carve-tar.py

#!/usr/bin/python3

import argparse
import datetime
import os
import sys
import tarfile


def get_modes(info):
    s = ""

    if info.type == tarfile.REGTYPE or info.type == tarfile.AREGTYPE:
        s += "-"
    elif info.type == tarfile.LNKTYPE:
        s += "l"
    elif info.type == tarfile.SYMTYPE:
        s += "s"
    elif info.type == tarfile.CHRTYPE:
        s += "c"
    elif info.type == tarfile.BLKTYPE:
        s += "b"
    elif info.type == tarfile.DIRTYPE:
        s += "d"
    elif info.type == tarfile.FIFOTYPE:
        s += "p"
    else:
        s += "?"

    # Owner
    if info.mode & 0o400:
        s += "r"
    else:
        s += "-"
    if info.mode & 0o200:
        s += "w"
    else:
        s += "-"
    if info.mode & 0o100:
        s += "x"
    else:
        s += "-"

    # Group
    if info.mode & 0o040:
        s += "r"
    else:
        s += "-"
    if info.mode & 0o020:
        s += "w"
    else:
        s += "-"
    if info.mode & 0o010:
        s += "x"
    else:
        s += "-"

    # Other
    if info.mode & 0o004:
        s += "r"
    else:
        s += "-"
    if info.mode & 0o002:
        s += "w"
    else:
        s += "-"
    if info.mode & 0o001:
        s += "x"
    else:
        s += "-"

    return s


parser = argparse.ArgumentParser(
    description="carve files from tar archive",
    formatter_class=argparse.ArgumentDefaultsHelpFormatter,
)

parser.add_argument("--verbose", action="store_true", help="enable verbose output")
parser.add_argument("--errors", action="store_true", help="output tar header errors")
parser.add_argument("--offsets", action="store_true", help="output tar header offsets")
parser.add_argument("--path", default=os.getcwd(), help="output directory")
parser.add_argument(
    "--attrs",
    action=argparse.BooleanOptionalAction,
    default=True,
    help="set output mtime and mode",
)
parser.add_argument(
    "-t", "--list", action="store_true", help="list contents of archive"
)
parser.add_argument(
    "-x", "--extract", action="store_true", help="extract files from archive"
)
parser.add_argument("file", nargs="?", help="path to file")

args = parser.parse_args()

if not args.list and not args.extract:
    print("Option '--list' or '--extract' required.", file=sys.stderr)
    print(f"Try '{sys.argv[0]} --help' for more information.", file=sys.stderr)
    sys.exit(2)

if args.list and args.extract:
    print(
        "You may not specify more than one '--list' or '--extract' option.",
        file=sys.stderr,
    )
    print(f"Try '{sys.argv[0]} --help' for more information.", file=sys.stderr)
    sys.exit(2)

if args.file and args.file != "-":
    f = open(args.file, "rb")
else:
    f = sys.stdin.buffer

while True:
    data = f.read(tarfile.BLOCKSIZE)
    if len(data) != tarfile.BLOCKSIZE:
        break

    try:
        info = tarfile.TarInfo.frombuf(data, "utf-8", "strict")
    except (tarfile.HeaderError, UnicodeDecodeError) as e:
        if args.errors:
            print(f"Invalid tar header at {f.tell() - tarfile.BLOCKSIZE}: {e}")
        continue
    except Exception as e:
        print(f"Unexpected error at {f.tell() - tarfile.BLOCKSIZE}", file=sys.stderr)
        raise e

    if args.offsets:
        print(f"Valid tar header at {f.tell() - tarfile.BLOCKSIZE}")

    data_len = (info.size + (tarfile.BLOCKSIZE - 1)) & ~(tarfile.BLOCKSIZE - 1)

    name = info.name.lstrip("/")
    output = os.path.join(args.path, name)

    if args.list:
        if args.verbose:
            modes = get_modes(info)

            size = str(info.size)
            uid = str(info.uid)
            gid = str(info.gid)

            width = 19
            pad = len(uid) + 1 + len(gid) + 1 + len(size)
            if pad > width:
                width = pad

            mtime = datetime.datetime.fromtimestamp(info.mtime, datetime.UTC)

            print(
                f"{modes} {uid}/{gid} {info.size: >{width - pad + len(size)}} {mtime} {name}"
            )
        else:
            print(name)
    else:
        if args.verbose:
            print(name)

        os.makedirs(os.path.dirname(output), mode=0o755, exist_ok=True)

        with open(output, "wb") as o:
            data = f.read(data_len)
            o.write(data[: info.size])

        if args.attrs:
            os.chmod(output, info.mode, follow_symlinks=False)
            os.utime(output, (info.mtime, info.mtime), follow_symlinks=False)