Merged in feature/llm_lambda_uat (pull request #199)

Feature/llm lambda uat

Approved-by: Umang Mistry

textract-pipeline/src/lambda/prompt-orchestrator/index.py

@@ -20,7 +20,7 @@ BEDROCK_REGION = "us-east-1"
 SNOWFLAKE_SCHEMA = "STG"
 MAX_PROMPT_LENGTH_PER_GROUP = 20000  # Adjust as needed
 LLM_SELECTED = "Claude 3 - Sonnet"
-SNOWFLAKE_CONN_SECRET_NAME = "doczy_dev_db_creds"
+SNOWFLAKE_CONN_SECRET_NAME = "doczy-dev-db-svc-acc"
 
 # Get variable ENVIRONMENT from env variables using os
 env = os.environ.get('ENVIRONMENT')
@@ -70,8 +70,8 @@ sf_secrets = json.loads(get_secret())
 snowflake_conn = snowflake.connector.connect(
     user=sf_secrets.get('user'),
     password=sf_secrets.get('password'),
-    account=sf_secrets.get('account_alias'),
-    role=sf_secrets.get('ROLE'),
+    account=sf_secrets.get('account_locator'),
+    role=sf_secrets.get('role'),
     warehouse=sf_secrets.get('warehouse'),
     database=sf_secrets.get('database'),
     schema=SNOWFLAKE_SCHEMA
@@ -132,6 +132,8 @@ def lambda_handler(event, context):
         # Process each message from the SQS event
         for record in records['Records']:
             # Extract the message body from the record
+            logger.info(f"RECORD:: {record}")
+
             sqs_record = record['s3']
             bucket_name = sqs_record['bucket']['name']
             object_key = unquote_plus(sqs_record['object']['key'])
@@ -494,7 +496,7 @@ def main(bucket, object_key, field_groups):
 
         question = prepare_prompt(*field_groups)
         response_text = invoke_llm(context, question, llm_selected=LLM_SELECTED)
-        logger.info(f"Response text:  {response_text}, CONTEXT: {context}")
+        # logger.info(f"Response text:  {response_text}, CONTEXT: {context}")
         field_l, answer_l, page_no_l, location_l, snippet_l = json_parsing(response_text, context)
         field_list.extend(field_l)
         answer_list.extend(answer_l)

textract-pipeline/terraform/modules/aws-textract-pipeline-part3/main.tf

@@ -649,13 +649,14 @@ module "prompt_orchestrator_lambda_function" {
 
   layers = [
     module.anthropic_sf_lambda_layer.lambda_layer_arn,
+    # module.pandas_numpy_lambda_layer.lambda_layer_arn,
     module.pydantic_lambda_layer.lambda_layer_arn,
     var.pandas_numpy_lambda_layer_additional_arn
   ]
 
   environment_variables = merge(local.environment_variables, local.prompt_orchestrator_lambda.environment_variables, {
     PROMPT_PROCESSOR_QUEUE_URL = var.prompt_processor_queue_url
-    MAX_PROMPT_LENGTH_PER_GROUP = "5000"
+    MAX_PROMPT_LENGTH_PER_GROUP = "50000"
   })
 
   source_path   = local.prompt_orchestrator_lambda.source_path