Merge pull request #133 from neo4j-labs/fix-flaky-pytests

Fix flaky python tests

crates/mate/tests/ds_test.py

@@ -1,12 +1,12 @@
 import numpy as np
 import pandas as pd
 
-from graph_mate import DiGraph, Graph
+from graph_mate import DiGraph, Graph, Layout
 
 
 def test_numpy_graph():
     el = np.array([[0, 1], [2, 3], [4, 1]], dtype=np.uint32)
-    g = Graph.from_numpy(el)
+    g = Graph.from_numpy(el, layout=Layout.Sorted)
 
     assert g.node_count() == 5
     assert g.edge_count() == 3
@@ -20,7 +20,7 @@ def test_numpy_graph():
 
 def test_pandas_graph():
     df = pd.DataFrame({"source": [0, 2, 4], "target": [1, 3, 1]})
-    g = Graph.from_pandas(df)
+    g = Graph.from_pandas(df, layout=Layout.Sorted)
 
     assert g.node_count() == 5
     assert g.edge_count() == 3
@@ -34,7 +34,7 @@ def test_pandas_graph():
 
 def test_numpy_digraph():
     el = np.array([[0, 1], [2, 3], [4, 1]], dtype=np.uint32)
-    g = DiGraph.from_numpy(el)
+    g = DiGraph.from_numpy(el, layout=Layout.Sorted)
 
     assert g.node_count() == 5
     assert g.edge_count() == 3
@@ -49,7 +49,7 @@ def test_numpy_digraph():
 
 def test_pandas_digraph():
     df = pd.DataFrame({"source": [0, 2, 4], "target": [1, 3, 1]})
-    g = DiGraph.from_pandas(df)
+    g = DiGraph.from_pandas(df, layout=Layout.Sorted)
 
     assert g.node_count() == 5
     assert g.edge_count() == 3

crates/mate/tests/graph_test.py

@@ -12,28 +12,33 @@ def test_load_graph(g: DiGraph):
 
 
 def test_to_undirected(g: DiGraph, ug: Graph):
-    g = g.to_undirected()
+    undirected = g.to_undirected()
 
-    for n in range(g.node_count()):
-        assert set(g.copy_neighbors(n)) == set(ug.copy_neighbors(n))
+    for n in range(undirected.node_count()):
+        assert set(undirected.copy_neighbors(n)) == set(ug.copy_neighbors(n))
 
 
 def test_to_undirected_with_layout():
     g = DiGraph.from_numpy(
         np.array([[0, 1], [0, 1], [0, 2], [1, 2], [2, 1], [0, 3]], dtype=np.uint32)
     )
 
+    def compare_unsorted(expect, actual):
+        sorted = expect.copy()
+        sorted.sort()
+        return np.array_equal(sorted, actual)
+
     ug = g.to_undirected()
-    assert np.array_equal(ug.neighbors(0), [1, 1, 2, 3])
-    assert np.array_equal(ug.neighbors(1), [2, 0, 0, 2])
-    assert np.array_equal(ug.neighbors(2), [1, 0, 1])
-    assert np.array_equal(ug.neighbors(3), [0])
+    assert compare_unsorted(ug.neighbors(0), [1, 1, 2, 3])
+    assert compare_unsorted(ug.neighbors(1), [0, 0, 2, 2])
+    assert compare_unsorted(ug.neighbors(2), [0, 1, 1])
+    assert compare_unsorted(ug.neighbors(3), [0])
 
     ug = g.to_undirected(Layout.Unsorted)
-    assert np.array_equal(ug.neighbors(0), [1, 1, 2, 3])
-    assert np.array_equal(ug.neighbors(1), [2, 0, 0, 2])
-    assert np.array_equal(ug.neighbors(2), [1, 0, 1])
-    assert np.array_equal(ug.neighbors(3), [0])
+    assert compare_unsorted(ug.neighbors(0), [1, 1, 2, 3])
+    assert compare_unsorted(ug.neighbors(1), [0, 0, 2, 2])
+    assert compare_unsorted(ug.neighbors(2), [0, 1, 1])
+    assert compare_unsorted(ug.neighbors(3), [0])
 
     ug = g.to_undirected(Layout.Sorted)
     assert np.array_equal(ug.neighbors(0), [1, 1, 2, 3])