mortafix / flashfacts

A flash news website that uses AI to condense news from YouTube

Flash Facts

Flash Facts is a mini-news website that uses AI to condense news/facts from YT videos.

Flash Facts provides quick and concise news summaries by leveraging artificial intelligence. It gathers news from various sources on YT and delivers them in a condensed format, making it easy to stay updated with minimal effort.

Features

• AI-Powered Summarization: Uses AI models (like ChatGPT and Gemini) to summarize news articles.
• Diverse News Sources: Aggregates news from multiple YT channels and/or playlists.
• User-Friendly Interface: Simple and modern web GUI.

Prerequisites

Before the first run, you’ll need the following API keys:

1. Gemini API key: required for generating news summaries.
2. YouTube Data API V3 key: required for fetching videos from YouTube.
3. [Optional] Unsplash API key: needed, if you want, to get automatically images from Unsplash.

Installation & Setup

To get started, follow these steps

# clone the repo
git clone https://github.com/yourusername/FlashFacts.git
cd FlashFacts

# create a virtual environment
python -m venv venv
source venv/bin/activate # linux and macos
.\venv\Scripts\activate # windows

# install the requirements
pip install -r requirements.txt

Use the --setup argument and fill up the information needed

python app.py --setup

Usage

There are two ways to run the script to fetch and generate news:

1. Provide a CSV of sources

python app.py -s examples/sources.csv

# example of sources
category    language    type        channel     id
news        en          channel     BBC         @BBCNews
news        it          playlist    Euronews    PLGWKgaTMGTQs65mOM0B_UgwDUzL2BXGcd
...

Optional parameters:

• -d: Specify the number of days to go back for videos.

# get last 7 days videos of the sources provided
python app.py -s examples/sources.csv -d 7

2. Provide a folder containing transcripts in text format

python app.py -t examples/transcripts

General optional parameters

• -l: Specify the output language.

python app.py -s examples/sources.csv -l italian

• -o: Specify the output file path.

python app.py -s examples/sources.csv -o output_folder/result

• -f: Specify the output format.

python app.py -s examples/sources.csv -f json

• --save-on-mongo: Save the output in a MongoDB collection.

python app.py -s examples/sources.csv --save-on-mongo

To use this last feature you need to setup the MongoDB connection.

python app.py --setup mongo

Web GUI

In order too Run the web interface, you only need to run the script with --gui.
WARNING: Web GUI, for now, onyl works with data on MongoDB

python app.py --gui

and go to http://localhost:8200. You can change host and port with the setup

python app.py --setup gui

Check out a working example: news.moris.dev

Future Additions

• OpenAI support (ChatGPT API)
• Web GUI news from local folders