Package Tour
Almost all new functionality in Take2 is within intake.readers. Must of the functionality
from V1 should still work as it did before.
intake.config
Managing and persisting the config.
intake.readers.catalogs
Readers that produce catalogs: a few interesting serice endpoins:
Tiled
SQL
STAC (including paramterised search)
THREDDS
NASA Earthdata
and some collections of example data
huggingface hub
SKLearn examples
torch datasets
TF datasets
intake.readers.convert
Classes to convert between data representations without changing the data. Each converter specifies which types it acts on on, and what it produces.
Includes the Pipeline class used to store a sequence of steps, and a couple of utility functions for plotting making a graph of the available conversions and finding the shortest route from one type to another.
intake.readers.datatypes
All of the data prescription classes, subclassed from BaseData. Defines the minimum required information for an instance, and some ways to guess a type from a URL.
intake.readers.entry
Classes for the descriptions of data and readers that live inside catalogs, and the Catalog class itself.
intake.readers.importlist
How modules get imported when intake itself is imported; this is how subclasses of BaseData, BaseReader and BaseConverter are “registered”, rather than relying exclusively on entrypoints.
intake.readers.metadata
A loose descriptin of the fields expected in a metadata dictionary.
intake.readers.mixins
The magic that makes reader[..] and reader.<> work.
intake.readers.namespaces
Set of functions within a few popular packages, such as numpy, that you might expect
to automatically be available for tab-completion of a numpy-producing reader, something
like reader.np.abs would find the np.abs function and apply it.
intake.readers.output
Converters specialised for producing outputs, normally by side-effect. Most produce data objects, which you can also put in a catalog.
intake.readers.readers
The BaseReader class and all the readers derived from it. These are the things that do
the actual loading of data at runtime. Each one specified which datatype it can read,
what imports must be available, and what it produces. The doc() method Intake-specific
information, if any, and the docstring of the (main) function used for loading.
intake.readers.transform
Converters which actually change the data, but not normally the representation. The simplest would be column selection from a dataframe.
intake.readers.user_parameters
A few types that can be used to template data and reader descriptions in a Catalog. The extensible
type system allows for simple verification, and may in the future expand to something like
param or pydantic.