Add more tests for graph_diff (#1983)

This is to confirm that `graph_diff` does not work with quads.

Other changes:
- Added support to `GraphHelper` for collapsing `BNode`s to a specific IRI
  when converting to triple sets, this is in addition to the current
  functionality for ignoring BNodes. This makes it possible to more
  accurately compare Graphs that contain blank nodes.
- Added `assert_cgraph_isomorphic` to `GraphHelper`.
  This method adds some rudimentary support for checking if conjunctive
  graphs are isomorphic, it will ignore named graphs that have blank
  nodes as name.

test/test_graph/test_diff.py

@@ -1,12 +1,23 @@
-from test.utils import GraphHelper
-from typing import TYPE_CHECKING, Set
+from dataclasses import dataclass, field
+from test.utils import (
+    COLLAPSED_BNODE,
+    BNodeHandling,
+    GHQuad,
+    GHTriple,
+    GraphHelper,
+    MarksType,
+    MarkType,
+)
+from typing import TYPE_CHECKING, Collection, Set, Tuple, Type, Union, cast
 
 import pytest
+from _pytest.mark.structures import ParameterSet
 
 import rdflib
 from rdflib import Graph
 from rdflib.compare import graph_diff
-from rdflib.namespace import FOAF, RDF
+from rdflib.graph import ConjunctiveGraph, Dataset
+from rdflib.namespace import FOAF, RDF, Namespace
 from rdflib.term import BNode, Literal
 
 if TYPE_CHECKING:
@@ -15,7 +26,7 @@
 """Test for graph_diff - much more extensive testing
 would certainly be possible"""
 
-_TripleSetT = Set["_TripleType"]
+_TripleSetType = Set["_TripleType"]
 
 
 class TestDiff:
@@ -48,7 +59,7 @@ def test_subsets(self) -> None:
         triples only in `g0`, and that there are triples that occur in both
         `g0` and `g1`, and that there are triples only in `g1`.
         """
-        g0_ts: _TripleSetT = set()
+        g0_ts: _TripleSetType = set()
         bnode = BNode()
         g0_ts.update(
             {
@@ -59,7 +70,7 @@ def test_subsets(self) -> None:
         g0 = Graph()
         g0 += g0_ts
 
