You can not select more than 25 topics
Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
252 lines
8.2 KiB
252 lines
8.2 KiB
"""Base class for notebook tests."""
|
|
|
|
from __future__ import print_function
|
|
|
|
from binascii import hexlify
|
|
from contextlib import contextmanager
|
|
import errno
|
|
import os
|
|
import sys
|
|
from threading import Thread, Event
|
|
import time
|
|
from unittest import TestCase
|
|
|
|
pjoin = os.path.join
|
|
|
|
from unittest.mock import patch
|
|
|
|
import requests
|
|
from tornado.ioloop import IOLoop
|
|
import zmq
|
|
|
|
import jupyter_core.paths
|
|
from traitlets.config import Config
|
|
from ..notebookapp import NotebookApp, urlencode_unix_socket
|
|
from ..utils import url_path_join
|
|
from ipython_genutils.tempdir import TemporaryDirectory
|
|
|
|
MAX_WAITTIME = 30 # seconds to wait for notebook server to start
|
|
POLL_INTERVAL = 0.1 # time between attempts
|
|
|
|
# TimeoutError is a builtin on Python 3. This can be removed when we stop
|
|
# supporting Python 2.
|
|
class TimeoutError(Exception):
|
|
pass
|
|
|
|
class NotebookTestBase(TestCase):
|
|
"""A base class for tests that need a running notebook.
|
|
|
|
This create some empty config and runtime directories
|
|
and then starts the notebook server with them.
|
|
"""
|
|
|
|
port = 12341
|
|
config = None
|
|
# run with a base URL that would be escaped,
|
|
# to test that we don't double-escape URLs
|
|
url_prefix = '/a%40b/'
|
|
|
|
@classmethod
|
|
def wait_until_alive(cls):
|
|
"""Wait for the server to be alive"""
|
|
url = cls.base_url() + 'api/contents'
|
|
for _ in range(int(MAX_WAITTIME/POLL_INTERVAL)):
|
|
try:
|
|
cls.fetch_url(url)
|
|
except Exception as e:
|
|
if not cls.notebook_thread.is_alive():
|
|
raise RuntimeError("The notebook server failed to start") from e
|
|
time.sleep(POLL_INTERVAL)
|
|
else:
|
|
return
|
|
|
|
raise TimeoutError("The notebook server didn't start up correctly.")
|
|
|
|
@classmethod
|
|
def wait_until_dead(cls):
|
|
"""Wait for the server process to terminate after shutdown"""
|
|
cls.notebook_thread.join(timeout=MAX_WAITTIME)
|
|
if cls.notebook_thread.is_alive():
|
|
raise TimeoutError("Undead notebook server")
|
|
|
|
@classmethod
|
|
def auth_headers(cls):
|
|
headers = {}
|
|
if cls.token:
|
|
headers['Authorization'] = 'token %s' % cls.token
|
|
return headers
|
|
|
|
@staticmethod
|
|
def fetch_url(url):
|
|
return requests.get(url)
|
|
|
|
@classmethod
|
|
def request(cls, verb, path, **kwargs):
|
|
"""Send a request to my server
|
|
|
|
with authentication and everything.
|
|
"""
|
|
headers = kwargs.setdefault('headers', {})
|
|
headers.update(cls.auth_headers())
|
|
response = requests.request(verb,
|
|
url_path_join(cls.base_url(), path),
|
|
**kwargs)
|
|
return response
|
|
|
|
@classmethod
|
|
def get_patch_env(cls):
|
|
return {
|
|
'HOME': cls.home_dir,
|
|
'PYTHONPATH': os.pathsep.join(sys.path),
|
|
'IPYTHONDIR': pjoin(cls.home_dir, '.ipython'),
|
|
'JUPYTER_NO_CONFIG': '1', # needed in the future
|
|
'JUPYTER_CONFIG_DIR' : cls.config_dir,
|
|
'JUPYTER_DATA_DIR' : cls.data_dir,
|
|
'JUPYTER_RUNTIME_DIR': cls.runtime_dir,
|
|
}
|
|
|
|
@classmethod
|
|
def get_argv(cls):
|
|
return []
|
|
|
|
@classmethod
|
|
def get_bind_args(cls):
|
|
return dict(port=cls.port)
|
|
|
|
@classmethod
|
|
def setup_class(cls):
|
|
cls.tmp_dir = TemporaryDirectory()
|
|
def tmp(*parts):
|
|
path = os.path.join(cls.tmp_dir.name, *parts)
|
|
try:
|
|
os.makedirs(path)
|
|
except OSError as e:
|
|
if e.errno != errno.EEXIST:
|
|
raise
|
|
return path
|
|
|
|
cls.home_dir = tmp('home')
|
|
data_dir = cls.data_dir = tmp('data')
|
|
config_dir = cls.config_dir = tmp('config')
|
|
runtime_dir = cls.runtime_dir = tmp('runtime')
|
|
cls.notebook_dir = tmp('notebooks')
|
|
cls.env_patch = patch.dict('os.environ', cls.get_patch_env())
|
|
cls.env_patch.start()
