- Generic class that can hold and accumulate data to be passed
The purpose of this class is to provide a flexible data structure that works a bit like a dict or blank Python class, but prevents the user from accessing members that have not been defined a priori (more like a C struct), and also keeps metadata information such as a description, defaults, and units for each item in the container.
Containers can transform the data into a
as_dictmethod. This allows them to be written to an output table for example, where each Field defines a column. The
dictconversion can be made recursively and even flattened so that a nested set of Containers can be translated into a set of columns in a flat table without naming conflicts (the name of the parent Field is pre-pended).
Only members of instance
Fieldwill be used as output. For hierarchical data structures, Field can use
Containersubclasses or a
Mapas the default value.
>>> import astropy.units as u >>> class MyContainer(Container): ... x = Field(100, "The X value") ... energy = Field(-1, "Energy measurement", unit=u.TeV) ... >>> cont = MyContainer() >>> print(cont.x) 100 >>> # metadata will become header keywords in an output file: >>> cont.meta["KEY"] = "value"
Fields inside Containers can contain instances of other containers, to allow for a hierarchy of containers, and can also contain a
Mapfor the case where one wants e.g. a set of sub-classes indexed by a value like the
telescope_id. Examples of this can be found in
Containerworks by shadowing all class variables (which must be instances of
Field) with instance variables of the same name that hold the actual data. If
resetis called, all instance variables are reset to their default values as defined in the class.
Finally, a Container can have associated metadata via its
metaattribute, which is a
dictof keywords to values.
dict of attached metadata
Prefix attached to column names when saved to a table or file