feat!: reworking api versions logic (#67)

.env.example

@@ -1,3 +1,2 @@
-AZURE_API_VERSION=2023-03-15-preview
 MODEL_ALIASES={}
 LOG_LEVEL=INFO

README.md

@@ -55,15 +55,15 @@ Copy `.env.example` to `.env` and customize it for your environment:
 |---|---|---|
 |LOG_LEVEL|INFO|Log level. Use DEBUG for dev purposes and INFO in prod|
 |WEB_CONCURRENCY|1|Number of workers for the server|
-|AZURE_API_VERSION|2023-03-15-preview|The version API for requests to Azure OpenAI API|
-|MODEL_ALIASES|{}|Mapping request's deployment_id to [model name of tiktoken](https://github.com/openai/tiktoken/blob/main/tiktoken/model.py) for correct calculate of tokens. Example: `{"gpt-35-turbo":"gpt-3.5-turbo-0301"}`|
+|MODEL_ALIASES|`{}`|Mapping request's deployment_id to [model name of tiktoken](https://github.com/openai/tiktoken/blob/main/tiktoken/model.py) for correct calculate of tokens. Example: `{"gpt-35-turbo":"gpt-3.5-turbo-0301"}`|
 |DIAL_USE_FILE_STORAGE|False|Save image model artifacts to DIAL File storage (DALL-E images are uploaded to the files storage and its base64 encodings are replaced with links to the storage)|
 |DIAL_URL||URL of the core DIAL server (required when DIAL_USE_FILE_STORAGE=True)|
 |DALLE3_DEPLOYMENTS|``|Comma-separated list of deployments that support DALL-E 3 API. Example: `dall-e-3,dalle3,dall-e`|
 |GPT4_VISION_DEPLOYMENTS|``|Comma-separated list of deployments that support GPT-4V API. Example: `gpt-4-vision-preview,gpt-4-vision`|
 |GPT4_VISION_MAX_TOKENS|1024|Default value of `max_tokens` parameter for GPT-4V when it wasn't provided in the request|
 |ACCESS_TOKEN_EXPIRATION_WINDOW|10|Expiration window of access token in seconds|
 |AZURE_OPEN_AI_SCOPE|https://cognitiveservices.azure.com/.default|Provided scope of access token to Azure OpenAI services|
+|API_VERSIONS_MAPPING|`{}`|The mapping of versions API for requests to Azure OpenAI API. Example: `{"2023-03-15-preview": "2023-05-15", "": "2024-02-15-preview"}`. An empty key sets the default api version for the case when the user didn't pass it in the request|
 
 ### Docker
 

aidial_adapter_openai/app.py

@@ -24,25 +24,23 @@
     parse_deployment_list,
     parse_upstream,
 )
-from aidial_adapter_openai.utils.request_classifier import (
-    does_request_use_functions_or_tools,
-)
 from aidial_adapter_openai.utils.sse_stream import to_openai_sse_stream
 from aidial_adapter_openai.utils.storage import create_file_storage
 from aidial_adapter_openai.utils.streaming import generate_stream, map_stream
 from aidial_adapter_openai.utils.tokens import Tokenizer, discard_messages
-from aidial_adapter_openai.utils.versions import compare_versions
 
 logging.config.dictConfig(LogConfig().dict())
 app = FastAPI()
 model_aliases: Dict[str, str] = json.loads(os.getenv("MODEL_ALIASES", "{}"))
-azure_api_version = os.getenv("AZURE_API_VERSION", "2023-03-15-preview")
 dalle3_deployments = parse_deployment_list(
     os.getenv("DALLE3_DEPLOYMENTS") or ""
 )
 gpt4_vision_deployments = parse_deployment_list(
     os.getenv("GPT4_VISION_DEPLOYMENTS") or ""
 )
+api_versions_mapping: Dict[str, str] = json.loads(
+    os.getenv("API_VERSIONS_MAPPING", "{}")
+)
 
 
 async def handle_exceptions(call):
@@ -58,6 +56,12 @@ async def handle_exceptions(call):
         )
 
 
+def get_api_version(request: Request):
+    api_version = request.query_params.get("api-version", "")
+
+    return api_versions_mapping.get(api_version, api_version)
+
+
 @app.post("/openai/deployments/{deployment_id}/chat/completions")
 async def chat_completion(deployment_id: str, request: Request):
     data = await parse_body(request)
@@ -67,11 +71,25 @@ async def chat_completion(deployment_id: str, request: Request):
     api_type, api_key = await get_credentials(request)
 
     upstream_endpoint = request.headers["X-UPSTREAM-ENDPOINT"]
+    api_version = get_api_version(request)
+
+    if api_version == "":
+        raise HTTPException(
+            "Api version is a required query parameter",
+            400,
+            "invalid_request_error",
+        )
 
     if deployment_id in dalle3_deployments:
         storage = create_file_storage("images", request.headers)
         return await dalle3_chat_completion(
-            data, upstream_endpoint, api_key, is_stream, storage, api_type
+            data,
+            upstream_endpoint,
+            api_key,
+            is_stream,
+            storage,
+            api_type,
+            api_version,
         )
     elif deployment_id in gpt4_vision_deployments:
         storage = create_file_storage("images", request.headers)
@@ -83,6 +101,7 @@ async def chat_completion(deployment_id: str, request: Request):
             is_stream,
             storage,
             api_type,
+            api_version,
         )
 
     openai_model_name = model_aliases.get(deployment_id, deployment_id)
@@ -92,17 +111,6 @@ async def chat_completion(deployment_id: str, request: Request):
         upstream_endpoint, ApiType.CHAT_COMPLETION
     )
 
