def get_config():
try:
flowauth_fernet_key = environ["FLOWAUTH_FERNET_KEY"].encode()
_ = Fernet(flowauth_fernet_key) # Error if fernet key is bad
log_level = getattr(logging,
getenv("FLOWAUTH_LOG_LEVEL", "error").upper(),
logging.ERROR)
db_uri = getenv("DB_URI", os.getenv("DB_URI",
"sqlite:////tmp/test.db"))
db_uri = db_uri.format(getenv("FLOWAUTH_DB_PASSWORD", ""))
return dict(
PRIVATE_JWT_SIGNING_KEY=load_private_key(
environ["PRIVATE_JWT_SIGNING_KEY"]),
LOG_LEVEL=log_level,
ADMIN_USER=environ["FLOWAUTH_ADMIN_USERNAME"],
ADMIN_PASSWORD=environ["FLOWAUTH_ADMIN_PASSWORD"],
SQLALCHEMY_DATABASE_URI=db_uri,
SQLALCHEMY_ENGINE_OPTIONS=dict(pool_recycle=3600),
SECRET_KEY=environ["SECRET_KEY"],
SESSION_PROTECTION="strong",
SQLALCHEMY_TRACK_MODIFICATIONS=False,
FLOWAUTH_FERNET_KEY=flowauth_fernet_key,
DEMO_MODE=True if getenv("DEMO_MODE") is not None else False,
RESET_DB=True
if getenv("RESET_FLOWAUTH_DB") is not None else False,
DB_IS_SET_UP=Event(),
CACHE_BACKEND=get_cache_backend(),
)
except KeyError as e:
raise UndefinedConfigOption(
f"Undefined configuration option: '{e.args[0]}'. Please set docker secret or environment variable."
)
def get_cache_backend() -> CacheRegion:
"""
Get a dogpilecache cache region.
Returns
-------
CacheRegion
"""
cache_backend = getenv("FLOWAUTH_CACHE_BACKEND", "FILE").upper()
if cache_backend == "REDIS":
backend = "dogpile.cache.redis"
cache_args = dict(
host=environ["FLOWAUTH_REDIS_HOST"],
port=int(getenv("FLOWAUTH_REDIS_PORT", "6379")),
db=int(getenv("FLOWAUTH_REDIS_DB", "0")),
redis_expiration_time=32,
distributed_lock=True,
password=getenv("FLOWAUTH_REDIS_PASSWORD", None),
)
elif cache_backend == "FILE":
backend = "dogpile.cache.dbm"
cache_args = dict(filename=environ["FLOWAUTH_CACHE_FILE"])
else:
backend = "dogpile.cache.memory"
cache_args = {}
return make_region().configure(backend=backend,
expiration_time=30,
arguments=cache_args)
def get_config():
try:
jwt_public_key = load_public_key(environ["PUBLIC_JWT_SIGNING_KEY"])
log_level = logging.getLevelName(
getenv("FLOWAPI_LOG_LEVEL", "error").upper())
flowmachine_host = environ["FLOWMACHINE_HOST"]
flowmachine_port = environ["FLOWMACHINE_PORT"]
flowdb_user = environ["FLOWAPI_FLOWDB_USER"]
flowdb_password = environ["FLOWAPI_FLOWDB_PASSWORD"]
flowdb_host = environ["FLOWDB_HOST"]
flowdb_port = environ["FLOWDB_PORT"]
flowapi_server_id = environ["FLOWAPI_IDENTIFIER"]
except KeyError as e:
raise UndefinedConfigOption(
f"Undefined configuration option: '{e.args[0]}'. Please set docker secret or environment variable."
)
return dict(
JWT_PUBLIC_KEY=jwt_public_key,
JWT_ALGORITHM="RS256",
FLOWAPI_LOG_LEVEL=log_level,
FLOWMACHINE_HOST=flowmachine_host,
FLOWMACHINE_PORT=flowmachine_port,
FLOWDB_DSN=
f"postgres://{flowdb_user}:{flowdb_password}@{flowdb_host}:{flowdb_port}/flowdb",
JWT_DECODE_AUDIENCE=flowapi_server_id,
)
def main(run_on_schedule: bool = True):
"""
Main function. Creates output directories, initialises the database, parses
a workflows definition file to define workflows and configure the available
dates sensor, and runs the available dates sensor.
