make image inputs compatible with langchain_ollama (langchain-ai#24619)

caseyclements · Jul 27, 2024 · 152427e · 152427e
1 parent 0535d72
commit 152427e
Show file tree

Hide file tree

Showing 2 changed files with 32 additions and 7 deletions.
diff --git a/libs/partners/ollama/langchain_ollama/chat_models.py b/libs/partners/ollama/langchain_ollama/chat_models.py
@@ -346,7 +346,7 @@ def _convert_messages_to_ollama_messages(
     ) -> Sequence[Message]:
         ollama_messages: List = []
         for message in messages:
-            role = ""
+            role: Literal["user", "assistant", "system", "tool"]
             tool_call_id: Optional[str] = None
             tool_calls: Optional[List[Dict[str, Any]]] = None
             if isinstance(message, HumanMessage):
@@ -383,11 +383,13 @@ def _convert_messages_to_ollama_messages(
                         image_url = None
                         temp_image_url = content_part.get("image_url")
                         if isinstance(temp_image_url, str):
-                            image_url = content_part["image_url"]
+                            image_url = temp_image_url
                         elif (
-                            isinstance(temp_image_url, dict) and "url" in temp_image_url
+                            isinstance(temp_image_url, dict)
+                            and "url" in temp_image_url
+                            and isinstance(temp_image_url["url"], str)
                         ):
-                            image_url = temp_image_url
+                            image_url = temp_image_url["url"]
                         else:
                             raise ValueError(
                                 "Only string image_url or dict with string 'url' "
@@ -408,15 +410,16 @@ def _convert_messages_to_ollama_messages(
                             "Must either have type 'text' or type 'image_url' "
                             "with a string 'image_url' field."
                         )
-            msg = {
+            # Should convert to ollama.Message once role includes tool, and tool_call_id is in Message # noqa: E501
+            msg: dict = {
                 "role": role,
                 "content": content,
                 "images": images,
             }
+            if tool_calls:
+                msg["tool_calls"] = tool_calls  # type: ignore
             if tool_call_id:
                 msg["tool_call_id"] = tool_call_id
-            if tool_calls:
-                msg["tool_calls"] = tool_calls
             ollama_messages.append(msg)
 
         return ollama_messages

diff --git a/libs/partners/ollama/tests/integration_tests/test_chat_models.py b/libs/partners/ollama/tests/integration_tests/test_chat_models.py
@@ -2,6 +2,8 @@
 
 from typing import Type
 
+import pytest
+from langchain_core.language_models import BaseChatModel
 from langchain_standard_tests.integration_tests import ChatModelIntegrationTests
 
 from langchain_ollama.chat_models import ChatOllama
@@ -15,3 +17,23 @@ def chat_model_class(self) -> Type[ChatOllama]:
     @property
     def chat_model_params(self) -> dict:
         return {"model": "llama3-groq-tool-use"}
+
+    @property
+    def supports_image_inputs(self) -> bool:
+        return True
+
+    @pytest.mark.xfail(
+        reason=(
+            "Fails with 'AssertionError'. Ollama does not support 'tool_choice' yet."
+        )
+    )
+    def test_structured_output(self, model: BaseChatModel) -> None:
+        super().test_structured_output(model)
+
+    @pytest.mark.xfail(
+        reason=(
+            "Fails with 'AssertionError'. Ollama does not support 'tool_choice' yet."
+        )
+    )
+    def test_structured_output_pydantic_2_v1(self, model: BaseChatModel) -> None:
+        super().test_structured_output_pydantic_2_v1(model)