aryn-ai · baitsguy · Jan 31, 2025 · Jan 30, 2025 · Jan 30, 2025 · Jan 30, 2025
diff --git a/lib/sycamore/sycamore/connectors/opensearch/opensearch_reader.py b/lib/sycamore/sycamore/connectors/opensearch/opensearch_reader.py
@@ -41,10 +41,10 @@ def __init__(self, client: "OpenSearch"):
     @classmethod
     @requires_modules("opensearchpy", extra="opensearch")
     def from_client_params(cls, params: BaseDBReader.ClientParams) -> "OpenSearchReaderClient":
-        from opensearchpy import OpenSearch
+        from sycamore.connectors.opensearch.utils import OpenSearchClientWithLogging
 
         assert isinstance(params, OpenSearchReaderClientParams)
-        client = OpenSearch(**params.os_client_args)
+        client = OpenSearchClientWithLogging(**params.os_client_args)
         return OpenSearchReaderClient(client)
 
     def read_records(self, query_params: BaseDBReader.QueryParams) -> "OpenSearchReaderQueryResponse":
@@ -276,7 +276,7 @@ def __init__(
     ):
         assert isinstance(
             query_params, OpenSearchReaderQueryParams
-        ), f"Wrong kind of query parameters found: {self._query_params}"
+        ), f"Wrong kind of query parameters found: {query_params}"
 
         super().__init__(client_params, query_params, **kwargs)
         self._client_params = client_params

diff --git a/lib/sycamore/sycamore/connectors/opensearch/opensearch_writer.py b/lib/sycamore/sycamore/connectors/opensearch/opensearch_writer.py
@@ -100,13 +100,13 @@ def __init__(self, os_client: "OpenSearch"):
     @classmethod
     @requires_modules(["opensearchpy", "opensearchpy.helpers"], extra="opensearch")
     def from_client_params(cls, params: BaseDBWriter.ClientParams) -> "OpenSearchWriterClient":
-        from opensearchpy import OpenSearch
+        from sycamore.connectors.opensearch.utils import OpenSearchClientWithLogging
 
         assert isinstance(
             params, OpenSearchWriterClientParams
         ), f"Provided params was not of type OpenSearchWriterClientParams:\n{params}"
         paramsdict = asdict(params)
-        os_client = OpenSearch(**paramsdict)
+        os_client = OpenSearchClientWithLogging(**paramsdict)
         os_client.ping()
         return OpenSearchWriterClient(os_client)
 

diff --git a/lib/sycamore/sycamore/connectors/opensearch/utils.py b/lib/sycamore/sycamore/connectors/opensearch/utils.py
@@ -1,10 +1,26 @@
-from typing import Optional
+import logging
+from typing import Optional, Any
+
+from opensearchpy import OpenSearch
 
 from sycamore import Context
 from sycamore.context import context_params
 from sycamore.transforms import Embedder
 
 
+logger = logging.getLogger("opensearch")
+
+
+class OpenSearchClientWithLogging(OpenSearch):
+    def search(self, **kwargs) -> Any:
+        """Helper method to execute OpenSearch search queries, and silent errors."""
+        response = super().search(**kwargs)
+        shards = response.get("_shards", {})
+        if shards.get("total") != shards.get("successful"):
+            logger.error(f"OpenSearch query skipped shards: {response}")
+        return response
+
+
 @context_params("opensearch")
 def get_knn_query(
     text_embedder: Embedder,

diff --git a/lib/sycamore/sycamore/query/client.py b/lib/sycamore/sycamore/query/client.py
@@ -18,6 +18,7 @@
 import structlog
 import yaml
 from rich.console import Console
+
 from sycamore.schema import Schema
 
 import sycamore
@@ -131,7 +132,7 @@ def __init__(
         llm: Optional[Union[LLM, str]] = None,
         query_plan_strategy: Optional[QueryPlanStrategy] = None,
     ):
-        from opensearchpy import OpenSearch
+        from sycamore.connectors.opensearch.utils import OpenSearchClientWithLogging
 
