Merge pull request #244 from awslabs/chore/update-get-bedrock-client

chore: update bedrock client retrieval and aoss index creation

apidocs/classes/LangchainCommonLayer.md

@@ -9,11 +9,11 @@ LangchainCommonLayer allows developers to instantiate a llm client adapter on be
 **`Example`**
 
 ```ts
+import boto3
 from genai_core.adapters.registry import registry
-from genai_core.clients import get_bedrock_client
 
 adapter = registry.get_adapter(f"{provider}.{model_id}")
-bedrock_client = get_bedrock_client()
+bedrock_client = boto3.client('bedrock-runtime')
 ```
 
 ## Hierarchy

lambda/aws-qa-appsync-opensearch/question_answering/src/llms/text_generation_llm_selector.py

@@ -25,43 +25,9 @@
 tracer = Tracer(service="QUESTION_ANSWERING")
 metrics = Metrics(namespace="question_answering", service="QUESTION_ANSWERING")
 
-sts_client = boto3.client("sts")
-
-aws_region = boto3.Session().region_name
-
-def get_bedrock_client(service_name="bedrock-runtime"):
-    config = {}
-    bedrock_config = config.get("bedrock", {})
-    bedrock_enabled = bedrock_config.get("enabled", False)
-    if not bedrock_enabled:
-        print("bedrock not enabled")
-        return None
-
-    bedrock_config_data = {"service_name": service_name}
-    region_name = bedrock_config.get("region")
-    endpoint_url = bedrock_config.get("endpointUrl")
-    role_arn = bedrock_config.get("roleArn")
-
-    if region_name:
-        bedrock_config_data["region_name"] = region_name
-    if endpoint_url:
-        bedrock_config_data["endpoint_url"] = endpoint_url
-
-    if role_arn:
-        assumed_role_object = sts_client.assume_role(
-            RoleArn=role_arn,
-            RoleSessionName="AssumedRoleSession",
-        )
-
-        credentials = assumed_role_object["Credentials"]
-        bedrock_config_data["aws_access_key_id"] = credentials["AccessKeyId"]
-        bedrock_config_data["aws_secret_access_key"] = credentials["SecretAccessKey"]
-        bedrock_config_data["aws_session_token"] = credentials["SessionToken"]
-
-    return boto3.client(**bedrock_config_data)
 
 def get_llm(callbacks=None):
-    bedrock = get_bedrock_client(service_name="bedrock-runtime")
+    bedrock = boto3.client('bedrock-runtime')
 
     params = {
         "max_tokens_to_sample": 600,
@@ -85,7 +51,7 @@ def get_llm(callbacks=None):
     return Bedrock(**kwargs)
 
 def get_embeddings_llm():
-    bedrock = get_bedrock_client(service_name="bedrock-runtime")
+    bedrock = boto3.client('bedrock-runtime')
     return BedrockEmbeddings(client=bedrock, model_id="amazon.titan-embed-text-v1")
 
 def get_max_tokens():

lambda/aws-rag-appsync-stepfn-opensearch/embeddings_job/src/helpers/opensearch_helper.py

@@ -23,39 +23,6 @@
 tracer = Tracer(service="INGESTION_EMBEDDING_JOB")
 metrics = Metrics(namespace="ingestion_pipeline", service="INGESTION_EMBEDDING_JOB")
 
-aws_region = boto3.Session().region_name
-sts_client = boto3.client("sts")
-
-def get_bedrock_client(service_name="bedrock-runtime"):
-    config = {}
-    bedrock_config = config.get("bedrock", {})
-    bedrock_enabled = bedrock_config.get("enabled", False)
-    if not bedrock_enabled:
-        print("bedrock not enabled")
-        return None
-
-    bedrock_config_data = {"service_name": service_name}
-    region_name = bedrock_config.get("region")
-    endpoint_url = bedrock_config.get("endpointUrl")
-    role_arn = bedrock_config.get("roleArn")
-
-    if region_name:
-        bedrock_config_data["region_name"] = region_name
-    if endpoint_url:
-        bedrock_config_data["endpoint_url"] = endpoint_url
-
-    if role_arn:
-        assumed_role_object = sts_client.assume_role(
-            RoleArn=role_arn,
-            RoleSessionName="AssumedRoleSession",
-        )
-
-        credentials = assumed_role_object["Credentials"]
-        bedrock_config_data["aws_access_key_id"] = credentials["AccessKeyId"]
-        bedrock_config_data["aws_secret_access_key"] = credentials["SecretAccessKey"]
-        bedrock_config_data["aws_session_token"] = credentials["SessionToken"]
-
-    return boto3.client(**bedrock_config_data)
 
