fix(kernel): forward full OAuth U2M app bundle into kernel (PECOBLR-4040)

CHANGELOG.md

@@ -1,5 +1,8 @@
 # Release History
 
+# Unreleased
+- Kernel backend (`use_kernel=True`): OAuth U2M with `auth_type="databricks-oauth"` now forwards the connector's `databricks-sql-python` OAuth-app bundle (`client_id` + `sql offline_access` scopes + redirect port) into the kernel, so a bare U2M connection authenticates as `databricks-sql-python` — parity with the Thrift path — instead of inheriting the kernel's own `databricks-sql-connector` default. A caller-supplied `oauth_client_id` (with its coupled `oauth_redirect_port`) is honored; scopes are fixed to match the Thrift path. `auth_type="azure-oauth"` (Azure AD) is not yet supported on the kernel path and raises `NotSupportedError` — use the Thrift backend for it (PECOBLR-4040; Azure tracked by PECOBLR-4120)
+
 # 4.4.0 (2026-07-22)
 - Raised the minimum supported Python version to 3.10, dropping the end-of-life 3.8/3.9, to update the lockfile and clear CVE-flagged dependencies in the repo (databricks/databricks-sql-python#798)
 - Fix: `REMOVE` staging operations no longer require `staging_allowed_local_path` to be set, since removing a remote file does not touch the local filesystem (databricks/databricks-sql-python#726)

src/databricks/sql/backend/kernel/auth_bridge.py

@@ -15,11 +15,16 @@
   connector's own OAuth provider because the kernel re-mints tokens
   itself and the client secret is not recoverable from a built
   provider.
-- **OAuth U2M** — for ``auth_type`` ``databricks-oauth`` /
-  ``azure-oauth`` (the browser authorization-code flow), the optional
-  ``oauth_client_id`` / ``oauth_redirect_port`` are forwarded to the
-  kernel's ``auth_type='oauth-u2m'`` and the kernel runs the browser
-  flow itself.
+- **OAuth U2M** — for ``auth_type`` ``databricks-oauth`` (the browser
+  authorization-code flow), the connector's ``databricks-sql-python``
+  app bundle (``client_id`` + ``redirect_port``, with the optional
+  ``oauth_client_id`` / ``oauth_redirect_port`` overriding it) is
+  forwarded to the kernel's ``auth_type='oauth-u2m'`` and the kernel
+  runs the browser flow itself. ``azure-oauth`` (Azure AD) is **not yet
+  supported** on the kernel path and is rejected with
+  ``NotSupportedError`` — the kernel resolves OAuth endpoints only from
+  the workspace-native OIDC config and cannot drive the Azure AD flow
+  (PECOBLR-4120).
 
 ``identity_federation_client_id`` is forwarded with whichever auth shape
 wins resolution. It selects mandatory SP-wide workload-identity token
@@ -48,6 +53,11 @@
 import re
 from typing import Any, Dict, Optional
 
