fix: remove use of JSON LD from CQL JSON parsing

prez/dependencies.py

@@ -150,11 +150,8 @@ 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=body, queryable_props=queryable_props
         )
         cql_parser.generate_jsonld()
         cql_parser.parse()
@@ -175,16 +172,14 @@ 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=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.")

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"
       ]
     }
   ]

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 ] ]
     }
   ]
-}
+}

prez/repositories/base.py

@@ -27,14 +27,17 @@ async def send_queries(
         tabular_queries: List[Tuple[URIRef | None, str]] = None,
     ) -> Tuple[Graph, List]:
         # Common logic to send both query types in parallel
-        results = await asyncio.gather(
-            *[self.rdf_query_to_graph(query) for query in rdf_queries if query],
-            *[
-                self.tabular_query_to_table(query, context)
-                for context, query in tabular_queries
-                if query
-            ],
-        )
+        try:
+            results = await asyncio.gather(
+                *[self.rdf_query_to_graph(query) for query in rdf_queries if query],
+                *[
+                    self.tabular_query_to_table(query, context)
+                    for context, query in tabular_queries
+                    if query
+                ],
+            )
+        except Exception as e:
+            print(e)
         g = Graph(namespace_manager=prefix_graph.namespace_manager)
         tabular_results = []
         for result in results:

prez/services/query_generation/cql.py

@@ -89,7 +89,6 @@ class CQLParser:
     def __init__(
         self,
         cql=None,
-        context: dict = None,
         cql_json: dict = None,
         crs=None,
         queryable_props=None,
@@ -98,7 +97,6 @@ def __init__(
         self.inner_select_gpnt_list = None
         self.inner_select_vars: list[Var] = []
         self.cql: dict = cql
-        self.context = context
         self.cql_json = cql_json
         self.var_counter = 0
         self.query_object = None
@@ -110,8 +108,7 @@ def __init__(
         self.queryable_props = queryable_props
 
     def generate_jsonld(self):
-        combined = {"@context": self.context, **self.cql}
-        self.cql_json = jsonld.expand(combined, options={"base": "h"})[0]
+        self.cql_json = self.cql
 
     def parse(self):
         root = self.cql_json
@@ -142,8 +139,8 @@ def parse(self):
     def parse_logical_operators(
         self, element, existing_ggps=None
     ) -> Generator[GroupGraphPatternSub, None, None]:
-        operator = element.get(str(CQL.operator))[0].get("@value")
-        args = element.get(str(CQL.args))
+        operator = element.get("op")
+        args = element.get("args")
 
         ggps = existing_ggps if existing_ggps is not None else GroupGraphPatternSub()
 
@@ -208,11 +205,11 @@ def _add_triple(self, ggps, subject, predicate, object):
     def _handle_comparison(self, operator, args, existing_ggps=None):
         ggps, object = self._add_tss_tssp(args, existing_ggps)
 
-        val = args[1].get("@value")
+        val = args[1]
         if not val:  # then should be an IRI
-            val = args[1].get("@id")
+            val = args[1]
             value = IRI(value=val)
-        elif val.startswith("http"):  # hack
+        elif isinstance(val, str) and val.startswith("http"):  # hack
             value = IRI(value=val)
         elif isinstance(val, str):  # literal string
             value = RDFLiteral(value=val)
@@ -240,7 +237,7 @@ def _handle_comparison(self, operator, args, existing_ggps=None):
     def _add_tss_tssp(self, args, existing_ggps):
         self.var_counter += 1
         ggps = existing_ggps if existing_ggps is not None else GroupGraphPatternSub()
-        prop = args[0].get(str(CQL.property))[0].get("@id")
+        prop = args[0].get("property")
         if prop in self.queryable_props:
             object = self._handle_shacl_defined_prop(prop)
         else:
@@ -255,7 +252,6 @@ def _handle_like(self, args, existing_ggps=None):
 
         value = (
             args[1]
-            .get("@value")
             .replace("%", ".*")
             .replace("_", ".")
             .replace("\\", "\\\\")
@@ -286,16 +282,16 @@ def _handle_spatial(self, operator, args, existing_ggps=None):
         self.var_counter += 1
         ggps = existing_ggps if existing_ggps is not None else GroupGraphPatternSub()
 
