A commit ID mapping creator for a hg repository and a (converted) git repository based on commit timestamps

create-hg-git-rev-mapping.py

#! /usr/bin/env python3
import os, sys
import subprocess
from datetime import datetime

def execute(cmd):
    proc = subprocess.Popen(cmd, shell=True, stdout=subprocess.PIPE)
    stdout, stderr = proc.communicate()
    return stdout.decode('utf-8')

def parse_logitem(line):
    pieces = line.split(' ')
    timestamp = datetime.strptime(' '.join(pieces[:3]), '%Y-%m-%d %H:%M:%S %z')
    tt = datetime.utctimetuple(timestamp)  # normalize the datetime object to naive UTC one.
    timestamp_utc = datetime(tt[0], tt[1], tt[2], tt[3], tt[4], tt[5], tzinfo=None)
    commit_id = pieces[3]
    short_desc = ' '.join(pieces[4:])
    return timestamp_utc, commit_id, short_desc

if __name__ == '__main__':
    hg_path = sys.argv[1]
    git_path = sys.argv[2]

    basedir = os.path.abspath(os.path.dirname(__file__))
    os.chdir(os.path.join(basedir, hg_path))
    print('reading hg log...', file=sys.stderr)
    hg_log_output = execute("hg log --template '{date|isodatesec} {node} {desc|firstline}\n' -r :")
    hg_log = {}
    for line in hg_log_output.splitlines():
        timestamp, commit_id, short_desc = parse_logitem(line)
        hg_log[timestamp] = (commit_id, short_desc)
    os.chdir(os.path.join(basedir, git_path))
    print('reading git log...', file=sys.stderr)
    git_log_output = execute("git --no-pager log --format='%ai %H %s' --reverse --all")
    git_log = {}
    for line in git_log_output.splitlines():
        timestamp, commit_id, short_desc = parse_logitem(line)
        git_log[timestamp] = (commit_id, short_desc)

    count_match, count_mismatch = 0, 0
    for hg_time, hg_info in hg_log.items():
        if hg_time in git_log:
            print('{0} => {1}'.format(hg_info[0], git_log[hg_time][0]))
            count_match += 1
        else:
            print('Mismatch: {0} at {2:%Y-%m-%d %H:%M:%S %z} - {1}'.format(hg_info[0], hg_info[1], hg_time))
            count_mismatch += 1
    print('Matched {0} commits, mismatched {1} commits.'.format(count_match, count_mismatch), file=sys.stderr)

    # Just to make it sure...
    count_match, count_mismatch = 0, 0
    for git_time, git_info in git_log.items():
        if git_time in hg_log:
            count_match += 1
        else:
            count_mismatch += 1
    print('In the reverse: Matched {0} commits, mismatched {1} commits.'.format(count_match, count_mismatch), file=sys.stderr)