mirror of
https://github.com/hwchase17/langchain.git
synced 2025-07-30 16:24:24 +00:00
Merge f764f6ea2d into 0e287763cd
This commit is contained in:
Leva
GitHub
commit
c0b50211da
19
docs/docs/integrations/providers/desearch.mdx
Normal file
19
docs/docs/integrations/providers/desearch.mdx
Normal file
@ -0,0 +1,19 @@
|
||||
# Desearch
|
||||
|
||||
[Desearch](https://desearch.ai/) is a static-site generator which provides out-of-the-box documentation features.
|
||||
|
||||
|
||||
## Installation and Setup
|
||||
|
||||
|
||||
```bash
|
||||
pip install -U langchain-desearch
|
||||
```
|
||||
|
||||
## Document Loader
|
||||
|
||||
See a [usage example](/docs/integrations/tools/desearch).
|
||||
|
||||
```python
|
||||
from langchain_desearch.tools import DesearchTool, BasicWebSearchTool, BasicTwitterSearchTool
|
||||
```
|
||||
319
docs/docs/integrations/tools/desearch.ipynb
Normal file
319
docs/docs/integrations/tools/desearch.ipynb
Normal file
@ -0,0 +1,319 @@
|
||||
{
|
||||
"cells": [
|
||||
{
|
||||
"cell_type": "markdown",
|
||||
"metadata": {},
|
||||
"source": [
|
||||
"## Overview"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "markdown",
|
||||
"metadata": {},
|
||||
"source": [
|
||||
"AI-Powered Search Engine and API for Advanced Data Discovery [Desearch](https://desearch.ai).\n",
|
||||
"\n",
|
||||
"Welcome to the comprehensive guide on integrating Desearch with LangChain to enable advanced Retrieval-Augmented Generation (RAG) and agent-style workflows. Desearch is a cutting-edge, privacy-first search API that excels in delivering relevant, real-time web and Twitter content, specifically optimized for use in Large Language Model (LLM) applications.\n",
|
||||
"\n",
|
||||
"First, get an Desearch API key and add it as an environment variable. Get $5 free credit (plus more by completing certain actions like making your first search) by [signing up here](https://console.desearch.ai)."
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "markdown",
|
||||
"metadata": {},
|
||||
"source": [
|
||||
"## Setup\n"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "markdown",
|
||||
"metadata": {},
|
||||
"source": [
|
||||
".env DESEARCH_API_KEY=your_api_key_here"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"metadata": {},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"from dotenv import load_dotenv\n",
|
||||
"\n",
|
||||
"load_dotenv()\n",
|
||||
"# Set your API key as an environment variable"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "markdown",
|
||||
"metadata": {},
|
||||
"source": [
|
||||
"And install the integration package"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"metadata": {},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"%pip install -U langchain-desearch"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "markdown",
|
||||
"metadata": {},
|
||||
"source": [
|
||||
"## Instantiation"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "markdown",
|
||||
"metadata": {},
|
||||
"source": [
|
||||
"### 🛠️ Tools Overview\n",
|
||||
"\n",
|
||||
"Desearch provides a suite of modular tools compatible with LangChain, each designed for specific use cases:"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "markdown",
|
||||
"metadata": {},
|
||||
"source": [
|
||||
"### 🔎 Search Tools\n",
|
||||
"\n",
|
||||
"* **DesearchTool**: A comprehensive tool for searching across web, AI, and Twitter posts.\n",
|
||||
"* **BasicWebSearchTool**: A lightweight tool focused solely on web searches.\n",
|
||||
"* **BasicTwitterSearchTool**: A specialized tool for searching tweets with advanced filtering options."