-        coordinates_list = args[1].get("http://example.com/vocab/coordinates")
-        coordinates, geom_type = self._extract_spatial_info(coordinates_list, args)
-        if geom_type in ["Polygon", "MultiPolygon"]:
-            coordinates = [coordinates]
+        coordinates = args[1].get("coordinates")
+        geom_type = args[1].get("type")
+        if args[1].get("bbox"):
+            geom_type = "Polygon"
 
         if coordinates:
             wkt = get_wkt_from_coords(coordinates, geom_type)
             wkt_with_crs = f"<{self.crs}> {wkt}"
-            prop = args[0].get(str(CQL.property))[0].get("@id")
-            if prop == "http://example.com/geometry":
+            prop = args[0].get("property")
+            if prop == "geometry":
                 subject = Var(value="focus_node")
             else:
                 subject = IRI(value=prop)
@@ -327,11 +323,17 @@ def _handle_spatial(self, operator, args, existing_ggps=None):
     def _handle_in(self, args, existing_ggps=None):
         ggps, object = self._add_tss_tssp(args, existing_ggps)
 
-        literal_values = [item["@value"] for item in args if "@value" in item]
-        uri_values = [item["@id"] for item in args if "@id" in item]
-        for i, lit_val in enumerate(literal_values):
-            if lit_val.startswith("http"):  # hack
-                uri_values.append(literal_values.pop(i))
+        uri_values = []
+        literal_values = []
+        numeric_values = []
+        for arg in args[1]:
+            if isinstance(arg, str) and arg.startswith("http"):
+                uri_values.append(arg)
+            elif isinstance(arg, (int, float)):
+                numeric_values.append(arg)
+            else:
+                literal_values.append(arg)
+
         grammar_uri_values = [IRI(value=URIRef(value)) for value in uri_values]
         grammar_literal_values = []
         for val in literal_values:
@@ -367,15 +369,16 @@ def _extract_spatial_info(self, coordinates_list, args):
         coordinates = []
         geom_type = None
         if coordinates_list:
-            coordinates = [
-                [coordinates_list[i]["@value"], coordinates_list[i + 1]["@value"]]
-                for i in range(0, len(coordinates_list), 2)
-            ]
-            geom_type = args[1]["http://www.opengis.net/ont/sf#type"][0]["@value"]
-        bbox_list = args[1].get("http://example.com/vocab/bbox")
+            # coordinates = [
+            #     [coordinates_list[i], coordinates_list[i + 1]]
+            #     for i in range(0, len(coordinates_list), 2)
+            # ]
+            coordinates = coordinates_list
+            geom_type = args[1].get("type")
+        bbox_list = args[1].get("bbox")
         if bbox_list:
             geom_type = "Polygon"
-            bbox_values = [item["@value"] for item in bbox_list]
+            bbox_values = [item for item in bbox_list]
             coordinates = format_coordinates_as_wkt(bbox_values, coordinates)
         return coordinates, geom_type
 
@@ -390,35 +393,34 @@ def _handle_temporal(self, comp_func, args, existing_ggps=None):
         operands = {}
         for i, arg in enumerate(args, start=1):
             # check if the arg is an interval
-            interval_list = arg.get(str(CQL.interval))
+            interval_list = arg.get("interval")
             if interval_list:
                 for n, item in enumerate(interval_list):
                     label = "start" if n == 0 else "end"
-                    prop = item.get(str(CQL.property))
-                    if prop:
-                        self._triple_for_time_prop(ggps, i, label, prop, operands)
-                    date_val = item.get("@value")
-                    if date_val:
-                        self._dt_to_rdf_literal(i, date_val, label, operands)
+                    if isinstance(item, dict):
+                        prop = item.get("property")
+                        if prop:
+                            self._triple_for_time_prop(ggps, i, label, prop, operands)
+                    elif isinstance(item, str):
+                        self._dt_to_rdf_literal(i, item, label, operands)
                 continue
 
             # handle instants - prop and date
             label = "instant"
             # check if the arg is a property
-            prop = arg.get(str(CQL.property))
+            prop = arg.get("property")
             if prop:
                 self._triple_for_time_prop(ggps, i, label, prop, operands)
                 continue
 
             # check if the arg is a date
             date = (
-                arg.get(str(CQL.date))
-                or arg.get(str(CQL.datetime))
-                or arg.get(str(CQL.timestamp))
+                arg.get("date")
+                or arg.get("datetime")
+                or arg.get("timestamp")
             )
             if date:
-                date_val = date[0].get("@value")
-                self._dt_to_rdf_literal(i, date_val, label, operands)
+                self._dt_to_rdf_literal(i, date, label, operands)
 
         gpnt = self.process_temporal_function(comp_func, operands)
 
