From 1348138c3a91d91c452cae52d825de9dc62a09da Mon Sep 17 00:00:00 2001
From: Subham Sangwan <subhamsangwan26@gmail.com>
Date: Fri, 26 Jun 2026 18:22:49 +0530
Subject: [PATCH] Fix DatabricksSqlHook sqlalchemy_url and get_uri to include
 http_path from connection extra

---
 .../databricks/hooks/databricks_sql.py        | 32 ++++++++------
 .../databricks/hooks/test_databricks_sql.py   | 43 +++++++++++++++++++
 2 files changed, 62 insertions(+), 13 deletions(-)

diff --git a/providers/databricks/src/airflow/providers/databricks/hooks/databricks_sql.py b/providers/databricks/src/airflow/providers/databricks/hooks/databricks_sql.py
index ffc089607c33a..23227431cfded 100644
--- a/providers/databricks/src/airflow/providers/databricks/hooks/databricks_sql.py
+++ b/providers/databricks/src/airflow/providers/databricks/hooks/databricks_sql.py
@@ -184,20 +184,26 @@ def _get_sql_endpoint_by_name(self, endpoint_name) -> dict[str, Any]:
         else:
             return endpoint
 
+    def _resolve_http_path(self) -> str:
+        """Resolve http_path from explicit parameter, endpoint name, or connection extra."""
+        if self._http_path:
+            return self._http_path
+        if self._sql_endpoint_name:
+            endpoint = self._get_sql_endpoint_by_name(self._sql_endpoint_name)
+            self._http_path = endpoint["odbc_params"]["path"]
+            return self._http_path
+        if "http_path" in self.databricks_conn.extra_dejson:
+            self._http_path = self.databricks_conn.extra_dejson["http_path"]
+            return self._http_path
+        raise ValueError(
+            "http_path should be provided either explicitly, "
+            "or in extra parameter of Databricks connection, "
+            "or sql_endpoint_name should be specified"
+        )
+
     def get_conn(self) -> AirflowConnection:
         """Return a Databricks SQL connection object."""
-        if not self._http_path:
-            if self._sql_endpoint_name:
-                endpoint = self._get_sql_endpoint_by_name(self._sql_endpoint_name)
-                self._http_path = endpoint["odbc_params"]["path"]
-            elif "http_path" in self.databricks_conn.extra_dejson:
-                self._http_path = self.databricks_conn.extra_dejson["http_path"]
-            else:
-                raise AirflowException(
-                    "http_path should be provided either explicitly, "
-                    "or in extra parameter of Databricks connection, "
-                    "or sql_endpoint_name should be specified"
-                )
+        self._resolve_http_path()
 
         prev_token = self._token
         new_token = self._get_token(raise_error=True)
@@ -255,7 +261,7 @@ def sqlalchemy_url(self) -> URL:
             )
 
         url_query = {
-            "http_path": self._http_path,
+            "http_path": self._resolve_http_path(),
             "catalog": self.catalog,
             "schema": self.schema,
         }
diff --git a/providers/databricks/tests/unit/databricks/hooks/test_databricks_sql.py b/providers/databricks/tests/unit/databricks/hooks/test_databricks_sql.py
index cd3c00e2839b1..02034d865f398 100644
--- a/providers/databricks/tests/unit/databricks/hooks/test_databricks_sql.py
+++ b/providers/databricks/tests/unit/databricks/hooks/test_databricks_sql.py
@@ -150,6 +150,49 @@ def test_get_uri():
     assert uri == expected_uri
 
 
+CONN_ID_WITH_HTTP_PATH_EXTRA = "databricks_with_http_path_extra"
+
+
+@pytest.fixture
+def create_connection_with_http_path_extra(create_connection_without_db):
+    create_connection_without_db(
+        Connection(
+            conn_id=CONN_ID_WITH_HTTP_PATH_EXTRA,
+            conn_type="databricks",
+            host=HOST,
+            login=None,
+            password=TOKEN,
+            extra={"http_path": HTTP_PATH},
+        )
+    )
+
+
+def test_sqlalchemy_url_with_http_path_from_connection_extra(create_connection_with_http_path_extra):
+    hook = DatabricksSqlHook(databricks_conn_id=CONN_ID_WITH_HTTP_PATH_EXTRA, catalog=CATALOG, schema=SCHEMA)
+    url = hook.sqlalchemy_url.render_as_string(hide_password=False)
+    expected_url = (
+        f"databricks://token:{TOKEN}@{HOST}?"
+        f"catalog={CATALOG}&http_path={quote_plus(HTTP_PATH)}&schema={SCHEMA}"
+    )
+    assert url == expected_url
+
+
+def test_get_uri_with_http_path_from_connection_extra(create_connection_with_http_path_extra):
+    hook = DatabricksSqlHook(databricks_conn_id=CONN_ID_WITH_HTTP_PATH_EXTRA, catalog=CATALOG, schema=SCHEMA)
+    uri = hook.get_uri()
+    expected_uri = (
+        f"databricks://token:{TOKEN}@{HOST}?"
+        f"catalog={CATALOG}&http_path={quote_plus(HTTP_PATH)}&schema={SCHEMA}"
+    )
+    assert uri == expected_uri
+
+
+def test_resolve_http_path_raises_when_not_provided():
+    hook = DatabricksSqlHook(databricks_conn_id=DEFAULT_CONN_ID)
+    with pytest.raises(ValueError, match="http_path should be provided"):
+        hook._resolve_http_path()
+
+
 def get_cursor_descriptions(fields: list[str]) -> list[tuple[str]]:
     return [(field,) for field in fields]