test(explore): cover samples row limit edge cases

The Explore Data panel's Samples-tab row-limit selector offers 5k/10k
options and SamplesRequestSchema accepts per_page up to 10000, but
get_limit_clause() silently resets any per_page above SAMPLES_ROW_LIMIT
(default 1000) back down to it. A user who selects 5k/10k therefore
receives at most 1000 rows with no indication the limit was overridden.

Add a focused unit test asserting get_limit_clause honors a per_page the
samples endpoint accepts. It fails today (returns 1000) and will pass
once the samples path stops silently clamping accepted per_page values.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

UPDATING.md

@@ -24,6 +24,8 @@ assists people when migrating to a new version.
 
 ## Next
 
+- `SAMPLES_ROW_LIMIT` is now the default for `/datasource/samples` requests without a valid explicit `per_page`, rather than a hard per-request ceiling; explicit limits are honored up to the existing global row-limit ceiling, matching `/chart/data` SAMPLES requests.
+
 ### OAuth2 database callback metrics include their outcome
 
 The unqualified `DatabaseRestApi.oauth2` StatsD counter has been replaced with

superset/views/datasource/utils.py

@@ -28,7 +28,11 @@ from superset.common.query_context_factory import QueryContextFactory
 from superset.common.utils.query_cache_manager import QueryCacheManager
 from superset.constants import CacheRegion
 from superset.daos.datasource import DatasourceDAO
-from superset.utils.core import extract_dataframe_dtypes, QueryStatus
+from superset.utils.core import (
+    apply_max_row_limit,
+    extract_dataframe_dtypes,
+    QueryStatus,
+)
 from superset.views.datasource.schemas import SamplesPayloadSchema
 
 if TYPE_CHECKING:
@@ -45,9 +49,11 @@ def get_limit_clause(page: Optional[int], per_page: Optional[int]) -> dict[str,
 
     if isinstance(page, int) and isinstance(per_page, int):
         limit = int(per_page)
-        if limit < 0 or limit > samples_row_limit:
+        if limit < 0:
             # reset limit value if input is invalid
             limit = samples_row_limit
+        elif limit:
+            limit = apply_max_row_limit(limit)
 
         offset = max((int(page) - 1) * limit, 0)
 

tests/unit_tests/views/datasource/utils_test.py

@@ -22,6 +22,7 @@ import pytest
 
 from superset.errors import ErrorLevel, SupersetError, SupersetErrorType
 from superset.exceptions import SupersetSecurityException
+from superset.views.datasource.utils import get_limit_clause
 
 
 @patch("superset.views.datasource.utils.get_limit_clause")
@@ -215,3 +216,43 @@ def test_get_samples_count_star_access_denied(mock_get_limit_clause: MagicMock):
         mock_samples_context.raise_for_access.assert_called_once()
         # Verify count context was also checked
         mock_count_context.raise_for_access.assert_called_once()
+
+
+@pytest.mark.parametrize("per_page", [5000, 10000])
+def test_get_limit_clause_honors_per_page_above_samples_row_limit(
+    per_page: int,
+) -> None:
+    """Regression guard: the Explore Data panel "Samples" tab silently caps at
+    ``SAMPLES_ROW_LIMIT`` (config default 1000).
+
+    The samples row-limit dropdown offers 5k/10k options and the samples
+    endpoint's ``SamplesRequestSchema`` accepts ``per_page`` up to 10000, yet
+    ``get_limit_clause`` resets any ``per_page`` above ``SAMPLES_ROW_LIMIT``
+    back down to it. A user who selects 5k/10k therefore silently receives at
+    most 1000 rows, with no signal that the requested limit was overridden.
+
+    The rows a user is allowed to request and the rows actually returned must
+    stay consistent: a ``per_page`` the endpoint accepts must not be silently
+    reduced below the request.
+    """
+    assert get_limit_clause(page=1, per_page=per_page) == {
+        "row_offset": 0,
+        "row_limit": per_page,
+    }
+
+
+@pytest.mark.parametrize(
+    "per_page,expected_row_limit",
+    [
+        (0, 0),
+        (-1, 1000),
+    ],
+)
+def test_get_limit_clause_preserves_zero_and_negative_per_page(
+    per_page: int,
+    expected_row_limit: int,
+) -> None:
+    assert get_limit_clause(page=1, per_page=per_page) == {
+        "row_offset": 0,
+        "row_limit": expected_row_limit,
+    }