@@ -479,28 +481,11 @@ def process_temporal_function(self, comp_func, operands):
             )
 
     def _triple_for_time_prop(self, ggps, i, label, prop, operands):
-        prop_uri = prop[0].get("@id")
-        value = IRI(value=prop_uri)
+        value = IRI(value=prop)
         var = Var(value=f"dt_{i}_{label}")
         operands[f"t{i}_{label}"] = var
         self._add_triple(ggps, Var(value="focus_node"), value, var)
 
-    def _handle_interval_list(self, all_args, comparator_args, interval_list):
-        for item in interval_list:
-            if item.get(str(CQL.property)):
-                prop = item.get(str(CQL.property))[0].get("@id")
-                comparator_args.append(IRI(value=prop))
-            elif item.get("@value"):
-                val = item.get("@value")
-                # self._dt_to_rdf_literal(comparator_args, val)
-                dt, _ = parse_datetime(val)
-                comparator_args.append(
-                    RDFLiteral(
-                        value=dt.isoformat(),
-                        datatype=IRI(value="http://www.w3.org/2001/XMLSchema#dateTime"),
-                    )
-                )
-        all_args.append(comparator_args)
 
     def _dt_to_rdf_literal(self, i, dt_str, label, operands):
         if dt_str == "..":

test_data/cql/expected_generated_queries/additional_temporal_disjoint_instant.rq

@@ -1,7 +1,7 @@
 CONSTRUCT {
-?focus_node <http://example.com/updated_at> ?dt_1_instant
+?focus_node <ex:updated_at> ?dt_1_instant
 }
 WHERE {
-?focus_node <http://example.com/updated_at> ?dt_1_instant
+?focus_node <ex:updated_at> ?dt_1_instant
 FILTER (?dt_1_instant > "2012-08-10T05:30:00+00:00"^^<http://www.w3.org/2001/XMLSchema#dateTime> || ?dt_1_instant < "2012-08-10T05:30:00+00:00"^^<http://www.w3.org/2001/XMLSchema#dateTime>)
 }

test_data/cql/expected_generated_queries/additional_temporal_during_intervals.rq

@@ -1,10 +1,10 @@
 CONSTRUCT {
-?focus_node <http://example.com/ends_at> ?dt_1_end .
-?focus_node <http://example.com/starts_at> ?dt_1_start
+?focus_node <ex:ends_at> ?dt_1_end .
+?focus_node <ex:starts_at> ?dt_1_start
 }
 WHERE {
-?focus_node <http://example.com/ends_at> ?dt_1_end .
-?focus_node <http://example.com/starts_at> ?dt_1_start
+?focus_node <ex:ends_at> ?dt_1_end .
+?focus_node <ex:starts_at> ?dt_1_start
 
 FILTER (?dt_1_start > "2017-06-10T07:30:00+00:00"^^<http://www.w3.org/2001/XMLSchema#dateTime> && ?dt_1_end < "2017-06-11T10:30:00+00:00"^^<http://www.w3.org/2001/XMLSchema#dateTime>)
 }

test_data/cql/expected_generated_queries/clause7_12.rq

@@ -1,7 +1,7 @@
 CONSTRUCT {
-?focus_node <http://example.com/event_time> ?dt_1_instant
+?focus_node <ex:event_time> ?dt_1_instant
 }
 WHERE {
-?focus_node <http://example.com/event_time> ?dt_1_instant
+?focus_node <ex:event_time> ?dt_1_instant
 FILTER (! (?dt_1_instant > "1969-07-24T16:50:35+00:00"^^<http://www.w3.org/2001/XMLSchema#dateTime> || ?dt_1_instant < "1969-07-16T05:32:00+00:00"^^<http://www.w3.org/2001/XMLSchema#dateTime>))
 }

test_data/cql/expected_generated_queries/clause7_13.rq

@@ -1,10 +1,10 @@
 CONSTRUCT {
-?focus_node <http://example.com/liftOff> ?dt_1_end .
-?focus_node <http://example.com/touchdown> ?dt_1_start
+?focus_node <ex:liftOff> ?dt_1_end .
+?focus_node <ex:touchdown> ?dt_1_start
 }
 WHERE {
-?focus_node <http://example.com/liftOff> ?dt_1_end .
-?focus_node <http://example.com/touchdown> ?dt_1_start
+?focus_node <ex:liftOff> ?dt_1_end .
+?focus_node <ex:touchdown> ?dt_1_start
 
 FILTER (?dt_1_start > "1969-07-16T13:32:00+00:00"^^<http://www.w3.org/2001/XMLSchema#dateTime> && ?dt_1_end < "1969-07-24T16:50:35+00:00"^^<http://www.w3.org/2001/XMLSchema#dateTime>)
 }