Offline Document Classifier & Metadata Extractor

A documented Python script bundle that sorts a document into one of 10 types (invoice, credit note, purchase order, receipt, quote, contract, delivery note, bank statement, HR letter, meeting minutes) and extracts date/ID metadata — running entirely on your own machine. No cloud call, no API key, no data ever leaves your computer.

$39 one-time

Buy the Classifier Bundle

What's included

document_classifier.py — the classifier and metadata extractor, Python standard library only, no third-party packages required to run it
10 sample fixtures, one per supported document type, so you can verify correct behaviour before pointing it at your own documents
Setup instructions: how to shape your own PDF-to-text output into the expected input format and invoke the classifier
Scope/limitations note and a single-machine license

Tested, not guessed

10/10 correct on a synthetic set including two deliberately confusable pairs (credit note vs. invoice, quote vs. purchase order). Zero wrong-label events across 33 additional stress cases with OCR-style character noise and non-English text — below its confidence threshold it returns UNCLASSIFIED rather than a guessed label.

What this is not

Not ML-trained. Rule-based weighted-keyword scoring against a fixed taxonomy — it generalises only as far as the keyword patterns cover.
Not a PDF extraction tool. You need your own PDF/OCR-to-text step; this bundle classifies already-extracted text.
Not English-agnostic. English-language documents are the primary tested coverage; other languages typically (and correctly) return UNCLASSIFIED rather than a false match.

How you get it

After payment, email hello@tsaielectro.com with your order confirmation and we'll send the bundle the same day.