xtar

#!/usr/bin/env python3
#
# Extracts from tarfiles, but resolves symlinks
#

import argparse
import copy
import fnmatch
import os
import pathlib
import tarfile
import typing


class CustomTarfile(tarfile.TarFile):
    def makelink(self, tarinfo, targetpath):
        self._extract_member(self._find_link_target(tarinfo), targetpath)


def check_member(
    excludes, includes, member: tarfile.TarInfo
) -> typing.Optional[tarfile.TarInfo]:
    if ".." in member.name or member.name.startswith("/"):
        raise ValueError("Invalid member")
    for excl in excludes:
        if fnmatch.fnmatch(member.name, excl):
            return
    for incl in includes:
        if fnmatch.fnmatch(member.name, incl):
            return member


def filter_members(excludes, includes, tfp: CustomTarfile, strip: typing.Optional[int]):
    for member in tfp.getmembers():
        member = check_member(excludes, includes, member)
        if member:
            if strip:
                name = pathlib.PurePosixPath(member.name)
                parts = name.parts[strip:]
                if not parts:
                    continue
                while member.issym():
                    member = tfp._find_link_target(member)
                member = copy.deepcopy(member)
                member.name = "/".join(parts)
            yield member


if __name__ == "__main__":

    parser = argparse.ArgumentParser()
    parser.add_argument("--directory", "-C", default=".")
    parser.add_argument("--file", "-f", required=True)
    parser.add_argument("-x", required=True, action="store_true")
    parser.add_argument("--exclude", action="append", default=[])
    parser.add_argument("--strip", default=None, type=int)
    parser.add_argument("include", nargs="*", default=["*"])

    args = parser.parse_args()

    with CustomTarfile.open(args.file, "r:*") as tfp:

        tfp.extractall(
            path=args.directory,
            members=filter_members(args.exclude, args.include, tfp, args.strip),
            numeric_owner=True,
        )