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "markdown",
|
||||
"metadata": {},
|
||||
"source": [
|
||||
"## Invocation"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "markdown",
|
||||
"metadata": {},
|
||||
"source": [
|
||||
"### 🧪 Quick Examples"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "markdown",
|
||||
"metadata": {},
|
||||
"source": [
|
||||
"#### Basic Usage"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "markdown",
|
||||
"metadata": {},
|
||||
"source": [
|
||||
"Here's a simple example to demonstrate how to use the DesearchTool for web searches"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"metadata": {},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"from langchain_desearch.tools import DesearchTool\n",
|
||||
"\n",
|
||||
"# Initialize the tool\n",
|
||||
"tool = DesearchTool()\n",
|
||||
"\n",
|
||||
"# Run a search query\n",
|
||||
"response = tool._run(\n",
|
||||
" prompt=\"Bittensor network activity\",\n",
|
||||
" tool=[\"web\"],\n",
|
||||
" model=\"NOVA\",\n",
|
||||
" date_filter=\"PAST_24_HOURS\",\n",
|
||||
")\n",
|
||||
"\n",
|
||||
"# Print the response\n",
|
||||
"print(response)"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "markdown",
|
||||
"metadata": {},
|
||||
"source": [
|
||||
"#### Search Twitter\n",
|
||||
"\n",
|
||||
"To search Twitter for specific topics, use the BasicTwitterSearchTool"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"metadata": {},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"from langchain_desearch.tools import BasicTwitterSearchTool\n",
|
||||
"\n",
|
||||
"# Initialize the Twitter search tool\n",
|
||||
"tool = BasicTwitterSearchTool()\n",
|
||||
"\n",
|
||||
"# Execute a search query\n",
|
||||
"response = tool._run(query=\"AI governance\", sort=\"Top\", count=5)\n",
|
||||
"\n",
|
||||
"# Display the results\n",
|
||||
"print(response)"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "markdown",
|
||||
"metadata": {},
|
||||
"source": [
|
||||
"## Use within an agent"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "markdown",
|
||||
"metadata": {},
|
||||
"source": [
|
||||
"### 🤖 Building a RAG Chain\n",
|
||||
"\n",
|
||||
"Creating a RAG chain involves integrating Desearch with LangChain's prompt and LLM capabilities"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"metadata": {},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"from langchain_core.output_parsers import StrOutputParser\n",
|
||||
"from langchain_core.prompts import ChatPromptTemplate, PromptTemplate\n",
|
||||
"from langchain_core.runnables import RunnableParallel, RunnablePassthrough\n",
|
||||
"from langchain_deepseek import ChatDeepSeek\n",
|
||||
"from langchain_desearch.tools import DesearchTool\n",
|
||||
"\n",
|
||||
"# Initialize the Desearch tool\n",
|
||||
"tool = DesearchTool()\n",
|
||||
"\n",
|
||||
"\n",
|
||||
"# Define a function to fetch context\n",
|
||||
"def fetch_context(query):\n",
|
||||
" return tool._run(prompt=query, tool=\"desearch_web\", model=\"NOVA\")\n",
|
||||
"\n",
|
||||
"\n",
|
||||
"# Create a prompt template\n",
|
||||
"prompt = ChatPromptTemplate.from_messages(\n",
|
||||
" [\n",
|
||||
" (\"system\", \"You are a research assistant.\"),\n",
|
||||
" (\n",
|
||||
" \"human\",\n",
|
||||
" \"Answer this using the provided context:\\n\\n<context>{context}</context>\",\n",
|
||||
" ),\n",
|
||||
" ]\n",
|
||||
")\n",
|
||||
"\n",
|
||||
"# Set up the LLM\n",
|
||||
"llm = ChatDeepSeek(model=\"deepseek-chat\", temperature=0.7)\n",
|
||||
"parser = StrOutputParser()\n",
|
||||
"\n",
|
||||
"# Build the RAG chain\n",
|
||||
"chain = (\n",
|
||||
" RunnableParallel(\n",
|
||||
" {\n",
|
||||
" \"query\": RunnablePassthrough(),\n",
|
||||
" \"context\": lambda q: fetch_context(q[\"query\"]),\n",
|
||||
" }\n",
|
||||
" )\n",
|
||||
" | prompt\n",
|
||||
" | llm\n",
|
||||
" | parser\n",
|
||||
")\n",
|
||||
"\n",
|
||||
"# Invoke the chain with a query\n",
|
||||
"result = chain.invoke(\"What is the latest update on the EU AI Act?\")"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "markdown",
|
||||
"metadata": {},
|
||||
"source": [
|
||||
"### 🧠 Using Desearch in a LangChain Agent\n",
|
||||
"\n",
|
||||
"Integrate Desearch into a LangChain agent to automate complex workflows:"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"metadata": {},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"from langchain_deepseek import ChatDeepSeek\n",
|
||||
"from langchain_desearch.agent import create_search_agent\n",
|
||||
"\n",
|
||||
"# Initialize the LLM\n",
|
||||
"llm = ChatDeepSeek(model=\"deepseek-chat\", temperature=0.7)\n",
|
||||
"\n",
|
||||
"# Create a search agent\n",
|
||||
"agent = create_search_agent(llm=llm)\n",
|
||||
"\n",
|
||||
"# Invoke the agent with an input message\n",
|
||||
"response = agent.invoke({\"input_message\": \"Summarize current AI regulation trends\"})\n",
|
||||
"print(response[\"output\"])"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "markdown",
|
||||
"metadata": {},
|
||||
"source": [
|
||||
"## API reference\n",
|
||||
"\n",
|
||||
"For detailed API reference, please refer to the official documentation or the source code comments for each tool and method."
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "markdown",
|
||||
"metadata": {},
|
||||
"source": [
|
||||
"## ✅ Testing\n",
|
||||
"\n",
|
||||
"### Unit Tests (Mocked)\n",
|
||||
"\n",
|
||||
"To ensure your integration works as expected, run unit tests:"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"metadata": {},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"%pytest tests/integration_tests/test_tools.py"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"metadata": {},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"%pytest tests/unit_tests/test_imports.py"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "markdown",
|
||||
"metadata": {},
|
||||
"source": [
|
||||
"This structure adds the required headers and organizes the content under them. You can adjust the content under each header as needed to better fit the specific details of your project."
|
||||
]
|
||||
}
|
||||
],
|
||||
"metadata": {
|
||||
"language_info": {
|
||||
"name": "python"
|
||||
}
|
||||
},
|
||||
"nbformat": 4,
|
||||
"nbformat_minor": 2
|
||||
}
|
||||
Loading…
Reference in New Issue
Block a user