fix: skip prefix generation for predicates corresponding to base namespace (#3244)

* skip prefix generation for predicates corresponding to base namespace * fix: skip prefix generation for predicates corresponding to the base namespace. This applies to both turtle and longturtle Also added fix to escape brackets in pnames to longturtle. --------- Co-authored-by: bulricht <108867473+bulricht@users.noreply.github.com>

Author: edmondchuc

rdflib/plugins/serializers/longturtle.py

@@ -22,7 +22,7 @@
 
 from rdflib.compare import to_canonical_graph
 from rdflib.exceptions import Error
-from rdflib.graph import Graph
+from rdflib.graph import Graph, _TripleType
 from rdflib.namespace import RDF
 from rdflib.term import BNode, Literal, URIRef
 
@@ -149,11 +149,20 @@ def serialize(
 
  self.base = None
 
- def preprocessTriple(self, triple):
+ def preprocessTriple(self, triple: _TripleType) -> None:
  super(LongTurtleSerializer, self).preprocessTriple(triple)
  for i, node in enumerate(triple):
- if node in self.keywords:
- continue
+ if i == VERB:
+ if node in self.keywords:
+ # predicate is a keyword
+ continue
+ if (
+ self.base is not None
+ and isinstance(node, URIRef)
+ and node.startswith(self.base)
+ ):
+ # predicate corresponds to base namespace
+ continue
  # Don't use generated prefixes for subjects and objects
  self.getQName(node, gen_prefix=(i == VERB))
  if isinstance(node, Literal) and node.datatype:
@@ -180,6 +189,8 @@ def getQName(self, uri, gen_prefix=True):
 
  prefix, namespace, local = parts
 
+ local = local.replace(r"(", r"\(").replace(r")", r"\)")
+
  # QName cannot end with .
  if local.endswith("."):
  return None

rdflib/plugins/serializers/turtle.py

@@ -263,9 +263,17 @@ def serialize(
  def preprocessTriple(self, triple: _TripleType) -> None:
  super(TurtleSerializer, self).preprocessTriple(triple)
  for i, node in enumerate(triple):
- if i == VERB and node in self.keywords:
- # predicate is a keyword
- continue
+ if i == VERB:
+ if node in self.keywords:
+ # predicate is a keyword
+ continue
+ if (
+ self.base is not None
+ and isinstance(node, URIRef)
+ and node.startswith(self.base)
+ ):
+ # predicate corresponds to base namespace
+ continue
  # Don't use generated prefixes for subjects and objects
  self.getQName(node, gen_prefix=(i == VERB))
  if isinstance(node, Literal) and node.datatype:

test/test_serializers/test_serializer_turtle_base_namespace.py

@@ -0,0 +1,34 @@
+from rdflib import Graph, Namespace, URIRef
+
+# https://github.com/RDFLib/rdflib/issues/3237
+# Test that the ns1 prefix is not generated when the base is set.
+mns = Namespace("http://my-namespace.net/")
+
+
+def test_turtle():
+ g = Graph(base="http://my-base.net/")
+ g.add((mns.foo, URIRef("http://my-base.net/my-predicate"), mns.bar))
+ result = g.serialize(format="text/turtle")
+ assert (
+ result
+ == """@base <http://my-base.net/> .
+
+<http://my-namespace.net/foo> <my-predicate> <http://my-namespace.net/bar> .
+
+"""
+ )
+
+
+def test_longturtle():
+ g = Graph(base="http://my-base.net/")
+ g.add((mns.foo, URIRef("http://my-base.net/my-predicate"), mns.bar))
+ result = g.serialize(format="longturtle", canon=True)
+ assert (
+ result
+ == """BASE <http://my-base.net/>
+
+<http://my-namespace.net/foo>
+ <my-predicate> <http://my-namespace.net/bar> ;
+.
+"""
+ ), print(result)