git-annotate-log.py

#!/usr/bin/python -O
"""
Finds the branch all the commits have most likely been mode on
"""

import os
import sys
import re
from pprint import pprint

if len(sys.argv) < 2:
    print ("USAGE: {0} <list-of-branches>".format([sys.argv[0]]))
    exit(1)

rev_list = os.popen('git log --format="%H:%P:%s" ' + ' '.join(sys.argv[1:]))

heads = {}
for head in sys.argv[1:]:
    if head[0] == "^":
        continue
    sha = os.popen('git rev-parse {0}'.format(head)).read().strip()
    heads[head] = sha

commits = {}

# commit structure:
# [parents, children, msg, distances

for line in rev_list:
    line = line.strip().split(":")
    sha = line[0]
    if sha not in commits:
        commits[sha] = [[], [], "", {}]
    commit = commits[sha]

    # Build Tree
    for parent in line[1].split(' '):
        if parent.strip() == "":
            continue
        if parent not in commits:
            commits[parent] = [[], [], "", {}]
        parent = commits[parent]
        parent[1].append(commit)
        commit[0].append(parent)

    # Commit message
    if len(commit[0]) > 1:
        commit[2] = line[2]


def getBranchNames(message):
    m = re.match(r"Merge branch '([^']+)' into (.+)", message)
    if m is None:
        return None
    else:
        return [m.group(2), m.group(1)]

known = []

for (branch, sha) in heads.iteritems():
    todo = []
    c = commits[sha]
    d = 0
    while True:
        c[3][branch] = d
        d = d - 1
        if len(c[0]) == 2:
            names = getBranchNames(c[2])
            todo.append([c[0][1], d + 100 + 10000 * (names[1] == branch)])
            c = c[0][0]
            d = d + 10000 * (names[0] == branch)
            for (parent, name) in zip(c[0], names):
                if name not in heads:
                    continue
                for parent_child in parent[1]:
                    # Are we the first parent?
                    if parent_child[0][0] == parent:
                        known.append([name, parent_child])
        elif len(c[0]) > 0:
            c = c[0][0]
            todo.extend([x, d + 100] for x in c[0][1:])
        else:
            try:
                (c, d) = todo.pop()
            except IndexError:
                break

branchHeadIds = set(id(commits[sha]) for sha in heads.itervalues())

for (branch, c) in known:
    while True:
        if branch not in c[3]:
            break
        if id(c) in branchHeadIds:
            break
        c[3][branch] = c[3][branch] - 10000
        if len(c[1]) == 1:
            c = c[1][0]
        else:
            break

match = {}

for (sha, commit) in sorted(commits.iteritems()):
    match[sha] = min(commit[3].items(), key=lambda x: x[1])[0]
    #print sha + ": " + match[sha]

log = os.popen('git log --color --oneline --graph --decorate --abbrev=100 '
               + ' '.join(sys.argv[1:])).read().splitlines()

for line in log:
    try:
        sha = re.search('[a-f0-9]{40}', line).group(0)
    except AttributeError:
        print line
        continue
    print re.sub('([a-f0-9]{7})[a-f0-9]{33}',
                 r'\1 [{}]'.format(match[sha]), line)