BED-9505: canonicalize organization login before collection

src/openhound_github/resources/organization.py

@@ -101,6 +101,7 @@ class SourceContext:
     github_deployment_id: str = DEFAULT_GITHUB_DEPLOYMENT_ID
     github_web_origin: str = DEFAULT_GITHUB_WEB_ORIGIN
     cache_lock: Lock = field(default_factory=Lock)
+    organizations_cache: dict[str, dict[str, Any]] = field(default_factory=dict)
     app_cache: dict[str, dict[str, Any]] = field(default_factory=dict)
     actions_permissions_cache: dict[str, dict[str, Any]] = field(default_factory=dict)
     runner_permissions_cache: dict[str, dict[str, Any]] = field(default_factory=dict)
@@ -252,7 +253,13 @@ def organizations(ctx: SourceContext):
         org_name = org.org_name
         client = org.client
         try:
-            org_data = client.get(f"/orgs/{org_name}").json()
+            org_data = _cached_org_response(
+                ctx,
+                ctx.organizations_cache,
+                client,
+                org_name,
+                f"/orgs/{org_name}",
+            )
 
             actions = _actions_permissions(ctx, client, org_name)
             self_hosted_runners = _runner_permissions(ctx, client, org_name)

src/openhound_github/source.py

@@ -53,6 +53,7 @@ class SourceContext:
     github_deployment_id: str = DEFAULT_GITHUB_DEPLOYMENT_ID
     github_web_origin: str = DEFAULT_GITHUB_WEB_ORIGIN
     cache_lock: Lock = field(default_factory=Lock)
+    organizations_cache: dict[str, dict[str, Any]] = field(default_factory=dict)
     app_cache: dict[str, dict[str, Any]] = field(default_factory=dict)
     actions_permissions_cache: dict[str, dict[str, Any]] = field(default_factory=dict)
     runner_permissions_cache: dict[str, dict[str, Any]] = field(default_factory=dict)
@@ -63,6 +64,31 @@ def org_names(self) -> list[str]:
         return [org.org_name for org in self.organizations or []]
 
 
+def _canonicalize_org_names(ctx: SourceContext) -> None:
+    for org in ctx.organizations or []:
+        configured_name = org.org_name
+        try:
+            org_data = org.client.get(f"/orgs/{configured_name}").json()
+        except Exception as exc:
+            logger.warning(
+                "Unable to resolve canonical GitHub login for organization '%s': %s",
+                configured_name,
+                exc,
+            )
+            continue
+
+        canonical_name = org_data.get("login")
+        if not isinstance(canonical_name, str) or not canonical_name:
+            logger.warning(
+                "GitHub organization response for '%s' did not include a canonical login",
+                configured_name,
+            )
+            continue
+
+        org.org_name = canonical_name
+        ctx.organizations_cache[canonical_name] = org_data
+
+
 @configspec
 class GithubCredentials(CredentialsConfiguration):
     org_name: str | None = None
@@ -228,6 +254,7 @@ def token_client(token: str) -> RESTClient:
             )
         )
 
+        _canonicalize_org_names(ctx)
         return organization_resources(ctx)
 
     else:
@@ -255,4 +282,5 @@ def token_client(token: str) -> RESTClient:
                 github_web_origin=github_web_origin,
             )
         )
+        _canonicalize_org_names(ctx)
         return organization_resources(ctx)

