How to Use Search Tool Langchain: Simple Guide

To use the SearchTool in Langchain, import it and create an instance with a search function that returns relevant documents. Then, call the tool with a query string to get search results. This lets you integrate search capabilities easily in your Langchain applications.

Syntax

The SearchTool requires a search function that takes a query string and returns a list of results. You create the tool by passing this function to SearchTool.from_function(). Then, use the run() method with your query to get results.

• search_function: Your custom function to perform search.
• SearchTool.from_function(search_function): Creates the search tool.
• tool.run(query): Runs the search with the query string.

python

from langchain.tools import SearchTool

def search_function(query: str) -> list[str]:
    # Your search logic here
    return ["Result 1", "Result 2"]

search_tool = SearchTool.from_function(search_function)

results = search_tool.run("example query")
print(results)

Output

['Result 1', 'Result 2']

Example

This example shows how to create a simple search tool that returns matching items from a list. It demonstrates defining the search function, creating the tool, and running a query.

python

from langchain.tools import SearchTool

# Sample data to search
documents = [
    "Learn Python programming",
    "Introduction to Langchain",
    "Advanced search techniques",
    "Langchain search tool example"
]

def simple_search(query: str) -> list[str]:
    # Return documents containing the query (case-insensitive)
    return [doc for doc in documents if query.lower() in doc.lower()]

# Create the search tool
search_tool = SearchTool.from_function(simple_search)

# Run a search query
output = search_tool.run("langchain")
print(output)

Output

["Introduction to Langchain", "Langchain search tool example"]

Common Pitfalls

Common mistakes when using the Langchain search tool include:

• Not returning a list of strings from the search function, which causes errors.
• Using a search function that is too slow or blocking, which delays results.
• Forgetting to handle case sensitivity or empty queries, leading to missed results.

Always ensure your search function returns a list and handles queries gracefully.

python

from langchain.tools import SearchTool

# Wrong: returns a string instead of list
# def bad_search(query: str) -> str:
#     return "No results"

# Right: returns a list

def good_search(query: str) -> list[str]:
    return ["No results"]

search_tool = SearchTool.from_function(good_search)
print(search_tool.run("test"))

Output

['No results']

Quick Reference

Remember these key points when using the Langchain search tool:

• Define a search function that takes a string and returns a list of strings.
• Create the tool with SearchTool.from_function().
• Use run() to execute the search.
• Handle empty or invalid queries inside your search function.
• Keep the search function efficient for better performance.

Key Takeaways

Use a function that returns a list of strings as the search logic for SearchTool.

Create the search tool with SearchTool.from_function(your_search_function).

Call run(query) on the tool to get search results.

Ensure your search function handles queries efficiently and returns consistent results.

Avoid returning single strings or non-list types from the search function.