 
> comments from people online using or best served by the same solution:
>
> http://www.unix.com/unix-for-dummies-questions-and-answers/259428-find-replace-single-character-w-awk-given-conditions.html
> https://unix.stackexchange.com/questions/208888/sed-replace-at-a-specific-group-of-positions
> https://www.cpume.com/question/tfnznotf-formatting-output-in-bash-or-awk.html
> http://blog.bogojoker.com/tag/gawk/
> http://bida001.com/question/119893
> http://stackoverflow.com/q/35178883/1745001
> http://stackoverflow.com/q/26842477/1745001
> http://stackoverflow.com/q/43942619/1745001
> http://stackoverflow.com/q/24909124/1745001
> https://groups.google.com/d/msg/comp.lang.awk/uip6-Ll9Lj0/_wMoc7lEAwAJ
>
> Regards,
>
>      Ed.
>
> On 5/19/2017 4:13 AM, address@hidden wrote:
> > Is this request based on a theoretical discussion, or is someone
> > actually using FIELDWIDTHS do real work?
> >
> > Thanks,
> >
> > Arnold
> >
> > Ed Morton <address@hidden> wrote:
> >
> >> Could we get an extension for FIELDWIDTHS such that a "*" (or some other
> >> indicator) means "everything else"? It's very common to have to deal with 
> >> files
> >> like:
> >>
> >>       foo  bar    stuff
> >>       bill johnny other stuff of some different length
> >>
> >> where it's easy to see the length of the first 2 fields but the length of 
> >> the
> >> final field can't be determined based on any sample so we end up writing 
> >> code like:
> >>
> >>       FIELDWIDTHS = "5 7 9999"
> >>
> >> and pick some arbitrary number that we hope will be big enough to hold 
> >> whatever
> >> appears in that 3rd field. It would be clearer and more robust if we could
> >> instead write:
> >>
> >>       FIELDWIDTHS = "5 7 *"
> >>
> >> or similar.
> >>
> >> We've been having a discussion about the applicability of this at
> >> https://groups.google.com/forum/#!topic/comp.lang.awk/uip6-Ll9Lj0 in case 
> >> some
> >> more background is useful.
> >>
> >> Regards,
> >>
> >>       Ed.
>