refactor(core): implement lazy initialization for OpenAI clients

reme/__init__.py

@@ -20,7 +20,7 @@ __all__ = [
     "ReMeFs",
 ]
 
-__version__ = "0.3.0.0a8"
+__version__ = "0.3.0.0a9"
 
 
 """

reme/core/embedding/openai_embedding_model.py

@@ -15,16 +15,23 @@ class OpenAIEmbeddingModel(BaseEmbeddingModel):
         super().__init__(**kwargs)
         self.encoding_format: Literal["float", "base64"] = encoding_format
 
-        # Create client using factory method
-        self._client = self._create_client()
+        # Lazy client initialization
+        self._client = None
 
     def _create_client(self):
         """Create and return an internal AsyncOpenAI client instance."""
         return AsyncOpenAI(api_key=self.api_key, base_url=self.base_url)
 
+    @property
+    def client(self):
+        """Lazily create and return the AsyncOpenAI client."""
+        if self._client is None:
+            self._client = self._create_client()
+        return self._client
+
     async def _get_embeddings(self, input_text: list[str], **kwargs) -> list[list[float]]:
         """Fetch embeddings from the API for a batch of strings."""
-        completion = await self._client.embeddings.create(
+        completion = await self.client.embeddings.create(
             model=self.model_name,
             input=input_text,
             dimensions=self.dimensions,
@@ -40,4 +47,6 @@ class OpenAIEmbeddingModel(BaseEmbeddingModel):
 
     async def close(self):
         """Close the asynchronous OpenAI client and release network resources."""
-        await self._client.close()
+        if self._client is not None:
+            await self._client.close()
+            self._client = None

reme/core/embedding/openai_embedding_model_sync.py

@@ -14,7 +14,7 @@ class OpenAIEmbeddingModelSync(OpenAIEmbeddingModel):
 
     def _get_embeddings_sync(self, input_text: list[str], **kwargs) -> list[list[float]]:
         """Fetch embeddings synchronously from the API for a batch of strings."""
-        completion = self._client.embeddings.create(
+        completion = self.client.embeddings.create(
             model=self.model_name,
             input=input_text,
             dimensions=self.dimensions,
@@ -30,4 +30,6 @@ class OpenAIEmbeddingModelSync(OpenAIEmbeddingModel):
 
     def close_sync(self):
         """Close the synchronous OpenAI client and release network resources."""
-        self._client.close()
+        if self._client is not None:
+            self._client.close()
+            self._client = None

reme/core/llm/openai_llm.py

@@ -19,13 +19,20 @@ class OpenAILLM(BaseLLM):
         """Initialize the OpenAI async client with API credentials and model configuration."""
         super().__init__(**kwargs)
 
-        # Create client using factory method
-        self._client = self._create_client()
+        # Lazy client initialization
+        self._client = None
 
     def _create_client(self):
         """Create and return an instance of the AsyncOpenAI client."""
         return AsyncOpenAI(api_key=self.api_key, base_url=self.base_url)
 
+    @property
+    def client(self):
+        """Lazily create and return the AsyncOpenAI client."""
+        if self._client is None:
+            self._client = self._create_client()
+        return self._client
+
     def _build_stream_kwargs(
         self,
         messages: list[Message],
@@ -76,7 +83,7 @@ class OpenAILLM(BaseLLM):
     ) -> AsyncGenerator[StreamChunk, None]:
         """Generate a stream of chat completion chunks including text, reasoning content, and tool calls."""
         stream_kwargs = stream_kwargs or {}
-        completion = await self._client.chat.completions.create(**stream_kwargs)
+        completion = await self.client.chat.completions.create(**stream_kwargs)
         ret_tool_calls: list[ToolCall] = []
 
         async for chunk in completion:
@@ -102,4 +109,6 @@ class OpenAILLM(BaseLLM):
 
     async def close(self):
         """Asynchronously close the OpenAI client and release network resources."""
-        await self._client.close()
+        if self._client is not None:
+            await self._client.close()
+            self._client = None

reme/core/llm/openai_llm_sync.py

@@ -26,7 +26,7 @@ class OpenAILLMSync(OpenAILLM):
     ) -> Generator[StreamChunk, None, None]:
         """Synchronously generate a stream of chat completion chunks including text, reasoning, and tool calls."""
         stream_kwargs = stream_kwargs or {}
-        completion = self._client.chat.completions.create(**stream_kwargs)
+        completion = self.client.chat.completions.create(**stream_kwargs)
         ret_tool_calls: list[ToolCall] = []
 
         for chunk in completion:
@@ -52,4 +52,6 @@ class OpenAILLMSync(OpenAILLM):
 
     def close_sync(self):
         """Close the synchronous OpenAI client and release network resources."""
-        self._client.close()
+        if self._client is not None:
+            self._client.close()
+            self._client = None