Create a useless AF spreadsheet from git history

git-history-to-xlsx.py

#!/usr/bin/env python3

import xlsxwriter, os, re, sys
from os.path import join as pjoin, isdir
from datetime import timezone, datetime, timedelta
from pygit2 import Repository as GR, Tree, Blob, Diff, GIT_SORT_REVERSE, GIT_SORT_TIME
from functools import reduce

CONTEXT_LINES = 100
OUTFILE = 'winman-patches.xlsx'
REPLACE_CHARS = set([i for i in '<>;:,.()[]{}$%§"\'?`/´\\=*+~^°#|!& '])
MUTLIPLE_DASHES = re.compile(r'--+')

def get_patch_name(x, m):
    n = ['-' if i in REPLACE_CHARS else i for i in m] # normalize
    n = re.sub(MUTLIPLE_DASHES, '-', ''.join(n)) # replace consecutive - with single -
    return '%04d-%s.patch' % (x, n.lower())

def get_repo_path():
    for rn in ('.', '..'): # executing this from build-dirs
        if isdir(pjoin(rn, '.git')):
            return pjoin(rn, '.git')
    print('Could not find ".git"')
    sys.exit(1)

def main():
    with xlsxwriter.Workbook(OUTFILE) as wb:
        ws = wb.add_worksheet()
        header_format, std_format = wb.add_format({'bold': True}), wb.add_format({})
        r = GR(get_repo_path())
        n = 0 # tracking row for append(), first row
        def append(*args, **kwargs):
            nonlocal n
            for i, a in enumerate(args):
                ws.write(n, i, a, kwargs.get('fmt', std_format))
            n += 1
        append('Index', 'Date', 'Author', 'Short Description', 'Size', 'Insertions', 'Deletions', 'Files Changed', 'Patch File', fmt=header_format)
        for i, c in enumerate(r.walk(r.head.target, GIT_SORT_TIME | GIT_SORT_REVERSE)):
            firstline = c.message.split('\n', 1)[0]
            st = {'insertions': 0, 'deletions': 0, 'files_changed': 0}
            if len(c.parents) > 0:
                s = r.diff(c.parents[0], c).stats
                st['insertions'] = s.insertions
                st['deletions'] = s.deletions
                st['files_changed'] = s.files_changed
            else:
                def s(l):
                    return reduce(lambda x, y: (x[0] + y[0], x[1] + y[1]), (b(r[x.id]) for x in l), (0, 0))
                def b(bl):
                    if isinstance(bl, Blob):
                        d = bl.diff(old_as_path='/dev/null')
                        return (sum(d.line_stats), 1)  # there are only insertions
                    elif isinstance(bl, Tree):
                        return s(bl)
                    return (0, 0)
                ins, fs = s(c.tree)
                st['insertions'] = ins
                st['files_changed'] = fs
            p = get_patch_name(i, firstline)
            os.system('git show -U%d %s > %s' % (CONTEXT_LINES, c.oid, p)) # so, sue me!
            ps = os.stat(p).st_size
            committer = '%s <%s>' % (c.author.name, c.author.email)
            tzi = timezone(timedelta(minutes=c.author.offset))
            t = datetime.fromtimestamp(float(c.author.time), tzi)
            append(str(i+1), t.strftime('%Y-%m-%dT%H:%M:%S%z'), committer, firstline, ps, st['insertions'], st['deletions'], st['files_changed'], p)

__name__ == '__main__' and main()