langchain-b12 0.1.9__py3-none-any.whl → 0.1.11__py3-none-any.whl

langchain_b12/genai/genai.py

@@ -35,14 +35,7 @@ from langchain_core.tools import BaseTool
 from langchain_core.utils.function_calling import (
     convert_to_openai_tool,
 )
-from pydantic import BaseModel, ConfigDict, Field
-from tenacity import (
-    retry,
-    retry_if_exception_type,
-    stop_after_attempt,
-    stop_never,
-    wait_exponential_jitter,
-)
+from pydantic import BaseModel, ConfigDict, Field, field_validator, model_validator
 
 from langchain_b12.genai.genai_utils import (
     convert_messages_to_contents,
@@ -84,7 +77,9 @@ class ChatGenAI(BaseChatModel):
     seed: int | None = None
     """Random seed for the generation."""
     max_retries: int | None = Field(default=3)
-    """Maximum number of retries when generation fails. None disables retries."""
+    """Maximum number of retries. Prefer `http_retry_options`, but this is kept for compatibility."""
+    http_retry_options: types.HttpRetryOptions | None = Field(default=None)
+    """HTTP retry options for API requests. If not set, max_retries will be used to create default options."""
     safety_settings: list[types.SafetySetting] | None = None
     """The default safety settings to use for all generations.
 
@@ -107,6 +102,13 @@ class ChatGenAI(BaseChatModel):
         arbitrary_types_allowed=True,
     )
 
+    @model_validator(mode="after")
+    def _setup_retry_options(self) -> "ChatGenAI":
+        """Convert max_retries to http_retry_options if not explicitly set."""
+        if self.http_retry_options is None and self.max_retries is not None:
+            self.http_retry_options = types.HttpRetryOptions(attempts=self.max_retries)
+        return self
+
     @property
     def _llm_type(self) -> str:
         return "vertexai"
@@ -183,29 +185,10 @@ class ChatGenAI(BaseChatModel):
         run_manager: CallbackManagerForLLMRun | None = None,
         **kwargs: Any,
     ) -> ChatResult:
-        @retry(
-            reraise=True,
-            stop=stop_after_attempt(self.max_retries + 1)
-            if self.max_retries is not None
-            else stop_never,
-            wait=wait_exponential_jitter(initial=1, max=60),
-            retry=retry_if_exception_type(Exception),
-            before_sleep=lambda retry_state: logger.warning(
-                "ChatGenAI._generate failed (attempt %d/%s). "
-                "Retrying in %.2fs... Error: %s",
-                retry_state.attempt_number,
-                self.max_retries + 1 if self.max_retries is not None else "∞",
-                retry_state.next_action.sleep,
-                retry_state.outcome.exception(),
-            ),
+        stream_iter = self._stream(
+            messages, stop=stop, run_manager=run_manager, **kwargs
         )
-        def _generate_with_retry() -> ChatResult:
-            stream_iter = self._stream(
-                messages, stop=stop, run_manager=run_manager, **kwargs
-            )
-            return generate_from_stream(stream_iter)
-
-        return _generate_with_retry()
+        return generate_from_stream(stream_iter)
 
     async def _agenerate(
         self,
@@ -214,29 +197,10 @@ class ChatGenAI(BaseChatModel):
         run_manager: AsyncCallbackManagerForLLMRun | None = None,
         **kwargs: Any,
     ) -> ChatResult:
-        @retry(
-            reraise=True,
-            stop=stop_after_attempt(self.max_retries + 1)
-            if self.max_retries is not None
-            else stop_never,
-            wait=wait_exponential_jitter(initial=1, max=60),
-            retry=retry_if_exception_type(Exception),
-            before_sleep=lambda retry_state: logger.warning(
-                "ChatGenAI._agenerate failed (attempt %d/%s). "
-                "Retrying in %.2fs... Error: %s",
-                retry_state.attempt_number,
-                self.max_retries + 1 if self.max_retries is not None else "∞",
-                retry_state.next_action.sleep,
-                retry_state.outcome.exception(),
-            ),
+        stream_iter = self._astream(
+            messages, stop=stop, run_manager=run_manager, **kwargs
         )
-        async def _agenerate_with_retry() -> ChatResult:
-            stream_iter = self._astream(
-                messages, stop=stop, run_manager=run_manager, **kwargs
-            )
-            return await agenerate_from_stream(stream_iter)
-
-        return await _agenerate_with_retry()
+        return await agenerate_from_stream(stream_iter)
 