 @tracer.capture_method
 def check_if_index_exists(index_name: str, region: str, host: str, http_auth: Tuple[str, str]) -> OpenSearch:
@@ -72,13 +39,14 @@ def check_if_index_exists(index_name: str, region: str, host: str, http_auth: Tu
 
 def process_shard(shard, os_index_name, os_domain_ep, os_http_auth) -> int: 
     print(f'Starting process_shard of {len(shard)} chunks.')
-    bedrock_client = get_bedrock_client()
+    bedrock_client = boto3.client('bedrock-runtime')
     embeddings = BedrockEmbeddings(
         client=bedrock_client, 
         model_id="amazon.titan-embed-text-v1")
+    opensearch_url = os_domain_ep if os_domain_ep.startswith("https://") else f"https://{os_domain_ep}"
     docsearch = OpenSearchVectorSearch(index_name=os_index_name,
                                        embedding_function=embeddings,
-                                       opensearch_url=f"https://{os_domain_ep}",
+                                       opensearch_url=opensearch_url,
                                        http_auth=os_http_auth,
                                        use_ssl = True,
                                        verify_certs = True,

lambda/aws-rag-appsync-stepfn-opensearch/embeddings_job/src/lambda.py

@@ -22,7 +22,7 @@
 from helpers.update_ingestion_status import updateIngestionJobStatus
 from langchain_community.embeddings import BedrockEmbeddings
 from helpers.s3inmemoryloader import S3TxtFileLoaderInMemory
-from opensearchpy import OpenSearch, RequestsHttpConnection
+from opensearchpy import RequestsHttpConnection
 from langchain_community.vectorstores import OpenSearchVectorSearch
 from langchain.text_splitter import RecursiveCharacterTextSplitter
 import multiprocessing as mp
@@ -39,38 +39,7 @@
 aws_region = boto3.Session().region_name
 session = boto3.session.Session()
 credentials = session.get_credentials()
-sts_client = boto3.client("sts")
 
-def get_bedrock_client(service_name="bedrock-runtime"):
-    config = {}
-    bedrock_config = config.get("bedrock", {})
-    bedrock_enabled = bedrock_config.get("enabled", False)
-    if not bedrock_enabled:
-        print("bedrock not enabled")
-        return None
-
-    bedrock_config_data = {"service_name": service_name}
-    region_name = bedrock_config.get("region")
-    endpoint_url = bedrock_config.get("endpointUrl")
-    role_arn = bedrock_config.get("roleArn")
-
-    if region_name:
-        bedrock_config_data["region_name"] = region_name
-    if endpoint_url:
-        bedrock_config_data["endpoint_url"] = endpoint_url
-
-    if role_arn:
-        assumed_role_object = sts_client.assume_role(
-            RoleArn=role_arn,
-            RoleSessionName="AssumedRoleSession",
-        )
-
-        credentials = assumed_role_object["Credentials"]
-        bedrock_config_data["aws_access_key_id"] = credentials["AccessKeyId"]
-        bedrock_config_data["aws_secret_access_key"] = credentials["SecretAccessKey"]
-        bedrock_config_data["aws_session_token"] = credentials["SessionToken"]
-
-    return boto3.client(**bedrock_config_data)
 
 opensearch_secret_id = os.environ['OPENSEARCH_SECRET_ID']
 bucket_name = os.environ['OUTPUT_BUCKET']
@@ -88,7 +57,7 @@ def get_bedrock_client(service_name="bedrock-runtime"):
 INDEX_FILE="index_file"
 
 def process_documents_in_es(index_exists, shards, http_auth):
-    bedrock_client = get_bedrock_client()
+    bedrock_client = boto3.client('bedrock-runtime')
     embeddings = BedrockEmbeddings(client=bedrock_client)
 
