How split regions in half to get at most 1000 bp length?

Question

[star]

I have a table of genomic regions that I would like to split them in half until all regions are in at most 1000 bp long.

I have tried the bedops --chop 1000 data but it doesn`t work and it gave me some regions with 1 bp length.

Data : 
chr   start end   length   center of region
chr1   500   3000  2500         1750   
chr2   100   1100  1000         600
chr4   20    1900  1880         960


results:
chr    start end   length
chr1   500   1125    625
chr1   1125  1750    625
chr1   1750  2375    625
chr1   2375  3000    625
chr2   100   1100    1000
chr4   20    960     940
chr4   960    1900   940

Answer 1

[Pierre Lindenbaum]

with

function dump(C,S,E)
    {
    if( E-S <= 1000) {
        printf("%s\t%d\t%d\n",C,S,E);
        }
    else
        {
        dump(C,S,S+(E-S)/2);
        dump(C,S+(E-S)/2,E);
        }
    }

    {
    dump($1,int($2),int($3));
    }

usage

awk -f script.awk < input.bed