feat(collection-search): enable filter, sortby, and advanced freetext

Author: alambare

pyproject.toml

@@ -15,6 +15,7 @@ dependencies = [
     "orjson",
     "pydantic",
     "pydantic_core",
+    "cql2",
     "pygeofilter",
     "stac-fastapi.api >= 4.0",
     "stac-fastapi.extensions",
@@ -65,7 +66,13 @@ explicit_package_bases = true
 exclude = ["tests", ".venv"]
 
 [[tool.mypy.overrides]]
-module = ["geojson", "pygeofilter", "pygeofilter.*", "stac_fastapi", "stac_fastapi.*"]
+module = [
+    "geojson",
+    "pygeofilter",
+    "pygeofilter.*",
+    "stac_fastapi",
+    "stac_fastapi.*",
+]
 ignore_missing_imports = true
 
 [tool.pytest.ini_options]

stac_fastapi/eodag/app.py

@@ -38,6 +38,7 @@
 )
 from stac_fastapi.extensions.core import (
     CollectionSearchExtension,
+    CollectionSearchFilterExtension,
     FilterExtension,
     FreeTextExtension,
     QueryExtension,
@@ -114,9 +115,12 @@
 
 # collection_search extensions
 cs_extensions_map = {
-    "offset-pagination": OffsetPaginationExtension(),
-    "collection-search": CollectionSearchExtension(),
-    "free-text": FreeTextExtension(conformance_classes=[FreeTextConformanceClasses.COLLECTIONS]),
+    "sort": SortExtension(conformance_classes=[SortConformanceClasses.COLLECTIONS]),
+    "filter": CollectionSearchFilterExtension(client=FiltersClient()),
+    "free_text": FreeTextExtension(
+        conformance_classes=[FreeTextConformanceClasses.COLLECTIONS],
+    ),
+    "pagination": OffsetPaginationExtension(),
 }
 
 # item_collection extensions
@@ -127,11 +131,14 @@
     "filter": FilterExtension(client=FiltersClient(stac_metadata_model=stac_metadata_model)),
 }
 
+collection_search_extension = CollectionSearchExtension.from_extensions(cs_extensions_map.values())
+
 all_extensions = {
     **search_extensions_map,
     **cs_extensions_map,
     **itm_col_extensions_map,
     **{
+        "collection-search": collection_search_extension,
         "data-download": DataDownload(),
         "collection-order": CollectionOrderExtension(
             client=BaseCollectionOrderClient(stac_metadata_model=stac_metadata_model)
@@ -193,13 +200,6 @@ async def lifespan(app: FastAPI) -> AsyncGenerator[None, None]:
 search_post_model = create_post_request_model(search_extensions)
 search_get_model = create_get_request_model(search_extensions)
 
-collections_model = create_request_model(
-    "CollectionsRequest",
-    base_model=EmptyRequest,
-    extensions=get_enabled_extensions(cs_extensions_map),
-    request_type="GET",
-)
-
 item_collection_model = create_request_model(
     "ItemsRequest",
     base_model=ItemCollectionUri,
@@ -220,7 +220,7 @@ async def lifespan(app: FastAPI) -> AsyncGenerator[None, None]:
     response_class=ORJSONResponse,
     search_get_request_model=search_get_model,
     search_post_request_model=search_post_model,
-    collections_get_request_model=collections_model,
+    collections_get_request_model=collection_search_extension.GET,
     items_get_request_model=item_collection_model,
     middlewares=[
         Middleware(BrotliMiddleware),

stac_fastapi/eodag/core.py

@@ -26,13 +26,12 @@
 from urllib.parse import unquote_plus
 
 import attr
+import cql2
 import orjson
+import pygeofilter
 from fastapi import HTTPException
 from pydantic import ValidationError
 from pydantic_core import InitErrorDetails, PydanticCustomError
-from pygeofilter.backends.cql2_json import to_cql2
-from pygeofilter.parsers.cql2_json import parse as parse_json
-from pygeofilter.parsers.cql2_text import parse as parse_cql2_text
 from stac_fastapi.api.models import create_post_request_model
 from stac_fastapi.types.errors import NotFoundError
 from stac_fastapi.types.requests import get_base_url
@@ -202,7 +201,8 @@ async def all_collections(
         q: Optional[list[str]] = None,
         sortby: Optional[list[str]] = None,
         filter_expr: Optional[str] = None,
-        filter_lang: Optional[str] = "cql2-text",
+        filter_lang: Optional[str] = None,
+        **kwargs: Any,
     ) -> Collections:
         """
         Get all collections from EODAG.
@@ -229,10 +229,11 @@ async def all_collections(
         cql2_json = None
         if filter_expr:
             if filter_lang == "cql2-text":
-                filter_expr = to_cql2(parse_cql2_text(filter_expr))
-                filter_lang = "cql2-json"
-
-            cql2_json = str2json("filter_expr", filter_expr)
+                cql2_json = cql2.parse_text(filter_expr).to_json()
+            elif filter_lang == "cql2-json":
+                cql2_json = str2json("filter_expr", filter_expr)
+            else:
+                raise HTTPException(status_code=400, detail=f"Unsupported filter_lang {filter_lang}")
 
         collections = cast(
             CollectionsList,
@@ -241,10 +242,10 @@ async def all_collections(
                 geometry=bbox,
                 datetime=datetime,
                 limit=limit,
-                q=q,
+                q=" ".join(q) if q else None,
                 cql2_json=cql2_json,
                 sortby=sortby,
-            )
+            ),
         )
 
         number_matched = cast(int, collections.number_matched)
@@ -475,7 +476,7 @@ def _clean_search_args(
         """Clean up search arguments to match format expected by pgstac"""
         if filter_expr:
             if filter_lang == "cql2-text":
-                filter_expr = to_cql2(parse_cql2_text(filter_expr))
+                filter_expr = cql2.parse_text(filter_expr).to_json()
                 filter_lang = "cql2-json"
 
             base_args["filter"] = str2json("filter_expr", filter_expr)
@@ -495,12 +496,10 @@ def _clean_search_args(
             for sort in sortby:
                 sortparts = re.match(r"^([+-]?)(.*)$", sort)
                 if sortparts:
-                    sort_param.append(
-                        {
-                            "field": sortparts.group(2).strip(),
-                            "direction": "desc" if sortparts.group(1) == "-" else "asc",
-                        }
-                    )
+                    sort_param.append({
+                        "field": sortparts.group(2).strip(),
+                        "direction": "desc" if sortparts.group(1) == "-" else "asc",
+                    })
             base_args["sortby"] = sort_param
 
         # Remove None values from dict
@@ -645,7 +644,7 @@ def add_error(error_message: str) -> None:
 
     errors: list[InitErrorDetails] = []
     try:
-        parsing_result = EodagEvaluator().evaluate(parse_json(filter_))  # type: ignore
+        parsing_result = EodagEvaluator().evaluate(pygeofilter.parse(filter_))  # type: ignore
     except (ValueError, NotImplementedError) as e:
         add_error(str(e))
         raise ValidationError.from_exception_data(title="stac-fastapi-eodag", line_errors=errors) from e

stac_fastapi/eodag/dag.py

@@ -51,7 +51,7 @@ def fetch_external_stac_collections(
     ext_stac_collections: dict[str, dict[str, Any]] = {}
 
     for collection in collections:
-        file_path = getattr(collection, "eodag_stac_collection", None)
+        file_path = getattr(collection, "stacCollection", None)
         if not file_path:
             continue
         logger.info(f"Fetching external STAC collection for {collection.id}")