-    api_version = azure_api_version
-
-    if does_request_use_functions_or_tools(data):
-        request_api_version = request.query_params.get("api-version")
-
-        if request_api_version is not None:
-            # 2023-07-01-preview is the first azure api version that supports functions
-            compare_result = compare_versions(request_api_version, "2023-07-01")
-            if compare_result == 0 or compare_result == 1:
-                api_version = request_api_version
-
     discarded_messages = None
     if "max_prompt_tokens" in data:
         max_prompt_tokens = data["max_prompt_tokens"]
@@ -172,14 +180,22 @@ async def embedding(deployment_id: str, request: Request):
     api_base, upstream_deployment = parse_upstream(
         request.headers["X-UPSTREAM-ENDPOINT"], ApiType.EMBEDDING
     )
+    api_version = get_api_version(request)
+
+    if api_version == "":
+        raise HTTPException(
+            "Api version is a required query parameter",
+            400,
+            "invalid_request_error",
+        )
 
     return await handle_exceptions(
         Embedding().acreate(
             deployment_id=upstream_deployment,
             api_key=api_key,
             api_base=api_base,
             api_type=api_type,
-            api_version=azure_api_version,
+            api_version=api_version,
             request_timeout=(10, 600),  # connect timeout and total timeout
             **data,
         )

aidial_adapter_openai/dalle3.py

@@ -118,6 +118,7 @@ async def chat_completion(
     is_stream: bool,
     file_storage: Optional[FileStorage],
     api_type: str,
+    api_version: str,
 ) -> Response:
     if data.get("n", 1) > 1:
         raise HTTPException(
@@ -126,7 +127,7 @@ async def chat_completion(
             type="invalid_request_error",
         )
 
-    api_url = upstream_endpoint + "?api-version=2023-12-01-preview"
+    api_url = f"{upstream_endpoint}?api-version={api_version}"
     user_prompt = get_user_prompt(data)
     model_response = await generate_image(
         api_url, api_key, user_prompt, api_type

aidial_adapter_openai/gpt4_vision/chat_completion.py

@@ -267,6 +267,7 @@ async def chat_completion(
     is_stream: bool,
     file_storage: Optional[FileStorage],
     api_type: str,
+    api_version,
 ) -> Response:
     if request.get("n", 1) > 1:
         raise HTTPException(
@@ -283,7 +284,7 @@ async def chat_completion(
             type="invalid_request_error",
         )
 
-    api_url = upstream_endpoint + "?api-version=2023-12-01-preview"
+    api_url = f"{upstream_endpoint}?api-version={api_version}"
 
     result = await transform_messages(file_storage, messages)
 

tests/test_errors.py

@@ -50,7 +50,7 @@ async def test_error_during_streaming(aioresponses: aioresponses):
     test_app = AsyncClient(app=app, base_url="http://test.com")
 
     response = await test_app.post(
-        "/openai/deployments/gpt-4/chat/completions",
+        "/openai/deployments/gpt-4/chat/completions?api-version=2023-03-15-preview",
         json={
             "messages": [{"role": "user", "content": "Test content"}],
             "stream": True,
@@ -94,7 +94,7 @@ async def test_incorrect_upstream_url(aioresponses: aioresponses):
     test_app = AsyncClient(app=app, base_url="http://test.com")
 
     response = await test_app.post(
-        "/openai/deployments/gpt-4/chat/completions",
+        "/openai/deployments/gpt-4/chat/completions?api-version=2023-03-15-preview",
         json={"messages": [{"role": "user", "content": "Test content"}]},
         headers={
             "X-UPSTREAM-KEY": "TEST_API_KEY",
@@ -123,7 +123,7 @@ async def test_incorrect_format(aioresponses: aioresponses):
     test_app = AsyncClient(app=app, base_url="http://test.com")
 
     response = await test_app.post(
-        "/openai/deployments/gpt-4/chat/completions",
+        "/openai/deployments/gpt-4/chat/completions?api-version=2023-03-15-preview",
         json={"messages": [{"role": "user", "content": "Test content"}]},
         headers={
             "X-UPSTREAM-KEY": "TEST_API_KEY",
@@ -156,7 +156,7 @@ async def test_incorrect_streaming_request(aioresponses: aioresponses):
     test_app = AsyncClient(app=app, base_url="http://test.com")
 
     response = await test_app.post(
-        "/openai/deployments/gpt-4/chat/completions",
+        "/openai/deployments/gpt-4/chat/completions?api-version=2023-03-15-preview",
         json={
             "messages": [{"role": "user", "content": "Test content"}],
             "stream": True,

tests/test_streaming.py

@@ -10,7 +10,7 @@
 @pytest.mark.asyncio
 async def test_streaming(aioresponses: aioresponses):
     aioresponses.post(
-        "http://localhost:5001/openai/deployments/gpt-4/chat/completions?api-version=2023-03-15-preview",
+        "http://localhost:5001/openai/deployments/gpt-4/chat/completions?api-version=2023-06-15",
         status=200,
         body="data: "
         + json.dumps(