#!/usr/bin/pypy
#
# Generate a database of commits and major versions they went into.
#
# This is the painfully slow reworked brute-force version that
# takes forever to run, but which hopefully gets the right results
#
# committags [git-args]
#
# This code is part of the LWN git data miner.
#
# Copyright 2007-13 Eklektix, Inc.
# Copyright 2007-13 Jonathan Corbet <corbet@lwn.net>
#
# This file may be distributed under the terms of the GNU General
# Public License, version 2.
#
import sys
import re
import os
import pickle
import argparse

#
# Arg parsing stuff.
#
def setupargs():
    p = argparse.ArgumentParser()
    #
    # -d for the database
    # -l to load it before running
    #
    p.add_argument('-d', '--database', help = 'Database name',
                   required = False, default = 'committags.db')
    p.add_argument('-l', '--load', help = 'Load database at startup',
                   default = False, action = 'store_true')
    #
    # Args for git?
    #
    p.add_argument('-g', '--git', help = 'Arguments to git',
                   default = '')
    #
    # Where's the repo?
    #
    p.add_argument('-r', '--repository', help = 'Repository location',
                   default = '')
    return p


p = setupargs()
args = p.parse_args()

#
# Pull in an existing database if requested.
#
if args.load:
    DB = pickle.load(open(args.database, 'r'))
else:
    DB = { }
out = open(args.database, 'w')

#
# Time to fire up git.
#
git = 'git log --pretty=format:%H ' + args.git
if args.repository:
    os.chdir(args.repository)
input = os.popen(git, 'r')

nc = 0
for line in input.readlines():
    commit = line.strip()
    #
    # If we loaded a database and this commit is already there, we
    # can quit.
    #
    if args.load and DB.has_key(commit):
        break
    #
    # Figure out which version this one came from.
    #
    desc = os.popen('git describe --contains --match v\\* ' + commit, 'r')
    tag = desc.readline().strip()
    desc.close()
    dash = tag.find('-')
    if dash > 0:
        DB[commit] = tag[:dash]
    else:
        DB[commit] = tag
    #
    # Give them something to watch.
    #
    nc += 1
    if (nc % 25) == 0:
        print '%6d %s %s           \r' % (nc, commit[:8], tag),
        sys.stdout.flush()

print '\nFound %d/%d commits' % (nc, len(DB.keys()))
pickle.dump(DB, out)
out.close()