t407o / paperboy

A faithful deliverer to help watching new arrivals in messaging apps.

paperboy

A faithful deliverer to help watching new arrivals in messaging apps.

paperboy is designed to...

1. Fetch new arrivals from data sources
2. Summarize it by ChatGPT API
3. Post the summary to massaging apps

graph
    subgraph DS[Data sources]
        Feeds
        GS[Google Scholar]
    end

    subgraph MS[Messaging apps]
        Discord
        Gmail
    end
    
    DS --> |provide message|MS
    Feeds -.-> |uses| ChatGPT
    GS -.-> |uses| ChatGPT

Loading

Data souces

The followings are supported for now

1. Feeds - e.g. GitHub Blog, Spotify Engineering, etc..
2. Google Scholar - i.e. https://scholar.google.com/

Messaging apps

1. Discord
2. Gmail

Prerequisites

• Python 3.12+ (may work on older, but not assured)
• Open AI Account & API Key
• Discord Account or Gmail Account

CAUTION!!

You may be charged by OpenAI if you use ChatGPT API through this tool.
We do not take any responsibility for any damage or loss caused through this tool.

For instance, if there is a new arrival on a feed you watching, paperboy will pass entire the web page text to ChatGPT API. The cost depends on the amount of the text in the page but we can't manage it.

Also remember that if you watch more data souces, you may be charged more.
Please be sure to handle the cost by yourself.

NOTE: Setting soft/hard limit will be great help to manage your budget.

Preparation

1. Install python libraries

pip install -r requirements.txt

2. Create discord app

Let's try with the document! Don't forget to note the Bot Token and Channel Id.

3. Set environment variables

Environment Variable	Description
OPENAI_API_KEY	Your OpenAI API key.
DISCORD_BOT_TOKEN	Your Discord bot token.
DISCORD_CHANNEL_ID	The channel ID to post the paper.
DISCORD_NOTIFIEE	(Optional) The user name to be notified on finish (if one new arrival is there).

If you are working with devcontainer(VSCode plugin), .env file will be loaded when you start the container

4. Set configurations

paperboy will refer the configurations in config directory.

Basically, you have to create config directory by yourself, refering samples or the docs.
If you run without config dir, the directory will be generated from config_sample directory.

Run locally

python main.py

By default,

• only feeds will be refered as a data source
• ChatGPT's gpt-3.5-turbo-16k model will be used to summarize texts

Extentions

The extentions are effectively the implementation of the data sources.
By enabling the extentions, you can watching new arrivals from different data sources.

The extentions will be enabled by specifing by -e option.

python main.py -e google_scholar

Extention	Description
feeds	Get new articles from specified feeds. If -e option is omitted, this extention will be enabled automatically.
google_scholar	Get new papers from Google Scholar by searching by keywords

See the README.md files for each extentions for more details.