Parameters
----------
run_on_schedule : bool, default True
Set run_on_schedule=False to run the sensor only once, ignoring the schedule.
(useful for testing)
"""
# Initialise logger
# TODO: Use structlog (not sure whether it will be possible for the prefect logger)
log_level = os.environ["AUTOFLOW_LOG_LEVEL"]
logger = logging.getLogger(__name__)
handler = logging.StreamHandler()
formatter = logging.Formatter(
"[%(asctime)s] %(levelname)s - %(name)s | %(message)s"
) # Match prefect format for now
formatter.converter = time.gmtime
handler.setFormatter(formatter)
logger.addHandler(handler)
logger.setLevel(log_level)
logger.info(f"Log level for logger '{__name__}' set to '{log_level}'.")
# Make output directories
outputs_path = Path(os.environ["AUTOFLOW_OUTPUTS_DIR"])
logger.info(
f"Creating output directories '{outputs_path/'notebooks'}' and '{outputs_path/'reports'}'."
)
(outputs_path / "notebooks").mkdir(exist_ok=True)
(outputs_path / "reports").mkdir(exist_ok=True)
# Init DB
# Note: AUTOFLOW_DB_URI must be an env var so that it can be used in prefect.config, so we read it using os.environ.
# AUTOFLOW_DB_PASSWORD can (and should) be a docker secret, so we read it using get_secret_or_env_var.
db_uri = os.environ["AUTOFLOW_DB_URI"]
logger.info(f"Initialising database '{db_uri}'.")
init_db(db_uri.format(getenv("AUTOFLOW_DB_PASSWORD", "")))
# Create workflows according to workflow definition file
inputs_dir = os.environ["AUTOFLOW_INPUTS_DIR"]
logger.info(f"Creating workflows defined in '{Path(inputs_dir)/'workflows.yml'}'.")
workflow_storage, sensor_config = parse_workflows_yaml("workflows.yml", inputs_dir)
# Run available dates sensor
logger.info("Running available dates sensor.")
available_dates_sensor.schedule = sensor_config["schedule"]
available_dates_sensor.run(
workflow_configs=sensor_config["workflows"],
cdr_types=sensor_config["cdr_types"],
workflow_storage=workflow_storage,
run_on_schedule=run_on_schedule,
)
def get_session(db_uri: str) -> "sqlalchemy.orm.session.Session":
"""
Create a sqlalchemy session.
Parameters
----------
db_uri : str
Database URI
Returns
-------
Session
A sqlalchemy session
"""
# TODO: This seems like the wrong place to be reading a secret / env var,
# but we can't put a docker secret in the prefect config.
full_db_uri = db_uri.format(getenv("AUTOFLOW_DB_PASSWORD", ""))
engine = create_engine(full_db_uri)
return sessionmaker(bind=engine)()
def get_available_dates(
cdr_types: Optional[Sequence[str]] = None, ) -> List[pendulum.Date]:
"""
Task to return a union of the dates for which data is available in FlowDB for the specified set of CDR types.
Parameters
----------
cdr_types : list of str, optional
Subset of CDR types for which to find available dates.
If not provided, the union of available dates for all CDR types will be returned.
Returns
-------
list of pendulum.Date
List of available dates, in chronological order
"""
prefect.context.logger.info(
f"Getting available dates from FlowAPI at '{prefect.config.flowapi_url}'."
)
conn = flowclient.connect(
url=prefect.config.flowapi_url,
token=environ["FLOWAPI_TOKEN"],
ssl_certificate=getenv("SSL_CERTIFICATE_FILE"),
)
dates = flowclient.get_available_dates(connection=conn)
prefect.context.logger.debug(f"Available dates: {dates}")
if cdr_types is None:
prefect.context.logger.debug(
"No CDR types provided. Will return available dates for all CDR types."
