community: Azure Search Vector Store is missing Access Token Authentication (#24330)

Added Azure Search Access Token Authentication instead of API KEY auth.
Fixes Issue: https://github.com/langchain-ai/langchain/issues/24263
Dependencies: None
Twitter: @levalencia

@baskaryan

Could you please review? First time creating a PR that fixes some code.

---------

Co-authored-by: Erick Friis <erick@langchain.dev>

libs/community/extended_testing_deps.txt

@@ -16,7 +16,6 @@ cloudpickle>=2.0.0
 cohere>=4,<6
 databricks-vectorsearch>=0.21,<0.22
 datasets>=2.15.0,<3
-dedoc>=2.2.6,<3
 dgml-utils>=0.3.0,<0.4
 elasticsearch>=8.12.0,<9
 esprima>=4.0.1,<5

libs/community/langchain_community/vectorstores/azuresearch.py

@@ -5,6 +5,7 @@ import base64
 import itertools
 import json
 import logging
+import time
 import uuid
 from typing import (
     TYPE_CHECKING,
@@ -79,8 +80,9 @@ MAX_UPLOAD_BATCH_SIZE = 1000
 
 def _get_search_client(
     endpoint: str,
-    key: str,
     index_name: str,
+    key: Optional[str] = None,
+    azure_ad_access_token: Optional[str] = None,
     semantic_configuration_name: Optional[str] = None,
     fields: Optional[List[SearchField]] = None,
     vector_search: Optional[VectorSearch] = None,
@@ -95,7 +97,7 @@ def _get_search_client(
     async_: bool = False,
     additional_search_client_options: Optional[Dict[str, Any]] = None,
 ) -> Union[SearchClient, AsyncSearchClient]:
-    from azure.core.credentials import AzureKeyCredential
+    from azure.core.credentials import AccessToken, AzureKeyCredential, TokenCredential
     from azure.core.exceptions import ResourceNotFoundError
     from azure.identity import DefaultAzureCredential, InteractiveBrowserCredential
     from azure.search.documents import SearchClient
@@ -119,13 +121,23 @@ def _get_search_client(
 
     additional_search_client_options = additional_search_client_options or {}
     default_fields = default_fields or []
-    if key is None:
+    credential: Union[AzureKeyCredential, TokenCredential, InteractiveBrowserCredential]
-        credential = DefaultAzureCredential()
+
-    elif key.upper() == "INTERACTIVE":
+    # Determine the appropriate credential to use
-        credential = InteractiveBrowserCredential()
+    if key is not None:
-        credential.get_token("https://search.azure.com/.default")
+        if key.upper() == "INTERACTIVE":
+            credential = InteractiveBrowserCredential()
+            credential.get_token("https://search.azure.com/.default")
+        else:
+            credential = AzureKeyCredential(key)
+    elif azure_ad_access_token is not None:
+        credential = TokenCredential(
+            lambda *scopes, **kwargs: AccessToken(
+                azure_ad_access_token, int(time.time()) + 3600
+            )
+        )
     else:
-        credential = AzureKeyCredential(key)
+        credential = DefaultAzureCredential()
     index_client: SearchIndexClient = SearchIndexClient(
         endpoint=endpoint, credential=credential, user_agent=user_agent
     )
@@ -253,6 +265,7 @@ class AzureSearch(VectorStore):
         self,
         azure_search_endpoint: str,
         azure_search_key: str,
+        azure_ad_access_token: Optional[str],
         index_name: str,
         embedding_function: Union[Callable, Embeddings],
         search_type: str = "hybrid",
@@ -321,8 +334,9 @@ class AzureSearch(VectorStore):
             user_agent += " " + kwargs["user_agent"]
         self.client = _get_search_client(
             azure_search_endpoint,
-            azure_search_key,
             index_name,
+            azure_search_key,
+            azure_ad_access_token,
             semantic_configuration_name=semantic_configuration_name,
             fields=fields,
             vector_search=vector_search,
@@ -336,8 +350,9 @@ class AzureSearch(VectorStore):
         )
         self.async_client = _get_search_client(
             azure_search_endpoint,
-            azure_search_key,
             index_name,
+            azure_search_key,
+            azure_ad_access_token,
             semantic_configuration_name=semantic_configuration_name,
             fields=fields,
             vector_search=vector_search,
@@ -1387,6 +1402,7 @@ class AzureSearch(VectorStore):
         metadatas: Optional[List[dict]] = None,
         azure_search_endpoint: str = "",
         azure_search_key: str = "",
+        azure_ad_access_token: Optional[str] = None,
         index_name: str = "langchain-index",
         fields: Optional[List[SearchField]] = None,
         **kwargs: Any,
@@ -1395,6 +1411,7 @@ class AzureSearch(VectorStore):
         azure_search = cls(
             azure_search_endpoint,
             azure_search_key,
+            azure_ad_access_token,
             index_name,
             embedding,
             fields=fields,
@@ -1411,6 +1428,7 @@ class AzureSearch(VectorStore):
         metadatas: Optional[List[dict]] = None,
         azure_search_endpoint: str = "",
         azure_search_key: str = "",
+        azure_ad_access_token: Optional[str] = None,
         index_name: str = "langchain-index",
         fields: Optional[List[SearchField]] = None,
         **kwargs: Any,
@@ -1419,6 +1437,7 @@ class AzureSearch(VectorStore):
         azure_search = cls(
             azure_search_endpoint,
             azure_search_key,
+            azure_ad_access_token,
             index_name,
             embedding,
             fields=fields,

