How to use gemini api?

[warmshao]

It appears that langchain-google-genai has not yet released a version beyond 2.0.9. If you need to use a fix related to function calling or schema handling before an official release, you can apply a manual patch. This workaround involves directly modifying the installed package files based on changes made in a recent pull request.

1. Locate the langchain-google-genai Installation Path:

   • Use the command pip show langchain-google-genai to determine the installed location of the package.
   • For example, the output might show a path like D:\softwares\miniconda\envs\agent\Lib\site-packages\langchain_google_genai.

2. Navigate to the Installation Directory:

   • Open your file explorer or command line and go to the directory obtained in step 1.

3. Modify the _function_utils.py File:

   • Locate the file named _function_utils.py within the langchain_google_genai directory.
   • Open _function_utils.py with a text editor (like VS Code, Notepad++, etc.).
   • Find line 318, which likely looks like this:

     if properties_item.get("type_") == glm.Type.OBJECT:
        ... (original code)

   • Replace the original code block with the following:

     if properties_item.get("type_") == glm.Type.OBJECT:
         if (
                 v.get("anyOf")
                 and isinstance(v["anyOf"], list)
                 and isinstance(v["anyOf"][0], dict)
         ):
             v = v["anyOf"][0]
         v_properties = v.get("properties")
         if v_properties:
             properties_item["properties"] = _get_properties_from_schema_any(
                 v_properties
             )
             if isinstance(v_properties, dict):
                 properties_item["required"] = [
                     k for k, v in v_properties.items() if "default" not in v
                 ]
         else:
             # Providing dummy type for object without properties
             properties_item["type_"] = glm.Type.STRING

4. Save the Changes:

   • Save the modified _function_utils.py file.

refer: https://github.com/langchain-ai/langchain-google/pull/658/files

[warmshao]

Seems that langchain-google-genai have updated. Just use pip install langchain-google-genai -U will work fine