Merge localisation files

merge_localizations.py

#!/usr/bin/python

import re
import sys
import codecs

generated_regex = re.compile(r"/\*(?P<comment>.*?)\*/[\n\r]*\"(?P<key>.*?)\"\s*=\s*\"(?P<value>.*?)\";", re.S)
handcrafted_regex = re.compile(r"\"(?P<key>.*?)\"\s*=\s*\"(?P<value>.*?)\";(?P<comment>)", re.S)

def parse(file_name, regex, encoding = "utf-8"):
    with codecs.open(file_name, mode="rb", encoding = encoding) as file:
        contents = file.read();
        return [(m.group("key"), m.group("value"), m.group("comment")) for m in re.finditer(regex, contents)]

def merge(old, new):
    old_val = dict([(v[0], v[1]) for v in old])
    new_val = dict([(v[0], v[1]) for v in new])
    old_comment = dict([(v[0], v[2]) for v in old])
    new_comment = dict([(v[0], v[2]) for v in new])
    all_keys = set([v[0] for v in old] + [v[0] for v in new])
    return sorted([(key, \
                     new_val.has_key(key) and new_val[key] or old_val.has_key(key) and old_val[key], \
                     new_comment.has_key(key) and new_comment[key] or old_comment.has_key(key) and old_comment[key] or " No comment provided by engineer. "
                     ) for key in all_keys], key=lambda tup: tup[2])

def genstrings_print_triplet(file, triplet):
    print >> file, """/*%s*/\n"%s" = "%s";\n""" % (triplet[2], triplet[0], triplet[1])

def csv_print_triplet(file, triplet):
    print >> file, """%s\t%s\t%s""" % (triplet[0], triplet[2], triplet[1])

def print_content(file_name, triplets, formatter = genstrings_print_triplet):
    with codecs.open(file_name, "w", encoding="utf-8") as file:
        for t in triplets:
            formatter(file, t)

def main():
    if len(sys.argv) != 4:
        print "This script merges existing and new localization files by adding new key-value pairs (replacing old ones)."
        print "Usage: merge-localizations <old file> <new file> <output file>"
        return
    old_content = parse(sys.argv[1], generated_regex)
    print 'Old:', len(old_content)
    new_content = parse(sys.argv[2], handcrafted_regex)
    print 'New', len(new_content)
    #print set([v[0] for v in old_content]) - set([v[0] for v in new_content])
    merged_content = merge(old_content, new_content)
    print_content(sys.argv[3], merged_content)

if __name__ == '__main__':
    main()