path: root/modules/gitmirror/files/on-the-pull

#!/usr/bin/python3

import cgi
import http.server
import os
import pwd
import re
import subprocess
import sys
from optparse import OptionParser
from queue import Queue
from threading import Thread


GitUpdaterQueue = Queue(0)


# NB The following class and bits for running git commands were "liberated"
# from git_multimail.py

class CommandError(Exception):
    def __init__(self, cmd, retcode):
        self.cmd = cmd
        self.retcode = retcode
        Exception.__init__(
            self,
            'Command "%s" failed with retcode %s' % (' '.join(cmd), retcode,)
            )


# It is assumed in many places that the encoding is uniformly UTF-8,
# so changing these constants is unsupported.  But define them here
# anyway, to make it easier to find (at least most of) the places
# where the encoding is important.
ENCODING = 'UTF-8'


# The "git" program (this could be changed to include a full path):
GIT_EXECUTABLE = 'git'


# How "git" should be invoked (including global arguments), as a list
# of words.  This variable is usually initialized automatically by
# read_git_output() via choose_git_command(), but if a value is set
# here then it will be used unconditionally.
GIT_CMD = None


def choose_git_command():
    """Decide how to invoke git, and record the choice in GIT_CMD."""

    global GIT_CMD

    if GIT_CMD is None:
        try:
            # Check to see whether the "-c" option is accepted (it was
            # only added in Git 1.7.2).  We don't actually use the
            # output of "git --version", though if we needed more
            # specific version information this would be the place to
            # do it.
            cmd = [GIT_EXECUTABLE, '-c', 'foo.bar=baz', '--version']
            read_output(cmd)
            GIT_CMD = [GIT_EXECUTABLE, '-c', f'i18n.logoutputencoding={ENCODING}']
        except CommandError:
            GIT_CMD = [GIT_EXECUTABLE]


def read_git_output(args, inp=None, keepends=False, **kw):
    """Read the output of a Git command."""

    if GIT_CMD is None:
        choose_git_command()

    return read_output(GIT_CMD + args, inp=inp, keepends=keepends, **kw)


# NOTE: output is in bytes, not a string
def read_output(cmd, inp=None, keepends=False, **kw):
    if inp:
        stdin = subprocess.PIPE
    else:
        stdin = None
    p = subprocess.Popen(
        cmd, stdin=stdin, stdout=subprocess.PIPE, stderr=subprocess.PIPE, **kw
        )
    (out, err) = p.communicate(inp)
    retcode = p.wait()
    if retcode:
        raise CommandError(cmd, retcode)
    if not keepends:
        out = out.rstrip(b'\n\r')
    return out


def run_git_command(args, **kw):
    """Runs a git command, ignoring the output.
    """

    read_git_output(args, **kw)


def run_command(args, **kw):
    """Runs a git command, ignoring the output.
    """

    read_output(args, **kw)


class GitUpdater(Thread):
    def __init__(self, server, basedir, repoprefix, branch='master', cmd=''):
        Thread.__init__(self)
        self.server = server
        self.basedir = basedir
        self.repoprefix = repoprefix
        self.branch = branch
        self.cmd = cmd

    def run(self):
        while 42:
            repo = GitUpdaterQueue.get()
            if repo is None:
                break
            try:
                print(f"Got update request for '{repo}'", file=sys.stderr)
                clonefolder = os.path.join(self.basedir, repo)
                if self.repoprefix:
                    if not repo.startswith(self.repoprefix):
                        print(f"Ignoring repo '{repo}' due to invalid prefix", file=sys.stderr)
                        GitUpdaterQueue.task_done()
                        continue
                    clonefolder = os.path.join(self.basedir, repo[len(self.repoprefix):])
                command = []
                treeish = ''
                changed = True
                if not os.path.exists(clonefolder):
                    cloneparent = os.path.dirname(clonefolder)
                    if not os.path.exists(cloneparent):
                        os.makedirs(cloneparent)
                    cloneurl = self.server + '/' + repo
                    command = ['clone']
                    if '--mirror' == self.branch:
                        command.append('--mirror')
                    command.append(cloneurl)
                    command.append(clonefolder)
                    print(f"Cloning repo '{repo}' ('{cloneurl}' -> '{clonefolder}')", file=sys.stderr)

