Questions about 'PgSTACReader' and 'SimpleSTACReader'

[louisstuart96]

As far as I have found, 'Searches' and 'Collections' endpoints are built by titiler.pgstac.factory.MosaicTilerFactory, which uses titiler.pgstac.reader.SimpleSTACReader; 'Items' endpoints are built by titiler.core.factory.MultiBaseTilerFactory, which uses titiler.pgstac.reader.PgSTACReader (inherited from rio_tiler.io.stac.STACReader).

What are the differences between these two readers? Our team is trying to make some customizations, like support of HTTP and S3 links in parallel, and handling 'alterate_asset' STAC extensions. Do we need to customize both readers?

[vincentsarago]

• rio_tiler.io.stac.STACReader expects either a Path or a Pystac Item (internally it will fetch the stac Item if only a path is passed)

• titiler.pgstac.reader.PgSTACReader, used in /items expects a single input as a pystac Item

• titiler.pgstac.reader.SimpleSTACReader expects a Dict as input, from pgstac response directly

You'll need to customize each ones but some customization around alternate href is already available https://github.com/cogeotiff/rio-tiler/blob/main/rio_tiler/io/stac.py#L348-L350

I can't remind why it's not done the same in titiler-pgstac but we could certainly bring this into

titiler-pgstac/titiler/pgstac/reader.py

Lines 151 to 155 in d0d2f86

	asset_info = self.input["assets"][asset]
	info = AssetInfo(
	url=asset_info["href"],
	env={},
	)