mistralai[patch]: streaming tool calls (#19469)

libs/partners/mistralai/langchain_mistralai/chat_models.py

@@ -128,13 +128,13 @@ def _convert_delta_to_message_chunk(
     _delta: Dict, default_class: Type[BaseMessageChunk]
 ) -> BaseMessageChunk:
     role = _delta.get("role")
-    content = _delta.get("content", "")
+    content = _delta.get("content") or ""
     if role == "user" or default_class == HumanMessageChunk:
         return HumanMessageChunk(content=content)
     elif role == "assistant" or default_class == AIMessageChunk:
         additional_kwargs: Dict = {}
         if tool_calls := _delta.get("tool_calls"):
-            additional_kwargs["tool_calls"] = [tc.model_dump() for tc in tool_calls]
+            additional_kwargs["tool_calls"] = tool_calls
         return AIMessageChunk(content=content, additional_kwargs=additional_kwargs)
     elif role == "system" or default_class == SystemMessageChunk:
         return SystemMessageChunk(content=content)
@@ -355,8 +355,6 @@ class ChatMistralAI(BaseChatModel):
             if len(chunk["choices"]) == 0:
                 continue
             delta = chunk["choices"][0]["delta"]
-            if not delta["content"]:
-                continue
             new_chunk = _convert_delta_to_message_chunk(delta, default_chunk_class)
             # make future chunks same type as first chunk
             default_chunk_class = new_chunk.__class__
@@ -384,8 +382,6 @@ class ChatMistralAI(BaseChatModel):
             if len(chunk["choices"]) == 0:
                 continue
             delta = chunk["choices"][0]["delta"]
-            if not delta["content"]:
-                continue
             new_chunk = _convert_delta_to_message_chunk(delta, default_chunk_class)
             # make future chunks same type as first chunk
             default_chunk_class = new_chunk.__class__

libs/partners/mistralai/pyproject.toml

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "langchain-mistralai"
-version = "0.1.0rc1"
+version = "0.1.0rc2"
 description = "An integration package connecting Mistral and LangChain"
 authors = []
 readme = "README.md"

libs/partners/mistralai/tests/integration_tests/test_chat_models.py

@@ -1,5 +1,11 @@
 """Test ChatMistral chat model."""
 
+import json
+from typing import Any
+
+from langchain_core.messages import AIMessageChunk
+from langchain_core.pydantic_v1 import BaseModel
+
 from langchain_mistralai.chat_models import ChatMistralAI
 
 
@@ -83,3 +89,58 @@ def test_structured_output() -> None:
         "What weighs more a pound of bricks or a pound of feathers"
     )
     assert isinstance(result, dict)
+
+
+def test_streaming_structured_output() -> None:
+    llm = ChatMistralAI(model="mistral-large", temperature=0)
+
+    class Person(BaseModel):
+        name: str
+        age: int
+
+    structured_llm = llm.with_structured_output(Person)
+    strm = structured_llm.stream("Erick, 27 years old")
+    chunk_num = 0
+    for chunk in strm:
+        assert chunk_num == 0, "should only have one chunk with model"
+        assert isinstance(chunk, Person)
+        assert chunk.name == "Erick"
+        assert chunk.age == 27
+        chunk_num += 1
+
+
+def test_streaming_tool_call() -> None:
+    llm = ChatMistralAI(model="mistral-large", temperature=0)
+
+    class Person(BaseModel):
+        name: str
+        age: int
+
+    tool_llm = llm.bind_tools([Person])
+
+    # where it calls the tool
+    strm = tool_llm.stream("Erick, 27 years old")
+
+    additional_kwargs = None
+    for chunk in strm:
+        assert isinstance(chunk, AIMessageChunk)
+        assert chunk.content == ""
+        additional_kwargs = chunk.additional_kwargs
+
+    assert additional_kwargs is not None
+    assert "tool_calls" in additional_kwargs
+    assert len(additional_kwargs["tool_calls"]) == 1
+    assert additional_kwargs["tool_calls"][0]["function"]["name"] == "Person"
+    assert json.loads(additional_kwargs["tool_calls"][0]["function"]["arguments"]) == {
+        "name": "Erick",
+        "age": 27,
+    }
+
+    # where it doesn't call the tool
+    strm = tool_llm.stream("What is 2+2?")
+    acc: Any = None
+    for chunk in strm:
+        assert isinstance(chunk, AIMessageChunk)
+        acc = chunk if acc is None else acc + chunk
+    assert acc.content != ""
+    assert "tool_calls" not in acc.additional_kwargs