def test_do_blacklist_user_configured_folders(self):
"""
Files inside user configured folders should be blacklisted.
"""
folder_black_list = FolderBlackList(["/bar/some_folder"])
is_blacklisted = folder_black_list.is_blacklisted
self.assertTrue(is_blacklisted("/bar/some_folder/script.py"))
def test_do_not_blacklist(self):
"""
Ensure we're not accidentally blacklisting things we shouldn't be.
"""
folder_black_list = FolderBlackList([])
is_blacklisted = folder_black_list.is_blacklisted
self.assertFalse(is_blacklisted("/foo/not_blacklisted/script.py"))
self.assertFalse(is_blacklisted("/foo/not_blacklisted/.hidden_script.py"))
def test_do_blacklist(self):
"""
miniconda, anaconda, and .*/ folders should be blacklisted.
"""
folder_black_list = FolderBlackList([])
is_blacklisted = folder_black_list.is_blacklisted
self.assertTrue(is_blacklisted("/foo/miniconda2/script.py"))
self.assertTrue(is_blacklisted("/foo/miniconda3/script.py"))
self.assertTrue(is_blacklisted("/foo/anaconda2/script.py"))
self.assertTrue(is_blacklisted("/foo/anaconda3/script.py"))
self.assertTrue(is_blacklisted("/foo/.virtualenv/script.py"))
self.assertTrue(is_blacklisted("/foo/.venv/script.py"))
self.assertTrue(is_blacklisted("/foo/.random_hidden_folder/script.py"))
def _file_should_be_hashed(self, filename: str) -> bool:
global _FOLDER_BLACK_LIST
if not _FOLDER_BLACK_LIST:
_FOLDER_BLACK_LIST = FolderBlackList(
config.get_option("server.folderWatchBlacklist"))
filepath = os.path.abspath(filename)
file_is_blacklisted = _FOLDER_BLACK_LIST.is_blacklisted(filepath)
# Short circuiting for performance.
if file_is_blacklisted:
return False
return file_util.file_is_in_folder_glob(
filepath, self._get_main_script_directory()
) or file_util.file_in_pythonpath(filepath)
def __init__(self, session_data: SessionData):
self._session_data = session_data
self._on_file_changed: List[Callable[[], None]] = []
self._is_closed = False
# Blacklist for folders that should not be watched
self._folder_black_list = FolderBlackList(
config.get_option("server.folderWatchBlacklist"))
self._watched_modules: Dict[str, WatchedModule] = {}
self._register_watcher(
self._session_data.script_path,
module_name=None, # Only the root script has None here.
)
def __init__(self, report, on_file_changed):
self._report = report
self._on_file_changed = on_file_changed
self._is_closed = False
# Blacklist for folders that should not be watched
self._folder_black_list = FolderBlackList(
config.get_option("server.folderWatchBlacklist"))
# A dict of filepath -> WatchedModule.
self._watched_modules = {}
self._register_watcher(
self._report.script_path,
module_name=None, # Only the root script has None here.
)
def __init__(self, name="md5", hasher=None):
self.hashes = dict()
self.name = name
# The number of the bytes in the hash.
self.size = 0
# An ever increasing counter.
self._counter = 0
if hasher:
self.hasher = hasher
else:
self.hasher = hashlib.new(name)
self._folder_black_list = FolderBlackList(
config.get_option("server.folderWatchBlacklist"))
# If a dataframe has more than this many rows, we consider it large and hash a sample.
_PANDAS_ROWS_LARGE = 100000
_PANDAS_SAMPLE_SIZE = 10000
# Similar to dataframes, we also sample large numpy arrays.
_NP_SIZE_LARGE = 1000000
_NP_SAMPLE_SIZE = 100000
# Arbitrary item to denote where we found a cycle in a hashed object.
# This allows us to hash self-referencing lists, dictionaries, etc.
_CYCLE_PLACEHOLDER = b"streamlit-57R34ML17-hesamagicalponyflyingthroughthesky-CYCLE"
_FOLDER_BLACK_LIST = FolderBlackList(config.get_option("server.folderWatchBlacklist"))
# FFI objects (objects that interface with C libraries) can be any of these types:
_FFI_TYPE_NAMES = [
"_cffi_backend.FFI",
"builtins.CompiledFFI",
]
# KERAS objects can be any of these types:
_KERAS_TYPE_NAMES = [
"keras.engine.training.Model",
"tensorflow.python.keras.engine.training.Model",
"tensorflow.python.keras.engine.functional.Functional",
]