+from databricks.sql.auth.auth import (
+    PYSQL_OAUTH_CLIENT_ID,
+    PYSQL_OAUTH_REDIRECT_PORT_RANGE,
+    PYSQL_OAUTH_SCOPES,
+)
 from databricks.sql.auth.authenticators import AccessTokenAuthProvider, AuthProvider
 from databricks.sql.auth.token_federation import TokenFederationProvider
 from databricks.sql.exc import NotSupportedError, ProgrammingError
@@ -141,15 +151,23 @@ def kernel_auth_kwargs(
        rather than silently picking one flow (and failing later as a
        confusing 401 against the wrong principal):
        - a custom ``credentials_provider`` *and* M2M kwargs together;
-       - a U2M ``auth_type`` (``databricks-oauth`` / ``azure-oauth``)
-         *and* ``oauth_client_secret`` together.
+       - a U2M ``auth_type`` (``databricks-oauth``) *and*
+         ``oauth_client_secret`` together.
+
+    (``azure-oauth`` is rejected as unsupported before these guards —
+    PECOBLR-4120.)
     1. **OAuth M2M** — ``oauth_client_id`` + ``oauth_client_secret``
        both present → forward raw creds to the kernel's ``oauth-m2m``.
     2. **PAT** — the built provider is (or wraps) an
        ``AccessTokenAuthProvider`` → extract the bearer token.
-    3. **OAuth U2M** — ``auth_type`` is ``databricks-oauth`` /
-       ``azure-oauth`` → forward optional ``oauth_client_id`` /
-       ``oauth_redirect_port`` to the kernel's ``oauth-u2m``.
+    3. **OAuth U2M** — ``auth_type`` is ``databricks-oauth`` → forward the
+       connector's coupled ``databricks-sql-python`` bundle (``client_id``
+       + ``redirect_port``, with fixed ``PYSQL_OAUTH_SCOPES``) to the
+       kernel's ``oauth-u2m``, so a bare U2M connection authenticates as
+       ``databricks-sql-python`` — parity with the Thrift path — rather
+       than the kernel's own ``databricks-sql-connector`` default
+       (PECOBLR-4039/4040). ``azure-oauth`` is rejected as unsupported
+       (PECOBLR-4120).
     4. **Custom credentials_provider** → ``NotSupportedError`` (opaque
        token source; no raw creds for the kernel to own).
     5. Anything else → ``NotSupportedError``.
@@ -169,6 +187,25 @@ def kernel_auth_kwargs(
     auth_type = opts.get("auth_type")
     has_m2m = bool(client_id and client_secret)
 
+    # azure-oauth (Azure AD U2M) is not yet supported on the kernel path.
+    # Reject it up front — before any M2M/U2M routing — so ANY azure-oauth
+    # request gets a clear "not supported" error rather than being silently
+    # misrouted (e.g. azure-oauth + client_id + secret would otherwise look
+    # like M2M). The kernel resolves OAuth endpoints only from the
+    # workspace-native OIDC config and has no Azure AD path, so the Thrift
+    # azure-oauth flow (AAD token endpoint + /user_impersonation scope, see
+    # AzureOAuthEndpointCollection) cannot be reproduced here. Forwarding an
+    # azure bundle would authenticate against the wrong endpoints, so we fail
+    # loudly at session-open. Tracked by PECOBLR-4120.
+    if auth_type == "azure-oauth":
+        raise NotSupportedError(
+            "use_kernel=True does not support auth_type='azure-oauth' (Azure "
+            "AD U2M) yet: the kernel resolves OAuth endpoints only from the "
+            "workspace-native OIDC configuration and cannot drive the Azure AD "
+            "authorization/token flow. Use the Thrift backend (default) for "
+            "azure-oauth. Tracked by PECOBLR-4120."
+        )
+
     # 0. Ambiguity guards — fail before any flow is chosen.
     if client_secret and opts.get("credentials_provider") is not None:
         raise NotSupportedError(
@@ -178,7 +215,7 @@ def kernel_auth_kwargs(
             "kernel-managed M2M, or use the Thrift backend (default) for "
             "credentials_provider."
         )
-    if client_secret and auth_type in ("databricks-oauth", "azure-oauth"):
+    if client_secret and auth_type == "databricks-oauth":
         raise NotSupportedError(
             f"Ambiguous auth on use_kernel=True: auth_type={auth_type!r} selects "
             "the U2M browser flow, but oauth_client_secret was also provided "
@@ -214,16 +251,29 @@ def kernel_auth_kwargs(
         return kwargs
 
