update readme, rework streaming and function support

CHANGELOG.md

@@ -1,5 +1,21 @@
 # Changelogs
 
+#### v0.0.35 (2023-10-06)
+
+**Additions**
+
+- Added `auto_retry` option to `OpenAI` client, which will automatically retry failed requests.
+- Added `RotatingClients` class which handles the rotation of multiple clients. This can be enabled by passing `rotating_clients=True` to the `OpenAI` client while configuring.
+- Added `OpenAI.chat_create` and `OpenAI.async_chat_create` methods which automatically handles rotating clients and retrying failed requests.
+- Added `azure_model_mapping` which allows automatically mapping of Azure models to OpenAI models when passing `openai` models as a parameter, it will automatically convert it to the Azure model. This is only done in `chat` implementation.
+
+**Fixes**
+
+- Fixed `api_version` Configuration handling.
+- Fixed parsing of `function_call` in streaming implementation.
+
+
+
 #### v0.0.34 (2023-10-06)
 
 **Changes** 

README.md

@@ -121,6 +121,178 @@ result: CompletionResponse = asyncio.run(
 
 ```
 
+### Initialize Clients Manually
+
+```python
+
+from async_openai import OpenAI
+
+# Configure your primary client (default)
+
+
+OpenAI.configure(
+    api_key = "sk-XXXX",
+    organization = "org-XXXX",
+    debug_enabled = False,
+
+    # Azure Configuration
+    azure_api_base = 'https://....openai.azure.com/',
+    azure_api_version = '2023-07-01-preview',
+    azure_api_key = '....',
+)
+
+# Returns the default client (openai)
+oai = OpenAI.init_api_client()
+
+# Configure your secondary client (azure) and use it directly
+az = OpenAI.init_api_client('az', set_as_default = False, debug_enabled = True)
+result = az.completions.create(
+    prompt = 'say this is a test',
+    max_tokens = 4,
+    stream = True
+)
+
+
+# Use the default client (openai)
+result = OpenAI.completions.create(
+    prompt = 'say this is a test',
+    max_tokens = 4,
+    stream = True
+)
+# Or 
+result = oai.completions.create(
+    prompt = 'say this is a test',
+    max_tokens = 4,
+    stream = True
+)
+
+```
+
+### Handling Errors, Retries, and Rotations
+
+The below will show you how to rotate between multiple clients when you hit an error.
+
+**Important** Auto-rotation is only supported with `chat_create` and `async_chat_create` methods. Otherwise, you should handle the rotation manually.
+
+```python
+
+import asyncio
+from async_openai import OpenAI, ChatResponse
+from async_openai.utils import logger
+
+OpenAI.configure(
+    api_key = "sk-XXXX",
+    organization = "org-XXXX",
+    debug_enabled = False,
+
+    # Azure Configuration
+    azure_api_base = 'https://....openai.azure.com/',
+    azure_api_version = '2023-07-01-preview',
+    azure_api_key = '....',
+
+    # This will allow you to auto rotate clients when you hit an error.
+    # But only if you have multiple clients configured and are using `OpenAI.chat_create`
+    enable_rotating_clients = True, 
+
+    # This will prioritize Azure over OpenAI when using `OpenAI.chat_create`
+    prioritize = "azure",
+)
+
+# Display the current client
+OpenAI.get_current_client_info(verbose = True)
+
+# Rotate to the next client
+# OpenAI.rotate_client(verbose = True)
+
+###
+# [Sync] create a completion with auto-rotation and auto-retry
+###
+
+result: ChatResponse = OpenAI.chat_create(
+    model = "gpt-3.5-turbo-16k",
+    messages = [
+        {"role": "user", "content": "Translate the following English text to French: “Multiple models, each with different capabilities and price points. Prices are per 1,000 tokens. You can think of tokens as pieces of words, where 1,000 tokens is about 750 words. This paragraph is 35 tokens”"}
+    ],
+    auto_retry = True,
+
+)
+
+logger.info(f'Result Chat Message: {result.messages}')
+logger.info(f'Result Usage: {result.usage}')
+logger.info(f'Result Consumption: {result.consumption}')
+
+###
+# [Async] create a completion with auto-rotation and auto-retry
+###
+
+result: ChatResponse = asyncio.run(
+    OpenAI.async_chat_create(
+        model = "gpt-3.5-turbo-16k",
+        messages = [
