git_dumper.py

#!/usr/bin/env python3
from contextlib import closing
import argparse
import multiprocessing
import os
import os.path
import re
import socket
import subprocess
import sys
import traceback
import urllib.parse
import urllib3

import bs4
import dulwich.index
import dulwich.objects
import dulwich.pack
import requests
import socks


def printf(fmt, *args, file=sys.stdout):
    if args:
        fmt = fmt % args

    file.write(fmt)
    file.flush()


def is_html(response):
    """ Return True if the response is a HTML webpage """
    return (
        "Content-Type" in response.headers
        and "text/html" in response.headers["Content-Type"]
    )


def is_safe_path(path):
    """ Prevent directory traversal attacks """
    if path.startswith("/"):
        return False

    safe_path = os.path.expanduser("~")
    return (
        os.path.commonpath(
            (os.path.realpath(os.path.join(safe_path, path)), safe_path)
        )
        == safe_path
    )


def get_indexed_files(response):
    """ Return all the files in the directory index webpage """
    html = bs4.BeautifulSoup(response.text, "html.parser")
    files = []

    for link in html.find_all("a"):
        url = urllib.parse.urlparse(link.get("href"))

        if (
            url.path
            and is_safe_path(url.path)
            and not url.scheme
            and not url.netloc
        ):
            files.append(url.path)

    return files


def verify_response(response):
    if response.status_code != 200:
        return (
            False,
            "[-] %s/%s responded with status code {code}\n".format(
                code=response.status_code
            ),
        )
    elif (
        "Content-Length" in response.headers
        and response.headers["Content-Length"] == 0
    ):
        return False, "[-] %s/%s responded with a zero-length body\n"
    elif (
        "Content-Type" in response.headers
        and "text/html" in response.headers["Content-Type"]
    ):
        return False, "[-] %s/%s responded with HTML\n"
    else:
        return True, True


def create_intermediate_dirs(path):
    """ Create intermediate directories, if necessary """

    dirname, basename = os.path.split(path)

    if dirname and not os.path.exists(dirname):
        try:
            os.makedirs(dirname)
        except FileExistsError:
            pass  # race condition


def get_referenced_sha1(obj_file):
    """ Return all the referenced SHA1 in the given object file """
    objs = []

    if isinstance(obj_file, dulwich.objects.Commit):
        objs.append(obj_file.tree.decode())

        for parent in obj_file.parents:
            objs.append(parent.decode())
    elif isinstance(obj_file, dulwich.objects.Tree):
        for item in obj_file.iteritems():
            objs.append(item.sha.decode())
    elif isinstance(obj_file, dulwich.objects.Blob):
        pass
    elif isinstance(obj_file, dulwich.objects.Tag):
        pass
    else:
        printf(
            "error: unexpected object type: %r\n" % obj_file, file=sys.stderr
        )
        sys.exit(1)

    return objs


class Worker(multiprocessing.Process):
    """ Worker for process_tasks """

    def __init__(self, pending_tasks, tasks_done, args):
        super().__init__()
        self.daemon = True
        self.pending_tasks = pending_tasks
        self.tasks_done = tasks_done
        self.args = args

    def run(self):
        # initialize process
        self.init(*self.args)

        # fetch and do tasks
        while True:
            task = self.pending_tasks.get(block=True)

            if task is None:  # end signal
                return

            try:
                result = self.do_task(task, *self.args)
            except Exception:
                printf("Task %s raised exception:\n", task, file=sys.stderr)
                traceback.print_exc()
                result = []

            assert isinstance(
                result, list
            ), "do_task() should return a list of tasks"

            self.tasks_done.put(result)

    def init(self, *args):
        raise NotImplementedError

    def do_task(self, task, *args):
        raise NotImplementedError


def process_tasks(initial_tasks, worker, jobs, args=(), tasks_done=None):
    """ Process tasks in parallel """

    if not initial_tasks:
        return

    tasks_seen = set(tasks_done) if tasks_done else set()
    pending_tasks = multiprocessing.Queue()
    tasks_done = multiprocessing.Queue()
    num_pending_tasks = 0

