add framework, WIP

prez/dependencies.py

@@ -35,7 +35,7 @@
 from prez.services.query_generation.concept_hierarchy import ConceptHierarchyQuery
 from prez.services.query_generation.cql import CQLParser
 from prez.services.query_generation.search_default import SearchQueryRegex
-from prez.services.query_generation.search_fts import SearchQueryFusekiFTS
+from prez.services.query_generation.search_fuseki_fts import SearchQueryFusekiFTS
 from prez.services.query_generation.shacl import NodeShape
 from prez.services.query_generation.sparql_escaping import escape_for_lucene_and_sparql
 
@@ -188,7 +188,10 @@ async def cql_get_parser_dependency(
             )
 
 
-async def generate_search_query(request: Request):
+async def generate_search_query(
+        request: Request,
+        system_repo: Repo = Depends(get_system_repo),
+):
     term = request.query_params.get("q")
     if term:
         escaped_term = escape_for_lucene_and_sparql(term)
@@ -206,8 +209,26 @@ async def generate_search_query(request: Request):
                 offset=offset,
             )
         elif settings.search_method == SearchMethod.FTS_FUSEKI:
+            predicates = predicates if predicates else settings.search_predicates
+            shacl_shape_predicates = []
+            non_shacl_predicates = []
+            for pred in predicates:
+                try:
+                    URIRef(pred)
+                    non_shacl_predicates.append(pred)
+                except ValueError:
+                    shacl_shape_predicates.append(pred)
+                # print('9')
+                # shacl_pred_query = f"""
+                # DESCRIBE ?shape {{
+                #     ?shape a <https://prez.dev/ont/fts#> ;
+                #       <http://purl.org/dc/terms/identifier> {}.
+                # }}"""
+                # for predicate in shacl_shape_predicates:
+                #     pass
+            fts_gpnt = []
             return SearchQueryFusekiFTS(
-                term=escaped_term, predicates=predicates, limit=limit, offset=offset
+                term=escaped_term, predicates=predicates, limit=limit, offset=offset, fts_gpnt=fts_gpnt
             )
         else:
             raise NotImplementedError(

prez/services/query_generation/shacl.py

@@ -269,7 +269,7 @@ def optional_gpnt(depth):
 class PropertyShape(Shape):
     uri: URIRef | BNode  # URI of the shape
     graph: Graph
-    kind: TypingLiteral["endpoint", "profile"]
+    kind: TypingLiteral["endpoint", "profile", "fts"]
     focus_node: Union[IRI, Var]
     # inputs
     shape_number: int = 0
@@ -385,7 +385,7 @@ def to_grammar(self):
         # label nodes in the inner select and profile part of the query differently for clarity.
         if self.kind == "endpoint":
             path_or_prop = "path"
-        elif self.kind == "profile":
+        elif (self.kind == "profile") or (self.kind == "fts"):
             path_or_prop = f"prof_{self.shape_number + 1}"
 
         # set up the path nodes - either from supplied values or set as variables
@@ -535,17 +535,27 @@ def process_property_paths(self, property_paths, path_or_prop, tssp_list, pp_i):
                 else:
                     pred = IRI(value=property_path.value)
                     obj = path_node_1
-                triple = (self.focus_node, pred, obj)
+                if self.kind == "fts":
+                    triple = (self.focus_node, pred, Var(value="fts_search_node"))
+                else:
+                    triple = (self.focus_node, pred, obj)
                 self.tss_list.append(TriplesSameSubject.from_spo(*triple))
                 current_tssp.append(TriplesSameSubjectPath.from_spo(*triple))
                 pp_i += 1
 
             elif isinstance(property_path, InversePath):
-                triple = (
-                    path_node_1,
-                    IRI(value=property_path.value.value),
-                    self.focus_node,
-                )
+                if self.kind == "fts":
+                    triple = (
+                        Var(value="fts_search_node"),
+                        IRI(value=property_path.value.value),
+                        self.focus_node,
+                    )
+                else:
+                    triple = (
+                        path_node_1,
+                        IRI(value=property_path.value.value),
+                        self.focus_node,
+                    )
                 self.tss_list.append(TriplesSameSubject.from_spo(*triple))
                 current_tssp.append(TriplesSameSubjectPath.from_spo(*triple))
                 pp_i += 1
@@ -568,13 +578,14 @@ def process_property_paths(self, property_paths, path_or_prop, tssp_list, pp_i):
 