     # 3. OAuth U2M — browser authorization-code flow; the kernel runs it.
-    if auth_type in ("databricks-oauth", "azure-oauth"):
-        kwargs = {"auth_type": "oauth-u2m"}
-        if client_id:
-            kwargs["client_id"] = client_id
+    #    Only databricks-oauth reaches here (azure-oauth rejected up front).
+    #    Forward the connector's own databricks-sql-python bundle instead of
+    #    the kernel's databricks-sql-connector default, for parity with the
+    #    Thrift path. client_id + redirect_port are coupled per app (each
+    #    registers its own redirect URI): a caller port only overrides the
+    #    default when an explicit client_id is also supplied. A caller may
+    #    override oauth_scopes; absent one we forward PYSQL_OAUTH_SCOPES as
+    #    the default.
+    if auth_type == "databricks-oauth":
         redirect_port = opts.get("oauth_redirect_port")
-        if redirect_port is not None:
-            kwargs["redirect_port"] = int(redirect_port)
+        # Honor a caller-supplied oauth_scopes (normalized to a list of
+        # strings); fall back to the connector default when none is given.
         scopes = _normalize_scopes(opts.get("oauth_scopes"))
-        if scopes is not None:
-            kwargs["oauth_scopes"] = scopes
+        kwargs = {
+            "auth_type": "oauth-u2m",
+            "client_id": client_id or PYSQL_OAUTH_CLIENT_ID,
+            "redirect_port": (
+                int(redirect_port)
+                if client_id and redirect_port is not None
+                else PYSQL_OAUTH_REDIRECT_PORT_RANGE[0]
+            ),
+            "oauth_scopes": scopes if scopes is not None else list(PYSQL_OAUTH_SCOPES),
+        }
         if federation_client_id:
             kwargs["identity_federation_client_id"] = federation_client_id
         return kwargs
@@ -251,7 +301,7 @@ def kernel_auth_kwargs(
     raise NotSupportedError(
         f"use_kernel=True requires PAT (access_token), OAuth M2M "
         f"(oauth_client_id + oauth_client_secret), or OAuth U2M "
-        f"(auth_type='databricks-oauth' / 'azure-oauth'), but got "
+        f"(auth_type='databricks-oauth'), but got "
         f"{provider_desc} with auth_type={auth_type!r}. Use the Thrift "
         "backend (default) for other auth flows."
     )

tests/unit/test_kernel_auth_bridge.py

@@ -8,8 +8,9 @@
     look through the wrapper).
   - OAuth M2M (``oauth_client_id`` + ``oauth_client_secret``) routes
     through ``auth_type='oauth-m2m'`` with the raw creds forwarded.
-  - OAuth U2M (``auth_type='databricks-oauth'`` / ``'azure-oauth'``)
-    routes through ``auth_type='oauth-u2m'``.
+  - OAuth U2M (``auth_type='databricks-oauth'``) routes through
+    ``auth_type='oauth-u2m'``. ``azure-oauth`` (Azure AD) is not yet
+    supported on the kernel path and is rejected (PECOBLR-4120).
   - A custom ``credentials_provider`` and any other non-PAT shape raise
     ``NotSupportedError`` with a clear, actionable message.
 """
@@ -26,6 +27,11 @@
 # require the kernel wheel). So this test can run on the
 # default-deps CI matrix without any extras. No importorskip needed.
 
+from databricks.sql.auth.auth import (
+    PYSQL_OAUTH_CLIENT_ID,
+    PYSQL_OAUTH_SCOPES,
+    PYSQL_OAUTH_REDIRECT_PORT_RANGE,
+)
 from databricks.sql.auth.authenticators import (
     AccessTokenAuthProvider,
     AuthProvider,
@@ -240,34 +246,142 @@ def test_client_id_without_secret_does_not_trigger_m2m(self):
 
