Updates the icml files for the handbook's chapters based on a diff file from txt extracts

update-icmls.py

import re
import fileinput
import os

from glob import glob

ICML_DIR = "icml"
OUT_DIR = "output"

# Get icml files in memory
data = {}
for path in glob(ICML_DIR + '/*.icml'):
  with open(path, 'r') as ficml:
    data[path] = ''.join(ficml.readlines())

# Get all diffs to process
diffinfo = ''.join(fileinput.input())
diffs = re.split(r"\n(?=diff\b)", diffinfo)

# Go through the diffs and apply them to all files
for diff in diffs:
  chapter = re.match(r'diff \s --git \s "? a/(\d+)[^\s]+ "? \s "? b/[^\s]+"?', diff, re.X | re.I).group(1)
  for line in diff.split("\n"):
    if line.startswith("-") and not line.startswith("---"):
      oldstr = line[1:]
    elif line.startswith("+") and not line.startswith("+++"):
      newstr = line[1:]
  
  for path, contents in data.iteritems():  
    data[path] = contents.replace(oldstr, newstr)

# Write all files to output
os.makedirs('output')
for path, contents in data.iteritems():
  with open(path.replace(ICML_DIR + '/', OUT_DIR + '/'), 'w') as out:
    out.write(contents)