     if index_exists is False:
@@ -136,52 +105,32 @@ def process_documents_in_es(index_exists, shards, http_auth):
                     os_http_auth=http_auth)
 
 def process_documents_in_aoss(index_exists, shards, http_auth):
+    # Reference: https://python.langchain.com/docs/integrations/vectorstores/opensearch#using-aoss-amazon-opensearch-service-serverless
+    bedrock_client = boto3.client('bedrock-runtime')
+    embeddings = BedrockEmbeddings(client=bedrock_client)
+
+    shard_start_index = 0
     if index_exists is False:
-        vector_db = OpenSearch(
-            hosts = [{'host': opensearch_domain.replace("https://", ""), 'port': 443}],
-            http_auth = http_auth,
-            use_ssl = True,
-            verify_certs = True,
-            connection_class = RequestsHttpConnection
+        OpenSearchVectorSearch.from_documents(
+            shards[0],
+            embeddings,
+            opensearch_url=opensearch_domain,
+            http_auth=http_auth,
+            timeout=300,
+            use_ssl=True,
+            verify_certs=True,
+            connection_class=RequestsHttpConnection,
+            index_name=opensearch_index,
+            engine="faiss",
         )
-        index_body = {
-            'settings': {
-                "index.knn": True
-            },
-            "mappings": {
-                "properties": {
-                    "vector_field": {
-                        "type": "knn_vector",
-                        "dimension": 1536,
-                        "method": {
-                            "engine": "nmslib",
-                            "space_type": "cosinesimil",
-                            "name": "hnsw",
-                            "parameters": {},
-                        }
-                    },
-                    "id": {
-                        "type": "text",
-                        "fields": {"keyword": {"type": "keyword", "ignore_above": 256}},
-                    },
-                }
-            }
-        }
-        response = vector_db.indices.create(opensearch_index, body=index_body)
-        print(response)
+        # we now need to start the loop below for the second shard
+        shard_start_index = 1
 
-    print(f"index={opensearch_index} Adding Documents")
-    bedrock_client = get_bedrock_client()
-    embeddings = BedrockEmbeddings(client=bedrock_client, model_id="amazon.titan-embed-text-v1")
-    docsearch = OpenSearchVectorSearch(index_name=opensearch_index,
-                                       embedding_function=embeddings,
-                                       opensearch_url=opensearch_domain,
-                                       http_auth=http_auth,
-                                       use_ssl = True,
-                                       verify_certs = True,
-                                       connection_class = RequestsHttpConnection)
-    for shard in shards:
-        docsearch.add_documents(documents=shard)
+    for shard in shards[shard_start_index:]:
+        results = process_shard(shard=shard,
+                    os_index_name=opensearch_index,
+                    os_domain_ep=opensearch_domain,
+                    os_http_auth=http_auth)
 
 @logger.inject_lambda_context(log_event=True)
 @tracer.capture_lambda_handler

lambda/aws-summarization-appsync-stepfn/summary_generator/lambda.py

@@ -37,8 +37,6 @@
 transformed_bucket_name = os.environ["ASSET_BUCKET_NAME"]
 chain_type = os.environ["SUMMARY_LLM_CHAIN_TYPE"]
 
-aws_region = boto3.Session().region_name
-
 params = {
         "max_tokens_to_sample": 4000,
         "temperature": 0, 
@@ -47,11 +45,7 @@
         "stop_sequences": ["\\n\\nHuman:"],
     }
 
-bedrock_client = boto3.client(
-        service_name='bedrock-runtime', 
-        region_name=aws_region,
-        endpoint_url=f'https://bedrock-runtime.{aws_region}.amazonaws.com'
-    )
+bedrock_client = boto3.client('bedrock-runtime')
 
 @logger.inject_lambda_context(log_event=True)
 @tracer.capture_lambda_handler

layers/langchain-common-layer/python/genai_core/adapters/bedrock/ai21_j2.py

@@ -10,7 +10,7 @@
 # OR CONDITIONS OF ANY KIND, express or implied. See the License for the specific language governing permissions
 # and limitations under the License.
 #
-import genai_core.clients
+import boto3
 from langchain.llms import Bedrock
 from langchain.prompts.prompt import PromptTemplate
 
