fix: remove use of JSON LD from CQL JSON parsing (#316)

* fix: remove use of JSON LD from CQL JSON parsing * Change all filtering to use FILTER so as to avoid equality issues for numeric and string comparisons. e.g. FILTER(?var = 2000) will match on 2000, "2000"^^xsd:decimal etc. whereas VALUES ?var { 2000 } will only match on the exact same value. Use a triple pattern match for "=" filtering where the RHS operand is a URI. This should be more performant. * Make enums optional in SHACL defined CQL Queryables. Black code. * Add geo:hasGeometry geo:asWKT sequence path to OGCFeaturesAllProps profile in order to support geometries that are specified on named node sequence paths (rather than the common specification of geometries using blank nodes).
RDFLib · Jan 8, 2025 · 59ba6a8 · 59ba6a8
1 parent 165ee61
commit 59ba6a8
Show file tree

Hide file tree

Showing 117 changed files with 382 additions and 374 deletions.
diff --git a/dev/dev-setup.py b/dev/dev-setup.py
@@ -50,4 +50,4 @@ def setup():
             print(f"Successfully loaded {file_name}")
 
 
-setup()
+setup()
diff --git a/main.py b/main.py
@@ -10,5 +10,11 @@
 
     port = int(environ.get("PREZ_DEV_SERVER_PORT", 8000))
 
-    uvicorn.run("prez.app:assemble_app", factory=True, port=port, reload=True,
-                proxy_headers=True, forwarded_allow_ips='*')
+    uvicorn.run(
+        "prez.app:assemble_app",
+        factory=True,
+        port=port,
+        reload=True,
+        proxy_headers=True,
+        forwarded_allow_ips="*",
+    )
diff --git a/prez/dependencies.py b/prez/dependencies.py
@@ -26,7 +26,10 @@
     SearchMethod,
     SPARQLQueryMediaType,
 )
-from prez.exceptions.model_exceptions import NoEndpointNodeshapeException, URINotFoundException
+from prez.exceptions.model_exceptions import (
+    NoEndpointNodeshapeException,
+    URINotFoundException,
+)
 from prez.enums import SearchMethod
 from prez.models.query_params import QueryParams
 from prez.reference_data.prez_ns import ALTREXT, EP, OGCE, OGCFEAT, ONT
@@ -150,12 +153,7 @@ async def cql_post_parser_dependency(
 ) -> CQLParser:
     try:
         body = await request.json()
-        context = json.load(
-            (Path(__file__).parent / "reference_data/cql/default_context.json").open()
-        )
-        cql_parser = CQLParser(
-            cql=body, context=context, queryable_props=queryable_props
-        )
+        cql_parser = CQLParser(cql=body, queryable_props=queryable_props)
         cql_parser.generate_jsonld()
         cql_parser.parse()
         return cql_parser
@@ -175,16 +173,12 @@ async def cql_get_parser_dependency(
         try:
             crs = query_params.filter_crs
             query = json.loads(query_params.filter)
-            context = json.load(
-                (
-                    Path(__file__).parent / "reference_data/cql/default_context.json"
-                ).open()
-            )
-            cql_parser = CQLParser(
-                cql=query, context=context, crs=crs, queryable_props=queryable_props
-            )
+            cql_parser = CQLParser(cql=query, crs=crs, queryable_props=queryable_props)
             cql_parser.generate_jsonld()
-            cql_parser.parse()
+            try:
+                cql_parser.parse()
+            except Exception as e:
+                raise e
             return cql_parser
         except json.JSONDecodeError:
             raise HTTPException(status_code=400, detail="Invalid JSON format.")

diff --git a/prez/examples/cql/geo_contains_inverse.json b/prez/examples/cql/geo_contains_inverse.json
@@ -52,9 +52,9 @@
       "op": "=",
       "args": [
         {
-          "property": "^http://www.w3.org/2000/01/rdf-schema#member"
+          "property": "http://www.w3.org/2000/01/rdf-schema#member"
         },
-        { "@id": "http://example.com/datasets/sandgate/facilities" }
+        "http://example.com/datasets/sandgate/facilities"
       ]
     }
   ]

