Extract structured data from real-world PDF invoices and export it to a formatted Excel report — automatically, using Python.
This project is a production-style example of how to process invoices you did not generate yourself:
- accounting automation
- accounts payable pipelines
- ERP data entry replacement
- batch invoice processing
- financial reporting workflows
Most PDF extraction tutorials work on clean, predictable files. This example works on a real invoice from an external source — a PDF the script has never seen before.
The extraction is dynamic: instead of hardcoding coordinates, the script locates each section of the document by finding known anchor words and computing bounding boxes at runtime.
Step 1 — Input: a real invoice PDF
Step 2 — Debug: bounding boxes visualised during development
Each region (header, details, line items, totals) is detected and cropped dynamically.
Step 3 — Output: a formatted Excel report
| Field | Source |
|---|---|
| Vendor | Document header |
| Invoice number | Document header |
| Invoice date | Document header |
| Client | Bill To section |
| PO Number | Details section |
| Due Date | Payment section |
| Line items | Table (Item, Quantity, Rate, Amount) |
| Subtotal / VAT / Total | Summary section |
git clone https://github.com/hasff/python-invoice-pdf-data-extractor.git
cd python-invoice-pdf-data-extractor
python -m venv venv
source venv/bin/activate # Windows: venv\Scripts\activate
pip install -r requirements.txt
python extract_invoice.pyThe script will read from:
input/Invoice_sample.pdf
And generate:
output/example_output.xlsx
Rather than relying on table detection or AI-based extraction, this project uses a coordinate-based region extraction strategy:
- Anchor words (e.g.
BILL TO,ITEM,Subtotal) are located by scanning the word list - Bounding boxes are computed dynamically from those anchor positions
- Each region is cropped and parsed independently using
pdfplumber - Words within each region are grouped into lines and cells based on spatial proximity
This approach is predictable, debuggable, and does not require an internet connection or API key.
I help companies automate document processing pipelines:
- invoice and receipt extraction
- batch processing of large PDF archives
- integration with databases, APIs and ERP systems
- OCR for scanned documents
- export to Excel, CSV or JSON
📩 Contact: hugoferro.business(at)gmail.com
🌐 Courses and professional tools: https://hasff.github.io/site/
This repository is based on techniques covered in my full course:
Python PDF Handling: From Beginner to Winner (PyMuPDF)
The course covers text and image extraction, page manipulation, OCR from scans, high-speed batch processing, and how to debug complex extraction problems when documentation runs out.
The repository is fully usable on its own — the course provides the deeper understanding behind the decisions made here.