Added
- `select_records`, which allows for a single API for selecting attributes (result_identifiers) given filter_conditions and/or columns
- `retrieve_one`, and `retrieve_many` which allows for selecting one or multiple records given record_identifier
- pipestat reader submodule to read DB results via FastAPI endpoints: `pipestat serve --config "config.yaml"`
- ability to create `SamplePipestatManager` and `ProjectSamplePipestatManager` which are sample/project specific PipestatManager objects.
- PipestatBoss wrapper class which holds `SamplePipestatManager` and `ProjectSamplePipestatManager` classes.
- `to_dict` methods to parsed schema object.
- `select_distinct` function which retrieves unique results for a list of attributes.
- `pipestat link` which creates a directory of symlinks for reported results.
- `list_recent_results` which allows for retrieving records filtered via a start and end time.
- reporting and retrieving results via item access,e.g. `psm["sample1", "name_of_something"] = "name_of_something_string"` or `result = psm["sample1"]`
- pipestat now supports one results file per sample by using `{record_identifier}` in the results_file_path.
Fixed
- path expansion when creating database url.
- added jinja2 requirement.
- `pipeline_name` column not populating in postgres db backend.
- pipestat will now create subdirectories during results_file creation.
- fixed bugs and polished both report generation and stats table generation per https://github.com/pepkit/pipestat/pull/131
Changed
- Removed `retrieve`, `get_one_record`, `get_records function`.
- Removed `get_orm` and replace with `get_model`.
- Removed `get_table_name` function.
- Refactor:
- `sample_name` -> `record_identifier`.
- `pipeline_type` has been removed from most functions.
- added optional dependencies for the database backend and pipestat reader, e.g. `pip install pipestat[dbbackend]`.