I have an idea ... datastore.getView( filter ) makes use of ....
Where DefaultView (donated from GeoServer - thanks!) carefully combines queries as needed:
I would like to create something similar this - used to "partition" a large shapefile or database table that has had a type system imposed on it. Specifically there is an attribute (say "type") that is used to partition table contents and mark rows that belong to a specific class; the fact that they all have same attributes is an accident of history; they still represent different content.
I am not sure if this is a common enough case to be useful? It is more that I would like to do this code in GeoTools in the data module rather then hidden away in uDig.
The idea would be to do what DefaultView does; using:
- attribute name to partition on
- value "selected" for the partition being returned
Rather then simply use DefaultView - I would like the result to be a FeatureStore (ie editable) for all rows where type="foo".
So for FeatureStore methods:
- addFeatures - would need to add the "selected" value for the partition attribute to the supplied values (ie type="foo")
- modifyFeatures - would need to include the selected value in the filter
- deleteFeatures - would need to include the selected value in the filter
And for FeatureSource methods:
- getSchema - should hide the partition attribute - and possibly renamed so getSchema indicates what the partition value is
- getFeatures - should strip out the partition attribute
PartitionFeatureStore could be wrapped around any FeatureStore. We would probably need a second implementation to wrap around FeatureSource for consistency.