dulwich find commit

gistfile1.py

#!/usr/bin/env python2

from dulwich.repo import Repo
from collections import defaultdict, deque


def changes_file(key):
    def matches(commit, entry):
        changes = entry.changes()

        if len(commit.parents) < 2:
            changes = [changes]

        for base in changes:
            for ch in base:
                if ch.new.path == key:
                    return True

        return False

    return matches


def find_merge(repo, children, id):
    queue = deque(children[id])

    while len(queue) > 0:
        commit = repo[queue.popleft()]

        if len(commit.parents) > 1:
            return commit

        queue.append(commit.parents[0])


def find_commits(repo, key):
    children = defaultdict(list)

    for entry in repo.get_walker():
        commit = entry.commit

        if key(commit, entry):
            yield commit, find_merge(repo, children, commit.id)

        for parent in commit.parents:
            children[parent].append(commit.id)


repo = Repo('/home/keis/git/data/.git')
key = 'somefile.txt'

for commit, merge in find_commits(repo, changes_file(key)):
    print '**'
    print merge
    print '--'
    print commit