class UploadedFileManagerTest(unittest.TestCase):
def setUp(self):
self.mgr = UploadedFileManager()
self.filemgr_events = []
self.mgr.on_files_added.connect(self._on_files_added)
def _on_files_added(self, file_list, **kwargs):
self.filemgr_events.append(file_list)
def test_add_file(self):
self.assertIsNone(self.mgr.get_files("non-report", "non-widget"))
event1 = UploadedFileList("session", "widget", [file1])
event2 = UploadedFileList("session", "widget", [file2])
self.mgr.add_files("session", "widget", [file1])
self.assertEqual([file1], self.mgr.get_files("session", "widget"))
self.assertEqual([event1], self.filemgr_events)
# Add another file with the same ID
self.mgr.add_files("session", "widget", [file2])
self.assertEqual([file2], self.mgr.get_files("session", "widget"))
self.assertEqual([event1, event2], self.filemgr_events)
def test_remove_file(self):
# This should not error.
self.mgr.remove_files("non-report", "non-widget")
self.mgr.add_files("session", "widget", [file1])
self.assertEqual([file1], self.mgr.get_files("session", "widget"))
self.mgr.remove_files("session", "widget")
self.assertIsNone(self.mgr.get_files("session", "widget"))
def test_remove_all_files(self):
# This should not error.
self.mgr.remove_session_files("non-report")
# Add two files with different session IDs, but the same widget ID.
self.mgr.add_files("session1", "widget", [file1])
self.mgr.add_files("session2", "widget", [file1])
event1 = UploadedFileList("session1", "widget", [file1])
event2 = UploadedFileList("session2", "widget", [file1])
self.mgr.remove_session_files("session1")
self.assertIsNone(self.mgr.get_files("session1", "widget"))
self.assertEqual([file1], self.mgr.get_files("session2", "widget"))
self.assertEqual([event1, event2], self.filemgr_events)
class Server:
_singleton: Optional["Server"] = None
@classmethod
def get_current(cls) -> "Server":
"""
Returns
-------
Server
The singleton Server object.
"""
if Server._singleton is None:
raise RuntimeError("Server has not been initialized yet")
return Server._singleton
def __init__(self, ioloop: IOLoop, script_path: str, command_line: Optional[str]):
"""Create the server. It won't be started yet."""
if Server._singleton is not None:
raise RuntimeError("Server already initialized. Use .get_current() instead")
Server._singleton = self
_set_tornado_log_levels()
self._ioloop = ioloop
self._script_path = script_path
self._command_line = command_line
# Mapping of ReportSession.id -> SessionInfo.
self._session_info_by_id: Dict[str, SessionInfo] = {}
self._must_stop = threading.Event()
self._state = State.INITIAL
self._message_cache = ForwardMsgCache()
self._uploaded_file_mgr = UploadedFileManager()
self._uploaded_file_mgr.on_files_updated.connect(self.on_files_updated)
self._report: Optional[Report] = None
self._preheated_session_id: Optional[str] = None
def __repr__(self) -> str:
return util.repr_(self)
@property
def script_path(self) -> str:
return self._script_path
def get_session_by_id(self, session_id: str) -> Optional[ReportSession]:
"""Return the ReportSession corresponding to the given id, or None if
no such session exists."""
session_info = self._get_session_info(session_id)
if session_info is None:
return None
return session_info.session
def on_files_updated(self, session_id: str) -> None:
"""Event handler for UploadedFileManager.on_file_added.
Ensures that uploaded files from stale sessions get deleted.
"""
session_info = self._get_session_info(session_id)
if session_info is None:
# If an uploaded file doesn't belong to an existing session,
# remove it so it doesn't stick around forever.
self._uploaded_file_mgr.remove_session_files(session_id)
def _get_session_info(self, session_id: str) -> Optional[SessionInfo]:
"""Return the SessionInfo with the given id, or None if no such
session exists.
"""
return self._session_info_by_id.get(session_id, None)
def start(self, on_started: Callable[["Server"], Any]) -> None:
"""Start the server.
Parameters
----------
on_started : callable
A callback that will be called when the server's run-loop
has started, and the server is ready to begin receiving clients.