)
cdr_types = dates.keys()
else:
prefect.context.logger.debug(
f"Returning available dates for CDR types {cdr_types}.")
unknown_cdr_types = set(cdr_types).difference(dates.keys())
if unknown_cdr_types:
warnings.warn(
f"No data available for CDR types {unknown_cdr_types}.")
dates_union = set.union(*[
set(pendulum.parse(date, exact=True) for date in dates[cdr_type])
for cdr_type in cdr_types if cdr_type in dates.keys()
])
return sorted(list(dates_union))
def _do_connect(
*,
log_level: Optional[str] = None,
flowdb_port: Optional[int] = None,
flowdb_user: Optional[str] = None,
flowdb_password: Optional[str] = None,
flowdb_host: Optional[str] = None,
flowdb_connection_pool_size: Optional[int] = None,
flowdb_connection_pool_overflow: Optional[int] = None,
redis_host: Optional[str] = None,
redis_port: Optional[int] = None,
redis_password: Optional[str] = None,
conn: Optional[Connection] = None,
) -> Tuple[Connection, ThreadPoolExecutor, StrictRedis]:
"""
Connects flowmachine to a database, and performs initial set-up routines.
You may provide a Settings object here, which can specify the database
you wish to connect to, logging behaviour, available tables and so on.
Parameters
----------
log_level : str, default "error"
Level to log at
flowdb_port : int, default 9000
Port number to connect to flowdb
flowdb_user : str, default "flowmachine"
Name of user to connect to flowdb as
flowdb_password : str
Password to connect to flowdb
flowdb_host : str, default "localhost"
Hostname of flowdb server
flowdb_connection_pool_size : int, default 5
Default number of database connections to use
flowdb_connection_pool_overflow : int, default 1
Number of extra database connections to allow
redis_host : str, default "localhost"
Hostname for redis server.
redis_port : int, default 6379
Port the redis server is available on
redis_password : str
Password for the redis instance
conn : flowmachine.core.Connection
Optionally provide an existing Connection object to use, overriding any the db options specified here.
Returns
-------
Connection
Notes
-----
All parameters can also be provided as environment variables.
If a parameter is provided, and an environment variable is set,
then the provided value is used. If neither is provided, the defaults as given
in the docstring are used.
Parameters can _also_ be set using Docker secrets, in which case a file with the name
of the parameter in upper case should be present at /run/secrets/THE_PARAM.
If a secret is available, the secret takes precedence over both the environment variable, and
the default.
"""
try:
log_level = (getenv("FLOWMACHINE_LOG_LEVEL", "error")
if log_level is None else log_level)
flowdb_port = int(
getenv("FLOWDB_PORT", "9000"
) if flowdb_port is None else flowdb_port)
flowdb_user = (getenv("FLOWMACHINE_FLOWDB_USER", "flowmachine")
if flowdb_user is None else flowdb_user)
flowdb_password = (environ["FLOWMACHINE_FLOWDB_PASSWORD"]
if flowdb_password is None else flowdb_password)
flowdb_host = (getenv("FLOWDB_HOST", "localhost")
if flowdb_host is None else flowdb_host)
flowdb_connection_pool_size = (int(
getenv("DB_CONNECTION_POOL_SIZE", "5"))
if flowdb_connection_pool_size is None
else flowdb_connection_pool_size)
flowdb_connection_pool_overflow = int(
getenv("DB_CONNECTION_POOL_OVERFLOW", "1"
) if flowdb_connection_pool_overflow is None else
flowdb_connection_pool_overflow)
redis_host = (getenv("REDIS_HOST", "localhost")
if redis_host is None else redis_host)
redis_port = int(
getenv("REDIS_PORT", "6379") if redis_port is None else redis_port)
redis_password = (environ["REDIS_PASSWORD"]
if redis_password is None else redis_password)
except KeyError as e:
raise ValueError(
f"You must provide a secret named {e.args[0]}, set an environment variable named {e.args[0]}, or provide the value as a parameter."
)
set_log_level("flowmachine.debug", log_level)
if conn is None:
conn = Connection(
host=flowdb_host,
port=flowdb_port,
user=flowdb_user,
password=flowdb_password,
database="flowdb",
pool_size=flowdb_connection_pool_size,
overflow=flowdb_connection_pool_overflow,
)
redis_connection = redis.StrictRedis(host=redis_host,
port=redis_port,
password=redis_password)
thread_pool = ThreadPoolExecutor(flowdb_connection_pool_size)
conn.available_dates
print(f"FlowMachine version: {flowmachine.__version__}")
print(
f"Flowdb running on: {flowdb_host}:{flowdb_port}/flowdb (connecting user: {flowdb_user})"
)
return conn, thread_pool, redis_connection
