edvm / gpti Goto Github PK

If -some option (think about a good name) is passed along with a text (string) then use that as a prompt, instead of checking if its defined in the .toml config file.

Take config file from ~/.gpti.toml, if not found, check if it was defined by -c (command line argument). If not found, print an error and exit the program.

Right now, model is hardcoded in the source code:

    let chat_completion = ChatCompletion::builder("gpt-3.5-turbo", messages.clone())
        .create()
        .await
        .unwrap();

Move it to .toml config file so users are free to set the model they want to use

if -cb is given, then copy the output we get from OpenAI to the clipboard. This make it easier to later paste the response we got from OpenAI.

Would be a nice touch to add "spinners" so while we wait to get a response from OpenAI API we show a spinner in the terminal: https://crates.io/crates/spinners

On match expresisons, on Err branches, use early returns to make the code return an error (now probably its doing .unrwap() because I'm learning Rust heh)