跳转到主内容

PIP 包

PyPI version Open In Colab

🌟 激动人心的消息！ 现在，您可以将 gpt-researcher 无缝集成到您的应用程序中！

安装 GPT Researcher 的步骤​

按照以下简单步骤开始

  1. 先决条件：确保您的机器上已安装 Python 3.10+ 💻
  2. 安装 gpt-researcher：从 PyPi 获取官方包。
pip install gpt-researcher
  1. 环境变量： 创建一个包含您的 OpenAI API 密钥的 .env 文件，或直接导出它
export OPENAI_API_KEY={Your OpenAI API Key here}
export TAVILY_API_KEY={Your Tavily API Key here}
  1. 开始在您自己的代码库中使用 GPT Researcher

使用示例​

from gpt_researcher import GPTResearcher
import asyncio

async def get_report(query: str, report_type: str):
    researcher = GPTResearcher(query, report_type)
    research_result = await researcher.conduct_research()
    report = await researcher.write_report()
    
    # Get additional information
    research_context = researcher.get_research_context()
    research_costs = researcher.get_costs()
    research_images = researcher.get_research_images()
    research_sources = researcher.get_research_sources()
    
    return report, research_context, research_costs, research_images, research_sources

if __name__ == "__main__":
    query = "what team may win the NBA finals?"
    report_type = "research_report"

    report, context, costs, images, sources = asyncio.run(get_report(query, report_type))
    
    print("Report:")
    print(report)
    print("\nResearch Costs:")
    print(costs)
    print("\nNumber of Research Images:")
    print(len(images))
    print("\nNumber of Research Sources:")
    print(len(sources))

具体示例​

示例 1：研究报告​

query = "Latest developments in renewable energy technologies"
report_type = "research_report"

示例 2：资源报告​

query = "List of top AI conferences in 2023"
report_type = "resource_report"

示例 3：大纲报告​

query = "Outline for an article on the impact of AI in education"
report_type = "outline_report"

与 Web 框架集成​

FastAPI 示例​

from fastapi import FastAPI
from gpt_researcher import GPTResearcher
import asyncio

app = FastAPI()

@app.get("/report/{report_type}")
async def get_report(query: str, report_type: str) -> dict:
    researcher = GPTResearcher(query, report_type)
    research_result = await researcher.conduct_research()
    report = await researcher.write_report()
    
    source_urls = researcher.get_source_urls()
    research_costs = researcher.get_costs()
    research_images = researcher.get_research_images()
    research_sources = researcher.get_research_sources()
    
    return {
        "report": report,
        "source_urls": source_urls,
        "research_costs": research_costs,
        "num_images": len(research_images),
        "num_sources": len(research_sources)
    }

# Run the server
# uvicorn main:app --reload

Flask 示例​

先决条件：使用 async extra 安装 flask。

pip install 'flask[async]'
from flask import Flask, request, jsonify
from gpt_researcher import GPTResearcher

app = Flask(__name__)

@app.route('/report/<report_type>', methods=['GET'])
async def get_report(report_type):
    query = request.args.get('query')
    researcher = GPTResearcher(query, report_type)
    research_result = await researcher.conduct_research()
    report = await researcher.write_report()
    
    source_urls = researcher.get_source_urls()
    research_costs = researcher.get_costs()
    research_images = researcher.get_research_images()
    research_sources = researcher.get_research_sources()
    
    return jsonify({
        "report": report,
        "source_urls": source_urls,
        "research_costs": research_costs,
        "num_images": len(research_images),
        "num_sources": len(research_sources)
    })

# Run the server
# flask run

运行服务器

flask run

请求示例

curl -X GET "https://:5000/report/research_report?query=what team may win the nba finals?"

Getter 和 Setter​

GPT Researcher 提供了几种方法来检索有关研究过程的附加信息

获取研究来源​

来源是用于收集研究信息的 URL。

source_urls = researcher.get_source_urls()

获取研究上下文​

上下文是研究中检索到的所有信息。它包括来源及其相应的内容。

research_context = researcher.get_research_context()

获取研究成本​

成本是在研究过程中消耗的令牌数量。

research_costs = researcher.get_costs()

获取研究图片​

检索研究过程中找到的图片列表。

research_images = researcher.get_research_images()

获取研究来源​

检索研究来源列表，包括标题、内容和图片。

research_sources = researcher.get_research_sources()

设置详细模式 (Verbose)​

您可以设置详细模式以获取更详细的日志。

researcher.set_verbose(True)

添加成本​

如果您想跟踪外部使用的成本，也可以将成本添加到研究过程中。

researcher.add_costs(0.22)

高级用法​

自定义研究过程​

您可以通过在初始化 GPTResearcher 时传递附加参数来自定义研究过程的各个方面

researcher = GPTResearcher(
    query="Your research query",
    report_type="research_report",
    report_format="APA",
    tone="formal and objective",
    max_subtopics=5,
    verbose=True
)

处理研究结果​

进行研究后，您可以通过多种方式处理结果

# Conduct research
research_result = await researcher.conduct_research()

# Generate a standard report
report = await researcher.write_report()

# Generate a customized report with specific formatting requirements
custom_report = await researcher.write_report(custom_prompt="Answer in short, 2 paragraphs max without citations.")

# Generate a focused report for a specific audience
executive_summary = await researcher.write_report(custom_prompt="Create an executive summary focused on business impact and ROI. Keep it under 500 words.")

# Generate a report with specific structure requirements
technical_report = await researcher.write_report(custom_prompt="Create a technical report with problem statement, methodology, findings, and recommendations sections.")

# Generate a conclusion
conclusion = await researcher.write_report_conclusion(report)

# Get subtopics
subtopics = await researcher.get_subtopics()

# Get draft section titles for a subtopic
draft_titles = await researcher.get_draft_section_titles("Subtopic name")

使用自定义提示自定义报告生成​

write_report 方法接受一个 custom_prompt 参数，让您完全控制研究的呈现方式

# After conducting research
research_result = await researcher.conduct_research()

# Generate a report with a custom prompt
report = await researcher.write_report(
    custom_prompt="Based on the research, provide a bullet-point summary of the key findings."
)

自定义提示可用于多种目的

  1. 格式控制：指定报告的结构、长度或风格

    report = await researcher.write_report(
        custom_prompt="Write a blog post in a conversational tone using the research. Include headings and a conclusion."
    )

  2. 受众定位：为特定读者量身定制内容

    report = await researcher.write_report(
        custom_prompt="Create a report for technical stakeholders, focusing on methodologies and implementation details."
    )

  3. 专门输出：生成特定类型的内容

    report = await researcher.write_report(
        custom_prompt="Create a FAQ section based on the research with at least 5 questions and detailed answers."
    )

自定义提示将与研究上下文相结合，以生成您的定制报告。

使用研究上下文​

您可以使用研究上下文进行进一步处理或分析

# Get the full research context
context = researcher.get_research_context()

# Get similar written contents based on draft section titles
similar_contents = await researcher.get_similar_written_contents_by_draft_section_titles(
    current_subtopic="Subtopic name",
    draft_section_titles=["Title 1", "Title 2"],
    written_contents=some_written_contents,
    max_results=10
)

这份全面的文档应能帮助用户理解和充分利用 GPT Researcher 包的全部功能。