@spec id_list([Indexed.record()], any()) :: [Indexed.id()]

Return a list of all ids from the collection.

@spec pre(keyword()) :: Indexed.t()

For a set of entities, "prewarm" the cache by validating and normalizing all options and creating the ETS tables. Data and indexes will be created during the main "warm" step which comes next.

Argument is a keyword list where the top level has:

• :entities - A keyword list, configuring each entity to be stored. One or more of these are expected. Options within listed below.
• :namespace - Atom to prefix the ETS table names with. If nil (default) then named tables will not be used.

OR if the :entities key is not present, the whole keyword list is presumed to be the option.

entity-options

Entity Options

The :entities keyword list main option described above has entity name atoms as keys and keyword lists of options are values. Allowed options are as follows:

• :fields - List of field name atoms to index by. At least one required.
  • If field is a DateTime, use sort: {:my_field, sort: :datetime}.
  • Ascending and descending will be indexed for each field.
• :id_key - Primary key to use in indexes and for accessing the records of this entity. See Indexed.Entity.t/0. Default: :id.
• :lookups - See Indexed.Entity.t/0.
• :namespace - Atom name of the ETS table when a named table is desired. Useful for accessing the data in a process without the table ref. When this is non-nil, named tables will be used instead of references and the namespace is used as a prefix for them.
• :prefilters - List of field name atoms which should be prefiltered on. This means that separate indexes will be managed for each unique value for each of these fields, across all records of this entity type. While field name nil refers to the indexes where no prefilter is used (all records) and it is included by default, it may be defined in the arguments if further options are needed. Default [{nil, []}]. If options are needed, 2-element tuples may be used in place of the atom where the the first element is the field name atom, and the second is a keyword list of any of the following options:
  • :maintain_unique - List of field name atoms for which a list of unique values under the prefilter will be managed. If the nil prefilter is defined, leave the other prefilter fields off the :maintain_unique option as these are automatically included. These lists can be fetched via get_uniques_list/4.

@spec record_sort_fn(Indexed.Entity.field()) :: (any(), any() -> boolean())

From a field, make a compare function, suitable for Enum.sort/2.

@spec resolve_data_opt(data_opt(), atom(), [Indexed.Entity.field()]) ::
  {atom(), atom(), [Indexed.record()]}

Normalize warm/1's data option.

@spec resolve_fields_opt([atom() | Indexed.Entity.field()], atom()) :: [
  Indexed.Entity.field()
]

Normalize fields.

@spec resolve_prefilters_opt([atom() | keyword()] | nil) :: keyword(keyword())

Normalize the prefilters option to tuples, adding nil prefilter if needed.

@spec run(keyword()) :: Indexed.t()

See pre/1. The only difference is that the :data entity option is allowed:

• :data - List of maps (with id key) -- the data to index and cache. Required. May take one of the following forms:
  • {direction, field, list} - data list, with a hint that it is already sorted by field (atom) and direction (:asc or :desc), data_tuple/0.
  • list - data list with unknown ordering; must be sorted for every field.

@spec warm_sorted(t(), Indexed.prefilter(), Indexed.Entity.field(), data_tuple()) ::
  true

Create the asc and desc indexes for one field.