Merge pull request #528 from pchampin/fix-issue-381

SPARQL parser can deal with spurious ";", fixes #381

rdflib/plugins/sparql/parser.py

@@ -51,10 +51,11 @@ def expandTriples(terms):
         l = len(terms)
         for i, t in enumerate(terms):
             if t == ',':
-                res.append(res[i - 3])
-                res.append(res[i - 2])
+                res.extend([res[-3], res[-2]])
             elif t == ';':
-                res.append(res[i - 3])
+                if i+1 == len(terms) or terms[i+1] == ";" or terms[i+1] == ".":
+                    continue # this semicolon is spurious
+                res.append(res[-3])
             elif isinstance(t, list):
                 # BlankNodePropertyList
                 # is this bnode the object of previous triples?

test/test_issue381.py

@@ -0,0 +1,119 @@
+from rdflib import BNode, Graph, Namespace
+from rdflib.compare import isomorphic
+
+NS = Namespace("http://example.org/")
+
+def test_no_spurious_semicolon():
+    sparql = """
+      PREFIX : <http://example.org/>
+      CONSTRUCT {
+        :a :b :c ; :d :e .
+      } WHERE {}
+    """
+    expected = Graph()
+    expected.addN( t+(expected,) for t in [
+        (NS.a, NS.b, NS.c),
+        (NS.a, NS.d, NS.e),
+    ])
+    got = Graph().query(sparql).graph
+    assert isomorphic(got, expected), got.serialize(format="turtle")
+
+def test_one_spurious_semicolon():
+    sparql = """
+      PREFIX : <http://example.org/>
+      CONSTRUCT {
+        :a :b :c ; :d :e ; .
+      } WHERE {}
+    """
+    expected = Graph()
+    expected.addN( t+(expected,) for t in [
+        (NS.a, NS.b, NS.c),
+        (NS.a, NS.d, NS.e),
+    ])
+    got = Graph().query(sparql).graph
+    assert isomorphic(got, expected), got.serialize(format="turtle")
+
+def test_one_spurious_semicolon_no_perdiod():
+    sparql = """
+      PREFIX : <http://example.org/>
+      CONSTRUCT {
+        :a :b :c ; :d :e ;
+      } WHERE {}
+    """
+    expected = Graph()
+    expected.addN( t+(expected,) for t in [
+        (NS.a, NS.b, NS.c),
+        (NS.a, NS.d, NS.e),
+    ])
+    got = Graph().query(sparql).graph
+    assert isomorphic(got, expected), got.serialize(format="turtle")
+
+def test_two_spurious_semicolons_no_period():
+    sparql = """
+      PREFIX : <http://example.org/>
+      CONSTRUCT {
+        :a :b :c ; :d :e ; ;
+      } WHERE {}
+    """
+    expected = Graph()
+    expected.addN( t+(expected,) for t in [
+        (NS.a, NS.b, NS.c),
+        (NS.a, NS.d, NS.e),
+    ])
+    got = Graph().query(sparql).graph
+    assert isomorphic(got, expected), got.serialize(format="turtle")
+
+def test_one_spurious_semicolons_bnode():
+    sparql = """
+      PREFIX : <http://example.org/>
+      CONSTRUCT {
+        [ :b :c ; :d :e ; ]
+      } WHERE {}
+    """
+    expected = Graph()
+    expected.addN( t+(expected,) for t in [
+        (BNode("a"), NS.b, NS.c),
+        (BNode("a"), NS.d, NS.e),
+    ])
+    got = Graph().query(sparql).graph
+    assert isomorphic(got, expected), got.serialize(format="turtle")
+
+def test_pathological():
+    """
+    This test did not raise an exception,
+    but generated a graph completely different from the expected one.
+    """
+    sparql = """
+      PREFIX : <http://example.org/>
+      CONSTRUCT {
+        :a :b :c ; ;
+           :d :e ; ;
+           :f :g ;
+      } WHERE {}
+    """
+    expected = Graph()
+    expected.addN( t+(expected,) for t in [
+        (NS.a, NS.b, NS.c),
+        (NS.a, NS.d, NS.e),
+        (NS.a, NS.f, NS.g),
+    ])
+    got = Graph().query(sparql).graph
+    assert isomorphic(got, expected), got.serialize(format="turtle")
+
+def test_mixing_spurious_semicolons_and_commas():
+    sparql = """
+      PREFIX : <http://example.org/>
+      CONSTRUCT {
+        :a :b :c ; ;
+           :d :e, :f
+      } WHERE {}
+    """
+    expected = Graph()
+    expected.addN( t+(expected,) for t in [
+        (NS.a, NS.b, NS.c),
+        (NS.a, NS.d, NS.e),
+        (NS.a, NS.d, NS.f),
+    ])
+    got = Graph().query(sparql).graph
+    assert isomorphic(got, expected), got.serialize(format="turtle")
+