Type mapping for Collections bbox

[ccancellieri]

I'm proceeding into the development of the collections search and I encountered the following schema definition:

"extent.spatial.bbox": {"type": "long"},
 "extent.temporal.interval": {"type": "date"},

stac-fastapi-elasticsearch-opensearch/stac_fastapi/opensearch/stac_fastapi/opensearch/database_logic.py

Line 141 in 106fabb

"extent.spatial.bbox": {"type": "long"},

This is preventing me to apply any spatial filter on this field, so I was wondering if we can consider to change this schema.

The standard says:
https://github.com/radiantearth/stac-spec/blob/master/collection-spec/collection-spec.md#spatial-extent-object

So it should be a [[number]] which may be mapped to a long or double rater than a long, don't you think so?
https://www.elastic.co/guide/en/elasticsearch/reference/current/number.html

Moreover the standard talks about interval as [[string|null]] ( https://github.com/radiantearth/stac-spec/blob/master/collection-spec/collection-spec.md#temporal-extent-object )
and this has been mapped as 'date' instead.

This is great so we can search and filter by date not using generic strings.

But so, why don't we map the spatial extent with a geo_shape?

"extent.spatial.bbox": {"type": "geo_shape"},

so we can start properly search collections with s_intersection filters?

Thanks

[jamesfisher-geo]

This sounds like a good approach to me. Here are a couple thoughts:

• You'll probably need some code to reformat the bounding box from STAC to a format that ES/OS can index into a geo_shape. It looks like ES/OS can index a WKT BBOX to a geo_shape, or you could convert to a geojson polygon and index that. https://www.elastic.co/guide/en/elasticsearch/reference/current/geo-shape.html#:~:text=The%20geo_shape%20data%20type%20facilitates,necessary%20to%20use%20this%20mapping.
• There would also need to be code to translate the geo_shape back to a bbox to fit the STAC spec
• The collection spatial extent is actually a list of bounding boxes. Not sure if that will cause an issue, or if you can search based on only the first bbox geo_shape in the list.

[ccancellieri]

Thanks @jamesfisher-geo for your reply.

Let me try to reply to your points:
Point 1: I think we could reuse what has been done for the items, it may not be a challenge.
Point 2: same as above.
Point 3: well this is actually an interesting point, I think we should also look on how the time is defined but definitely this could add a level of complexity to properly manage it:
We will first need to create an index that includes nested fields for our catalog structure and geo_shape fields for geospatial data. This will allow us to perform properly searches on each bbox of the list but it's a quite complex approach.
Another option, probably easier, is to encode the multiple extents as a single multi-polygon (multiple bboxes) so the search will be accurate and performed on all of the extents in one shot.

Experimenting More:

I'm currently testing my extension with:

{"limit":120,
"filter-lang":"cql2-text",
"filter":"S_INTERSECTS(\"extent.spatial.bbox\", BBOX(\"-73.80811779512624, -43.580390855607845, 80.0639090544287, 43.068887774169625\"))",
"sortBy":[{"field":"id","direction":"asc"}]}

And this is the resulting query:

{
 'geo_shape': 
 {
  'extent.spatial.bbox': 
  {
   'shape': {
     'function': 'bbox',
     'args': [{'property': '-73.80811779512624, -43.580390855607845, 80.0639090544287, 43.068887774169625'}]}, 
     'relation': 'intersects'
   }
  }
}

It is unfortunately not working probably because of the type of the extent (not geo_shape):

File "/usr/local/lib/python3.10/site-packages/stac_fastapi/api/routes.py", line 65, in _endpoint
    return _wrap_response(await func(request_data, request=request))
  File "/app/stac_fastapi/core/stac_fastapi/core/core.py", line 1045, in post_all_collections
    return await self.post_search(search_request=search_request, request=request)
  File "/app/stac_fastapi/core/stac_fastapi/core/core.py", line 632, in post_search
    items, maybe_count, next_token = await self.database.execute_search(
  File "/app/stac_fastapi/elasticsearch/stac_fastapi/elasticsearch/database_logic.py", line 704, in execute_search
    es_response = await search_task
  File "/usr/local/lib/python3.10/site-packages/elasticsearch/_async/client/__init__.py", line 3735, in search
    return await self.perform_request(  # type: ignore[return-value]
  File "/usr/local/lib/python3.10/site-packages/elasticsearch/_async/client/_base.py", line 320, in perform_request
    raise HTTP_EXCEPTIONS.get(meta.status, ApiError)(
elasticsearch.BadRequestError: BadRequestError(400, 'illegal_argument_exception', 'Required [type]')