Count words changed in plain text by parsing color codes in git diff --color-words output

words_changed.py

#!/usr/bin/env python

import sys, re
import datetime
from subprocess import Popen, PIPE

rev_cmd = 'git rev-list -n1 --before="%s" master'
diff_cmd = 'git diff --color-words $(%s) -- *.md' % rev_cmd # FIXME: hard coded to markdown files

def count_changed(stream):
    total_added, total_removed = 0,0
    
    for line in stream:
        added = re.findall("\x1b\[32m([^\x1b]+)\x1b\[m", line)
        removed = re.findall("\x1b\[31m([^\x1b]+)\x1b\[m", line)
        total_added += sum([len(s.split()) for s in added])
        total_removed += sum([len(s.split()) for s in removed])
    
    return total_added,total_removed

# NOTE: set up to compute changes since last 4am local time. Change rev_cmd or date computation here for diffs against something else
tod = datetime.time(4)
timestring = tod.strftime('%H:%M%p')
if datetime.datetime.now().time() < tod:
    timestring += ' yesterday'

cmd = diff_cmd % timestring
p = Popen(cmd, shell=True, stdout=PIPE)

total_added,total_removed = count_changed(p.stdout)

print >>sys.stderr, 'Added:',total_added
print >>sys.stderr, 'Removed:',total_removed

print '%d,%d' % (total_added,total_removed)