             elif isinstance(property_path, SequencePath):
                 preds_pathmods_inverse = []
+                seq_path_len = len(property_path.value)
                 for j, path in enumerate(property_path.value):
                     if isinstance(path, Path):
                         if self.kind == "endpoint":
                             preds_pathmods_inverse.append(
                                 (IRI(value=path.value), None, False)
                             )
-                        elif self.kind == "profile":
+                        elif (self.kind == "profile") or (self.kind == "fts"):
                             if j == 0:
                                 triple = (
                                     self.focus_node,
@@ -592,7 +603,7 @@ def process_property_paths(self, property_paths, path_or_prop, tssp_list, pp_i):
                             preds_pathmods_inverse.append(
                                 (IRI(value=path.value.value), None, True)
                             )
-                        elif self.kind == "profile":
+                        elif (self.kind == "profile") or (self.kind == "fts"):
                             if j == 0:
                                 triple = (
                                     path_node_1,
@@ -619,6 +630,14 @@ def process_property_paths(self, property_paths, path_or_prop, tssp_list, pp_i):
                     if self.kind == "profile":
                         self.tss_list.append(TriplesSameSubject.from_spo(*triple))
                         current_tssp.append(TriplesSameSubjectPath.from_spo(*triple))
+                    elif self.kind == "fts":
+                        if j == seq_path_len - 1:  # we're at the end of the sequence path
+                            new_triple = triple[:2] + (Var(value="fts_search_node"),)
+                            self.tss_list.append(TriplesSameSubject.from_spo(*new_triple))
+                            current_tssp.append(TriplesSameSubjectPath.from_spo(*new_triple))
+                        else:
+                            self.tss_list.append(TriplesSameSubject.from_spo(*triple))
+                            current_tssp.append(TriplesSameSubjectPath.from_spo(*triple))
                 pp_i += len(property_path.value)
                 if self.kind == "endpoint":
                     tssp = _tssp_for_sequence(

prez/services/query_generation/umbrella.py

@@ -30,7 +30,7 @@
 from prez.services.query_generation.cql import CQLParser
 from prez.services.query_generation.datetime_filter import generate_datetime_filter
 from prez.services.query_generation.search_default import SearchQueryRegex
-from prez.services.query_generation.search_fts import SearchQueryFusekiFTS
+from prez.services.query_generation.search_fuseki_fts import SearchQueryFusekiFTS
 from prez.services.query_generation.shacl import NodeShape
 
 

test_data/fts_property_shapes.ttl

@@ -0,0 +1,46 @@
+@prefix fts: <https://prez.dev/ont/fts#> .
+@prefix dcterms: <http://purl.org/dc/terms/> .
+@prefix dwc: <http://rs.tdwg.org/dwc/terms/> .
+@prefix ex: <http://example.com/> .
+@prefix sh: <http://www.w3.org/ns/shacl#> .
+@prefix sname: <https://fake-scientific-name-id.com/name/afd/> .
+@prefix sosa: <http://www.w3.org/ns/sosa/> .
+@prefix xsd: <http://www.w3.org/2001/XMLSchema#> .
+@prefix prez: <https://prez.dev/> .
+
+
+ex:FTSSequenceShape
+    a sh:PropertyShape ;
+    a fts:PropertyShape ;
+    sh:path
+        (
+            ex:hasSimpleResult ex:hasFeatureOfInterest ex:testProp ex:testProp2
+        ) ;
+.
+
+ex:FTSInverseShape
+    a sh:PropertyShape ;
+    a fts:PropertyShape ;
+    sh:path [ sh:inversePath ex:hasFeatureOfInterest ] ;
+    sh:name "Inverse" ;
+    dcterms:identifier "inv" ;
+.
+
+ex:FTSSequenceShape
+    a sh:PropertyShape ;
+    a fts:PropertyShape ;
+    sh:path
+        (
+            ex:hasSimpleResult ex:hasFeatureOfInterest ex:testProp ex:testProp2
+        ) ;
+    sh:name "Sequence" ;
+    dcterms:identifier "seq" ;
+.
+
+ex:FTSInverseSequenceShape
+    a sh:PropertyShape ;
+    a fts:PropertyShape ;
+    sh:path ( ex:hasSimpleResult [ sh:inversePath ex:hasFeatureOfInterest ] ) ;
+    sh:name "Sequence Inverse" ;
+    dcterms:identifier "seqinv" ;
+.

tests/test_search_fuseki_fts_class.py

@@ -0,0 +1,35 @@
+from pathlib import Path
+
+from rdflib import RDFS, URIRef, Graph
+from sparql_grammar_pydantic import Var
+
+from prez.services.query_generation.search_fuseki_fts import SearchQueryFusekiFTS
+from prez.services.query_generation.shacl import PropertyShape
+
+
+def test_query_gen():
+    query_obj = SearchQueryFusekiFTS(term="test", limit=10, offset=0, predicates=[RDFS.label, RDFS.comment])
+    query_string = query_obj.to_string()
+    print(query_string)
+
+def test_combo_query_gen():
+    file = Path(__file__).parent.parent / "test_data" / "fts_property_shapes.ttl"
+    ps_g = Graph().parse(file)
+    ps1 = PropertyShape(
+        uri=URIRef("FTSInverseSequenceShape"),
+        graph=ps_g,
+        kind="fts",  # "profile" would expand these out to plain triple pattern matches
+        focus_node=Var(value="focus_node"),
+        shape_number=100
+    )
+    ps2 = PropertyShape(
+        uri=URIRef("FTSInverseSequenceShape"),
+    )
+    query = SearchQueryFusekiFTS(
+        term="test",
+        limit=10,
+        offset=0,
+        predicates=[RDFS.label, RDFS.comment],
+        gpnt_list=gpnt_list
+    )
+    print('')