fix: RecursiveSerializer- outputs undeclared prefix for predicates that contains the base as a substring (#3267)

* fix: turtle, longturtle, and n3 serializers - outputs undeclared prefix when using base on a predicate that contains the base as a substring

Fixes: #3160

* chore: remove print

* chore: formatting

---------

Co-authored-by: Nicholas Car <nick@kurrawong.net>

Author: edmondchuc

rdflib/plugins/serializers/longturtle.py

@@ -160,6 +160,8 @@ def preprocessTriple(self, triple: _TripleType) -> None:
                     self.base is not None
                     and isinstance(node, URIRef)
                     and node.startswith(self.base)
+                    and "#" not in node.replace(self.base, "")
+                    and "/" not in node.replace(self.base, "")
                 ):
                     # predicate corresponds to base namespace
                     continue

rdflib/plugins/serializers/turtle.py

@@ -17,6 +17,8 @@
     Optional,
     Sequence,
     Tuple,
+    TypeVar,
+    Union,
 )
 
 from rdflib.exceptions import Error
@@ -25,6 +27,8 @@
 from rdflib.serializer import Serializer
 from rdflib.term import BNode, Literal, Node, URIRef
 
+_StrT = TypeVar("_StrT", bound=str)
+
 if TYPE_CHECKING:
     from rdflib.graph import _PredicateType, _SubjectType, _TripleType
 
@@ -169,6 +173,18 @@ def write(self, text: str) -> None:
         # type error: Item "None" of "Optional[IO[bytes]]" has no attribute "write"
         self.stream.write(text.encode(self.encoding, "replace"))  # type: ignore[union-attr]
 
+    def relativize(self, uri: _StrT) -> Union[_StrT, URIRef]:
+        base = self.base
+        if (
+            base is not None
+            and uri.startswith(base)
+            and "#" not in uri.replace(base, "")
+            and "/" not in uri.replace(base, "")
+        ):
+            # type error: Incompatible types in assignment (expression has type "str", variable has type "Node")
+            uri = URIRef(uri.replace(base, "", 1))  # type: ignore[assignment]
+        return uri
+
 
 SUBJECT = 0
 VERB = 1
@@ -271,6 +287,8 @@ def preprocessTriple(self, triple: _TripleType) -> None:
                     self.base is not None
                     and isinstance(node, URIRef)
                     and node.startswith(self.base)
+                    and "#" not in node.replace(self.base, "")
+                    and "/" not in node.replace(self.base, "")
                 ):
                     # predicate corresponds to base namespace
                     continue

test/test_n3.py

@@ -123,8 +123,10 @@ def test_base_serialize(self):
                 URIRef("http://example.com/people/Linda"),
             )
         )
-        s = g.serialize(base="http://example.com/", format="n3", encoding="latin-1")
-        assert b"<people/Bob>" in s
+        s = g.serialize(
+            base="http://example.com/people/", format="n3", encoding="latin-1"
+        )
+        assert b"<Bob>" in s
         g2 = Dataset()
         g2.parse(data=s, format="n3")
         assert list(g) == list(g2.triples((None, None, None)))

test/test_serializers/test_serializer_longturtle.py

@@ -1,5 +1,6 @@
 import difflib
 from pathlib import Path
+from textwrap import dedent
 
 from rdflib import Graph, Namespace
 from rdflib.namespace import GEO, SDO
@@ -181,3 +182,31 @@ def test_longturtle():
     diff = "\n".join(list(difflib.unified_diff(target.split("\n"), output.split("\n"))))
 
     assert not diff, diff
+
+
+def test_longturtle_undeclared_prefix_when_using_base():
+    """
+    See https://github.com/RDFLib/rdflib/issues/3160
+    """
+    from rdflib import Graph, Literal, URIRef
+
+    g = Graph()
+    g.add(
+        (
+            URIRef("https://example.com/subject"),
+            URIRef("https://example.com/p/predicate"),
+            Literal("object"),
+        )
+    )
+    output = g.serialize(format="longturtle", base="https://example.com/")
+    expected = dedent(
+        """
+        BASE <https://example.com/>
+        PREFIX ns1: <https://example.com/p/>
+
+        <subject>
+            ns1:predicate "object" ;
+        .
+    """
+    )
+    assert output.strip() == expected.strip()

test/test_serializers/test_serializer_turtle.py

@@ -1,3 +1,5 @@
+from textwrap import dedent
+
 from rdflib import RDF, RDFS, BNode, Graph, Literal, Namespace, URIRef
 from rdflib.collection import Collection
 from rdflib.plugins.serializers.turtle import TurtleSerializer
@@ -113,3 +115,29 @@ def test_turtle_namespace():
     assert "GENO:0000385" in output
     assert "SERIAL:0167-6423" in output
     assert r"EX:name_with_\(parenthesis\)" in output
+
+
+def test_turtle_undeclared_prefix_when_using_base():
+    """
+    See https://github.com/RDFLib/rdflib/issues/3160
+    """
+    from rdflib import Graph, Literal, URIRef
+
+    g = Graph()
+    g.add(
+        (
+            URIRef("https://example.com/subject"),
+            URIRef("https://example.com/p/predicate"),
+            Literal("object"),
+        )
+    )
+    output = g.serialize(format="turtle", base="https://example.com/")
+    expected = dedent(
+        """
+        @base <https://example.com/> .
+        @prefix ns1: <https://example.com/p/> .
+
+        <subject> ns1:predicate "object" .
+    """
+    )
+    assert output.strip() == expected.strip()