README

This is a proof of concept repository for querying pdf files. It allows the user to interrogate a pdf file on the CLI and returns a response in the CLI.

Pre-requisites

1. You will need python 3.11 or greater.
2. You will need pipx. Install it using the following commands in your CLI.

> py -3 -m pip install --user pipx
> py -3 -m pipx ensurepath

3. You will need poetry. Install it using the following commands in your CLI.

> pipx install poetry
> poetry config virtualenvs.in-project = true

4. You will need to use your openai API key. For now set it as an environment variable through your CLI. For some reason using .env file along with python-dotenv does not work.

>$end:$env:OPENAI_API_KEY="your-key-here"

Note: All the CLi commands stated assume you are Windows with PowerShell.

Installation

Use the following command in your CLI.

poetry install

Environment activation

Use the following command in your CLI.

poetry shell

Use instructions

In your CLI do the following.

askpdf --file "<path-to-pdf-file>" --question "Your question here"
# or use abbreviations
askpdf -f "<path-to-pdf-file>" -q "Your question here"
# if you ommit the --f parameter then the sample pdf file in this repository 
# will be used
askpdf -q "Your question here"
# get help 
askpdf -h
askpdf --help 

TODOs

• Fix issue with openai API KEY not being recognised when using python-dotenv along with a .env file.
• Use local large language model rather than sending requests to openai.