"""
if self._state != State.INITIAL:
raise RuntimeError("Server has already been started")
LOGGER.debug("Starting server...")
app = self._create_app()
start_listening(app)
port = config.get_option("server.port")
LOGGER.debug("Server started on port %s", port)
self._ioloop.spawn_callback(self._loop_coroutine, on_started)
def get_debug(self) -> Dict[str, Dict[str, Any]]:
if self._report:
return {"report": self._report.get_debug()}
return {}
def _create_app(self) -> tornado.web.Application:
"""Create our tornado web app."""
base = config.get_option("server.baseUrlPath")
routes = [
(
make_url_path_regex(base, "stream"),
_BrowserWebSocketHandler,
dict(server=self),
),
(
make_url_path_regex(base, "healthz"),
HealthHandler,
dict(callback=lambda: self.is_ready_for_browser_connection),
),
(make_url_path_regex(base, "debugz"), DebugHandler, dict(server=self)),
(make_url_path_regex(base, "metrics"), MetricsHandler),
(
make_url_path_regex(base, "message"),
MessageCacheHandler,
dict(cache=self._message_cache),
),
(
make_url_path_regex(
base,
UPLOAD_FILE_ROUTE,
),
UploadFileRequestHandler,
dict(
file_mgr=self._uploaded_file_mgr,
get_session_info=self._get_session_info,
),
),
(
make_url_path_regex(base, "assets/(.*)"),
AssetsFileHandler,
{"path": "%s/" % file_util.get_assets_dir()},
),
(make_url_path_regex(base, "media/(.*)"), MediaFileHandler, {"path": ""}),
(
make_url_path_regex(base, "component/(.*)"),
ComponentRequestHandler,
dict(registry=ComponentRegistry.instance()),
),
]
if config.get_option("server.scriptHealthCheckEnabled"):
routes.extend(
[
(
make_url_path_regex(base, "script-health-check"),
HealthHandler,
dict(callback=lambda: self.does_script_run_without_error()),
)
]
)
if config.get_option("global.developmentMode"):
LOGGER.debug("Serving static content from the Node dev server")
else:
static_path = file_util.get_static_dir()
LOGGER.debug("Serving static content from %s", static_path)
routes.extend(
[
(
make_url_path_regex(base, "(.*)"),
StaticFileHandler,
{"path": "%s/" % static_path, "default_filename": "index.html"},
),
(make_url_path_regex(base, trailing_slash=False), AddSlashHandler),
]
)
return tornado.web.Application(
routes, # type: ignore[arg-type]
cookie_secret=config.get_option("server.cookieSecret"),
xsrf_cookies=config.get_option("server.enableXsrfProtection"),
**TORNADO_SETTINGS, # type: ignore[arg-type]
)
def _set_state(self, new_state: State) -> None:
LOGGER.debug("Server state: %s -> %s" % (self._state, new_state))
self._state = new_state
@property
async def is_ready_for_browser_connection(self) -> Tuple[bool, str]:
if self._state not in (State.INITIAL, State.STOPPING, State.STOPPED):
return True, "ok"
return False, "unavailable"
async def does_script_run_without_error(self) -> Tuple[bool, str]:
"""Load and execute the app's script to verify it runs without an error.
Returns
-------
(True, "ok") if the script completes without error, or (False, err_msg)
if the script raises an exception.
"""
session = ReportSession(
ioloop=self._ioloop,
script_path=self._script_path,
command_line=self._command_line,
uploaded_file_manager=self._uploaded_file_mgr,
)
try:
session.request_rerun(None)
now = time.perf_counter()
while (
SCRIPT_RUN_WITHOUT_ERRORS_KEY not in session.session_state
and (time.perf_counter() - now) < SCRIPT_RUN_CHECK_TIMEOUT
):
await tornado.gen.sleep(0.1)
if SCRIPT_RUN_WITHOUT_ERRORS_KEY not in session.session_state:
return False, "timeout"
ok = session.session_state[SCRIPT_RUN_WITHOUT_ERRORS_KEY]
msg = "ok" if ok else "error"
return ok, msg
finally:
session.shutdown()
@property
def browser_is_connected(self) -> bool:
return self._state == State.ONE_OR_MORE_BROWSERS_CONNECTED
@property
def is_running_hello(self) -> bool:
from streamlit.hello import hello
return self._script_path == hello.__file__
@tornado.gen.coroutine
def _loop_coroutine(
self, on_started: Optional[Callable[["Server"], Any]] = None
) -> Generator[Any, None, None]:
try:
if self._state == State.INITIAL:
self._set_state(State.WAITING_FOR_FIRST_BROWSER)
elif self._state == State.ONE_OR_MORE_BROWSERS_CONNECTED:
pass
else:
raise RuntimeError("Bad server state at start: %s" % self._state)
if on_started is not None:
on_started(self)
while not self._must_stop.is_set():
if self._state == State.WAITING_FOR_FIRST_BROWSER:
pass
elif self._state == State.ONE_OR_MORE_BROWSERS_CONNECTED:
# Shallow-clone our sessions into a list, so we can iterate
# over it and not worry about whether it's being changed
# outside this coroutine.
session_infos = list(self._session_info_by_id.values())
for session_info in session_infos:
if session_info.ws is None:
# Preheated.
continue
msg_list = session_info.session.flush_browser_queue()
for msg in msg_list:
try:
self._send_message(session_info, msg)
except tornado.websocket.WebSocketClosedError:
self._close_report_session(session_info.session.id)
yield
yield
elif self._state == State.NO_BROWSERS_CONNECTED:
pass
else:
# Break out of the thread loop if we encounter any other state.
break
yield tornado.gen.sleep(0.01)
# Shut down all ReportSessions
for session_info in list(self._session_info_by_id.values()):
session_info.session.shutdown()
self._set_state(State.STOPPED)
except Exception:
# Can't just re-raise here because co-routines use Tornado
# exceptions for control flow, which appears to swallow the reraised
# exception.
traceback.print_exc()
LOGGER.info(
"""
Please report this bug at https://github.com/streamlit/streamlit/issues.
"""
)
finally:
self._on_stopped()
def _send_message(self, session_info: SessionInfo, msg: ForwardMsg) -> None:
"""Send a message to a client.
If the client is likely to have already cached the message, we may
instead send a "reference" message that contains only the hash of the
message.
Parameters
----------
session_info : SessionInfo
The SessionInfo associated with websocket
msg : ForwardMsg
The message to send to the client
"""
msg.metadata.cacheable = is_cacheable_msg(msg)
msg_to_send = msg
if msg.metadata.cacheable:
populate_hash_if_needed(msg)
if self._message_cache.has_message_reference(
msg, session_info.session, session_info.report_run_count
):
# This session has probably cached this message. Send
# a reference instead.
LOGGER.debug("Sending cached message ref (hash=%s)" % msg.hash)
msg_to_send = create_reference_msg(msg)
# Cache the message so it can be referenced in the future.
# If the message is already cached, this will reset its
# age.
LOGGER.debug("Caching message (hash=%s)" % msg.hash)
self._message_cache.add_message(
msg, session_info.session, session_info.report_run_count
)
# If this was a `report_finished` message, we increment the
# report_run_count for this session, and update the cache
if (
msg.WhichOneof("type") == "report_finished"
and msg.report_finished == ForwardMsg.FINISHED_SUCCESSFULLY
):
LOGGER.debug(
"Report finished successfully; "
"removing expired entries from MessageCache "
"(max_age=%s)",
config.get_option("global.maxCachedMessageAge"),
)
session_info.report_run_count += 1
self._message_cache.remove_expired_session_entries(
session_info.session, session_info.report_run_count
)
# Ship it off!
if session_info.ws is not None:
session_info.ws.write_message(
serialize_forward_msg(msg_to_send), binary=True
)
def stop(self) -> None:
click.secho(" Stopping...", fg="blue")
self._set_state(State.STOPPING)
self._must_stop.set()
def _on_stopped(self) -> None:
"""Called when our runloop is exiting, to shut down the ioloop.
This will end our process.
(Tests can patch this method out, to prevent the test's ioloop
from being shutdown.)
"""
self._ioloop.stop()
def add_preheated_report_session(self) -> None:
"""Register a fake browser with the server and run the script.
This is used to start running the user's script even before the first
browser connects.
"""
session = self._create_or_reuse_report_session(ws=None)
session.handle_rerun_script_request(is_preheat=True)
def _create_or_reuse_report_session(
self, ws: Optional[WebSocketHandler]
) -> ReportSession:
"""Register a connected browser with the server.
Parameters
----------
ws : _BrowserWebSocketHandler or None
The newly-connected websocket handler or None if preheated
connection.
Returns
-------
ReportSession
The newly-created ReportSession for this browser connection.
"""
if self._preheated_session_id is not None:
assert len(self._session_info_by_id) == 1
assert ws is not None
session_id = self._preheated_session_id
self._preheated_session_id = None
session_info = self._session_info_by_id[session_id]
session_info.ws = ws
session = session_info.session
LOGGER.debug(
"Reused preheated session for ws %s. Session ID: %s", id(ws), session_id
)
else:
session = ReportSession(
ioloop=self._ioloop,
script_path=self._script_path,
command_line=self._command_line,
uploaded_file_manager=self._uploaded_file_mgr,
)
LOGGER.debug(
"Created new session for ws %s. Session ID: %s", id(ws), session.id
)
assert session.id not in self._session_info_by_id, (
"session.id '%s' registered multiple times!" % session.id
)
self._session_info_by_id[session.id] = SessionInfo(ws, session)
if ws is None:
self._preheated_session_id = session.id
else:
self._set_state(State.ONE_OR_MORE_BROWSERS_CONNECTED)
return session
def _close_report_session(self, session_id: str) -> None:
"""Shutdown and remove a ReportSession.
