zulip/tools/run-dev

#!/usr/bin/env python3
import argparse
import asyncio
import errno
import logging
import os
import pwd
import signal
import subprocess
import sys
from typing import List

TOOLS_DIR = os.path.dirname(os.path.abspath(__file__))
sys.path.insert(0, os.path.dirname(TOOLS_DIR))

# check for the venv
from tools.lib import sanity_check

sanity_check.check_venv(__file__)

import aiohttp
from aiohttp import hdrs, web
from aiohttp.http_exceptions import BadStatusLine
from returns.curry import partial

from tools.lib.test_script import add_provision_check_override_param, assert_provisioning_status_ok

if "posix" in os.name and os.geteuid() == 0:
    raise RuntimeError("run-dev should not be run as root.")

DESCRIPTION = """
Starts the app listening on localhost, for local development.

This script launches the Django and Tornado servers, then runs a reverse proxy
which serves to both of them.  After it's all up and running, browse to

    http://localhost:9991/

Note that, while runserver and runtornado have the usual auto-restarting
behavior, the reverse proxy itself does *not* automatically restart on changes
to this file.
"""

parser = argparse.ArgumentParser(
    description=DESCRIPTION, formatter_class=argparse.RawTextHelpFormatter
)

parser.add_argument("--test", action="store_true", help="Use the testing database and ports")
parser.add_argument("--minify", action="store_true", help="Minifies assets for testing in dev")
parser.add_argument("--interface", help="Set the IP or hostname for the proxy to listen on")
parser.add_argument(
    "--no-clear-memcached",
    action="store_false",
    dest="clear_memcached",
    help="Do not clear memcached on startup",
)
parser.add_argument("--streamlined", action="store_true", help="Avoid process_queue, etc.")
parser.add_argument(
    "--behind-https-proxy",
    action="store_true",
    help="Start app server in HTTPS mode, using reverse proxy",
)
add_provision_check_override_param(parser)
options = parser.parse_args()

assert_provisioning_status_ok(options.skip_provision_check)

if options.interface is None:
    user_id = os.getuid()
    user_name = pwd.getpwuid(user_id).pw_name
    if user_name in ["vagrant", "zulipdev"]:
        # In the Vagrant development environment, we need to listen on
        # all ports, and it's safe to do so, because Vagrant is only
        # exposing certain guest ports (by default just 9991) to the
        # host.  The same argument applies to the remote development
        # servers using username "zulipdev".
        options.interface = None
    else:
        # Otherwise, only listen to requests on localhost for security.
        options.interface = "127.0.0.1"
elif options.interface == "":
    options.interface = None

runserver_args: List[str] = []
base_port = 9991
if options.test:
    base_port = 9981
    settings_module = "zproject.test_settings"
    # Don't auto-reload when running Puppeteer tests
    runserver_args = ["--noreload"]
    runtornado_command = ["./manage.py", "runtornado"]
else:
    settings_module = "zproject.settings"
    runtornado_command = [
        "-m",
        "tornado.autoreload",
        "--until-success",
        "./manage.py",
        "runtornado",
        "--autoreload",
    ]

manage_args = [f"--settings={settings_module}"]
os.environ["DJANGO_SETTINGS_MODULE"] = settings_module

if options.behind_https_proxy:
    os.environ["BEHIND_HTTPS_PROXY"] = "1"

from scripts.lib.zulip_tools import CYAN, ENDC

proxy_port = base_port
django_port = base_port + 1
tornado_port = base_port + 2
webpack_port = base_port + 3

os.chdir(os.path.join(os.path.dirname(__file__), ".."))

if options.clear_memcached:
    subprocess.check_call("./scripts/setup/flush-memcached")

# Set up a new process group, so that we can later kill run{server,tornado}
# and all of the processes they spawn.
os.setpgrp()

# Save pid of parent process to the pid file. It can be used later by
# tools/stop-run-dev to kill the server without having to find the
# terminal in question.

