SImplify core model

[jokiefer]

status quo

1. There are coexisting model instances

... like metadata model classes for the following resource types:

• Service (wms, wfs, csw, ...)
• Layer
• FeatureType

All instances of this types will allways has one related metadata instance.

code example

class ServiceMetadata(MetadataTermsOfUse, AbstractMetadata):
    """Concrete model class to store the parsed metadata information from the capabilities document of a given layer
        OR to store the information of the parsed iso metadata which was linked in the capabilities document.
    """
    described_object = models.OneToOneField(to=Service,
                                            on_delete=models.CASCADE,
                                            editable=False,
                                            related_name="metadata",
                                            related_query_name="metadata",
                                            verbose_name=_("described service"),
                                            help_text=_("choice the service you want to describe with this metadata"))
    ...

disadvantages

• reading data from db: everytime a service (or layer/featuretype) shall be displayed by a table (or something else), the metadata attributes like title must be collected by a join on the related metadata object
• writing data to db: two different instances of different object types needed to be stored
• misunderstanding issues like what is a layer metadata, what is a featuretype metadata. This data modeling matches no real world problem

simplification

1. Merge the models together

... so that the Layer model for example holds all information (the layer information and metadata information)

code example

Switch to abstract metadata models like:

class LayerMetadata(AbstractMetadata):
    """ no longer needed
    described_object = models.OneToOneField(to=Layer,
                                            on_delete=models.CASCADE,
                                            related_name="metadata",
                                            related_query_name="metadata",
                                            verbose_name=_("described layer"),
                                            help_text=_("choice the layer you want to describe with this metadata"))
    """
    ...
    preview_image = models.ImageField(null=True, blank=True)
    ...

    class Meta:
        abstract = True

class Layer(LayerMetadata, ServiceElement, MPTTModel):
   ...

2. Decouple information collection from service registry

Provide one model to store MD_Metadata records for hierachyLevel service and dataset.

There are different ways where this records are coming from:

1. harvested by a remote service
2. created by a user

advantages

• decouples the information collection fully from service registry.

@MrSnyder lets discuss this again on coming monday to convert it to a well formed issue