I built a small Python tool for a practical invoice-automation problem: extracting embedded text from digital PDF invoices and exporting candidate JSON/CSV for human review.
What it does
- Extracts embedded text from digital invoice PDFs
- Sends the extracted content to a chat-completions provider that you configure
- Exports candidate JSON/CSV with decimal arithmetic checks
- Keeps a human review step before anything is accepted
What it does not do
This is an honest beta, not a production accuracy claim:
- No OCR for scanned/image-only PDFs
- No automatic spreadsheet write
- No automatic approval or posting
- No accuracy guarantee or live commercial-provider benchmark yet
The intended workflow is simple: inspect the extracted text, review the candidate fields, correct exceptions, and then use the result in your own process. The tool is designed to keep the provider and credentials under the user's control.
Why build it?
Invoice extraction looks straightforward until layouts, decimal values, tax fields, currencies, and missing values create edge cases. A small, transparent pipeline is useful when you want to inspect the intermediate result instead of trusting a black-box import.
The beta includes synthetic test PDFs and offline tests. It is aimed at developers and automation engineers who want a starting point they can adapt, rather than a claim that every invoice format is solved.
The beta is available here for anyone who wants to try it: https://tariq.gumroad.com/l/invoice-pdf-csv-python-beta
Feedback on failure cases and invoice layouts would be more useful than generic praise.
Top comments (0)