    # add all initial tasks in the queue
    for task in initial_tasks:
        assert task is not None

        if task not in tasks_seen:
            pending_tasks.put(task)
            num_pending_tasks += 1
            tasks_seen.add(task)

    # initialize processes
    processes = [worker(pending_tasks, tasks_done, args) for _ in range(jobs)]

    # launch them all
    for p in processes:
        p.start()

    # collect task results
    while num_pending_tasks > 0:
        task_result = tasks_done.get(block=True)
        num_pending_tasks -= 1

        for task in task_result:
            assert task is not None

            if task not in tasks_seen:
                pending_tasks.put(task)
                num_pending_tasks += 1
                tasks_seen.add(task)

    # send termination signal (task=None)
    for _ in range(jobs):
        pending_tasks.put(None)

    # join all
    for p in processes:
        p.join()


class DownloadWorker(Worker):
    """ Download a list of files """

    def init(self, url, directory, retry, timeout, http_headers):
        self.session = requests.Session()
        self.session.verify = False
        self.session.headers = http_headers
        self.session.mount(
            url, requests.adapters.HTTPAdapter(max_retries=retry)
        )

    def do_task(self, filepath, url, directory, retry, timeout, http_headers):
        if os.path.isfile(os.path.join(directory, filepath)):
            printf("[-] Already downloaded %s/%s\n", url, filepath)
            return []

        with closing(
            self.session.get(
                "%s/%s" % (url, filepath),
                allow_redirects=False,
                stream=True,
                timeout=timeout,
            )
        ) as response:
            printf(
                "[-] Fetching %s/%s [%d]\n",
                url,
                filepath,
                response.status_code,
            )

            valid, error_message = verify_response(response)
            if not valid:
                printf(error_message, url, filepath, file=sys.stderr)
                return []

            abspath = os.path.abspath(os.path.join(directory, filepath))
            create_intermediate_dirs(abspath)

            # write file
            with open(abspath, "wb") as f:
                for chunk in response.iter_content(4096):
                    f.write(chunk)

            return []


class RecursiveDownloadWorker(DownloadWorker):
    """ Download a directory recursively """

    def do_task(self, filepath, url, directory, retry, timeout, http_headers):
        if os.path.isfile(os.path.join(directory, filepath)):
            printf("[-] Already downloaded %s/%s\n", url, filepath)
            return []

        with closing(
            self.session.get(
                "%s/%s" % (url, filepath),
                allow_redirects=False,
                stream=True,
                timeout=timeout,
            )
        ) as response:
            printf(
                "[-] Fetching %s/%s [%d]\n",
                url,
                filepath,
                response.status_code,
            )

            if (
                response.status_code in (301, 302)
                and "Location" in response.headers
                and response.headers["Location"].endswith(filepath + "/")
            ):
                return [filepath + "/"]

            if filepath.endswith("/"):  # directory index
                assert is_html(response)

                return [
                    filepath + filename
                    for filename in get_indexed_files(response)
                ]
            else:  # file
                valid, error_message = verify_response(response)
                if not valid:
                    printf(error_message, url, filepath, file=sys.stderr)
                    return []

                abspath = os.path.abspath(os.path.join(directory, filepath))
                create_intermediate_dirs(abspath)

                # write file
                with open(abspath, "wb") as f:
                    for chunk in response.iter_content(4096):
                        f.write(chunk)

                return []


class FindRefsWorker(DownloadWorker):
    """ Find refs/ """

    def do_task(self, filepath, url, directory, retry, timeout, http_headers):
        response = self.session.get(
            "%s/%s" % (url, filepath), allow_redirects=False, timeout=timeout
        )
        printf(
            "[-] Fetching %s/%s [%d]\n", url, filepath, response.status_code
        )

        valid, error_message = verify_response(response)
        if not valid:
            printf(error_message, url, filepath, file=sys.stderr)
            return []

        abspath = os.path.abspath(os.path.join(directory, filepath))
        create_intermediate_dirs(abspath)

