News Fetcher API

This is a FastAPI backend for fetching and summarizing news articles using Google News and AI models like OpenAI's GPT and GeminiAI.

Features

• Fetch top news articles using Google News.
• Summarize articles using OpenAI GPT-3 or GeminiAI.
• Caching of fetched articles and summaries to improve performance.
• Asynchronous operations to handle multiple requests efficiently.

Requirements

• Defined in requirements.txt

Setup

1. Clone the repository:

   git clone https://github.com/marcodestefano/newsfetcher.git
   cd newsfetcher

2. Create a virtual environment and activate it:

   python -m venv venv
   source venv/bin/activate  # On Windows use `venv\Scripts\activate`

3. Install the required dependencies:

   pip install -r requirements.txt

4. Create a .env file in the root directory and add your configuration:

   PORT=8000
   TIMEOUT=43200  # 12 hours in seconds
   ARTICLES=5
   MAX_ARTICLES=15
   LANGUAGE=it
   OPENAI_API_KEY=your_openai_api_key
   GEMINI_API_KEY=your_gemini_api_key
   

Usage

Run the FastAPI server:

fastapi dev main.py

Endpoints

Default Status Check

• URL: /
• Method: GET
• Response: JSON object with status "OK".

Fetch News URLs

• URL: /news
• Method: POST
• Parameters:
  • num (int): Number of articles to fetch. Default is 5.
  • language (str): Language of the news. Default is 'it'.
• Response: JSON array of articles with "url" and "title".

Fetch Article content

• URL: /article
• Method: POST
• Parameters:
  • url (str): URL of the article to fetch.
  • ai (str): AI service to use ('openai' or 'gemini'). Default is None. If ai is not defined, the article is not summarized.
  • model (str): Model to use with the AI service. Default models are used if not specified: gpt-3.5-turbo-0125 for openai and gemini-1.5-flash-latest for gemini
  • aikey (str): API key for the AI service. Default is None.
• Response: JSON object with "title" and "content".

Example Request

Using Postman or any HTTP client, send a POST request to /news:

{
  "num": 1,
  "language": "en"
}

And to /article:

{
  "url": "https://example.com/article",
  "ai": "openai",
  "model": "gpt-3.5-turbo",
  "aikey": "your_openai_api_key"
}

Deployment

The app is ready for deploy on Vercel.

Contributing

Contributions are welcome, please create a pull request with a description of your changes.

License

This project is licensed under the Apache 2.0 License.