|
|
# Patch systemwide & user-wide data & config directories, to isolate
|
|
# the tests from oddities of the local setup. But leave Python env
|
|
# locations alone, so data files for e.g. nbconvert are accessible.
|
|
# If this isolation isn't sufficient, you may need to run the tests in
|
|
# a virtualenv or conda env.
|
|
cls.path_patch = patch.multiple(
|
|
jupyter_core.paths,
|
|
SYSTEM_JUPYTER_PATH=[tmp('share', 'jupyter')],
|
|
SYSTEM_CONFIG_PATH=[tmp('etc', 'jupyter')],
|
|
)
|
|
cls.path_patch.start()
|
|
|
|
config = cls.config or Config()
|
|
config.NotebookNotary.db_file = ':memory:'
|
|
|
|
cls.token = hexlify(os.urandom(4)).decode('ascii')
|
|
|
|
started = Event()
|
|
def start_thread():
|
|
try:
|
|
bind_args = cls.get_bind_args()
|
|
app = cls.notebook = NotebookApp(
|
|
port_retries=0,
|
|
open_browser=False,
|
|
config_dir=cls.config_dir,
|
|
data_dir=cls.data_dir,
|
|
runtime_dir=cls.runtime_dir,
|
|
notebook_dir=cls.notebook_dir,
|
|
base_url=cls.url_prefix,
|
|
config=config,
|
|
allow_root=True,
|
|
token=cls.token,
|
|
**bind_args
|
|
)
|
|
if 'asyncio' in sys.modules:
|
|
app._init_asyncio_patch()
|
|
import asyncio
|
|
asyncio.set_event_loop(asyncio.new_event_loop())
|
|
# don't register signal handler during tests
|
|
app.init_signal = lambda : None
|
|
# clear log handlers and propagate to root for nose to capture it
|
|
# needs to be redone after initialize, which reconfigures logging
|
|
app.log.propagate = True
|
|
app.log.handlers = []
|
|
app.initialize(argv=cls.get_argv())
|
|
app.log.propagate = True
|
|
app.log.handlers = []
|
|
loop = IOLoop.current()
|
|
loop.add_callback(started.set)
|
|
app.start()
|
|
finally:
|
|
# set the event, so failure to start doesn't cause a hang
|
|
started.set()
|
|
app.session_manager.close()
|
|
cls.notebook_thread = Thread(target=start_thread)
|
|
cls.notebook_thread.daemon = True
|
|
cls.notebook_thread.start()
|
|
started.wait()
|
|
cls.wait_until_alive()
|
|
|
|
@classmethod
|
|
def teardown_class(cls):
|
|
cls.notebook.stop()
|
|
cls.wait_until_dead()
|
|
cls.env_patch.stop()
|
|
cls.path_patch.stop()
|
|
cls.tmp_dir.cleanup()
|
|
# cleanup global zmq Context, to ensure we aren't leaving dangling sockets
|
|
def cleanup_zmq():
|
|
zmq.Context.instance().term()
|
|
t = Thread(target=cleanup_zmq)
|
|
t.daemon = True
|
|
t.start()
|
|
t.join(5) # give it a few seconds to clean up (this should be immediate)
|
|
# if term never returned, there's zmq stuff still open somewhere, so shout about it.
|
|
if t.is_alive():
|
|
raise RuntimeError("Failed to teardown zmq Context, open sockets likely left lying around.")
|
|
|
|
@classmethod
|
|
def base_url(cls):
|
|
return 'http://localhost:%i%s' % (cls.port, cls.url_prefix)
|
|
|
|
|
|
class UNIXSocketNotebookTestBase(NotebookTestBase):
|
|
# Rely on `/tmp` to avoid any Linux socket length max buffer
|
|
# issues. Key on PID for process-wise concurrency.
|
|
sock = '/tmp/.notebook.%i.sock' % os.getpid()
|
|
|
|
@classmethod
|
|
def get_bind_args(cls):
|
|
return dict(sock=cls.sock)
|
|
|
|
@classmethod
|
|
def base_url(cls):
|
|
return '%s%s' % (urlencode_unix_socket(cls.sock), cls.url_prefix)
|
|
|
|
@staticmethod
|
|
def fetch_url(url):
|
|
# Lazily import so it is not required at the module level
|
|
if os.name != 'nt':
|
|
import requests_unixsocket
|
|
with requests_unixsocket.monkeypatch():
|
|
return requests.get(url)
|
|
|
|
|
|
@contextmanager
|
|
def assert_http_error(status, msg=None):
|
|
try:
|
|
yield
|
|
except requests.HTTPError as e:
|
|
real_status = e.response.status_code
|
|
assert real_status == status, \
|
|
"Expected status %d, got %d" % (status, real_status)
|
|
if msg:
|
|
assert msg in str(e), e
|
|
else:
|
|
assert False, "Expected HTTP error status"
|