init public release

tests/unit/test_graph_creation.py

@@ -0,0 +1,136 @@
+"""Unit tests for ``graphCreation.create_graph``.
+
+This module turns a JSON bundle into a networkx DiGraph:
+- a dict value becomes a child node ``{parent}.{key}`` (label = key) with an
+  edge carrying ``edge_type = key``;
+- a list with no dicts is stored as a plain attribute on the parent;
+- a list containing dicts becomes one child node per dict (indexed when the
+  list has more than one element);
+- any other scalar becomes an attribute on the parent.
+
+Pure logic, only networkx -- no stubs or mocks needed. These tests import the
+real module directly (conftest prefers the real package when present).
+"""
+
+import networkx as nx
+import pytest
+
+from graphCreation.create_graph import (
+    add_json_to_networkx,
+    add_nodes_from_dict,
+    process_dictionaries,
+)
+
+
+def _graph_from(json_data, name="root"):
+    g = nx.DiGraph()
+    add_json_to_networkx(json_data, name, g)
+    return g
+
+
+# --- entry point / guards --------------------------------------------------
+
+def test_rejects_non_digraph():
+    with pytest.raises(ValueError):
+        add_json_to_networkx({}, "root", nx.Graph())  # undirected, not DiGraph
+
+
+def test_root_node_created_with_root_label():
+    g = _graph_from({})
+    assert g.nodes["root_bundle"]["label"] == "root"
+
+
+def test_bundle_suffix_is_always_appended():
+    # SHARP EDGE: the function appends '_bundle' to whatever name it's given.
+    # bundles_to_graph already passes ``id + '_bundle'``, so the in-graph root
+    # ends up doubly suffixed (e.g. 'patient1_bundle_bundle'). Harmless while
+    # consistent, but surprising if anything looks the root up by id.
+    g = _graph_from({}, name="patient1_bundle")
+    assert "patient1_bundle_bundle" in g.nodes
+
+
+# --- scalars ---------------------------------------------------------------
+
+def test_scalar_values_become_parent_attributes():
+    g = _graph_from({"id": "123", "active": True, "count": 5})
+    attrs = g.nodes["root_bundle"]
+    assert attrs["id"] == "123"
+    assert attrs["active"] is True
+    assert attrs["count"] == 5
+
+
+# --- nested dicts ----------------------------------------------------------
+
+def test_nested_dict_becomes_child_node_with_edge():
+    g = _graph_from({"name": {"family": "Mueller", "given": "Anna"}})
+    assert "root_bundle.name" in g.nodes
+    assert g.nodes["root_bundle.name"]["label"] == "name"
+    assert g.nodes["root_bundle.name"]["family"] == "Mueller"
+    assert g.edges["root_bundle", "root_bundle.name"]["edge_type"] == "name"
+
+
+def test_deeply_nested_dicts_chain_through_nodes():
+    g = _graph_from({"a": {"b": {"c": 1}}})
+    assert {"root_bundle", "root_bundle.a", "root_bundle.a.b"} <= set(g.nodes)
+    assert g.nodes["root_bundle.a.b"]["c"] == 1
+    assert g.has_edge("root_bundle.a", "root_bundle.a.b")
+
+
+# --- lists -----------------------------------------------------------------
+
+def test_list_of_scalars_stored_as_attribute():
+    g = _graph_from({"tags": ["a", "b", "c"]})
+    assert g.nodes["root_bundle"]["tags"] == ["a", "b", "c"]
+    assert list(g.nodes) == ["root_bundle"]  # no child nodes created
+
+
+def test_empty_list_stored_as_attribute():
+    g = _graph_from({"items": []})
+    assert g.nodes["root_bundle"]["items"] == []
+
+
+def test_single_dict_list_has_no_index_suffix():
+    g = _graph_from({"contact": [{"phone": "555"}]})
+    assert "root_bundle.contact" in g.nodes
+    assert "root_bundle.contact[0]" not in g.nodes
+    assert g.nodes["root_bundle.contact"]["phone"] == "555"
+
+
+def test_multi_dict_list_creates_indexed_nodes():
+    g = _graph_from({"entries": [{"x": 1}, {"y": 2}]})
+    assert "root_bundle.entries[0]" in g.nodes
+    assert "root_bundle.entries[1]" in g.nodes
+    assert g.nodes["root_bundle.entries[0]"]["x"] == 1
+    assert g.nodes["root_bundle.entries[1]"]["y"] == 2
+
+
+def test_mixed_list_silently_drops_non_dict_items():
+    # FINDING: when a list contains at least one dict, process_dictionaries
+    # runs and *skips* every non-dict element -- it is neither stored as an
+    # attribute nor turned into a node. The scalar "lost" below disappears.
+    # Pinned as current behaviour; flip if such data should be retained.
+    g = _graph_from({"mixed": ["lost", {"k": "v"}]})
+    assert "root_bundle.mixed[1]" in g.nodes      # the dict survives (indexed)
+    assert g.nodes["root_bundle.mixed[1]"]["k"] == "v"
+    assert "mixed" not in g.nodes["root_bundle"]  # not stored as an attribute
+    # The scalar appears nowhere in the graph.
+    all_attr_values = [v for _, data in g.nodes(data=True) for v in data.values()]
+    assert "lost" not in all_attr_values
+
+
+# --- helpers used directly -------------------------------------------------
+
+def test_add_nodes_from_dict_attaches_to_given_parent():
+    g = nx.DiGraph()
+    g.add_node("p", label="root")
+    add_nodes_from_dict(g, "p", {"status": "final"})
+    assert g.nodes["p"]["status"] == "final"
+
+
+def test_process_dictionaries_skips_when_no_dicts_present():
+    # Defensive: passing a dict-free list straight to process_dictionaries
+    # creates nothing (the inner isinstance check is never satisfied).
+    g = nx.DiGraph()
+    g.add_node("p", label="root")
+    process_dictionaries(["a", "b"], "p", "vals", g)
+    assert list(g.nodes) == ["p"]