simpleml.save_patterns.libcloud module¶
Module for cloud save pattern definitions Uses Apache Libcloud as a universal engine
-
class
simpleml.save_patterns.libcloud.
CloudBase
[source]¶ Bases:
simpleml.save_patterns.base.BaseSavePattern
,simpleml.save_patterns.libcloud.CloudSavePatternMixin
Implementation class for extended base save patterns to/from the cloud via Apache Libcloud
-
class
simpleml.save_patterns.libcloud.
CloudHDF5SavePattern
[source]¶ Bases:
simpleml.save_patterns.libcloud.CloudBase
Save pattern implementation to save objects to Cloud in HDF5 format
-
SAVE_PATTERN
= 'cloud_hdf5'¶
-
-
class
simpleml.save_patterns.libcloud.
CloudKerasHDF5SavePattern
[source]¶ Bases:
simpleml.save_patterns.libcloud.CloudBase
Save pattern implementation to save objects to Cloud in Keras HDF5 format
-
SAVE_PATTERN
= 'cloud_keras_hdf5'¶
-
-
class
simpleml.save_patterns.libcloud.
CloudPickleSavePattern
[source]¶ Bases:
simpleml.save_patterns.libcloud.CloudBase
Save pattern implementation to save objects to Cloud in pickled format
-
SAVE_PATTERN
= 'cloud_pickled'¶
-
-
class
simpleml.save_patterns.libcloud.
CloudSavePatternMixin
[source]¶ Bases:
object
Mixin class to save/load objects via Apache Libcloud
Generic api for all cloud providers so naming convention is extremely important to follow in the config. Please reference libcloud documentation for supported input parameters
``` [cloud] section = name of the config section to use, ex: s3
[s3] param = value –> normal key:value syntax. match these to however they are referenced later, examples: key = abc123 secret = superSecure region = us-east-1 something_specific_to_s3 = s3_parameter — Below are internally referenced SimpleML params — driver = S3 –> this must be the Apache Libcloud provider (https://github.com/apache/libcloud/blob/trunk/libcloud/storage/types.py) connection_params = key,secret,region,something_specific_to_s3 –> this determines the key: value params passed to the constructor (it can be different for each provider) path = simpleml/specific/root –> similar to disk based home directory, cloud home directory will start relative to here container = simpleml –> the cloud bucket or container name ```
How this gets used: ``` from libcloud.storage.types import Provider from libcloud.storage.providers import get_driver
cloud_section = CONFIG.get(CLOUD_SECTION, ‘section’) connection_params = CONFIG.getlist(cloud_section, ‘connection_params’) root_path = CONFIG.get(cloud_section, ‘path’, fallback=’’)
driver_cls = get_driver(getattr(Provider, CONFIG.get(cloud_section, ‘driver’))) driver = driver_cls(**{param: CONFIG.get(cloud_section, param) for param in connection_params}) container = driver.get_container(container_name=CONFIG.get(cloud_section, ‘container’)) extra = {‘content_type’: ‘application/octet-stream’}
- obj = driver.upload_object(LOCAL_FILE_PATH,
- container=container, object_name=root_path + simpleml_folder_path + filename, extra=extra)
- obj = driver.download_object(CLOUD_OBJECT,
- destination_path=LOCAL_FILE_PATH, overwrite_existing=True, delete_on_failure=True)
-
CLOUD_DRIVER
= None¶
-
download_from_cloud
(folder: str, filename: str) → None[source]¶ Download any file from cloud to disk
-
driver
¶ “classproperty” to return and optionally globally set the cloud provider