#!/usr/bin/env python3

# Get target branch from the corresponding mbox
#
# NOTE: this script was based on patches coming to the openembedded-core
# where target branch is defined inside brackets as subject prefix
# i.e. [master], [rocko], etc.
#
# Copyright (C) 2016 Intel Corporation
#
# SPDX-License-Identifier: GPL-2.0-only
#

import mailbox
import argparse
import re
import git

re_prefix = re.compile(r"(\[.*\])", re.DOTALL)

def get_branch(filepath_repo, filepath_mbox, default_branch):
    branch = None

    # get all remotes branches
    gitbranches = git.Git(filepath_repo).branch('-a').splitlines()

    # from gitbranches, just get the names
    branches = [b.split('/')[-1] for b in gitbranches]

    subject = ' '.join(mailbox.mbox(filepath_mbox)[0]['subject'].splitlines())

    # we expect that patches will have somewhere between one and three
    # consecutive sets of square brackets with tokens inside, e.g.:
    # 1. [PATCH]
    # 2. [OE-core][PATCH]
    # 3. [OE-core][kirkstone][PATCH]
    # Some of them may also be part of a series, in which case the PATCH
    # token will be formatted like:
    # [PATCH 1/4]
    # or they will be revisions to previous patches, where it will be:
    # [PATCH v2]
    # Or they may contain both:
    # [PATCH v2 3/4]
    # In any case, we want mprefix to contain all of these tokens so
    # that we can search for branch names within them. 
    mprefix = re.findall(r'\[.*?\]', subject)
    found_branch = None
    if mprefix:
        # Iterate over the tokens and compare against the branch list to
        # figure out which one the patch is targeting
        for token in mprefix:
             stripped = token.lower().strip('[]')
             if default_branch in stripped:
                 found_branch = default_branch
                 break
             else:
                 for branch in branches:
                     # ignore branches named "core"
                     if branch != "core" and stripped.rfind(branch) != -1:
                         found_branch = token.split(' ')[0].strip('[]')
                         break

    # if there's no mprefix content or no known branches were found in
    # the tokens, assume the target is master
    if found_branch is None:
        found_branch = "master"

    return (subject, found_branch)

if __name__ == '__main__':

    parser = argparse.ArgumentParser()
    parser.add_argument('repo', metavar='REPO', help='Main repository')
    parser.add_argument('mbox', metavar='MBOX', help='mbox filename')
    parser.add_argument('--default-branch', metavar='DEFAULT_BRANCH', default='master', help='Use this branch if no one is found')
    parser.add_argument('--separator', '-s', metavar='SEPARATOR', default=' ', help='Char separator for output data')
    args = parser.parse_args()

    subject, branch = get_branch(args.repo, args.mbox, args.default_branch)
    print("branch: %s" % branch)