if options.test:
    pid_file_path = os.path.join(os.path.join(os.getcwd(), "var/puppeteer/run_dev.pid"))
else:
    pid_file_path = os.path.join(os.path.join(os.getcwd(), "var/run/run_dev.pid"))

# Required for compatibility python versions.
if not os.path.exists(os.path.dirname(pid_file_path)):
    os.makedirs(os.path.dirname(pid_file_path))
with open(pid_file_path, "w+") as f:
    f.write(str(os.getpgrp()) + "\n")


def server_processes() -> List[List[str]]:
    main_cmds = [
        [
            "./manage.py",
            "rundjangoserver",
            *manage_args,
            *runserver_args,
            f"127.0.0.1:{django_port}",
        ],
        [
            "env",
            "PYTHONUNBUFFERED=1",
            "python3",
            *runtornado_command,
            *manage_args,
            f"127.0.0.1:{tornado_port}",
        ],
    ]

    if options.streamlined:
        # The streamlined operation allows us to do many
        # things, but search/etc. features won't work.
        return main_cmds

    other_cmds = [
        ["./manage.py", "process_queue", "--all", *manage_args],
        [
            "env",
            "PGHOST=127.0.0.1",  # Force password authentication using .pgpass
            "./puppet/zulip/files/postgresql/process_fts_updates",
            "--quiet",
        ],
        ["./manage.py", "deliver_scheduled_messages"],
    ]

    # NORMAL (but slower) operation:
    return main_cmds + other_cmds


def do_one_time_webpack_compile() -> None:
    # We just need to compile webpack assets once at startup, not run a daemon,
    # in test mode.  Additionally, webpack-dev-server doesn't support running 2
    # copies on the same system, so this model lets us run the Puppeteer tests
    # with a running development server.
    subprocess.check_call(["./tools/webpack", "--quiet", "--test"])


def start_webpack_watcher() -> "subprocess.Popen[bytes]":
    webpack_cmd = ["./tools/webpack", "--watch", f"--port={webpack_port}"]
    if options.minify:
        webpack_cmd.append("--minify")
    if options.interface is None:
        # If interface is None and we're listening on all ports, we also need
        # to disable the webpack host check so that webpack will serve assets.
        webpack_cmd.append("--disable-host-check")
    if options.interface:
        webpack_cmd.append(f"--host={options.interface}")
    else:
        webpack_cmd.append("--host=0.0.0.0")
    return subprocess.Popen(webpack_cmd)


session: aiohttp.ClientSession

# https://datatracker.ietf.org/doc/html/rfc2616#section-13.5.1
HOP_BY_HOP_HEADERS = {
    hdrs.CONNECTION,
    hdrs.KEEP_ALIVE,
    hdrs.PROXY_AUTHENTICATE,
    hdrs.PROXY_AUTHORIZATION,
    hdrs.TE,
    hdrs.TRAILER,
    hdrs.TRANSFER_ENCODING,
    hdrs.UPGRADE,
}

# Headers that aiohttp would otherwise generate by default
SKIP_AUTO_HEADERS = {
    hdrs.ACCEPT,
    hdrs.ACCEPT_ENCODING,
    hdrs.CONTENT_TYPE,
    hdrs.USER_AGENT,
}


async def forward(upstream_port: int, request: web.Request) -> web.StreamResponse:
    try:
        upstream_response = await session.request(
            request.method,
            request.url.with_host("127.0.0.1").with_port(upstream_port),
            headers=[
                (key, value)
                for key, value in request.headers.items()
                if key not in HOP_BY_HOP_HEADERS
            ],
            data=request.content.iter_any() if request.body_exists else None,
            allow_redirects=False,
            auto_decompress=False,
            compress=False,
            skip_auto_headers=SKIP_AUTO_HEADERS,
        )
    except aiohttp.ClientError as error:
        logging.error(
            "Failed to forward %s %s to port %d: %s",
            request.method,
            request.url.path,
            upstream_port,
            error,
        )
        raise web.HTTPBadGateway from error

