Cleaning up and refactoring causal_learns has_directed_path

Signed-off-by: Nicholas Parente <parentenickj@gmail.com>

dowhy/causal_graph.py

@@ -5,6 +5,7 @@
 import networkx as nx
 
 from dowhy.gcm.causal_models import ProbabilisticCausalModel
+from dowhy.graph import has_directed_path
 from dowhy.utils.api import parse_state
 from dowhy.utils.graph_operations import daggity_to_dot
 from dowhy.utils.plotting import plot
@@ -449,11 +450,11 @@ def get_all_directed_paths(self, nodes1, nodes2):
     def has_directed_path(self, nodes1, nodes2):
         """Checks if there is any directed path between two sets of nodes.
 
-        Currently only supports singleton sets.
+        Returns True if and only if every one of the treatments has at least one direct
+        path to one of the outcomes. And, every one of the outcomes has a direct path from
+        at least one of the treatments.
         """
-        # dpaths = self.get_all_directed_paths(nodes1, nodes2)
-        # return len(dpaths) > 0
-        return nx.has_path(self._graph, nodes1[0], nodes2[0])
+        return has_directed_path(self._graph, nodes1[0], nodes2[0])
 
     def get_adjacency_matrix(self, *args, **kwargs):
         """

dowhy/graph.py

@@ -277,8 +277,6 @@ def has_directed_path(graph: nx.DiGraph, nodes1, nodes2):
     path to one of the outcomes. And, every one of the outcomes has a direct path from
     at least one of the treatments.
     """
-    # dpaths = self.get_all_directed_paths(nodes1, nodes2)
-    # return len(dpaths) > 0
     outcome_node_candidates = set()
     action_node_candidates = set()
     for node in nodes1:

tests/causal_identifiers/test_auto_identifier.py

@@ -4,17 +4,14 @@
 
 
 class TestAutoIdentification(object):
-    def test_auto_identify_consistently_checks_for_directed_paths(self):
+    def test_auto_identify_identifies_no_directed_path(self):
+        # Test added for issue #1250
         graph = build_graph_from_str("digraph{T->Y;A->Y;A->B;}")
         identifier = AutoIdentifier(estimand_type=EstimandType.NONPARAMETRIC_ATE)
-        identified_estimand = identifier.identify_effect(
+
+        assert identifier.identify_effect(
             graph, action_nodes=["T", "B"], outcome_nodes=["Y"], observed_nodes=["T", "Y", "A", "B"]
-        )
-        identified_estimand_swapped_action_order = identifier.identify_effect(
+        ).no_directed_path
+        assert identifier.identify_effect(
             graph, action_nodes=["B", "T"], outcome_nodes=["Y"], observed_nodes=["T", "Y", "A", "B"]
-        )
-        backdoor_vars = identified_estimand.get_backdoor_variables()
-        backdoor_vars_swapped_action_order = identified_estimand_swapped_action_order.get_backdoor_variables()
-
-        assert len(backdoor_vars) == 0
-        assert len(backdoor_vars_swapped_action_order) == 0
+        ).no_directed_path

tests/test_causal_graph.py

@@ -123,4 +123,5 @@ def test_build_graph_from_str(self):
     def test_has_path(self):
         assert has_directed_path(self.nx_graph, ["X0"], ["y"])
         assert has_directed_path(self.nx_graph, ["X0", "X1", "X2"], ["y", "v0"])
+        assert not has_directed_path(self.nx_graph, [], ["y"])
         assert not has_directed_path(self.nx_graph, ["X0", "X1", "X2"], ["y", "v0", "Z0"])