+            {"role": "user", "content": "Translate the following English text to French: “Multiple models, each with different capabilities and price points. Prices are per 1,000 tokens. You can think of tokens as pieces of words, where 1,000 tokens is about 750 words. This paragraph is 35 tokens”"}
+        ],
+        auto_retry = True,
+    )
+)
+
+```
+
+### Function Calls
+
+The latest version of the API allows for function calls to be made. This is currently only supported in `Chat` and requires api version: `2023-07-01-preview` for `azure`.
+
+Function calls support using `pydantic` models to auto-generate the schemas
+
+```python
+
+import asyncio
+from enum import Enum
+from client_rotate import OpenAI
+from async_openai.utils import logger
+from pydantic import BaseModel, Field
+
+class Unit(str, Enum):
+    celsius = "celsius"
+    fahrenheit = "fahrenheit"
+
+class Weather(BaseModel):
+    location: str = Field(..., description="The city and state, e.g. San Francisco, CA.")
+    unit: Unit = Field(Unit.fahrenheit)
+
+functions = [ 
+  {
+    "name": "get_current_weather",
+    "description": "Get the current weather in a given location",
+    "parameters": Weather,
+  }
+]
+
+result: ChatResponse = OpenAI.chat_create(
+    model = "gpt-3.5-turbo-16k",
+    messages = [
+        {"role": "user", "content": "What's the weather like in Boston today?"}
+    ],
+    functions = functions,
+    auto_retry = True,
+)
+
+logger.info(f'Result Chat Message: {result.messages}')
+logger.info(f'Result Chat Function: {result.function_results}')
+logger.info(f'Result Usage: {result.usage}')
+logger.info(f'Result Consumption: {result.consumption}')
+
+"""
+Result:
+
+> Result Chat Message: [ChatMessage(content='', role='assistant', function_call=FunctionCall(name='get_current_weather', arguments={'location': 'Boston, MA'}), name=None)]
+> Result Chat Function: [FunctionCall(name='get_current_weather', arguments={'location': 'Boston, MA'})]
+> Result Usage: prompt_tokens=16 completion_tokens=19 total_tokens=35
+> Result Consumption: 0.00012399999999999998
+"""
+
+```
+
+
 
 ---
 

async_openai/client.py

@@ -58,6 +58,7 @@ class OpenAIClient:
     settings: Optional[OpenAISettings] = None
     name: Optional[str] = None
     is_azure: Optional[bool] = None
+    azure_model_mapping: Optional[Dict[str, str]] = None
 
     auth: Optional[OpenAIAuth] = None
     _client: Optional[aiohttpx.Client] = None
@@ -118,6 +119,7 @@ def configure_params(
         settings: Optional[OpenAISettings] = None,
         name: Optional[str] = None,
         is_azure: Optional[bool] = None,
+        azure_model_mapping: Optional[Dict[str, str]] = None,
         auth: Optional[OpenAIAuth] = None,
         **kwargs
     ):  # sourcery skip: low-code-quality
@@ -136,10 +138,6 @@ def configure_params(
             self.api_type = api_type
         elif self.api_type is None:
             self.api_type = self.settings.api_type
-        if api_version is not None:
-            self.api_version = api_version
-        elif self.api_version is None:
-            self.api_version = self.settings.api_version
         if organization is not None:
             self.organization = organization
         elif self.organization is None:
@@ -214,10 +212,17 @@ def configure_params(
             self.is_azure = is_azure
         elif self.is_azure is None:
             self.is_azure = isinstance(self.settings, AzureOpenAISettings)
+        if azure_model_mapping is not None:
+            self.azure_model_mapping = azure_model_mapping
         if name is not None:
             self.name = name
         elif self.name is None:
             self.name = 'default'
+        if api_version is not None:
+            self.api_version = api_version
+        elif self.api_version is None:
+            self.api_version = self.settings.api_version
+
 
         if auth is not None:
             self.auth = auth
@@ -264,6 +269,7 @@ def configure_routes(self, **kwargs):
             max_retries = self.max_retries,
             settings = self.settings,
             is_azure = self.is_azure,
+            azure_model_mapping = self.azure_model_mapping,
             disable_retries = self.disable_retries,
             retry_function = self.retry_function,
             **kwargs