        # write file
        with open(abspath, "w") as f:
            f.write(response.text)

        # find refs
        tasks = []

        for ref in re.findall(
            r"(refs(/[a-zA-Z0-9\-\.\_\*]+)+)", response.text
        ):
            ref = ref[0]
            if not ref.endswith("*") and is_safe_path(ref):
                tasks.append(".git/%s" % ref)
                tasks.append(".git/logs/%s" % ref)

        return tasks


class FindObjectsWorker(DownloadWorker):
    """ Find objects """

    def do_task(self, obj, url, directory, retry, timeout, http_headers):
        filepath = ".git/objects/%s/%s" % (obj[:2], obj[2:])

        if os.path.isfile(os.path.join(directory, filepath)):
            printf("[-] Already downloaded %s/%s\n", url, filepath)
        else:
            response = self.session.get(
                "%s/%s" % (url, filepath),
                allow_redirects=False,
                timeout=timeout,
            )
            printf(
                "[-] Fetching %s/%s [%d]\n",
                url,
                filepath,
                response.status_code,
            )

            valid, error_message = verify_response(response)
            if not valid:
                printf(error_message, url, filepath, file=sys.stderr)
                return []

            abspath = os.path.abspath(os.path.join(directory, filepath))
            create_intermediate_dirs(abspath)

            # write file
            with open(abspath, "wb") as f:
                f.write(response.content)

        abspath = os.path.abspath(os.path.join(directory, filepath))
        # parse object file to find other objects
        obj_file = dulwich.objects.ShaFile.from_path(abspath)
        return get_referenced_sha1(obj_file)


def fetch_git(url, directory, jobs, retry, timeout, http_headers):
    """ Dump a git repository into the output directory """

    assert os.path.isdir(directory), "%s is not a directory" % directory
    assert jobs >= 1, "invalid number of jobs"
    assert retry >= 1, "invalid number of retries"
    assert timeout >= 1, "invalid timeout"

    session = requests.Session()
    session.verify = False
    session.headers = http_headers
    session.mount(url, requests.adapters.HTTPAdapter(max_retries=retry))

    if os.listdir(directory):
        printf("Warning: Destination '%s' is not empty\n", directory)

    # find base url
    url = url.rstrip("/")
    if url.endswith("HEAD"):
        url = url[:-4]
    url = url.rstrip("/")
    if url.endswith(".git"):
        url = url[:-4]
    url = url.rstrip("/")

    # check for /.git/HEAD
    printf("[-] Testing %s/.git/HEAD ", url)
    response = session.get("%s/.git/HEAD" % url, allow_redirects=False)
    printf("[%d]\n", response.status_code)

    valid, error_message = verify_response(response)
    if not valid:
        printf(error_message, url, "/.git/HEAD", file=sys.stderr)
        return 1
    elif not re.match(r"^(ref:.*|[0-9a-f]{40}$)", response.text.strip()):
        printf(
            "error: %s/.git/HEAD is not a git HEAD file\n",
            url,
            file=sys.stderr,
        )
        return 1

    # check for directory listing
    printf("[-] Testing %s/.git/ ", url)
    response = session.get("%s/.git/" % url, allow_redirects=False)
    printf("[%d]\n", response.status_code)

    if (
        response.status_code == 200
        and is_html(response)
        and "HEAD" in get_indexed_files(response)
    ):
        printf("[-] Fetching .git recursively\n")
        process_tasks(
            [".git/", ".gitignore"],
            RecursiveDownloadWorker,
            jobs,
            args=(url, directory, retry, timeout, http_headers),
        )

        printf("[-] Running git checkout .\n")
        os.chdir(directory)
        subprocess.check_call(["git", "checkout", "."])
        return 0

