Re: csplit - split by content of field

[Pádraig Brady]

On 02/06/2013 10:09 PM, Assaf Gordon wrote:
> Hello,
>
> Attach is a patch that gives 'csplit' the ability to split files by content of 
> a field.
> A typical usage is:
>
>      ## the "@1" pattern means "start a new file when field 1 changes"
>      $ printf "A\nA\nB\nB\nB\nC\n" | csplit - @1 {*}
>      $ wc -l xx*
>      2 xx00
>      3 xx01
>      1 xx02
>      6 total
>      $ head xx*
>      ==> xx00 <==
>      A
>      A
>
>      ==> xx01 <==
>      B
>      B
>      B
>
>      ==> xx02 <==
>      C
>
>
>
> This is just a proof of concept, and the pattern specification can be changed (I think 
> "@N" doesn't conflict with any existing pattern).
>
> The same can probably be achieved using other programs (awk comes to mind), but 
> it won't be as simple and clean (with all of csplit's output features).
>
> Let me know if you're willing to consider such addition.

Yes such a feature is useful, though maybe in conjuntion with uniq:
http://lists.gnu.org/archive/html/coreutils/2011-03/msg00000.html

So basically the proposal there is to support --suppress-matched
so that you could then do:

uniq -w1 --unique=separated --all-repeated=separated |
csplit --suppress-matched '/^$/' '{*}'

The caveat with that though is that uniq would
benefit from better field selection, which is
also on the TODO list.

cheers,
Pádraig.