         self.llm_cache_dir = llm_cache_dir
         self.os_config = os_config
@@ -155,7 +156,7 @@ def __init__(
 
         assert self.context.params, "Could not find required params in Context"
         self.os_client_args = self.context.params.get("opensearch", {}).get("os_client_args", os_client_args)
-        self._os_client = OpenSearch(**self.os_client_args)
+        self._os_client = OpenSearchClientWithLogging(**self.os_client_args)
         self._os_query_executor = OpenSearchQueryExecutor(self.os_client_args)
 
     def get_opensearch_indices(self) -> List[str]:

diff --git a/lib/sycamore/sycamore/tests/integration/connectors/opensearch/test_html_to_opensearch.py b/lib/sycamore/sycamore/tests/integration/connectors/opensearch/test_html_to_opensearch.py
@@ -2,9 +2,9 @@
 import os
 import tempfile
 
-from opensearchpy import OpenSearch
 import sycamore
 from sycamore.connectors.file.file_scan import JsonManifestMetadataProvider
+from sycamore.connectors.opensearch.utils import OpenSearchClientWithLogging
 from sycamore.tests.config import TEST_DIR
 from sycamore.transforms.embed import SentenceTransformerEmbedder
 from sycamore.transforms.partition import HtmlPartitioner
@@ -75,4 +75,4 @@ def test_html_to_opensearch(exec_mode):
         ds.write.opensearch(os_client_args=os_client_args, index_name="toyindex", index_settings=index_settings)
     finally:
         tmp_manifest.close()
-        OpenSearch(**os_client_args).indices.delete("toyindex")
+        OpenSearchClientWithLogging(**os_client_args).indices.delete("toyindex")
diff --git a/lib/sycamore/sycamore/tests/integration/connectors/opensearch/test_opensearch_read.py b/lib/sycamore/sycamore/tests/integration/connectors/opensearch/test_opensearch_read.py
@@ -5,7 +5,6 @@
 from typing import Optional, Dict, Any
 
 import pytest
-from opensearchpy import OpenSearch
 
 import sycamore
 from sycamore import EXEC_LOCAL, ExecMode
@@ -21,7 +20,9 @@
 
 @pytest.fixture(scope="class")
 def os_client():
-    client = OpenSearch(**TestOpenSearchRead.OS_CLIENT_ARGS)
+    from sycamore.connectors.opensearch.utils import OpenSearchClientWithLogging
+
+    client = OpenSearchClientWithLogging(**TestOpenSearchRead.OS_CLIENT_ARGS)
     yield client
 
 

diff --git a/lib/sycamore/sycamore/tests/integration/connectors/opensearch/test_pdf_to_opensearch.py b/lib/sycamore/sycamore/tests/integration/connectors/opensearch/test_pdf_to_opensearch.py
@@ -3,9 +3,9 @@
 
 import boto3
 from urllib.parse import urlparse
-from opensearchpy import OpenSearch
 
 import sycamore
+from sycamore.connectors.opensearch.utils import OpenSearchClientWithLogging
 from sycamore.context import OperationTypes, ExecMode
 from sycamore.functions import HuggingFaceTokenizer
 from sycamore.llms import OpenAIModels, OpenAI
@@ -143,7 +143,7 @@ def test_pdf_to_opensearch_with_llm_caching():
         )
         ds.write.opensearch()
 
-        OpenSearch(**os_client_args).indices.delete("toyindex")
+        OpenSearchClientWithLogging(**os_client_args).indices.delete("toyindex")
 