    # no directory listing
    printf("[-] Fetching common files\n")
    tasks = [
        ".gitignore",
        ".git/COMMIT_EDITMSG",
        ".git/description",
        ".git/hooks/applypatch-msg.sample",
        ".git/hooks/commit-msg.sample",
        ".git/hooks/post-commit.sample",
        ".git/hooks/post-receive.sample",
        ".git/hooks/post-update.sample",
        ".git/hooks/pre-applypatch.sample",
        ".git/hooks/pre-commit.sample",
        ".git/hooks/pre-push.sample",
        ".git/hooks/pre-rebase.sample",
        ".git/hooks/pre-receive.sample",
        ".git/hooks/prepare-commit-msg.sample",
        ".git/hooks/update.sample",
        ".git/index",
        ".git/info/exclude",
        ".git/objects/info/packs",
    ]
    process_tasks(
        tasks,
        DownloadWorker,
        jobs,
        args=(url, directory, retry, timeout, http_headers),
    )

    # find refs
    printf("[-] Finding refs/\n")
    tasks = [
        ".git/FETCH_HEAD",
        ".git/HEAD",
        ".git/ORIG_HEAD",
        ".git/config",
        ".git/info/refs",
        ".git/logs/HEAD",
        ".git/logs/refs/heads/master",
        ".git/logs/refs/remotes/origin/HEAD",
        ".git/logs/refs/remotes/origin/master",
        ".git/logs/refs/stash",
        ".git/packed-refs",
        ".git/refs/heads/master",
        ".git/refs/remotes/origin/HEAD",
        ".git/refs/remotes/origin/master",
        ".git/refs/stash",
        ".git/refs/wip/wtree/refs/heads/master",  # Magit
        ".git/refs/wip/index/refs/heads/master",  # Magit
    ]

    process_tasks(
        tasks,
        FindRefsWorker,
        jobs,
        args=(url, directory, retry, timeout, http_headers),
    )

    # find packs
    printf("[-] Finding packs\n")
    tasks = []

    # use .git/objects/info/packs to find packs
    info_packs_path = os.path.join(
        directory, ".git", "objects", "info", "packs"
    )
    if os.path.exists(info_packs_path):
        with open(info_packs_path, "r") as f:
            info_packs = f.read()

        for sha1 in re.findall(r"pack-([a-f0-9]{40})\.pack", info_packs):
            tasks.append(".git/objects/pack/pack-%s.idx" % sha1)
            tasks.append(".git/objects/pack/pack-%s.pack" % sha1)

    process_tasks(
        tasks,
        DownloadWorker,
        jobs,
        args=(url, directory, retry, timeout, http_headers),
    )

    # find objects
    printf("[-] Finding objects\n")
    objs = set()
    packed_objs = set()

    # .git/packed-refs, .git/info/refs, .git/refs/*, .git/logs/*
    files = [
        os.path.join(directory, ".git", "packed-refs"),
        os.path.join(directory, ".git", "info", "refs"),
        os.path.join(directory, ".git", "FETCH_HEAD"),
        os.path.join(directory, ".git", "ORIG_HEAD"),
    ]
    for dirpath, _, filenames in os.walk(
        os.path.join(directory, ".git", "refs")
    ):
        for filename in filenames:
            files.append(os.path.join(dirpath, filename))
    for dirpath, _, filenames in os.walk(
        os.path.join(directory, ".git", "logs")
    ):
        for filename in filenames:
            files.append(os.path.join(dirpath, filename))

    for filepath in files:
        if not os.path.exists(filepath):
            continue

        with open(filepath, "r") as f:
            content = f.read()

        for obj in re.findall(r"(^|\s)([a-f0-9]{40})($|\s)", content):
            obj = obj[1]
            objs.add(obj)

    # use .git/index to find objects
    index_path = os.path.join(directory, ".git", "index")
    if os.path.exists(index_path):
        index = dulwich.index.Index(index_path)

        for entry in index.iterobjects():
            objs.add(entry[1].decode())

    # use packs to find more objects to fetch, and objects that are packed
    pack_file_dir = os.path.join(directory, ".git", "objects", "pack")
    if os.path.isdir(pack_file_dir):
        for filename in os.listdir(pack_file_dir):
            if filename.startswith("pack-") and filename.endswith(".pack"):
                pack_data_path = os.path.join(pack_file_dir, filename)
                pack_idx_path = os.path.join(
                    pack_file_dir, filename[:-5] + ".idx"
                )
                pack_data = dulwich.pack.PackData(pack_data_path)
                pack_idx = dulwich.pack.load_pack_index(pack_idx_path)
                pack = dulwich.pack.Pack.from_objects(pack_data, pack_idx)

                for obj_file in pack.iterobjects():
                    packed_objs.add(obj_file.sha().hexdigest())
                    objs |= set(get_referenced_sha1(obj_file))

