def up(self) -> None:
"""Ensures that Airflow is running."""
if self.is_running:
logger.info("Airflow is already running.")
self._log_webserver_credentials()
return
if not fileio.file_exists(self.dags_directory):
fileio.create_dir_recursive_if_not_exists(self.dags_directory)
from airflow.cli.commands.standalone_command import StandaloneCommand
command = StandaloneCommand()
# Run the daemon with a working directory inside the current
# zenml repo so the same repo will be used to run the DAGs
daemon.run_as_daemon(
command.run,
pid_file=self.pid_file,
log_file=self.log_file,
working_directory=fileio.get_zenml_dir(),
)
while not self.is_running:
# Wait until the daemon started all the relevant airflow processes
time.sleep(0.1)
self._log_webserver_credentials()
def clean(git_examples_handler: GitExamplesHandler) -> None:
"""Deletes the ZenML examples directory from your current working
directory."""
examples_directory = os.path.join(os.getcwd(), "zenml_examples")
if (fileio.file_exists(examples_directory)
and fileio.is_dir(examples_directory) and confirmation(
"Do you wish to delete the ZenML examples directory? \n"
f"{examples_directory}")):
git_examples_handler.clean_current_examples()
declare(
"ZenML examples directory was deleted from your current working "
"directory.")
elif not fileio.file_exists(examples_directory) and not fileio.is_dir(
examples_directory):
logger.error(f"Unable to delete the ZenML examples directory - "
f"{examples_directory} - "
"as it was not found in your current working directory.")
def pull(
git_examples_handler: GitExamplesHandler,
example_name: str,
force: bool,
version: str,
) -> None:
"""Pull examples straight into your current working directory.
Add the flag --force or -f to redownload all the examples afresh.
Use the flag --version or -v and the version number to specify
which version of ZenML you wish to use for the examples."""
git_examples_handler.pull(force=force, version=version)
destination_dir = os.path.join(os.getcwd(), "zenml_examples")
fileio.create_dir_if_not_exists(destination_dir)
examples = (git_examples_handler.examples if not example_name else [
Example(
example_name,
Path(
os.path.join(
git_examples_handler.examples_repo.examples_dir,
example_name,
)),
)
])
for example in examples:
if not fileio.file_exists(str(example.path)):
error(
f"Example {example.name} does not exist! Available examples: "
f"{[e.name for e in git_examples_handler.examples]}")
return
example_destination_dir = os.path.join(destination_dir, example.name)
if fileio.file_exists(example_destination_dir):
if confirmation(f"Example {example.name} is already pulled. "
f"Do you wish to overwrite the directory?"):
fileio.rm_dir(example_destination_dir)
else:
warning(f"Example {example.name} not overwritten.")
continue
declare(f"Pulling example {example.name}...")
git_examples_handler.copy_example(example, example_destination_dir)
declare(f"Example pulled in directory: {example_destination_dir}")
def read_json(file_path: str) -> Any:
"""Read JSON on file path and returns contents as dict.
Args:
file_path: Path to JSON file.
"""
if fileio.file_exists(file_path):
contents = fileio.read_file_contents_as_string(file_path)
return json.loads(contents)
else:
raise FileNotFoundError(f"{file_path} does not exist.")
def __init__(self, **data: Any):
"""We persist the attributes in the config file. For the global
config, we want to persist the data as soon as it is initialized for
the first time."""
super().__init__(**data)
# At this point, if the serialization file does not exist we should
# create it and dump our data.
f = self.get_serialization_full_path()
if not fileio.file_exists(str(f)):
self._dump()
def readme_content(self) -> str:
"""Returns the readme content associated with a particular example."""
readme_file = os.path.join(self.path, "README.md")
try:
with open(readme_file) as readme:
readme_content = readme.read()
return readme_content
except FileNotFoundError:
if fileio.file_exists(str(self.path)) and fileio.is_dir(
str(self.path)):
raise ValueError(f"No README.md file found in {self.path}")
else:
raise FileNotFoundError(
f"Example {self.name} is not one of the available options."
f"\nTo list all available examples, type: `zenml example "
f"list`")
def read_yaml(file_path: str) -> Any:
"""Read YAML on file path and returns contents as dict.
Args:
file_path: Path to YAML file.
Returns:
Contents of the file in a dict.
Raises:
FileNotFoundError if file does not exist.
"""
if fileio.file_exists(file_path):
contents = fileio.read_file_contents_as_string(file_path)
return yaml.load(contents, Loader=yaml.FullLoader)
else:
raise FileNotFoundError(f"{file_path} does not exist.")
def _save_backup_file_if_required(self) -> None:
"""Saves a backup of the config file if the schema changed."""
if self._superfluous_options:
logger.warning(
"Found superfluous configuration values for class `%s`: %s",
self.__class__.__name__,
set(self._superfluous_options),
)
config_path = self.get_serialization_full_path()
if fileio.file_exists(config_path):
backup_path = config_path + ".backup"
fileio.copy(config_path, backup_path, overwrite=True)
logger.warning("Saving backup configuration to '%s'.",
backup_path)
# save the updated file without the extra options
self.update()
def _log_webserver_credentials(self):
"""Logs URL and credentials to login to the airflow webserver.
Raises:
FileNotFoundError: If the password file does not exist.
"""
if fileio.file_exists(self.password_file):
with open(self.password_file) as file:
password = file.read().strip()
else:
raise FileNotFoundError(
f"Can't find password file '{self.password_file}'")
logger.info(
"To inspect your DAGs, login to http://0.0.0.0:8080 "
"with username: admin password: %s",
password,
)
def json_config_settings_source(settings: BaseSettings) -> Dict[str, Any]:
"""
A simple settings source that loads variables from a YAML file
at the project's root.
Here we happen to choose to use the `env_file_encoding` from Config
when reading the config json file.
Args:
settings (BaseSettings): BaseSettings from pydantic.
Returns:
A dict with all configuration, empty dict if config not found.
"""
full_path = Path(config_dir) / config_name
logger.debug(f"Parsing file: {full_path}")
if fileio.file_exists(str(full_path)):
return cast(Dict[str, Any], yaml_utils.read_json(str(full_path)))
return {}
def _copy_to_dag_directory_if_necessary(self, dag_filepath: str):
"""Copies the DAG module to the airflow DAGs directory if it's not
already located there.
Args:
dag_filepath: Path to the file in which the DAG is defined.
"""
dags_directory = fileio.resolve_relative_path(self.dags_directory)
if dags_directory == os.path.dirname(dag_filepath):
logger.debug("File is already in airflow DAGs directory.")
else:
logger.debug("Copying dag file '%s' to DAGs directory.",
dag_filepath)
destination_path = os.path.join(dags_directory,
os.path.basename(dag_filepath))
if fileio.file_exists(destination_path):
logger.info(
"File '%s' already exists, overwriting with new DAG file",
destination_path,
)
fileio.copy(dag_filepath, destination_path, overwrite=True)
def _create_serialization_file_if_not_exists(self) -> None:
"""Creates the serialization file if it does not exist."""
f = self.get_serialization_full_path()
if not fileio.file_exists(str(f)):
fileio.create_file_if_not_exists(str(f))
def test_global_config_file_creation():
"""A simple test to check whether the global config is created."""
GlobalConfig()
# Raw config should now exist
assert fileio.file_exists(os.path.join(APP_DIR, GLOBAL_CONFIG_NAME))
