fix: have EnterpriseOrganization check if the organization is present in the lookup

src/openhound_github/models/enterprise_organization.py

@@ -55,6 +55,8 @@ def node_id(self) -> str:
 
     @property
     def as_node(self) -> GHNode:
+        # Only collected if the org was explicitly collected (present in the lookup).
+        collected = self._lookup.org_login_for_id(self.node_id) is not None
         return GHNode(
             kinds=[nk.ORGANIZATION],
             properties=GHEnterpriseOrganizationProperties(
@@ -64,7 +66,7 @@ def as_node(self) -> GHNode:
                 environmentid=self.node_id,
                 environment_name=self.login,
                 login=self.login,
-                collected=False,
+                collected=collected,
                 query_enterprise=f"MATCH p=(:GH_Enterprise {{node_id:'{self.enterprise_node_id}'}})-[:GH_Contains]->(:GH_Organization {{node_id:'{self.node_id}'}}) RETURN p",
             ),
         )

tests/test_enterprise_organization.py

@@ -0,0 +1,69 @@
+"""
+Tests for EnterpriseOrganization.as_node `collected` resolution.
+
+An organization discovered via the enterprise is a stub node. Its `collected`
+flag must be derived from the organizations lookup table so that it agrees with
+the full Organization node regardless of ingestion order:
+
+    - explicitly collected org (present in lookup) -> collected=True
+    - enterprise-only discovered org (absent)       -> collected=False
+"""
+from unittest.mock import MagicMock
+
+from openhound_github.models import EnterpriseOrganization
+
+
+def _make_org(node_id: str = "ORG_NODE_1", login: str = "acme") -> EnterpriseOrganization:
+    org = EnterpriseOrganization(
+        id=node_id,
+        login=login,
+        enterprise_node_id="ENT_NODE_1",
+        enterprise_slug="acme-enterprise",
+    )
+    return org
+
+
+def test_collected_true_when_org_in_lookup() -> None:
+    """An org present in the organizations lookup is marked collected=True."""
+    org = _make_org(node_id="ORG_NODE_1", login="acme")
+    lookup = MagicMock()
+    lookup.org_login_for_id.return_value = "acme"
+    org._lookup = lookup
+
+    node = org.as_node
+
+    assert node.properties.collected is True
+    lookup.org_login_for_id.assert_called_once_with("ORG_NODE_1")
+
+
+def test_collected_false_when_org_not_in_lookup() -> None:
+    """An enterprise-only discovered org (absent from lookup) is collected=False."""
+    org = _make_org(node_id="ORG_NODE_2", login="beta")
+    lookup = MagicMock()
+    lookup.org_login_for_id.return_value = None
+    org._lookup = lookup
+
+    node = org.as_node
+
+    assert node.properties.collected is False
+    lookup.org_login_for_id.assert_called_once_with("ORG_NODE_2")
+
+
+def test_collected_resolved_per_node_id_in_multi_org_enterprise() -> None:
+    """Each stub resolves `collected` against its own node_id independently."""
+    collected_ids = {"ORG_NODE_COLLECTED"}
+
+    def fake_lookup(node_id: str):
+        return "login" if node_id in collected_ids else None
+
+    lookup = MagicMock()
+    lookup.org_login_for_id.side_effect = fake_lookup
+
+    collected_org = _make_org(node_id="ORG_NODE_COLLECTED", login="collected")
+    collected_org._lookup = lookup
+
+    stub_org = _make_org(node_id="ORG_NODE_STUB", login="stub")
+    stub_org._lookup = lookup
+
+    assert collected_org.as_node.properties.collected is True
+    assert stub_org.as_node.properties.collected is False