Super-Entities

Introduction

Sometimes it is useful to share the same component across multiple resources. To achieve this in a generic way, S3 supports a special type of link tables - so-called Super-Entities:

Instead of having several foreign keys for different primary resources, the shared component contains only one foreign key to the link table, the so-called super-key.

The primary resource (=instance table) contains the same super-key field as the component, thus the forward join of resource-to-component can be established as a natural join, i.e. without accessing the link table.

For the (seldom needed) backward join component-to-resource, the link table must be involved because it contains the table name of the primary resource in the field {{instance_type}}.

Disadvantage of the link table method is that the link table has to be updated whenever a primary resource record is created, updated or deleted - which generates some (minimal) extra load on write, however, it is recommended to not use super-entities for resources where the extra load on write can give a serious performance problem (e.g. messages), or at least to keep the super-entity table lean and free of extra references.

• code examples to follow...

API

To simplify the handling of super-entities in the model, S3Model provides a unified API for super-entities:

Defining a Super-Entity

To define a super-entity, you can use the super_entity() function:

table = s3db.super_entity(tablename, "sit_id", situation_types,
                          Field("datetime", "datetime"), # shared field
                          location_id())                 # shared field

You can define so-called shared fields in the super-entity, which are mirrored from the respective resource record. This allows to easily access these data from the component without the need to involve the primay instance table in the backward-join:

In case the names of the shared fields in your instance table differ from those of the super-entity, you can define a field mapping:

Defining an Instance of a Super-Entity

To make a table an instance of a super-entity, you can use s3db.configure():

s3db.configure(table, super_entity = db.sit_situation)

By default, all fields that the table and the super-entity have in common will be mirrored in the super-entity ("shared fields"). You can override this by specifying a list of fields to be mirrored by the super-entity.

s3db.configure(table,
               super_entity = db.sit_situation,
               sit_situation_fields = ["datetime"])

In case your table uses different names for the shared fields, you can use a dict instead to specify a mapping:

s3db.configure(table,
               super_entity = db.sit_situation,
               sit_situation_fields = dict(datetime="timestmp", location_id="location_id"))

Linking to a Super-Entity

Both, instance tables as well as shared components need to be linked to the super-entity. This can be done by inserting a super-key field into the instance/component table which must have the same name as the primary key of the super-entity. You can use the super_link() function as a DRY method for this:

tablename = "pr_presence"
table = db.define_table(tablename,
                        super_link("pe_id", "pr_pentity"),
                        super_link("sit_id", "sit_situation"), # sit_id
                        ...)

Note: super_link generates a Field instance - if you just need the name of the super-key field, you can use:

sk = s3db.super_key(db.pr_pentity) # returns the string "pe_id"

Updating a Super-Entity

If you use s3_rest_controller() for CRUD, it will automatically create, update and delete super-entities as necessary.

Otherwise, or if you manipulate records outside s3_rest_controller, you have to update all super-entities which an instance table implements whenever you create, update or delete a record in that table.

To do this, you can use:

s3db.update_super(table, record)

where table is the instance table and record the newly created/updated record.

In case you're going to delete a record from an instance table, you may use:

s3db.delete_super(table, record)

before you delete the record, where table is the instance table and record the record to be deleted.

---

DeveloperGuidelines