Adding sources

The access_nri_intake.catalog sub-package contains tools to create/extend/trim intake-dataframe-catalogs of Intake-ESM datastores. The access_nri_intake.catalog.manager.CatalogManager class can be used to create a new intake-dataframe-catalog or load an existing one. Intake-ESM datastore sources can be built (using an access_nri_intake.source.builders Builder) or loaded and then added to the catalog. Translators are specified to translate the metadata in source datastores to be compatible with the catalog schema.

When access_nri_intake.catalog is first imported, it downloads and parses the a specific commit of the schema at ACCESS-NRI/schema. The raw schema is stored in the variable access_nri_intake.catalog.EXP_JSONSCHEMA (more on this later) and a version with the “required” field replaced with access_nri_intake.catalog.CORE_COLUMNS is stored in the variable access_nri_intake.catalog.CATALOG_JSONSCHEMA (this is to allow this field to be customized). The latter defines what metadata must be included in the intake-dataframe-catalog, and what types and fields are allowed. Subsequent imports read the downloaded schema, unless the schema is changed (see Updating schema), in which case the new schema is downloaded.

Translators

Translators receive an Intake source to translate from and a list of metadata columns to target (these are the columns in the intake-dataframe-catalog), and return a dataframe of translated data when their translate method is called. The returned dataframe has rows containing tuples of unique values of the translated metadata after grouping by the metadata columns specified in access_nri_intake.catalog.TRANSLATOR_GROUPBY_COLUMNS.

When a source is added to the catalog and no translator is specified, the translator defaults to access_nri_intake.catalog.translators.DefaultTranslator which operates as follows:

• If the input source is an Intake-ESM datastore, the translator will first look for the column in the esmcat.df attribute, casting iterable columns to tuples. If the source is not an Intake-ESM datastore, this step is skipped.

• If that fails, the translator will then look for the column name as an attribute on the source itself

• If that fails, the translator will then look for the column name in the metadata attribute of the source

The access_nri_intake.catalog.translators.DefaultTranslator is appropriate for Intake-ESM datastore sources built using access_nri_intake.source.builders because the schema used to validate the datastores is consistent with the schema used to validate the catalog.

When adding a pre-generated Intake-ESM datastore to the catalog, a dedicated Translator may be required. For example, the CMIP5 and CMIP6 NCI-managed Intake-ESM datastores that are included in the ACCESS-NRI catalog use dedicated translators which implement specific translations from the CMIP vocabulary used in the Intake-ESM datastore to the vocabulary used in the catalog schema (e.g. see access_nri_intake.catalog.translators.Cmip6Translator).

Creating a new Translator

New Translators should inherit from access_nri_intake.catalog.translators.DefaultTranslator. The general approach to creating a new translator is to create a specific translator method for each input column that cannot use the default translator. These methods should return a dataframe object. Take a look at the existing Translator class implementations for examples.

API for access_nri_intake.catalog

This documentation has been auto-generated using sphinx-autoapi

• access_nri_intake.catalog
  • Submodules
    • access_nri_intake.catalog.manager
      • Module Contents
        • Classes
          • CatalogManagerError
          • CatalogManager
    • access_nri_intake.catalog.translators
      • Module Contents
        • Classes
          • TranslatorError
          • DefaultTranslator
          • Cmip6Translator
          • Cmip5Translator
          • EraiTranslator