GoogleSearch tool

[khteh]

Checked other resources

• I added a very descriptive title to this question.
• I searched the LangChain documentation with the integrated search.
• I used the GitHub search to find a similar question and didn't find it.

Commit to Help

• I commit to help with one of those options 👆

Example Code

from google import genai
from google.genai import types
from google.genai.types import Tool, GenerateContentConfig, GoogleSearch

@tool
async def GoogleSearch(
    query: str, *, config: Annotated[RunnableConfig, InjectedToolArg]
)-> Optional[list[str]]:
    """Search for general web results.

    This function performs a search using the Google search engine, which is designed
    to provide comprehensive, accurate, and trusted results. It's particularly useful
    for answering questions about current events.
    """
    client = genai.Client(api_key=os.environ.get("GEMINI_API_KEY"))
    model_id = "gemini-2.0-flash"
    google_search_tool = Tool(
        google_search = GoogleSearch()
    )
    response = client.models.generate_content(
        model=model_id,
        contents=query,
        config=GenerateContentConfig(
            tools=[google_search_tool],
            response_modalities=["TEXT"],
        )
    )
    return [content.text for content in response.candidates[0].content.parts]

Warning:

/usr/src/Python/rag-agent/src/rag_agent/Tools.py:51: LangChainDeprecationWarning: The method `BaseTool.__call__` was deprecated in langchain-core 0.1.47 and will be removed in 1.0. Use :meth:`~invoke` instead.

Error:

Error: TypeError("BaseTool.__call__() missing 1 required positional argument: 'tool_input'")

Description

I am trying to use Google GenAI GoogleSearch tool with google_vertexai/gemini-2.0-flash and hits the error.

System Info

langchain==0.3.20
langchain-community==0.3.19
langchain-core==0.3.43
langchain-google-genai==2.0.11
langchain-google-vertexai==2.0.14
langchain-openai==0.3.8
langchain-text-splitters==0.3.6

3.12.7

[PragyanTiwari]

Although your system has the latest version of langchain-core, the tool GoogleSearch isn't utilizing it.
Try navigating to the directory /usr/src/Python/rag-agent/src/rag_agent and run the update on langchain-core.

Therefore, you might be encountering the error because it is using the deprecated calling method instead of the invoke method.

Also, check which python interpreter you're using and whether you're working with a particular virtual env or not.

[khteh]

It works now. I have to use API Key for VertexAI project name.

[khteh]

Any documentation for GenerateContentConfig response_modalities? The example uses TEXT and it is markdown text.

[PragyanTiwari]

This is the link you can know through the documentation here

It is indeed a markdown text. Since, you're using python, you can use the markdown library and pass the generated content to it.

Here is the way to get it:

import re
import markdown

# replace the return value with this:
return [(re.sub('<.*?>', '', markdown.markdown(content.text))) for content in response.candidates[0].content.parts]

By default, google models always generate the markdown text, where the plain text becomes a part of it. If you're thinking is there any other argument other than TEXT to pass to parameter response_modalities, well that's the only argument which supports text, rest of the values are about video, image etc.

[khteh]

This is clearer: https://ai.google.dev/api/generate-content#Modality

[khteh]

Thanks!