-        g1_ts: _TripleSetT = set()
+        g1_ts: _TripleSetType = set()
         bnode = BNode()
         g1_ts.update(
             {
@@ -76,3 +87,154 @@ def test_subsets(self) -> None:
         assert in_first == set()
         assert len(in_second) > 0
         assert len(in_both) > 0
+
+
+_ElementSetType = Union[Collection[GHTriple], Collection[GHQuad]]
+
+_ElementSetTypeOrStr = Union[_ElementSetType, str]
+
+
+@dataclass
+class GraphDiffCase:
+    graph_type: Type[Graph]
+    format: str
+    lhs: str
+    rhs: str
+    expected_result: Tuple[
+        _ElementSetTypeOrStr, _ElementSetTypeOrStr, _ElementSetTypeOrStr
+    ]
+    marks: MarkType = field(default_factory=lambda: cast(MarksType, list()))
+
+    def as_element_set(self, value: _ElementSetTypeOrStr) -> _ElementSetType:
+        if isinstance(value, str):
+            graph = self.graph_type()
+            graph.parse(data=value, format=self.format)
+            if isinstance(graph, ConjunctiveGraph):
+                return GraphHelper.quad_set(graph, BNodeHandling.COLLAPSE)
+            else:
+                return GraphHelper.triple_set(graph, BNodeHandling.COLLAPSE)
+        return value
+
+    def expected_in_both_set(self) -> _ElementSetType:
+        return self.as_element_set(self.expected_result[0])
+
+    def expected_in_lhs_set(self) -> _ElementSetType:
+        return self.as_element_set(self.expected_result[1])
+
+    def expected_in_rhs_set(self) -> _ElementSetType:
+        return self.as_element_set(self.expected_result[2])
+
+    def as_params(self) -> ParameterSet:
+        return pytest.param(self, marks=self.marks)
+
+
+EGSCHEME = Namespace("example:")
+
+
+@pytest.mark.parametrize(
+    "test_case",
+    [
+        GraphDiffCase(
+            Graph,
+            format="turtle",
+            lhs="""
+            @prefix eg: <example:> .
+            _:a _:b _:c .
+            eg:o0 eg:p0 eg:s0 .
+            eg:o1 eg:p1 eg:s1 .
+            """,
+            rhs="""
+            @prefix eg: <example:> .
+            eg:o0 eg:p0 eg:s0 .
+            eg:o1 eg:p1 eg:s1 .
+            """,
+            expected_result=(
+                """
+            @prefix eg: <example:> .
+            eg:o0 eg:p0 eg:s0 .
+            eg:o1 eg:p1 eg:s1 .
+            """,
+                {(COLLAPSED_BNODE, COLLAPSED_BNODE, COLLAPSED_BNODE)},
+                "",
+            ),
+        ),
+        GraphDiffCase(
+            Graph,
+            format="turtle",
+            lhs="""
+            @prefix eg: <example:> .
+            eg:o0 eg:p0 eg:s0 .
+            eg:o1 eg:p1 eg:s1 .
+            """,
+            rhs="""
+            @prefix eg: <example:> .
+            eg:o0 eg:p0 eg:s0 .
+            eg:o1 eg:p1 eg:s1 .
+            """,
+            expected_result=(
+                """
+            @prefix eg: <example:> .
+            eg:o0 eg:p0 eg:s0 .
+            eg:o1 eg:p1 eg:s1 .
+            """,
+                "",
+                "",
+            ),
+        ),
+        GraphDiffCase(
+            Dataset,
+            format="trig",
+            lhs="""
+            @prefix eg: <example:> .
+            eg:o0 eg:p0 eg:s0 .
+            eg:o1 eg:p1 eg:s1 .
+            """,
+            rhs="""
+            @prefix eg: <example:> .
+            eg:o0 eg:p0 eg:s0 .
+            eg:o1 eg:p1 eg:s1 .
+            """,
+            expected_result=(
+                """
+            @prefix eg: <example:> .
+            eg:o0 eg:p0 eg:s0 .
+            eg:o1 eg:p1 eg:s1 .
+            """,
+                "",
+                "",
+            ),
+            marks=pytest.mark.xfail(
+                reason="quads are not supported", raises=ValueError
+            ),
+        ).as_params(),
+    ],
+)
+def test_assert_sets_equal(test_case: GraphDiffCase):
+    """
+    GraphHelper.sets_equals and related functions work correctly in both
+    positive and negative cases.
+    """
+    lhs_graph: Graph = test_case.graph_type()
+    lhs_graph.parse(data=test_case.lhs, format=test_case.format)
+
+    rhs_graph: Graph = test_case.graph_type()
+    rhs_graph.parse(data=test_case.rhs, format=test_case.format)
+
+    in_both, in_lhs, in_rhs = graph_diff(lhs_graph, rhs_graph)
+    in_both_set = GraphHelper.triple_or_quad_set(in_both, BNodeHandling.COLLAPSE)
+    in_lhs_set = GraphHelper.triple_or_quad_set(in_lhs, BNodeHandling.COLLAPSE)
+    in_rhs_set = GraphHelper.triple_or_quad_set(in_rhs, BNodeHandling.COLLAPSE)
+
+    assert test_case.expected_in_both_set() == in_both_set
+    assert test_case.expected_in_lhs_set() == in_lhs_set
+    assert test_case.expected_in_rhs_set() == in_rhs_set
+
+    # Diff should be symetric
+    in_rboth, in_rlhs, in_rrhs = graph_diff(rhs_graph, lhs_graph)
+    in_rboth_set = GraphHelper.triple_or_quad_set(in_rboth, BNodeHandling.COLLAPSE)
+    in_rlhs_set = GraphHelper.triple_or_quad_set(in_rlhs, BNodeHandling.COLLAPSE)
+    in_rrhs_set = GraphHelper.triple_or_quad_set(in_rrhs, BNodeHandling.COLLAPSE)
+
+    assert test_case.expected_in_both_set() == in_rboth_set
+    assert test_case.expected_in_rhs_set() == in_rlhs_set
+    assert test_case.expected_in_lhs_set() == in_rrhs_set

test/test_graph/test_graph_cbd.py

@@ -1,5 +1,5 @@
 from test.data import TEST_DATA_DIR
-from test.utils import GraphHelper
+from test.utils import BNodeHandling, GraphHelper
 
 import pytest
 
@@ -130,7 +130,7 @@ def test_cbd_example():
 
     query = "http://example.com/aReallyGreatBook"
     GraphHelper.assert_isomorphic(g.cbd(URIRef(query)), g_cbd)
-    GraphHelper.assert_sets_equals(g.cbd(URIRef(query)), g_cbd, exclude_blanks=True)
+    GraphHelper.assert_sets_equals(g.cbd(URIRef(query)), g_cbd, BNodeHandling.COLLAPSE)
     assert len(g.cbd(URIRef(query))) == (
         21
     ), "cbd() for aReallyGreatBook should return 21 triples"

test/test_roundtrip.py

@@ -3,7 +3,7 @@
 import os.path
 from pathlib import Path
 from test.data import TEST_DATA_DIR
-from test.utils import GraphHelper
+from test.utils import BNodeHandling, GraphHelper
 from typing import Callable, Iterable, List, Optional, Set, Tuple, Type, Union
 from xml.sax import SAXParseException
 
@@ -285,9 +285,13 @@ def roundtrip(
         GraphHelper.assert_isomorphic(g1, g2)
     if checks is not None:
         if Check.SET_EQUALS in checks:
-            GraphHelper.assert_sets_equals(g1, g2, exclude_blanks=False)
+            GraphHelper.assert_sets_equals(
+                g1, g2, bnode_handling=BNodeHandling.COLLAPSE
+            )
         if Check.SET_EQUALS_WITHOUT_BLANKS in checks:
-            GraphHelper.assert_sets_equals(g1, g2, exclude_blanks=True)
+            GraphHelper.assert_sets_equals(
+                g1, g2, bnode_handling=BNodeHandling.COLLAPSE
+            )
 
     if logger.isEnabledFor(logging.DEBUG):
         logger.debug("OK")