tests/test_source_org_canonicalization.py

@@ -0,0 +1,138 @@
+import importlib
+import inspect
+
+import pytest
+
+from openhound_github.resources.organization import organizations
+from openhound_github.source import (
+    GithubOrgAppCredentials,
+    GithubTokenCredentials,
+    OrgContext,
+    SourceContext,
+    _canonicalize_org_names,
+)
+
+
+class _FakeResponse:
+    def __init__(self, payload: dict):
+        self._payload = payload
+
+    def json(self) -> dict:
+        return self._payload
+
+
+class _FakeClient:
+    def __init__(self, payloads: dict[str, dict]):
+        self.payloads = payloads
+        self.get_calls: list[str] = []
+
+    def get(self, path: str) -> _FakeResponse:
+        self.get_calls.append(path)
+        return _FakeResponse(self.payloads[path])
+
+
+def test_canonicalize_org_names_rewrites_org_context_and_caches_response() -> None:
+    client = _FakeClient(
+        {
+            "/orgs/spectertst": {
+                "login": "SpecterTst",
+                "node_id": "O_kgDOCoV2OQ",
+            }
+        }
+    )
+    ctx = SourceContext(
+        organizations=[OrgContext(client=client, org_name="spectertst")]
+    )
+
+    _canonicalize_org_names(ctx)
+
+    assert ctx.organizations[0].org_name == "SpecterTst"
+    assert ctx.organizations_cache == {
+        "SpecterTst": {
+            "login": "SpecterTst",
+            "node_id": "O_kgDOCoV2OQ",
+        }
+    }
+
+
+def test_organizations_reuses_preflight_org_response() -> None:
+    client = _FakeClient(
+        {
+            "/orgs/spectertst": {
+                "login": "SpecterTst",
+                "node_id": "O_kgDOCoV2OQ",
+            },
+            "/orgs/SpecterTst/actions/permissions": {},
+            "/orgs/SpecterTst/actions/permissions/self-hosted-runners": {},
+            "/orgs/SpecterTst/actions/permissions/workflow": {},
+        }
+    )
+    ctx = SourceContext(
+        organizations=[OrgContext(client=client, org_name="spectertst")]
+    )
+
+    _canonicalize_org_names(ctx)
+    rows = list(inspect.unwrap(organizations._pipe.gen)(ctx))
+
+    assert rows[0]["login"] == "SpecterTst"
+    assert client.get_calls == [
+        "/orgs/spectertst",
+        "/orgs/SpecterTst/actions/permissions",
+        "/orgs/SpecterTst/actions/permissions/self-hosted-runners",
+        "/orgs/SpecterTst/actions/permissions/workflow",
+    ]
+
+
+@pytest.mark.parametrize(
+    "credentials",
+    (
+        GithubTokenCredentials(token="token", org_name="spectertst"),
+        GithubOrgAppCredentials(
+            client_id="Iv1.client-id",
+            install_id="12345",
+            key_path="/tmp/github-app.pem",
+            org_name="spectertst",
+        ),
+    ),
+)
+def test_org_only_sources_canonicalize_before_resource_fanout(
+    monkeypatch: pytest.MonkeyPatch,
+    credentials,
+) -> None:
+    source_module = importlib.import_module("openhound_github.source")
+    captured_ctx: dict[str, SourceContext] = {}
+
+    class FakeRESTClient:
+        def __init__(self, **kwargs) -> None:
+            pass
+
+        def get(self, path: str) -> _FakeResponse:
+            assert path == "/orgs/spectertst"
+            return _FakeResponse({"login": "SpecterTst", "node_id": "O_kgDOCoV2OQ"})
+
+    monkeypatch.setattr(source_module, "RESTClient", FakeRESTClient)
+    monkeypatch.setattr(source_module, "GithubInstallation", lambda **_: object())
+    monkeypatch.setattr(
+        source_module, "GitHubAppInstallationAuth", lambda **_: object()
+    )
+
+    def fake_organization_resources(ctx: SourceContext):
+        captured_ctx["ctx"] = ctx
+        return ()
+
+    monkeypatch.setattr(
+        source_module, "organization_resources", fake_organization_resources
+    )
+
+    resources = source_module.source.__wrapped__(
+        credentials=credentials,
+        host="https://api.github.com",
+        emit_legacy_scim_correlations=False,
+    )
+
+    assert resources == ()
+    assert captured_ctx["ctx"].organizations[0].org_name == "SpecterTst"
+    assert (
+        captured_ctx["ctx"].organizations_cache["SpecterTst"]["node_id"]
+        == "O_kgDOCoV2OQ"
+    )