libs/community/tests/integration_tests/vectorstores/test_azuresearch.py

@@ -13,6 +13,7 @@ model = os.getenv("OPENAI_EMBEDDINGS_ENGINE_DOC", "text-embedding-ada-002")
 # Vector store settings
 vector_store_address: str = os.getenv("AZURE_SEARCH_ENDPOINT", "")
 vector_store_password: str = os.getenv("AZURE_SEARCH_ADMIN_KEY", "")
+access_token: str = os.getenv("AZURE_SEARCH_ACCESS_TOKEN", "")
 index_name: str = "embeddings-vector-store-test"
 
 
@@ -25,6 +26,7 @@ def similarity_search_test() -> None:
     vector_store: AzureSearch = AzureSearch(
         azure_search_endpoint=vector_store_address,
         azure_search_key=vector_store_password,
+        azure_ad_access_token=access_token,
         index_name=index_name,
         embedding_function=embeddings.embed_query,
     )
@@ -68,6 +70,7 @@ def test_semantic_hybrid_search() -> None:
     vector_store: AzureSearch = AzureSearch(
         azure_search_endpoint=vector_store_address,
         azure_search_key=vector_store_password,
+        azure_ad_access_token=access_token,
         index_name=index_name,
         embedding_function=embeddings.embed_query,
         semantic_configuration_name="default",

libs/community/tests/unit_tests/vectorstores/test_azure_search.py

@@ -32,6 +32,7 @@ class FakeEmbeddingsWithDimension(FakeEmbeddings):
 DEFAULT_INDEX_NAME = "langchain-index"
 DEFAULT_ENDPOINT = "https://my-search-service.search.windows.net"
 DEFAULT_KEY = "mykey"
+DEFAULT_ACCESS_TOKEN = "myaccesstoken1"
 DEFAULT_EMBEDDING_MODEL = FakeEmbeddingsWithDimension()
 
 
@@ -127,6 +128,7 @@ def create_vector_store(
     return AzureSearch(
         azure_search_endpoint=DEFAULT_ENDPOINT,
         azure_search_key=DEFAULT_KEY,
+        azure_ad_access_token=DEFAULT_ACCESS_TOKEN,
         index_name=DEFAULT_INDEX_NAME,
         embedding_function=DEFAULT_EMBEDDING_MODEL,
         additional_search_client_options=additional_search_client_options,