diff --git a/prez/examples/cql/geo_crosses.json b/prez/examples/cql/geo_crosses.json
@@ -1,35 +1,17 @@
 {
   "op": "s_crosses",
   "args": [
-    {
-      "property": "geometry"
-    },
+    { "property": "geometry" },
     {
       "type": "LineString",
-      "coordinates": [
-        [
-          [
-            153.06307,
-            -27.3151243
-          ],
-          [
-            153.069877,
-            -27.3151243
-          ],
-          [
-            153.069877,
-            -27.2859541
-          ],
-          [
-            153.06307,
-            -27.2859541
-          ],
-          [
-            153.06307,
-            -27.3151243
-          ]
-        ]
-      ]
+      "coordinates": [ [ 172.03086, 1.5 ],
+                       [ 1.1, -90.0 ],
+                       [ -159.757695, 0.99999 ],
+                       [ -180.0, 0.5 ],
+                       [ -12.111235, 81.336403 ],
+                       [ -0.5, 64.43958 ],
+                       [ 0.0, 81.991815 ],
+                       [ -155.93831, 90.0 ] ]
     }
   ]
-}
+}
diff --git a/prez/reference_data/profiles/ogc_features.ttl b/prez/reference_data/profiles/ogc_features.ttl
@@ -87,7 +87,14 @@ prez:OGCFeaturesAllProps a
         "application/rdf+xml",
         "text/anot+turtle",
         "text/turtle" ;
-    sh:property [ sh:path shext:allPredicateValues ] ;
+    sh:property [
+        sh:path (
+            sh:union (
+                     shext:allPredicateValues
+                     ( geo:hasGeometry geo:asWKT )
+                   )
+          )
+        ] ;
     shext:bnode-depth 2 ;
 .
 

diff --git a/prez/services/listings.py b/prez/services/listings.py
@@ -186,9 +186,10 @@ async def ogc_features_listing_function(
         )
         if queryables:  # from shacl definitions
             content = io.BytesIO(
-                queryables.model_dump_json(exclude_none=True, by_alias=True).encode(
-                    "utf-8"
-                )
+                queryables.model_dump_json(
+                    exclude_none=True,
+                    by_alias=True,
+                ).encode("utf-8")
             )
         else:
             queryable_var = Var(value="queryable")
@@ -264,7 +265,9 @@ async def ogc_features_listing_function(
         # queries.append(feature_collection_query)
         # run the feature_collection_query by itself with caching as it will be the same for all paginated sets of features
 
-        fc_item_graph = await _cached_feature_collection_query(collection_uri, data_repo, feature_collection_query)
+        fc_item_graph = await _cached_feature_collection_query(
+            collection_uri, data_repo, feature_collection_query
+        )
 
     link_headers = None
     if selected_mediatype == "application/sparql-query":
@@ -484,14 +487,16 @@ async def generate_queryables_from_shacl_definition(
     CONSTRUCT {
     ?queryable cql:id ?id ;
     	cql:name ?title ;
+    	cql:description ?description ;
     	cql:datatype ?type ;
     	cql:enum ?enums .
     }
     WHERE {?queryable a cql:Queryable ;
         dcterms:identifier ?id ;
         sh:name ?title ;
-        sh:datatype ?type ;
-        sh:in/rdf:rest*/rdf:first ?enums ;
+        sh:description ?description ;
+        sh:datatype ?type .
+        OPTIONAL { ?queryable sh:in/rdf:rest*/rdf:first ?enums }
     }
     """
     g, _ = await system_repo.send_queries([query], [])
@@ -511,11 +516,15 @@ async def generate_queryables_from_shacl_definition(
             ].split("#")[
                 -1
             ],  # hack
-            "enum": [
-                enum_item["@id"]
-                for enum_item in item["http://www.opengis.net/doc/IS/cql2/1.0/enum"]
-            ],
+            "description": item["http://www.opengis.net/doc/IS/cql2/1.0/description"][
+                0
+            ]["@value"],
         }
+        enum = item.get(
+            "http://www.opengis.net/doc/IS/cql2/1.0/enum"
+        )  # enums are optional.
+        if enum:
+            queryable_props[id_value]["enum"] = [enum_item["@id"] for enum_item in enum]
     if endpoint_uri == OGCFEAT["queryables-global"]:
         title = "Global Queryables"
         description = (
@@ -536,9 +545,9 @@ async def generate_queryables_from_shacl_definition(
 
 
 @cached(ttl=600, key=lambda collection_uri: collection_uri)
-async def _cached_feature_collection_query(collection_uri, data_repo, feature_collection_query):
+async def _cached_feature_collection_query(
+    collection_uri, data_repo, feature_collection_query
+):
     """cache the feature collection information for 10 minutes as it is an expensive query at present"""
-    fc_item_graph, _ = await data_repo.send_queries(
-        [feature_collection_query], []
-    )
+    fc_item_graph, _ = await data_repo.send_queries([feature_collection_query], [])
     return fc_item_graph
Original file line number	Diff line number	Diff line change
Expand Up		@@ -50,4 +50,4 @@ def setup():
		print(f"Successfully loaded {file_name}")


		setup()
		setup()