@@ -25,7 +25,7 @@ def __init__(self, model_id, *args, **kwargs):
         super().__init__(*args, **kwargs)
 
     def get_llm(self, model_kwargs={}):
-        bedrock = genai_core.clients.get_bedrock_client()
+        bedrock = boto3.client('bedrock-runtime')
 
         params = {}
         if "temperature" in model_kwargs:

layers/langchain-common-layer/python/genai_core/adapters/bedrock/claude.py

@@ -10,7 +10,7 @@
 # OR CONDITIONS OF ANY KIND, express or implied. See the License for the specific language governing permissions
 # and limitations under the License.
 #
-import genai_core.clients
+import boto3
 
 from langchain.llms import Bedrock
 from langchain.prompts.prompt import PromptTemplate
@@ -26,7 +26,7 @@ def __init__(self, model_id, *args, **kwargs):
         super().__init__(*args, **kwargs)
 
     def get_llm(self, model_kwargs={}):
-        bedrock = genai_core.clients.get_bedrock_client()
+        bedrock = boto3.client('bedrock-runtime')
 
         params = {}
         if "temperature" in model_kwargs:

layers/langchain-common-layer/python/genai_core/adapters/bedrock/cohere.py

@@ -10,7 +10,7 @@
 # OR CONDITIONS OF ANY KIND, express or implied. See the License for the specific language governing permissions
 # and limitations under the License.
 #
-import genai_core.clients
+import boto3
 
 from langchain.llms import Bedrock
 from langchain.prompts.prompt import PromptTemplate
@@ -26,7 +26,7 @@ def __init__(self, model_id, *args, **kwargs):
         super().__init__(*args, **kwargs)
 
     def get_llm(self, model_kwargs={}):
-        bedrock = genai_core.clients.get_bedrock_client()
+        bedrock = boto3.client('bedrock-runtime')
 
         params = {}
         if "temperature" in model_kwargs:

layers/langchain-common-layer/python/genai_core/adapters/bedrock/titan.py

@@ -10,7 +10,7 @@
 # OR CONDITIONS OF ANY KIND, express or implied. See the License for the specific language governing permissions
 # and limitations under the License.
 #
-import genai_core.clients
+import boto3
 from langchain.prompts.prompt import PromptTemplate
 
 from langchain.llms import Bedrock
@@ -26,7 +26,7 @@ def __init__(self, model_id, *args, **kwargs):
         super().__init__(*args, **kwargs)
 
     def get_llm(self, model_kwargs={}):
-        bedrock = genai_core.clients.get_bedrock_client()
+        bedrock = boto3.client('bedrock-runtime')
 
         params = {}
         if "temperature" in model_kwargs:

layers/langchain-common-layer/python/genai_core/clients.py

@@ -15,8 +15,6 @@
 import openai
 from botocore.config import Config
 
-sts_client = boto3.client("sts")
-
 def get_openai_client():
     api_key = os.environ['OPEN_API_KEY']
     if not api_key:
@@ -33,34 +31,3 @@ def get_sagemaker_client():
     client = boto3.client("sagemaker-runtime", config=config)
 
     return client
-
-
-def get_bedrock_client(service_name="bedrock-runtime"):
-    config = {}
-    bedrock_config = config.get("bedrock", {})
-    bedrock_enabled = bedrock_config.get("enabled", False)
-    if not bedrock_enabled:
-        return None
-
-    bedrock_config_data = {"service_name": service_name}
-    region_name = bedrock_config.get("region")
-    endpoint_url = bedrock_config.get("endpointUrl")
-    role_arn = bedrock_config.get("roleArn")
-
-    if region_name:
-        bedrock_config_data["region_name"] = region_name
-    if endpoint_url:
-        bedrock_config_data["endpoint_url"] = endpoint_url
-
-    if role_arn:
-        assumed_role_object = sts_client.assume_role(
-            RoleArn=role_arn,
-            RoleSessionName="AssumedRoleSession",
-        )
-
-        credentials = assumed_role_object["Credentials"]
-        bedrock_config_data["aws_access_key_id"] = credentials["AccessKeyId"]
-        bedrock_config_data["aws_secret_access_key"] = credentials["SecretAccessKey"]
-        bedrock_config_data["aws_session_token"] = credentials["SessionToken"]
-
-    return boto3.client(**bedrock_config_data)