Usage:

    $ python2 diff-large-files.py fileA fileB

Pass -h to get options. Output is in unified diff format, possible to
pipe into `dwdiff --diff-input` to get word-diffs.

The script is meant to give visually inspectable diffs of large files
with few differences, to avoid the memory consumption issue with
diffing large files. When a difference is found, it moves forth in
both files a line at a time, looking for a line in one file that
matches a previously seen line in the other file.

This simple method works well for some types of files, but can easily
match the wrong lines if e.g. every other line is identical (or there
are a lot of identical lines in the files). In that case, you're
probably better off using `split` to split your input files and then
sending each file-pair into regular `diff`.