         # validate caching
 

diff --git a/lib/sycamore/sycamore/tests/integration/query/conftest.py b/lib/sycamore/sycamore/tests/integration/query/conftest.py
@@ -1,8 +1,8 @@
 import os
 import pytest
-from opensearchpy import OpenSearch
 
 import sycamore
+from sycamore.connectors.opensearch.utils import OpenSearchClientWithLogging
 from sycamore.functions import HuggingFaceTokenizer
 from sycamore.tests.config import TEST_DIR
 from sycamore.transforms.embed import SentenceTransformerEmbedder
@@ -69,7 +69,7 @@ def query_integration_test_index():
         index_name=QUERY_INTEGRATION_TEST_INDEX_NAME,
         index_settings=index_settings,
     )
-    osc = OpenSearch(**OS_CLIENT_ARGS)
+    osc = OpenSearchClientWithLogging(**OS_CLIENT_ARGS)
     osc.indices.refresh(QUERY_INTEGRATION_TEST_INDEX_NAME)
     yield QUERY_INTEGRATION_TEST_INDEX_NAME
     osc.indices.delete(QUERY_INTEGRATION_TEST_INDEX_NAME)
diff --git a/lib/sycamore/sycamore/tests/integration/query/test_planner.py b/lib/sycamore/sycamore/tests/integration/query/test_planner.py
@@ -1,5 +1,4 @@
-from opensearchpy import OpenSearch
-
+from sycamore.connectors.opensearch.utils import OpenSearchClientWithLogging
 from sycamore.tests.integration.query.conftest import OS_CLIENT_ARGS, OS_CONFIG
 from sycamore.query.planner import LlmPlanner
 from sycamore.query.schema import OpenSearchSchema, OpenSearchSchemaField
@@ -10,7 +9,7 @@ def test_simple_llm_planner(query_integration_test_index: str):
     Simple test ensuring nodes are being created and dependencies are being set.
     Using a simple query here for consistent query plans.
     """
-    os_client = OpenSearch(OS_CLIENT_ARGS)
+    os_client = OpenSearchClientWithLogging(OS_CLIENT_ARGS)
 
     schema = OpenSearchSchema(
         fields={

diff --git a/lib/sycamore/sycamore/tests/integration/query/test_query_opensearch.py b/lib/sycamore/sycamore/tests/integration/query/test_query_opensearch.py
@@ -1,12 +1,12 @@
 import pytest
 import sycamore
 from sycamore import ExecMode
+from sycamore.connectors.opensearch.utils import OpenSearchClientWithLogging
 from sycamore.data.document import OpenSearchQuery
 from sycamore.tests.config import TEST_DIR
 from sycamore.transforms.embed import SentenceTransformerEmbedder
 from sycamore.transforms.partition import UnstructuredPdfPartitioner
 from sycamore.transforms.query import OpenSearchQueryExecutor
-from opensearchpy import OpenSearch
 
 
 @pytest.fixture(scope="class")
@@ -47,7 +47,7 @@ def setup_index():
         index_name=TestQueryOpenSearch.INDEX,
         index_settings=index_settings,
     )
-    osc = OpenSearch(**TestQueryOpenSearch.OS_CLIENT_ARGS)
+    osc = OpenSearchClientWithLogging(**TestQueryOpenSearch.OS_CLIENT_ARGS)
     osc.indices.refresh(TestQueryOpenSearch.INDEX)
     yield TestQueryOpenSearch.INDEX
     osc.indices.delete(TestQueryOpenSearch.INDEX)

diff --git a/lib/sycamore/sycamore/transforms/query.py b/lib/sycamore/sycamore/transforms/query.py
@@ -1,5 +1,6 @@
 from abc import abstractmethod, ABC
 from typing import Any
+
 from sycamore.utils.import_utils import requires_modules
 
 from sycamore.data import OpenSearchQueryResult, Element, OpenSearchQuery
@@ -26,10 +27,10 @@ def __init__(self, os_client_args: dict) -> None:
 
     @requires_modules("opensearchpy", extra="opensearch")
     def query(self, query: OpenSearchQuery) -> OpenSearchQueryResult:
-        from opensearchpy import OpenSearch
+        from sycamore.connectors.opensearch.utils import OpenSearchClientWithLogging
 
         logger.debug("Executing OS query: " + str(query))
-        client = OpenSearch(**self._os_client_args)
+        client = OpenSearchClientWithLogging(**self._os_client_args)
 
         os_result = client.transport.perform_request(
             "POST",