• The cumulus-process-py no longer assumes the structure of input and config given to a processing sublcass. These assumptions have to be added in the __init__ section of the subclass (see example).
• All methods that have specific assumptions about the structure of the Cumulus message are deprecated. Subclasses should implement these methods in the subclass if they are needed
• Some of the deprecated methods are moved to a new helpers modules can be used by importing them directly from that module

Added

• Add deprecation warning to the followings methods of the Process class (these methods will be removed in v0.8.0)
• Add example folder with an example implementation of a Process sublcass
• Add localstack to tests

Changed

• Simplify the structure of Process class to support CUMULUS-456
• clean_all method now removes the whole temp folder at the end of the process (this solves the problem of lambda functions running out of temporary storage)
• CUMULUS-456 Use the fileStagingDir to create the URL paths and if none exists, use url_path (this is added to deprecation methods)
• CUMULUS-477 Updates to use the new bucket structure where each bucket is no longer a string, but an object with name and type (this is added to deprecation methods)

Changed

• Get cumulus-message-adapter-python from pypi instead of github

Changed

• Removed the built-in assumptions about the config and input to the Process class when the class is initialized. [CUMULUS-543]
• A new property called self.regex is introduced. This property should be set for the self.gid property to work properly.

• allow input_keys to be specified in payload
• move existing hardcoded regex to default_keys, used if none in payload
• backwards-compatible, but child classes should rename input_keys to default_keys or they will override inherited logic

• truncate error strings passed to sfn.send_task_failure()
• log exceptions raised in the handler

• Peg version of cumulus-message-adapter-python
• Update imports for Python3

• Fix activity handler for cumulus

• Fix CLI for processing payloads

• Fix passing in of a local path for saving files and output
• Properly clean up downloaded files

• Multiple bug fixes to work with Cumulus deployment

• Refactored Process class to remove any automatic functionality
• Update to use cumulus-message-adapter
• cumulus-py now public and open source, turned into standard package, docker files removed
• ci updated to auto tag and deploy to PyPi on merge to master

• Refactor Granule class into new Process class
• New payload format

• Payloads updated for Cumulus phase 3
• refactored granule class to be more abstracted, payload code moved to aws service module
• added support for multiple output granules
• use regex to determine which input files are which so order does not matter (only that all needed files are supplied, in any order)