Picking Random Genomic Positions

I do have a set of TF binding coordinates and want to see if there is any significant overlap with an open chromatin annotation.

Example of TF coord:
chr1 19280 19298
chr1 245920 245938
chr2 97290 97308
chr9 752910 752938
...

Example of open chrom. coord. (UCSC track):
chr2 33031543 33032779
chr3 2304169 2304825
chr5 330899 330940
...

I have checked the intersection with the Bedtools (open chrom. coord vs TF coord. -/+ 100bp) and now I want to check the intersection between random genomic coordinates and open chrom.

The idea is to:

1. Pick random genomic position (from the same chromosome as TF coordinate);
2. -/+9bp (binding site size);
3. -/+ 100bp;
4. Run this simulation for 1000 times (TF x 1000);
5. Bedtools;

Any ideas how can I do this simulation to pick random genomic positions from the same chromosome? I know a little bit of bash and Perl, but won't be able to write the script by myself.
Is it possible to measure the length of every chromosome;
Pick TF chromosome and from it's length get a random number which would represent a genomic position?

Can someone help me with the simulation and the pipeline.