                    run_git_command(command)
                    if not os.path.isdir(clonefolder):
                        raise Exception(f"Clone folder '{clonefolder}' is not a directory. Cloning failed or file in it's place?")
                    os.chdir(clonefolder)
                    if '--mirror' != self.branch and 'master' != self.branch:
                        command = ['checkout', '-t', 'origin/' + self.branch]
                        run_git_command(command)
                elif os.path.isdir(clonefolder):
                    os.chdir(clonefolder)
                    print(f"Updating existing repo '{repo}' ({clonefolder})", file=sys.stderr)
                    command = ['remote', 'update']
                    run_git_command(command)
                    if '--mirror' != self.branch:
                        sha1before = read_git_output(['rev-parse', 'refs/heads/' + self.branch])
                        sha1after = read_git_output(['rev-parse', 'refs/remotes/origin/' + self.branch])
                        if sha1before and sha1after:
                            if sha1before == sha1after:
                                changed = False
                                print(f"Repo '{repo}' update on branch '{self.branch}': No changed detected", file=sys.stderr)
                            else:
                                treeish = sha1before.decode(ENCODING) + '..' + sha1after.decode(ENCODING)
                                print(f"Repo '{repo}' update on branch '{self.branch}': Treeish '{treeish}'", file=sys.stderr)
                        else:
                            print(f"Repo '{repo}' update on branch '{self.branch}': Before or after sha1 could not be extracted.", file=sys.stderr)
                        command = ['update-ref', 'refs/heads/' + self.branch, 'refs/remotes/origin/' + self.branch]
                        run_git_command(command)
                        command = ['checkout', '-f', self.branch]
                        run_git_command(command)
                else:
                    raise Exception(f"Clone folder '{clonefolder}' appears to be a file :s")

                if changed and self.cmd:
                    # Update the info/web/last-modified file as used by cgit
                    os.chdir(clonefolder)
                    command = [self.cmd, repo]
                    if treeish:
                        command += [treeish]
                    run_command(command)

                print(f"Update for '{repo}' complete.", file=sys.stderr)
            except Exception as e:
                print(f"Error processing repo '{repo}'", file=sys.stderr)
                print(str(e), file=sys.stderr)

            GitUpdaterQueue.task_done()
            sys.stderr.flush()


class TimeoutServer(http.server.HTTPServer):
    def get_request(self):
        result = self.socket.accept()
        result[0].settimeout(10)
        return result


class PostHandler(http.server.BaseHTTPRequestHandler):
    def do_POST(self):
        ctype, pdict = cgi.parse_header(self.headers['content-type'])
        repo = ""
        try:
            if ctype != 'x-git/repo':
                self.send_response(415)
                self.end_headers()
                return

            # chunked mode is a legitimate reason there would be no content-length,
            # but it's easier to just insist on it
            length = int(self.headers['content-length']) if self.headers['content-length'] else 0
            if length < 1:
                self.send_response(411)
                self.end_headers()
                return
            if length > 1024:
                self.send_response(413)
                self.end_headers()
                return
            repo = self.rfile.read(length).decode(ENCODING)

            if re.match(r"^[-_/a-zA-Z0-9\+\.]+$", repo) is None:
                self.send_response(400)
                self.end_headers()
                return

