file node class in default schema

1 files changed, 8 insertions, 9 deletions

diff --git a/bsie/base/extractor.py b/bsie/base/extractor.py
index a5c7846..678dcec 100644
--- a/bsie/base/extractor.py
+++ b/bsie/base/extractor.py
@@ -35,6 +35,7 @@ SCHEMA_PREAMBLE = '''
 
     # essential nodes
     bsfs:Entity rdfs:subClassOf bsfs:Node .
+    bsfs:File rdfs:subClassOf bsfs:Entity .
 
     # common definitions
     xsd:string rdfs:subClassOf bsfs:Literal .
@@ -77,15 +78,13 @@ class Extractor(abc.ABC):
         return hash((type(self), self.CONTENT_READER, self.schema))
 
     def predicates(self) -> typing.Iterator[_schema.Predicate]:
-        """Return the predicates that may be part of extracted triples."""
-        # NOTE: Some predicates in the schema might not occur in actual triples,
-        # but are defined due to predicate class hierarchy. E.g., bsfs:Predicate
-        # is part of every schema but should not be used in triples.
-        # Announcing all predicates might not be the most efficient way, however,
-        # it is the most safe one. Concrete extractors that produce additional
-        # predicates (e.g. auxiliary nodes with their own predicates) should
-        # overwrite this method to only include the principal predicates.
-        return self.schema.predicates()
+        ent = self.schema.node(ns.bsfs.Entity)
+        return (
+            pred
+            for pred
+            in self.schema.predicates()
+            if pred.domain <= ent or (pred.range is not None and pred.range <= ent)
+            )
 
     @abc.abstractmethod
     def extract(