Skip to content

stac.md

Latest commit

 

History

History
34 lines (22 loc) · 2.47 KB

stac.md

File metadata and controls

34 lines (22 loc) · 2.47 KB

The STAC Extension

The stac extension provides a SpatioTemporal Asset Catalog (STAC) compliant interface. Its philosophy is to act as a rich, standardized, read-only discovery layer on top of the data managed by the core modules.

It is not intended for data modification. It purely exists to enable powerful search and discovery workflows for STAC-aware clients.

The stac_generator

The core logic of this extension resides in the stac_generator. Its operations manage the translation from Agro-Informatics Platform (AIP) - Catalog Services's internal, generic models to STAC-compliant JSON objects via the pystac standard library.

Workflow Example (get_stac_collection)

  1. Fetches the generic Collection model by calling catalog_module.get_collection.
  2. Executes parallel calls against shared_queries to fetch granular, dynamic DB queries computing true real-time spatial and temporal extents (bypassing static declarations).
  3. Punts everything to the generator. pystac builds a valid Collection map object merging dynamic endpoints and applying correct JSON schemas linking back into self and root objects establishing navigating structure.

STAC Search — Spatial Filter Implementation

POST /stac/search supports bbox and intersects filters. These are implemented as raw SQL clauses that reference the geom column in the geometry sidecar. Because the QueryOptimizer only JOINs a sidecar when its fields appear in SELECT, WHERE, or ORDER BY, the search layer forces the geometry sidecar into the query plan by appending a cheap selection:

FieldSelection(field="geom", transformation="ST_SRID", alias="_srid")

ST_SRID is included in the ALLOWED_TRANSFORMATIONS allowlist in query_builder.py. Removing it from the allowlist (or forgetting to add it) causes all spatial searches to return 400 Bad Request.

The asset_factory

A key feature of the generator is the add_dynamic_assets function. This is a forward-looking mechanism for service chaining.

If additional extensions are active in the system environment (for example, a tiles API):

  1. The generator iterates the context before finalize.
  2. Identifies a physical data pipeline capability.
  3. Automatically synthesizes a direct JSON HTTP link asset node in the collection root pointing directly at .../tiles/{collection_id}/{z}/{x}/{y}.

This creates self-assembling ecosystems where client scanners can autonomously figure out all ways to interact with data representations.