     def _stream(
         self,
@@ -246,10 +210,16 @@ class ChatGenAI(BaseChatModel):
         **kwargs: Any,
     ) -> Iterator[ChatGenerationChunk]:
         system_message, contents = self._prepare_request(messages=messages)
+        http_options = (
+            types.HttpOptions(retry_options=self.http_retry_options)
+            if self.http_retry_options
+            else None
+        )
         response_iter = self.client.models.generate_content_stream(
             model=self.model_name,
             contents=contents,
             config=types.GenerateContentConfig(
+                http_options=http_options,
                 system_instruction=system_message,
                 temperature=self.temperature,
                 top_k=self.top_k,
@@ -282,10 +252,16 @@ class ChatGenAI(BaseChatModel):
         **kwargs: Any,
     ) -> AsyncIterator[ChatGenerationChunk]:
         system_message, contents = self._prepare_request(messages=messages)
+        http_options = (
+            types.HttpOptions(retry_options=self.http_retry_options)
+            if self.http_retry_options
+            else None
+        )
         response_iter = self.client.aio.models.generate_content_stream(
             model=self.model_name,
             contents=contents,
             config=types.GenerateContentConfig(
+                http_options=http_options,
                 system_instruction=system_message,
                 temperature=self.temperature,
                 top_k=self.top_k,

{langchain_b12-0.1.9.dist-info → langchain_b12-0.1.11.dist-info}/METADATA

@@ -1,11 +1,10 @@
 Metadata-Version: 2.4
 Name: langchain-b12
-Version: 0.1.9
+Version: 0.1.11
 Summary: A reusable collection of tools and implementations for Langchain
 Author-email: Vincent Min <vincent.min@b12-consulting.com>
 Requires-Python: >=3.11
 Requires-Dist: langchain-core>=0.3.60
-Requires-Dist: tenacity>=9.1.2
 Description-Content-Type: text/markdown
 
 # Langchain B12

{langchain_b12-0.1.9.dist-info → langchain_b12-0.1.11.dist-info}/RECORD

@@ -2,8 +2,8 @@ langchain_b12/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 langchain_b12/py.typed,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 langchain_b12/citations/citations.py,sha256=ZQvYayjQXIUaRosJ0qwL3Nc7kC8sBzmaIkE-BOslaVI,12261
 langchain_b12/genai/embeddings.py,sha256=h0Z-5PltDW9q79AjSrLemsz-_QKMB-043XXDvYSRQds,3483
-langchain_b12/genai/genai.py,sha256=zZdGwkkaxobnA1jT07MvWHaIHKeBis4X1zJ8o5KrnHk,18841
+langchain_b12/genai/genai.py,sha256=u-QAH_4VauBj99dWuYBaxAMT3bNbqMdxM-rYgRKctLw,18074
 langchain_b12/genai/genai_utils.py,sha256=tA6UiJURK25-11vtaX4768UV47jDCYwVKIIWydD4Egw,10736
-langchain_b12-0.1.9.dist-info/METADATA,sha256=CpmYdZwkej43d-DvTOR8FU65QSHAZ4iiFu7Ale8Gy0I,1235
-langchain_b12-0.1.9.dist-info/WHEEL,sha256=WLgqFyCfm_KASv4WHyYy0P3pM_m7J5L9k2skdKLirC8,87
-langchain_b12-0.1.9.dist-info/RECORD,,
+langchain_b12-0.1.11.dist-info/METADATA,sha256=w4uaxeVl7hg1h2Zj3J9ZOukgUAXHeJcQA0rrtBnmdyg,1205
+langchain_b12-0.1.11.dist-info/WHEEL,sha256=WLgqFyCfm_KASv4WHyYy0P3pM_m7J5L9k2skdKLirC8,87
+langchain_b12-0.1.11.dist-info/RECORD,,