refactor: centralize OpenAI client initialization with configurable b…

config.py

@@ -19,6 +19,7 @@ class Config:
 
     TIMESCALE_SERVICE_URL = os.getenv('TIMESCALE_SERVICE_URL')
     OPENAI_API_KEY = os.getenv('OPENAI_API_KEY')
+    OPENAI_BASE_URL = os.getenv('OPENAI_BASE_URL')
     SecretKey = os.getenv('SecretKey')
     Encreaption_key = os.getenv('Encreaption_key')
     Secret_IV = os.getenv('Secret_IV')
@@ -55,4 +56,5 @@ class Config:
     DOCSTAR_ORG_ID = os.getenv('DOCSTAR_ORG_ID')
     DOCSTAR_COLLECTION_ID = os.getenv('DOCSTAR_COLLECTION_ID')
     AI_ML_APIKEY = os.getenv('AI_ML_APIKEY')
-    AI_MIDDLEWARE_PAUTH_KEY = os.getenv('AI_MIDDLEWARE_PAUTH_KEY')
+    AI_MIDDLEWARE_PAUTH_KEY = os.getenv('AI_MIDDLEWARE_PAUTH_KEY')
+    FIRECRAWL_API_KEY = os.getenv('FIRECRAWL_API_KEY')

src/services/commonServices/openAI/image_model.py

@@ -1,14 +1,14 @@
 import traceback
-from openai import AsyncOpenAI 
 import uuid
 import asyncio
 from src.services.utils.gcp_upload_service import uploadDoc
+from src.services.utils.openai_client import get_async_openai_client
 
 
 
 async def OpenAIImageModel(configuration, apiKey, execution_time_logs, timer):
     try:
-        openai_config = AsyncOpenAI(api_key=apiKey)
+        openai_config = get_async_openai_client(api_key=apiKey)
         timer.start()
         chat_completion = await openai_config.images.generate(**configuration)
         execution_time_logs.append({"step": "OpenAI image Processing time", "time_taken": timer.stop("OpenAI image Processing time")})

src/services/commonServices/openAI/openai_embedding_model.py

@@ -1,9 +1,9 @@
-from openai import AsyncOpenAI
+from src.services.utils.openai_client import get_async_openai_client
 
 
 async def embedding_model(configuration, apiKey):
     try:
-        openAI = AsyncOpenAI(api_key=apiKey)
+        openAI = get_async_openai_client(api_key=apiKey)
         embedding = await openAI.embeddings.create(**configuration)
         return {'success': True, 'response': embedding.to_dict()}
 

src/services/commonServices/openAI/openai_run_batch.py

@@ -1,13 +1,14 @@
 import json
 import os
-from openai import AsyncOpenAI
 import io
 
+from src.services.utils.openai_client import get_async_openai_client
+
 async def create_batch_file(data, apiKey):
     try:
         file_content = "\n".join(data)
         filelike_obj = io.BytesIO(file_content.encode("utf-8"))
-        openAI = AsyncOpenAI(api_key=apiKey)
+        openAI = get_async_openai_client(api_key=apiKey)
         batch_input_file = await openAI.files.create(
             file=filelike_obj,
             purpose="batch"
@@ -19,7 +20,7 @@ async def create_batch_file(data, apiKey):
 
 async def process_batch_file(batch_input_file, apiKey):
     try:
-        openAI = AsyncOpenAI(api_key=apiKey)
+        openAI = get_async_openai_client(api_key=apiKey)
         batch_input_file_id = batch_input_file.id
 
         result = await openAI.batches.create(

src/services/commonServices/openAI/runModel.py

@@ -1,9 +1,9 @@
-from openai import AsyncOpenAI
 import traceback
 import copy
 from ..api_executor import execute_api_call
 # from src.services.utils.unified_token_validator import validate_openai_token_limit
 from globals import *
+from src.services.utils.openai_client import get_async_openai_client
 
 
 def remove_duplicate_ids_from_input(configuration):
@@ -42,7 +42,7 @@ def remove_duplicate_ids_from_input(configuration):
 
 
 async def openai_test_model(configuration, api_key):
-    openAI = AsyncOpenAI(api_key=api_key)
+    openAI = get_async_openai_client(api_key=api_key)
     try:
         chat_completion = await openAI.chat.completions.create(**configuration)
         return {'success': True, 'response': chat_completion.to_dict()}
@@ -59,7 +59,7 @@ async def openai_response_model(configuration, apiKey, execution_time_logs, brid
         # model_name = configuration.get('model')
         # validate_openai_token_limit(configuration, model_name, 'openai_response')
 
-        client = AsyncOpenAI(api_key=apiKey)
+        client = get_async_openai_client(api_key=apiKey)
 
         # Define the API call function with retry mechanism for duplicate ID errors
         async def api_call_with_retry(config, max_retries=2):
@@ -137,7 +137,7 @@ async def api_call(config):
 async def openai_completion(configuration, apiKey, execution_time_logs, bridge_id, timer, message_id=None, org_id=None, name = "", org_name= "", service = "", count=0, token_calculator=None):
     try:
 
-        openAI = AsyncOpenAI(api_key=apiKey)
+        openAI = get_async_openai_client(api_key=apiKey)
 
         # Define the API call function
         async def api_call(config):

src/services/utils/batch_script.py

@@ -1,11 +1,11 @@
 from ..cache_service import find_in_cache_with_prefix, delete_in_cache
-from openai import AsyncOpenAI
 from ..utils.send_error_webhook import create_response_format
 from ..commonServices.baseService.baseService import sendResponse
 import asyncio
 import json
 from .ai_middleware_format import Response_formatter
 from globals import *
+from .openai_client import get_async_openai_client
 
 async def repeat_function():
     while True:
@@ -26,7 +26,7 @@ async def check_batch_status():
             batch_id = id.get('id')
             if webhook.get('url') is not None:
                 response_format = create_response_format(webhook.get('url'), webhook.get('headers'))
-            openAI = AsyncOpenAI(api_key=apikey)
+            openAI = get_async_openai_client(api_key=apikey)
             batch = await openAI.batches.retrieve(batch_id)
             if batch.status == "completed":
                 file = batch.output_file_id or batch.error_file_id

src/services/utils/guardrails_validator.py

@@ -3,10 +3,10 @@
 """
 import json
 import traceback
-from openai import AsyncOpenAI
 from globals import logger
 from exceptions.bad_request import BadRequestException
 from config import Config
+from src.services.utils.openai_client import get_async_openai_client
 
 
 # Guardrails template definitions
@@ -201,7 +201,7 @@ async def validate_with_guardrails(user_message: str, api_key: str, guardrails_c
         ]
 
         # Initialize OpenAI client
-        client = AsyncOpenAI(api_key=api_key)
+        client = get_async_openai_client(api_key=api_key)
 
         # Call OpenAI gpt-5-nano model
         response = await client.chat.completions.create(

src/services/utils/openai_client.py

@@ -0,0 +1,18 @@
+from openai import AsyncOpenAI
+
+from config import Config
+
+
+def get_async_openai_client(api_key, **kwargs):
+    """
+    Build an AsyncOpenAI client that honors the configured base URL override.
+    Falls back to the default OpenAI endpoint when OPENAI_BASE_URL is unset.
+    """
+    client_kwargs = dict(kwargs)
+    client_kwargs["api_key"] = api_key
+
+    base_url = getattr(Config, "OPENAI_BASE_URL", None)
+    if base_url:
+        client_kwargs["base_url"] = base_url.rstrip("/")
+
+    return AsyncOpenAI(**client_kwargs)