Environ

Module that initialize singleton environ object for the researcher component

[Environ][fedbiomed.common.environ] will be initialized after the object environ is imported from fedbiomed.researcher.environ

Typical use:

from fedbiomed.researcher.environ import environ

print(environ['RESEARCHER_ID'])

Descriptions of researcher Global Variables:

  • RESEARCHER_ID : id of the researcher
  • ID : equals to researcher id
  • TENSORBOARD_RESULTS_DIR : path for writing tensorboard log files
  • EXPERIMENTS_DIR : folder for saving experiments
  • MESSAGES_QUEUE_DIR : Path for writing queue files
  • SERVER_HOST : Hostname or IP of gRPC server
  • SERVER_PORT : TCP port of gRPC server
  • SERVER_SSL_KEY : Path to certificate private key file for gRPC
  • SERVER_SSL_CERT : Path to certificate PEM file for gRPC

Attributes

environ module-attribute 

environ = ResearcherEnviron()

Classes

ResearcherEnviron 

ResearcherEnviron(root_dir=None)

Bases: Environ

Source code in fedbiomed/researcher/environ.py 
def __init__(self, root_dir: str = None):
    """Constructs ResearcherEnviron object """
    super().__init__(root_dir=root_dir)

    self._config = ResearcherConfig(root_dir)

    logger.setLevel("DEBUG")
    # Set component type
    self._values["COMPONENT_TYPE"] = ComponentType.RESEARCHER
    # Setup environment variables
    self.set_environment()

Functions

info 
info()

Print useful information at environment creation

Source code in fedbiomed/researcher/environ.py 
def info(self):
    """Print useful information at environment creation"""

    logger.info("Component environment:")
    logger.info("type = " + str(self._values['COMPONENT_TYPE']))
set_environment 
set_environment()
Source code in fedbiomed/researcher/environ.py 
def set_environment(self):

    super().set_environment()

    # we may remove RESEARCHER_ID in the future (to simplify the code)
    # and use ID instead
    researcher_id = self._config.get('default', 'id')

    self._values['RESEARCHER_ID'] = os.getenv('RESEARCHER_ID', researcher_id)
    self._values['ID'] = self._values['RESEARCHER_ID']

    # more directories
    self._values['TENSORBOARD_RESULTS_DIR'] = os.path.join(self._values['ROOT_DIR'], TENSORBOARD_FOLDER_NAME)
    self._values['EXPERIMENTS_DIR'] = os.path.join(self._values['VAR_DIR'], "experiments")
    self._values['MESSAGES_QUEUE_DIR'] = os.path.join(self._values['VAR_DIR'], 'queue_messages')

    self._values["SERVER_HOST"] = os.getenv('RESEARCHER_SERVER_HOST',
                                            self._config.get('server', 'host'))
    self._values["SERVER_PORT"] = os.getenv('RESEARCHER_SERVER_PORT',
                                            self._config.get('server', 'port'))


    self._values["FBM_CERTIFICATE_KEY"] = os.path.join(
        self._values["CONFIG_DIR"], self._config.get('certificate', 'private_key')
    )
    self._values["FBM_CERTIFICATE_PEM"] = os.path.join(
        self._values["CONFIG_DIR"], self._config.get('certificate', 'public_key')
    )


    for _key in 'TENSORBOARD_RESULTS_DIR', 'EXPERIMENTS_DIR':
        dir = self._values[_key]
        if not os.path.isdir(dir):
            try:
                os.makedirs(dir)
            except FileExistsError as exp:
                raise FedbiomedEnvironError(
                    f"{ErrorNumbers.FB600.value}: path already exists but is not a "
                    f"directory {dir}"
                ) from exp
            except OSError as exp:
                raise FedbiomedEnvironError(
                    f"{ErrorNumbers.FB600.value}: cannot create environment subtree in: {dir}"
                ) from exp