 
 class TestKernelOAuthU2M:
-    @pytest.mark.parametrize("auth_type", ["databricks-oauth", "azure-oauth"])
-    def test_u2m_routes_to_kernel_u2m(self, auth_type):
+    """Only ``databricks-oauth`` U2M is supported on the kernel path.
+
+    The kernel core default U2M app is ``databricks-sql-connector`` /
+    ``sql offline_access`` / port 8030 (see PECOBLR-4039). The Python
+    connector is an OVERRIDE: on the kernel path it forwards its OWN
+    coupled ``client_id`` + ``redirect_port`` bundle so it authenticates
+    as ``databricks-sql-python`` rather than the kernel default. A caller
+    may override ``oauth_scopes``; absent one, ``PYSQL_OAUTH_SCOPES`` is
+    forwarded as the default.
+
+    ``azure-oauth`` (Azure AD) is deliberately NOT handled yet — the
+    kernel can't drive the Azure AD authorization/token flow — so it is
+    rejected up front (PECOBLR-4120)."""
+
+    def test_bare_databricks_oauth_forwards_full_python_bundle(self):
+        # No overrides → forward the databricks-sql-python bundle in full
+        # so the kernel does NOT fall back to its databricks-sql-connector
+        # default. This is the parity-with-Thrift acceptance criterion.
         kwargs = kernel_auth_kwargs(
             _FakeOAuthProvider(),
-            {"auth_type": auth_type},
+            {"auth_type": "databricks-oauth"},
         )
-        assert kwargs == {"auth_type": "oauth-u2m"}
+        assert kwargs == {
+            "auth_type": "oauth-u2m",
+            "client_id": PYSQL_OAUTH_CLIENT_ID,
+            "redirect_port": PYSQL_OAUTH_REDIRECT_PORT_RANGE[0],
+            "oauth_scopes": list(PYSQL_OAUTH_SCOPES),
+        }
 
-    def test_u2m_forwards_client_id_and_redirect_port(self):
+    @pytest.mark.parametrize(
+        "opts",
+        [
+            {"auth_type": "azure-oauth"},
+            {"auth_type": "azure-oauth", "oauth_client_id": "custom"},
+            {"auth_type": "azure-oauth", "oauth_redirect_port": 8030},
+        ],
+        ids=["bare", "with_client_id", "with_port"],
+    )
+    def test_azure_oauth_not_supported(self, opts):
+        # azure-oauth (Azure AD U2M) can't work through the kernel yet: the
+        # kernel resolves OAuth endpoints only from workspace-native OIDC
+        # discovery and has no Azure AD path. Fail loudly at session-open
+        # rather than forwarding a bundle that authenticates against the
+        # wrong endpoints. Tracked by PECOBLR-4120.
+        with pytest.raises(NotSupportedError, match="azure-oauth"):
+            kernel_auth_kwargs(_FakeOAuthProvider(), opts)
+
+    def test_u2m_custom_client_id_port_and_scopes_honored(self):
+        # A caller may override the coupled client_id + redirect_port and
+        # the oauth_scopes. All three are forwarded as supplied.
         kwargs = kernel_auth_kwargs(
             _FakeOAuthProvider(),
             {
                 "auth_type": "databricks-oauth",
                 "oauth_client_id": "custom-client",
-                "oauth_redirect_port": 8030,
+                "oauth_scopes": ["custom-scope", "offline_access"],
+                "oauth_redirect_port": 9999,
             },
         )
         assert kwargs == {
             "auth_type": "oauth-u2m",
             "client_id": "custom-client",
-            "redirect_port": 8030,
+            "redirect_port": 9999,
+            "oauth_scopes": ["custom-scope", "offline_access"],
         }
 
