Added fix for key error because of missing 'hits' key. (#616)

Updated CHANGELOG.md.



nox formatting applied.



Added new unit test for actions scan function.



Added type hints & nox formatting.



Added fix to async scan function & added matching unit tests for async.

Signed-off-by: Djcarrillo6 <[email protected]>

CHANGELOG.md

@@ -12,6 +12,7 @@ Inspired from [Keep a Changelog](https://keepachangelog.com/en/1.0.0/)
 ### Removed
 - Removed unnecessary `# -*- coding: utf-8 -*-` headers from .py files ([#615](https://github.com/opensearch-project/opensearch-py/pull/615), [#617](https://github.com/opensearch-project/opensearch-py/pull/617))
 ### Fixed
+- Fix KeyError when scroll return no hits ([#616](https://github.com/opensearch-project/opensearch-py/pull/616))
 ### Security
 
 ## [2.4.2]

opensearchpy/_async/helpers/actions.py

@@ -390,14 +390,17 @@ async def async_scan(
     scroll_id = resp.get("_scroll_id")
 
     try:
-        while scroll_id and resp["hits"]["hits"]:
+        while scroll_id and resp.get("hits", {}).get("hits"):
-            for hit in resp["hits"]["hits"]:
+            for hit in resp.get("hits", {}).get("hits", []):
                 yield hit
 
+            _shards = resp.get("_shards")
+
+            if _shards:
                 # Default to 0 if the value isn't included in the response
-            shards_successful = resp["_shards"].get("successful", 0)
+                shards_successful = _shards.get("successful", 0)
-            shards_skipped = resp["_shards"].get("skipped", 0)
+                shards_skipped = _shards.get("skipped", 0)
-            shards_total = resp["_shards"].get("total", 0)
+                shards_total = _shards.get("total", 0)
 
             # check if we have any errors
             if (shards_successful + shards_skipped) < shards_total:

opensearchpy/helpers/actions.py

@@ -586,14 +586,17 @@ def scan(
     scroll_id = resp.get("_scroll_id")
 
     try:
-        while scroll_id and resp["hits"]["hits"]:
+        while scroll_id and resp.get("hits", {}).get("hits"):
-            for hit in resp["hits"]["hits"]:
+            for hit in resp.get("hits", {}).get("hits", []):
                 yield hit
 
+            _shards = resp.get("_shards")
+
+            if _shards:
                 # Default to 0 if the value isn't included in the response
-            shards_successful = resp["_shards"].get("successful", 0)
+                shards_successful = _shards.get("successful", 0)
-            shards_skipped = resp["_shards"].get("skipped", 0)
+                shards_skipped = _shards.get("skipped", 0)
-            shards_total = resp["_shards"].get("total", 0)
+                shards_total = _shards.get("total", 0)
 
             # check if we have any errors
             if (shards_successful + shards_skipped) < shards_total:
@@ -614,6 +617,7 @@ def scan(
                             shards_total,
                         ),
                     )
+
             resp = client.scroll(
                 body={"scroll_id": scroll_id, "scroll": scroll}, **scroll_kwargs
             )

test_opensearchpy/test_async/test_server/test_helpers/test_actions.py

@@ -776,6 +776,34 @@ class TestScan(object):
                     }
                     assert async_client.scroll.call_args[1]["sort"] == "asc"
 
+    async def test_async_scan_with_missing_hits_key(
+        self, async_client: Any, scan_teardown: Any
+    ) -> None:
+        with patch.object(
+            async_client,
+            "search",
+            return_value=MockResponse({"_scroll_id": "dummy_scroll_id", "_shards": {}}),
+        ):
+            with patch.object(
+                async_client,
+                "scroll",
+                return_value=MockResponse(
+                    {"_scroll_id": "dummy_scroll_id", "_shards": {}}
+                ),
+            ):
+                with patch.object(
+                    async_client, "clear_scroll", return_value=MockResponse({})
+                ):
+                    async_scan_result = [
+                        hit
+                        async for hit in actions.async_scan(
+                            async_client, query={"query": {"match_all": {}}}
+                        )
+                    ]
+                    assert (
+                        async_scan_result == []
+                    ), "Expected empty results when 'hits' key is missing"
+
 
 @pytest.fixture(scope="function")  # type: ignore
 async def reindex_setup(async_client: Any) -> Any:

test_opensearchpy/test_helpers/test_actions.py

@@ -28,6 +28,7 @@
 import threading
 import time
 from typing import Any
+from unittest.mock import Mock
 
 import mock
 import pytest
@@ -270,3 +271,24 @@ class TestExpandActions(TestCase):
         self.assertEqual(
             ('{"index":{}}', "whatever"), helpers.expand_action("whatever")
         )
+
+
+class TestScanFunction(TestCase):
+    @mock.patch("opensearchpy.OpenSearch.clear_scroll")
+    @mock.patch("opensearchpy.OpenSearch.scroll")
+    @mock.patch("opensearchpy.OpenSearch.search")
+    def test_scan_with_missing_hits_key(
+        self, mock_search: Mock, mock_scroll: Mock, mock_clear_scroll: Mock
+    ) -> None:
+        # Simulate a response where the 'hits' key is missing
+        mock_search.return_value = {"_scroll_id": "dummy_scroll_id", "_shards": {}}
+
+        mock_scroll.side_effect = [{"_scroll_id": "dummy_scroll_id", "_shards": {}}]
+
+        mock_clear_scroll.return_value = None
+
+        client = OpenSearch()
+
+        # The test should pass without raising a KeyError
+        scan_result = list(helpers.scan(client, query={"query": {"match_all": {}}}))
+        assert scan_result == [], "Expected empty results when 'hits' key is missing"