fix: bad syntax event

Author: FlorentLvr

src/core/ontologies/ConsolidateOntology.py

@@ -1,10 +1,11 @@
 from pathlib import Path
-from rdflib import Graph, URIRef, RDFS
+from rdflib import Graph, URIRef, RDFS, RDF, OWL
 from abi import logger
 import json
 
 def consolidate_ontologies(ontologies_dir: str, output_file: str, mapping_file: str) -> None:
     """Consolidate all ontology files and save URI-label mapping to Python file.
+    Only keeps classes, data properties, object properties, and annotation properties.
     
     Args:
         ontologies_dir (str): Path to directory containing ontology files
@@ -13,6 +14,12 @@ def consolidate_ontologies(ontologies_dir: str, output_file: str, mapping_file:
     """
     # Create consolidated graph
     consolidated = Graph()
+    filtered = Graph()
+    
+    # Set CCO prefix
+    filtered.bind('cco', 'https://www.commoncoreontologies.org/')
+    filtered.bind('abi', 'http://ontology.naas.ai/abi/')
+    filtered.bind('bfo', 'http://purl.obolibrary.org/obo/')
 
     # Get all .ttl files recursively
     ontologies_path = Path(ontologies_dir)
@@ -31,16 +38,35 @@ def consolidate_ontologies(ontologies_dir: str, output_file: str, mapping_file:
             consolidated += g
         except Exception as e:
             logger.error(f"Error loading {ttl_file}: {e}")
+    
+    # Filter for desired types
+    desired_types = {
+        OWL.Class,
+        OWL.DatatypeProperty,
+        OWL.ObjectProperty,
+        OWL.AnnotationProperty
+    }
+    
+    # Add all triples where subject is of desired type
+    for s, p, o in consolidated.triples((None, RDF.type, None)):
+        if o in desired_types:
+            # Add the type triple
+            filtered.add((s, p, o))
+            # Add all triples where this subject is involved
+            for s2, p2, o2 in consolidated.triples((s, None, None)):
+                filtered.add((s2, p2, o2))
+            for s2, p2, o2 in consolidated.triples((None, None, s)):
+                filtered.add((s2, p2, o2))
 
-    # Save consolidated ontology and mapping
+    # Save filtered ontology and mapping
     try:
-        consolidated.serialize(destination=output_file, format="turtle")
-        logger.info(f"Saved consolidated ontology to {output_file}")
-        logger.info(f"Total triples: {len(consolidated)}")
+        filtered.serialize(destination=output_file, format="turtle")
+        logger.info(f"Saved filtered ontology to {output_file}")
+        logger.info(f"Total triples: {len(filtered)}")
 
         # Create URI-label mapping
         mapping = {}
-        for s, p, o in consolidated.triples((None, RDFS.label, None)):
+        for s, p, o in filtered.triples((None, RDFS.label, None)):
             if isinstance(s, URIRef):
                 mapping[str(s)] = str(o)
 
@@ -58,8 +84,7 @@ def consolidate_ontologies(ontologies_dir: str, output_file: str, mapping_file:
 
 if __name__ == "__main__":
     # Define paths relative to project root
-    project_root = Path(__file__).parent.parent.parent
-    ontologies_dir = project_root / "src" / "ontologies"
+    ontologies_dir = Path(__file__).parent
     output_file = ontologies_dir / "ConsolidatedOntology.ttl"
     mapping_file = ontologies_dir / "mapping.py"