amolangstats.py

"""
Takes CSV locale statistics from addons.mozilla.org and sums numbers
for same languages (e.g. es = es + es_ES + es_MX + ...).  Output is
similarly formatted CSV.
"""
import sys

languages = {}

for line in open(sys.argv[-1]):
    if line.startswith('# Fields'):
        line = line[11:-2]
        locales = line.split(';')[2:]
        for i, locale in enumerate(locales):
            chunks = locale.split('-')
            languages.setdefault(chunks[0], []).append(i)
        print 'date,total,' + ','.join(sorted(languages))
        continue
    if line.startswith('#'):
        continue

    data = line.split(',');
    date, total = data.pop(0), data.pop(0)
    data = [int(value) for value in data]
    print (
        date + ',' + total + ',' +
        ','.join(str(sum(data[i] for i in positions))
                 for lang, positions in sorted(languages.items()))
        )