This function may be called multiple times for the same session,
which is not an error. (Subsequent calls just no-op.)
Parameters
----------
session_id : str
The ReportSession's id string.
"""
if session_id in self._session_info_by_id:
session_info = self._session_info_by_id[session_id]
del self._session_info_by_id[session_id]
session_info.session.shutdown()
if len(self._session_info_by_id) == 0:
self._set_state(State.NO_BROWSERS_CONNECTED)
class UploadedFileManagerTest(unittest.TestCase):
def setUp(self):
self.mgr = UploadedFileManager()
self.filemgr_events = []
self.mgr.on_files_updated.connect(self._on_files_updated)
def _on_files_updated(self, file_list, **kwargs):
self.filemgr_events.append(file_list)
def test_added_file_id(self):
"""An added file should have a unique ID."""
f1 = self.mgr.add_file("session", "widget", FILE_1)
f2 = self.mgr.add_file("session", "widget", FILE_1)
self.assertNotEqual(FILE_1.id, f1.id)
self.assertNotEqual(f1.id, f2.id)
def test_added_file_properties(self):
"""An added file should maintain all its source properties
except its ID."""
added = self.mgr.add_file("session", "widget", FILE_1)
self.assertNotEqual(added.id, FILE_1.id)
self.assertEqual(added.name, FILE_1.name)
self.assertEqual(added.type, FILE_1.type)
self.assertEqual(added.data, FILE_1.data)
def test_retrieve_added_file(self):
"""After adding a file to the mgr, we should be able to get it back."""
self.assertEqual([], self.mgr.get_all_files("non-report",
"non-widget"))
file_1 = self.mgr.add_file("session", "widget", FILE_1)
self.assertEqual([file_1], self.mgr.get_all_files("session", "widget"))
self.assertEqual([file_1],
self.mgr.get_files("session", "widget", [file_1.id]))
self.assertEqual(len(self.filemgr_events), 1)
# Add another file
file_2 = self.mgr.add_file("session", "widget", FILE_2)
self.assertEqual([file_1, file_2],
self.mgr.get_all_files("session", "widget"))
self.assertEqual([file_1],
self.mgr.get_files("session", "widget", [file_1.id]))
self.assertEqual([file_2],
self.mgr.get_files("session", "widget", [file_2.id]))
self.assertEqual(len(self.filemgr_events), 2)
def test_remove_file(self):
# This should not error.
self.mgr.remove_files("non-report", "non-widget")
f1 = self.mgr.add_file("session", "widget", FILE_1)
self.mgr.remove_file("session", "widget", f1.id)
self.assertEqual([], self.mgr.get_all_files("session", "widget"))
# Remove the file again. It doesn't exist, but this isn't an error.
self.mgr.remove_file("session", "widget", f1.id)
self.assertEqual([], self.mgr.get_all_files("session", "widget"))
f1 = self.mgr.add_file("session", "widget", FILE_1)
f2 = self.mgr.add_file("session", "widget", FILE_2)
self.mgr.remove_file("session", "widget", f1.id)
self.assertEqual([f2], self.mgr.get_all_files("session", "widget"))
def test_remove_widget_files(self):
# This should not error.
self.mgr.remove_session_files("non-report")
# Add two files with different session IDs, but the same widget ID.
f1 = self.mgr.add_file("session1", "widget", FILE_1)
f2 = self.mgr.add_file("session2", "widget", FILE_1)
self.mgr.remove_files("session1", "widget")
self.assertEqual([], self.mgr.get_all_files("session1", "widget"))
self.assertEqual([f2], self.mgr.get_all_files("session2", "widget"))
def test_remove_session_files(self):
# This should not error.
self.mgr.remove_session_files("non-report")
