Adapt upstream Onya changes

Author: uogbuji

demo/kgraph/README.md

@@ -33,7 +33,9 @@ Basic `.onya` format example:
 ```onya
 # @docheader
 * @document: http://example.org/mydata
-* @base: http://example.org/entities/
+* @nodebase: http://example.org/entities/
+* @schema: https://schema.org/
+* @language: en
 
 # Alice [Person]
 * name: Alice Smith
@@ -45,7 +47,7 @@ Basic `.onya` format example:
 * occupation: Data Scientist
 ```
 
-There is a document header which declares namespaces and base IRIs (URIs). Node definitions are marked with `# NodeID [Type]`. The node ID is resolved against `@document` and the type against 
+There is a document header which declares namespaces and base IRIs (URIs). Node definitions are marked with `# NodeID [Type]`. Node IDs are resolved against `@nodebase`, while types and property labels are resolved against `@schema`.
 - **Properties**: Listed with `* property: value`
 - **Types**: Entities can have one or more types
 
@@ -61,12 +63,14 @@ There is a document header which declares namespaces and base IRIs (URIs). Node
    cat > my_knowledge/people.onya << 'EOF'
    # @docheader
    * @document: http://example.org/mykg
-   * @base: http://example.org/
+   * @nodebase: http://example.org/
+   * @schema: https://schema.org/
+   * @language: en
 
    # Person1 [Person]
    * name: Your Name
-   * role: Your Role
-   * expertise: Your Expertise
+   * jobTitle: Your Role
+   * knowsAbout: Your Expertise
    EOF
    ```
 

pylib/retrieval/graph.py

@@ -172,6 +172,37 @@ def _matches(self, value: str) -> bool:
 
         return False
 
+    @staticmethod
+    def _localname(iri_or_label: str) -> str:
+        '''
+        Best-effort extraction of a "local name" from an IRI.
+
+        Examples:
+            - "https://schema.org/bio" -> "bio"
+            - "http://example.org/people/Person" -> "Person"
+            - "name" -> "name"
+        '''
+        s = str(iri_or_label)
+        if '#' in s:
+            s = s.rsplit('#', 1)[-1]
+        if '/' in s:
+            s = s.rsplit('/', 1)[-1]
+        return s
+
+    def _property_label_matches(self, prop_label: object) -> bool:
+        '''
+        Allow matching properties regardless of whether labels are stored as
+        local names ("bio") or full IRIs ("https://schema.org/bio").
+        '''
+        want = str(self.property_label)
+        have = str(prop_label)
+        if have == want:
+            return True
+        # If caller provided a local name, compare against localname(have)
+        if '://' not in want and not want.startswith('urn:'):
+            return self._localname(have) == want
+        return False
+
     async def execute(
         self,
         query: str,
@@ -208,7 +239,7 @@ async def execute(
 
                 # Check properties
                 for prop in node.properties:
-                    if prop.label == self.property_label:
+                    if self._property_label_matches(prop.label):
                         if self._matches(prop.value):
                             matched = True
                             break

pylib/store/kgraph/onya_store.py

@@ -131,6 +131,7 @@ async def setup(self) -> None:
 
             # Parse into graph
             doc_iri = literate_lex.parse(onya_text, self._graph)
+
             self._loaded_files.append({
                 'path': str(onya_file),
                 'doc_iri': str(doc_iri) if doc_iri else None,
@@ -265,7 +266,30 @@ async def search_by_type(
         if not self._initialized:
             raise RuntimeError(f'KB {self.name} not initialized. Call setup() first.')
 
-        matched_nodes = list(self._graph.typematch(type_iri))
+        type_iri_str = str(type_iri)
+        type_iri_str_lower = type_iri_str.lower()
+
+        def _is_relative_iri(s: str) -> bool:
+            return '://' not in s and not s.startswith('urn:')
+
+        def _matches_type(t: Any) -> bool:
+            t_str = str(t)
+            if t_str == type_iri_str:
+                return True
+            # Back-compat: if Onya returns relative types (e.g. "Person") but the
+            # caller passes a full IRI, allow matching by localname.
+            if _is_relative_iri(t_str) and type_iri_str_lower.endswith(('/' + t_str.lower(), '#' + t_str.lower())):
+                return True
+            return False
+
+        matched_nodes = [
+            node
+            for node in self._graph.values()
+            if any(
+                _matches_type(t)
+                for t in getattr(node, 'types', ())
+            )
+        ]
 
         # Apply limit
         if limit > 0:
@@ -317,11 +341,19 @@ async def get_node(self, node_iri: str) -> dict | None:
             return None
 
         # Convert to dict representation
+        def _localname(iri_or_label: str) -> str:
+            s = str(iri_or_label)
+            if '#' in s:
+                s = s.rsplit('#', 1)[-1]
+            if '/' in s:
+                s = s.rsplit('/', 1)[-1]
+            return s
+
         return {
             'id': str(node.id),
             'types': [str(t) for t in node.types],
             'properties': {
-                prop.label: prop.value
+                _localname(prop.label): prop.value
                 for prop in node.properties
             }
         }

test/resource/onya/companies.onya

@@ -1,6 +1,7 @@
 # @docheader
 * @document: http://example.org/companies
-* @base: http://example.org/companies/
+* @nodebase: http://example.org/companies/
+* @schema: http://example.org/companies/
 
 # TechCorp [Organization]
 * name: TechCorp Inc.

test/resource/onya/people.onya

@@ -1,6 +1,7 @@
 # @docheader
 * @document: http://example.org/people
-* @base: http://example.org/people/
+* @nodebase: http://example.org/people/
+* @schema: http://example.org/people/
 
 # Alice [Person]
 * name: Alice Smith

test/store/kgraph/test_onya_store.py

@@ -33,10 +33,10 @@ async def test_onya_kb_setup():
     loaded_files = kb.get_loaded_files()
     assert len(loaded_files) == 2
 
-    # Should have multiple nodes
+    # Should have multiple nodes (includes 2 document nodes from Onya)
     node_count = kb.count_nodes()
     assert node_count > 0
-    assert node_count == 5  # 3 people + 2 companies
+    assert node_count == 7  # 3 people + 2 companies + 2 document nodes
 
     await kb.cleanup()