-    @pytest.mark.parametrize("auth_type", ["databricks-oauth", "azure-oauth"])
-    def test_u2m_forwards_scopes(self, auth_type):
+    def test_u2m_custom_client_id_only_falls_back_to_connector_defaults(self):
+        # A custom client_id without explicit scopes/port fills the
+        # remaining two from the connector defaults — a custom client_id
+        # still uses PYSQL_OAUTH_SCOPES and the default redirect-port range.
+        kwargs = kernel_auth_kwargs(
+            _FakeOAuthProvider(),
+            {
+                "auth_type": "databricks-oauth",
+                "oauth_client_id": "custom-client",
+            },
+        )
+        assert kwargs == {
+            "auth_type": "oauth-u2m",
+            "client_id": "custom-client",
+            "redirect_port": PYSQL_OAUTH_REDIRECT_PORT_RANGE[0],
+            "oauth_scopes": list(PYSQL_OAUTH_SCOPES),
+        }
+
+    def test_u2m_redirect_port_coerced_to_int(self):
+        # oauth_redirect_port may arrive as a string (e.g. from a DSN);
+        # the kernel binding wants an int. The port override is coupled to
+        # an explicit client_id (see the coupling test below), so supply
+        # one here to exercise the coercion path.
+        kwargs = kernel_auth_kwargs(
+            _FakeOAuthProvider(),
+            {
+                "auth_type": "databricks-oauth",
+                "oauth_client_id": "custom-client",
+                "oauth_redirect_port": "8021",
+            },
+        )
+        assert kwargs["redirect_port"] == 8021
+        assert isinstance(kwargs["redirect_port"], int)
+
+    def test_u2m_redirect_port_ignored_without_client_id(self):
+        # A bare oauth_redirect_port (no explicit client_id) must NOT be
+        # forwarded: it would be paired with the default databricks-sql-python
+        # app, whose registered redirect URIs only cover the default port
+        # range, so an arbitrary port would resolve to an unregistered URI
+        # and fail the U2M flow. This mirrors the Thrift path's coupling,
+        # where oauth_redirect_port_range is only overridden when both
+        # oauth_client_id and oauth_redirect_port are supplied.
+        kwargs = kernel_auth_kwargs(
+            _FakeOAuthProvider(),
+            {"auth_type": "databricks-oauth", "oauth_redirect_port": 9999},
+        )
+        assert kwargs["redirect_port"] == PYSQL_OAUTH_REDIRECT_PORT_RANGE[0]
+
+    def test_u2m_honors_custom_scopes(self):
+        # A caller-supplied oauth_scopes is forwarded to the kernel, even
+        # without an explicit client_id. Absent one, PYSQL_OAUTH_SCOPES is
+        # forwarded as the default (see the bare-bundle test above).
         kwargs = kernel_auth_kwargs(
             _FakeOAuthProvider(),
-            {"auth_type": auth_type, "oauth_scopes": ["all-apis", "offline_access"]},
+            {
+                "auth_type": "databricks-oauth",
+                "oauth_scopes": ["all-apis", "offline_access"],
+            },
+        )
+        assert kwargs["oauth_scopes"] == ["all-apis", "offline_access"]
+
+    def test_u2m_normalizes_space_delimited_scopes(self):
+        # A space-delimited oauth_scopes string is normalized to a list,
+        # mirroring the M2M path.
+        kwargs = kernel_auth_kwargs(
+            _FakeOAuthProvider(),
+            {
+                "auth_type": "databricks-oauth",
+                "oauth_scopes": "all-apis offline_access",
+            },
         )
         assert kwargs["oauth_scopes"] == ["all-apis", "offline_access"]
 
@@ -331,15 +445,15 @@ def _creds_provider():
                 },
             )
 
-    @pytest.mark.parametrize("auth_type", ["databricks-oauth", "azure-oauth"])
-    def test_u2m_auth_type_plus_client_secret_is_rejected(self, auth_type):
+    def test_u2m_auth_type_plus_client_secret_is_rejected(self):
         # User asked for U2M (browser) but also passed a secret (M2M).
-        # Don't silently route M2M against the wrong principal.
+        # Don't silently route M2M against the wrong principal. (azure-oauth
+        # is rejected earlier as unsupported, so it's not exercised here.)
         with pytest.raises(NotSupportedError, match="Ambiguous auth"):
             kernel_auth_kwargs(
                 _FakeOAuthProvider(),
                 {
-                    "auth_type": auth_type,
+                    "auth_type": "databricks-oauth",
                     "oauth_client_id": "id",
                     "oauth_client_secret": "sec",
                 },