# Add two files with different session IDs, but the same widget ID.
f1 = self.mgr.add_file("session1", "widget1", FILE_1)
f2 = self.mgr.add_file("session1", "widget2", FILE_1)
f3 = self.mgr.add_file("session2", "widget", FILE_1)
self.mgr.remove_session_files("session1")
self.assertEqual([], self.mgr.get_all_files("session1", "widget1"))
self.assertEqual([], self.mgr.get_all_files("session1", "widget2"))
self.assertEqual([f3], self.mgr.get_all_files("session2", "widget"))
def test_remove_orphaned_files(self):
"""Test the remove_orphaned_files behavior"""
f1 = self.mgr.add_file("session1", "widget1", FILE_1)
f2 = self.mgr.add_file("session1", "widget1", FILE_1)
f3 = self.mgr.add_file("session1", "widget1", FILE_1)
self.assertEqual([f1, f2, f3],
self.mgr.get_all_files("session1", "widget1"))
# Nothing should be removed here (all files are active).
self.mgr.remove_orphaned_files(
"session1",
"widget1",
newest_file_id=f3.id,
active_file_ids=[f1.id, f2.id, f3.id],
)
self.assertEqual([f1, f2, f3],
self.mgr.get_all_files("session1", "widget1"))
# Nothing should be removed here (no files are active, but they're all
# "newer" than newest_file_id).
self.mgr.remove_orphaned_files("session1",
"widget1",
newest_file_id=f1.id - 1,
active_file_ids=[])
self.assertEqual([f1, f2, f3],
self.mgr.get_all_files("session1", "widget1"))
# f2 should be removed here (it's not in the active file list)
self.mgr.remove_orphaned_files("session1",
"widget1",
newest_file_id=f3.id,
active_file_ids=[f1.id, f3.id])
self.assertEqual([f1, f3],
self.mgr.get_all_files("session1", "widget1"))
# remove_orphaned_files on an untracked session/widget should not error
self.mgr.remove_orphaned_files("no_session",
"no_widget",
newest_file_id=0,
active_file_ids=[])
class UploadedFileManagerTest(unittest.TestCase):
def setUp(self):
self.mgr = UploadedFileManager()
self.filemgr_events = []
self.mgr.on_files_updated.connect(self._on_files_updated)
def _on_files_updated(self, file_list, **kwargs):
self.filemgr_events.append(file_list)
def test_add_file(self):
self.assertIsNone(self.mgr.get_files("non-report", "non-widget"))
self.mgr.update_file_count("session", "widget", 1)
self.mgr.add_files("session", "widget", [file1])
self.assertEqual([file1], self.mgr.get_files("session", "widget"))
self.assertEqual(len(self.filemgr_events), 1)
# Add another file with the same ID
self.mgr.update_file_count("session", "widget", 2)
self.mgr.add_files("session", "widget", [file2])
self.assertEqual([file1, file2],
self.mgr.get_files("session", "widget"))
self.assertEqual(len(self.filemgr_events), 2)
def test_remove_file(self):
# This should not error.
self.mgr.remove_files("non-report", "non-widget")
self.mgr.add_files("session", "widget", [file1])
self.mgr.remove_file("session", "widget", file1.id)
self.assertEqual([], self.mgr.get_files("session", "widget"))
self.mgr.remove_file("session", "widget", file1.id)
self.assertEqual([], self.mgr.get_files("session", "widget"))
self.mgr.add_files("session", "widget", [file1])
self.mgr.add_files("session", "widget", [file2])
self.mgr.remove_file("session", "widget", file1.id)
self.assertEqual([file2], self.mgr.get_files("session", "widget"))
def test_remove_widget_files(self):
# This should not error.
self.mgr.remove_session_files("non-report")
# Add two files with different session IDs, but the same widget ID.
self.mgr.add_files("session1", "widget", [file1])
self.mgr.add_files("session2", "widget", [file1])
self.mgr.remove_files("session1", "widget")
self.assertIsNone(self.mgr.get_files("session1", "widget"))
self.assertEqual([file1], self.mgr.get_files("session2", "widget"))
def test_remove_session_files(self):
# This should not error.
self.mgr.remove_session_files("non-report")
# Add two files with different session IDs, but the same widget ID.
self.mgr.add_files("session1", "widget1", [file1])
self.mgr.add_files("session1", "widget2", [file1])
self.mgr.add_files("session2", "widget", [file1])
self.mgr.remove_session_files("session1")
self.assertIsNone(self.mgr.get_files("session1", "widget1"))
self.assertIsNone(self.mgr.get_files("session1", "widget2"))
self.assertEqual([file1], self.mgr.get_files("session2", "widget"))
def test_replace_widget_files(self):
self.mgr.add_files("session1", "widget", [file1])
self.mgr.replace_files("session1", "widget", [file2])
self.assertEqual(len(self.mgr.get_files("session1", "widget")), 1)
self.assertEqual([file2], self.mgr.get_files("session1", "widget"))
