git-ls-objects

#!/usr/bin/env python3

import subprocess
import sys
import argparse


def get_options():
    """Process system args and return a namespace with their values."""
    parser = argparse.ArgumentParser(description='List blob objects present in a git repository.')
    group = parser.add_argument_group("Select output")
    group.add_argument('--trees', action="store_true",
                       help="List trees instead of blobs")
    group.add_argument('--commits', action="store_true",
                       help="List commits instead of blobs")
    fgroup = parser.add_argument_group("Format output")
    fgroup.add_argument('--ids', action="store_true",
                        help="Print git ID")
    return parser.parse_args()


def get_str_output(program_and_args, encoding=None):
    """Like :py:func:`subprocess.check_output`, but returns the output as
    a :py:class:`str` rather than :py:class:`bytes`.

    :param list program_and_args: list of the program to be executed
        and any arguments to be passed to it. The call should *not*
        require a shell (see `shell` argument to
        :py:meth:`subprocess.POpen`)

    :param str encoding: specifies the encoding expected for the
        output; if omitted, defaults to the result of
        :py:func:`sys.getdefaultencoding`.

    """
    the_encoding = sys.getdefaultencoding() if encoding is None else encoding
    return subprocess.check_output(program_and_args).decode(the_encoding)


def git_object_type(sha1):
    """:returns: the type of the object with hash `sha1`
       :rtype: str

    """
    return get_str_output(["git", "cat-file", "-t", sha1]).replace("\n", "")


def main(opts):
    trees = set()  # each item is a tuple (hash, directory)
    commits = set()  # each item is a tuple (hash, '')
    blobs = set()  # each item is a tuple (hash, filename)
    # handy way to get set from objet type
    obj_sets = {"tree": trees, "commit": commits, "blob": blobs}

    all_objects_out = get_str_output(["git", "rev-list", "--objects", "--all"])

    for line in all_objects_out.split("\n"):
        parts = line.split(" ")
        if parts[0]:  # why do I get an empty line?
            typ = git_object_type(parts[0])
            # join to handle spaces in file names
            obj_sets[typ].add((parts[0], " ".join(parts[1:])))

    if opts.trees:
        for item in trees:
            print("{}\t{}".format(*item) if opts.ids else item[1])
    elif opts.commits:
        for item in commits:
            print(item[0])
    else:
        for item in blobs:
            print("{}\t{}".format(*item) if opts.ids else item[1])


if __name__ == '__main__':
    main(get_options())