    response = web.StreamResponse(status=upstream_response.status, reason=upstream_response.reason)
    response.headers.extend(
        (key, value)
        for key, value in upstream_response.headers.items()
        if key not in HOP_BY_HOP_HEADERS
    )
    assert request.remote is not None
    response.headers["X-Real-IP"] = request.remote
    response.headers["X-Forwarded-Port"] = str(proxy_port)
    await response.prepare(request)
    async for data in upstream_response.content.iter_any():
        await response.write(data)
    await response.write_eof()
    return response


app = web.Application()
app.add_routes(
    [
        web.route(
            hdrs.METH_ANY, r"/{path:json/events|api/v1/events}", partial(forward, tornado_port)
        ),
        web.route(hdrs.METH_ANY, r"/{path:webpack/.*}", partial(forward, webpack_port)),
        web.route(hdrs.METH_ANY, r"/{path:.*}", partial(forward, django_port)),
    ]
)


def print_listeners() -> None:
    # Since we can't import settings from here, we duplicate some
    # EXTERNAL_HOST logic from dev_settings.py.
    IS_DEV_DROPLET = pwd.getpwuid(os.getuid()).pw_name == "zulipdev"
    if IS_DEV_DROPLET:
        # Technically, the `zulip.` is a subdomain of the server, so
        # this is kinda misleading, but 99% of development is done on
        # the default/zulip subdomain.
        default_hostname = "zulip." + os.uname()[1].lower()
    else:
        default_hostname = "localhost"
    external_host = os.getenv("EXTERNAL_HOST", f"{default_hostname}:{proxy_port}")
    http_protocol = "https" if options.behind_https_proxy else "http"
    print(
        f"\nStarting Zulip on:\n\n\t{CYAN}{http_protocol}://{external_host}/{ENDC}\n\nInternal ports:"
    )
    ports = [
        (proxy_port, "Development server proxy (connect here)"),
        (django_port, "Django"),
        (tornado_port, "Tornado"),
    ]

    if not options.test:
        ports.append((webpack_port, "webpack"))

    for port, label in ports:
        print(f"   {port}: {label}")
    print()


def https_log_filter(record: logging.LogRecord) -> bool:
    # aiohttp emits an exception with a traceback when receiving an
    # https request (https://github.com/aio-libs/aiohttp/issues/8065).
    # Abbreviate it to a one-line message.
    if (
        record.exc_info is not None
        and isinstance(error := record.exc_info[1], BadStatusLine)
        and error.message.startswith(
            (
                "Invalid method encountered:\n\n  b'\\x16",
                'Invalid method encountered:\n\n  b"\\x16',
            )
        )
    ):
        record.msg = "Rejected https request (this development server only supports http)"
        record.exc_info = None
    return True


logging.getLogger("aiohttp.server").addFilter(https_log_filter)

runner: web.AppRunner
children: List["subprocess.Popen[bytes]"] = []


async def serve() -> None:
    global runner, session

    if options.test:
        do_one_time_webpack_compile()
    else:
        children.append(start_webpack_watcher())

    children.extend(subprocess.Popen(cmd) for cmd in server_processes())

    session = aiohttp.ClientSession()
    runner = web.AppRunner(app, auto_decompress=False, handler_cancellation=True)
    await runner.setup()
    site = web.TCPSite(runner, host=options.interface, port=proxy_port)
    try:
        await site.start()
    except OSError as e:
        if e.errno == errno.EADDRINUSE:
            print("\n\nERROR: You probably have another server running!!!\n\n")
        raise

    print_listeners()


loop = asyncio.new_event_loop()

try:
    loop.run_until_complete(serve())
    for s in (signal.SIGINT, signal.SIGTERM):
        loop.add_signal_handler(s, loop.stop)
    loop.run_forever()
finally:
    loop.run_until_complete(runner.cleanup())
    loop.run_until_complete(session.close())

    for child in children:
        child.terminate()

    print("Waiting for children to stop...")
    for child in children:
        child.wait()

    # Remove pid file when development server closed correctly.
    os.remove(pid_file_path)