    # fetch all objects
    printf("[-] Fetching objects\n")
    process_tasks(
        objs,
        FindObjectsWorker,
        jobs,
        args=(url, directory, retry, timeout, http_headers),
        tasks_done=packed_objs,
    )

    # git checkout
    printf("[-] Running git checkout .\n")
    os.chdir(directory)

    # ignore errors
    subprocess.call(["git", "checkout", "."], stderr=open(os.devnull, "wb"))

    return 0


def main():
    parser = argparse.ArgumentParser(
        usage="git-dumper [options] URL DIR",
        description="Dump a git repository from a website.",
    )
    parser.add_argument("url", metavar="URL", help="url")
    parser.add_argument("directory", metavar="DIR", help="output directory")
    parser.add_argument("--proxy", help="use the specified proxy")
    parser.add_argument(
        "-j",
        "--jobs",
        type=int,
        default=10,
        help="number of simultaneous requests",
    )
    parser.add_argument(
        "-r",
        "--retry",
        type=int,
        default=3,
        help="number of request attempts before giving up",
    )
    parser.add_argument(
        "-t",
        "--timeout",
        type=int,
        default=3,
        help="maximum time in seconds before giving up",
    )
    parser.add_argument(
        "-u",
        "--user-agent",
        type=str,
        default="Mozilla/5.0 (Windows NT 10.0; rv:78.0) Gecko/20100101 Firefox/78.0",
        help="user-agent to use for requests",
    )
    parser.add_argument(
        "-H",
        "--header",
        type=str,
        action="append",
        help="additional http headers, e.g `NAME=VALUE`",
    )
    args = parser.parse_args()

    # jobs
    if args.jobs < 1:
        parser.error("invalid number of jobs, got `%d`" % args.jobs)

    # retry
    if args.retry < 1:
        parser.error("invalid number of retries, got `%d`" % args.retry)

    # timeout
    if args.timeout < 1:
        parser.error("invalid timeout, got `%d`" % args.timeout)

    # header
    http_headers = {"User-Agent": args.user_agent}
    if args.header:
        for header in args.header:
            tokens = header.split("=", maxsplit=1)
            if len(tokens) != 2:
                parser.error(
                    "http header must have the form NAME=VALUE, got `%s`"
                    % header
                )
            name, value = tokens
            http_headers[name.strip()] = value.strip()

    # proxy
    if args.proxy:
        proxy_valid = False

        for pattern, proxy_type in [
            (r"^socks5:(.*):(\d+)$", socks.PROXY_TYPE_SOCKS5),
            (r"^socks4:(.*):(\d+)$", socks.PROXY_TYPE_SOCKS4),
            (r"^http://(.*):(\d+)$", socks.PROXY_TYPE_HTTP),
            (r"^(.*):(\d+)$", socks.PROXY_TYPE_SOCKS5),
        ]:
            m = re.match(pattern, args.proxy)
            if m:
                socks.setdefaultproxy(proxy_type, m.group(1), int(m.group(2)))
                socket.socket = socks.socksocket
                proxy_valid = True
                break

        if not proxy_valid:
            parser.error("invalid proxy, got `%s`" % args.proxy)

    # output directory
    if not os.path.exists(args.directory):
        os.makedirs(args.directory)

    if not os.path.isdir(args.directory):
        parser.error("`%s` is not a directory" % args.directory)

    urllib3.disable_warnings(urllib3.exceptions.InsecureRequestWarning)

    # fetch everything
    sys.exit(
        fetch_git(
            args.url,
            args.directory,
            args.jobs,
            args.retry,
            args.timeout,
            http_headers,
        )
    )


if __name__ == "__main__":
    main()