compare.py

"""Git branch comparator

Compares two git branches based on jira issue conventions. (eg. HADOOP-123). Prints out the commits which exist only on the first branch.

Usage:
   compare.py --repo=REPO <source> <destination>
"""

from git import Repo
import re
from docopt import docopt




def issue_match(commit):
    return (commit, re.search("([A-Z]+-[0-9]+)", commit.summary))

def find_match(x):
    return x[1].group(1)

def accept_true(x):
    return x[1]


if __name__ == '__main__':
    arguments = docopt(__doc__, version='Branch Comparator')
    r = Repo(arguments['--repo'])

    branchA = r.commit(arguments['<source>'])
    branchB = r.commit(arguments['<destination>'])
    merge_base = r.merge_base(branchA, branchB)[0]


    historyA = map(find_match, filter(accept_true, map(issue_match, r.iter_commits(merge_base.hexsha + ".." + branchA.hexsha))))
    historyB = map(find_match, filter(accept_true, map(issue_match, r.iter_commits(merge_base.hexsha + ".." + branchB.hexsha))))


    issue_to_commit =  dict(map((lambda x: (x[1].group(1), x[0].summary)), filter(accept_true, map(issue_match, r.iter_commits(merge_base.hexsha + ".." + branchA.hexsha)))))

    firstOnly = [val for val in historyA if val not in historyB]
    print("Numer of issues", len(firstOnly))

    print("\n".join([issue_to_commit[issue_id] for issue_id in firstOnly]))