            GitUpdaterQueue.put(repo)
            self.send_response(202)
            self.end_headers()

        except Exception as e:
            print("Error processing request", file=sys.stderr)
            print(str(e), file=sys.stderr)
            self.send_response(500)
            self.end_headers()

        sys.stderr.flush()


def Demote(pidfile, uid, gid):
    def result():
        piddir = os.path.dirname(pidfile)
        if not os.path.exists(piddir):
            os.makedirs(piddir)
        fd = open(pidfile, 'w')
        fd.write(str(os.getpid()))
        fd.close()

        if uid and gid:
            os.setgid(gid)
            os.setuid(uid)
    return result


def daemonise(options, serverprefix, basefolder):
    pw = None
    uid = False
    gid = False
    if options.user:
        pw = pwd.getpwnam(options.user)
        uid = pw.pw_uid
        gid = pw.pw_gid
    else:
        pw = pwd.getpwnam(os.getlogin())

    user = pw.pw_name
    dirname = pw.pw_dir
    env = {
      'HOME': dirname,
      'LOGNAME': user,
      'PWD': dirname,
      'USER': user,
    }
    if os.getenv('PATH') is not None:
        env['PATH'] = os.getenv('PATH')
    if os.getenv('PYTHONPATH') is not None:
        env['PYTHONPATH'] = os.getenv('PYTHONPATH')

    args = [os.path.abspath(sys.argv[0])]
    args.append('-a')
    args.append(options.addr)
    args.append('-p')
    args.append(str(options.port))
    args.append('-r')
    args.append(options.repoprefix)
    args.append('-b')
    args.append(options.branch)
    args.append('-c')
    args.append(options.cmd)
    args.append(serverprefix)
    args.append(basefolder)

    subprocess.Popen(
        args, preexec_fn=Demote(options.pidfile, uid, gid), cwd=dirname, env=env
    )
    exit(0)


def main():
    usage = "usage: %prog [options] <serverprefix> <basefolder>"
    description = """Listen for repository names being posted via a simple HTTP interface and clone/update them.
POST data simply via curl:
e.g. curl --header 'Content-Type: x-git/repo' --data 'my/repo/name' http://localhost:8000
"""
    parser = OptionParser(usage=usage, description=description)
    parser.add_option("-a", "--addr",
                      type="string", dest="addr", default="0.0.0.0",
                      help="The interface address to bind to")
    parser.add_option("-p", "--port",
                      type="int", dest="port", default=8000,
                      help="The port to bind to")
    parser.add_option("-r", "--repo-prefix",
                      type="string", dest="repoprefix", default="",
                      help="Only handle repositories with the following prefix. This SHOULD contain a trailing slash if it's a folder but SHOULD NOT include a leading slash")
    parser.add_option("-b", "--branch",
                      type="string", dest="branch", default="--mirror",
                      help="The branch to track on clone. If you pass '--mirror' (the default) as the branch name we will clone as a bare mirror")
    parser.add_option("-c", "--cmd",
                      type="string", dest="cmd", default="",
                      help="Third party command to execute after updates. It will execute in the "
                      "folder of the repo and if we're not in mirror mode, a treeish will be "
                      "passed as the only argument containing the refs that changed otherwise "
                      "the command will be run without any arguments")
    parser.add_option("-d", "--pid-file",
                      type="string", dest="pidfile", default="",
                      help="Daemonise and write pidfile")
    parser.add_option("-u", "--user",
                      type="string", dest="user", default="",
                      help="Drop privileges to the given user (must be run as root)")

    (options, args) = parser.parse_args()
    if len(args) < 2:
        parser.error("Both the <serverprefix> and <basefolder> arguments must be supplied.")
    if len(args) > 2:
        parser.print_usage()
        exit(1)

    serverprefix = args[0]
    basefolder = args[1]

    if options.pidfile:
        daemonise(options, serverprefix, basefolder)

    if options.user:
        parser.error("You can only specify a user if you're also deamonising (with a pid file).")

    print("Server started", file=sys.stderr)
    sys.stderr.flush()
    srvr = TimeoutServer((options.addr, options.port), PostHandler)
    updater = GitUpdater(serverprefix, basefolder, options.repoprefix, options.branch, options.cmd)
    updater.start()

    try:
        srvr.serve_forever()
    except KeyboardInterrupt:
        srvr.socket.close()
        GitUpdaterQueue.put(None)
        updater.join()


if __name__ == "__main__":
    main()