InvoiceNet VS tabula

You can also possibly look at invoice extraction tools such as https://github.com/naiveHobo/InvoiceNet. They solve a similar issue and are researched fairly well, since there is a big market for that.

Also, I would suggest you to use this codebase: https://github.com/naiveHobo/InvoiceNet

You can check out this repository. It contains an implementation of some recent research in deep learning for information extraction on invoices. https://github.com/naiveHobo/InvoiceNet

Another option would be JRuby. I routinely use an application called Tabula, which is built using JRuby and compiles to a Jar file. This, of course, requires Java on the target machine, but you can ship the Jar file and it will work. It's often easier to rely on a working Java environment than it is a working Ruby environment. Especially on Windows.

Idk if it has been suggested yet, tabulapdf is a self hosted solution to extract tables from PDF

For table extraction from pdfs, have a look at Tabula and Camelot, two open-source projects. They work well with clean tables, both the Tabula Python binding and Camelot allow you to export directly as a pandas dataframe. Otherwise AWS Textract API is very efficient at extracting tables from pdfs, regardless of how clean/messy they are.

No idea if it will work for you, but there is a git project that seems to do what you want https://github.com/tabulapdf/tabula

While trying to find a specific project I recalled, I encountered this list of projects which might be of interest: https://github.com/tstanislawek/awesome-document-understandi...

The project I had in mind was similar to this one but I can't remember the name currently: https://github.com/tabulapdf/tabula

However, if you're looking for a ML-based, invoice-specific project looks like the other comment to your reply might be more useful.