afipcaeqrdecode 0.0.9

Package to decode and extract invoice metadata from an AFIP CAE qr code link

PyPI python package

AFIP invoice pdf qr CAE extract and decode

This is a python package that uses pdf2image to convert the first page of your AFIP invoice with an AFIP CAE QR code to an image, and then run qreader on it in order to locate and decode the AFIP CAE QR code in order to extract relevant invoice metadata like:

• Invoice date
• CUIT of invoice creator
• AFIP electronic invoice point of sale (Punto de venta)
• Invoice number
• Amount
• Currency
• CUIT of inovoice recipient

And other less important properties.

Why convert to image instead of extracting

In its inception this library used PyMuPDF in order to extract all images inside the invoices and then run qreader on them, however, we came upon some invoices in which the qr code image failed to extract.

Why qreader instead of pyzbar

In its inception this library used just pyzbar, however we came upon some QR codes which did not decode succesfully using just pyzbar.

qreader depends on pyzbar, but uses a pre-trained AI model to detect and segment QR codes, using information extracted by this AI model, it applies different image preprocessing techniques that heavily increase the decoding rate by pyzbar

Example Usage and notes about metadata

Using the included sample files for demonstration (and ran from repository root using included sample file):

from afipcaeqrdecode import convert_pdf_to_image_and_detect_and_decode_qrs

invoice_metadata = convert_pdf_to_image_and_detect_and_decode_qrs('./tests/sample_files/2000005044986390.pdf')

Here, invoice metadata will evaluate to:

{
    "ver":1,
    "fecha":"2023-02-10", #I've found this field to be missing in some decodes
    "cuit":30710145764,
    "ptoVta":4,
    "tipoCmp":1,
    "nroCmp":25399,
    "importe":2460,
    "moneda":"PES",
    "ctz":1,
    "tipoDocRec":80,
    "nroDocRec":30717336905,
    "tipoCodAut":"E",
    "codAut":73064176949471
}

#The actual output will not be pretty printed, it will be stripped of all whitespace and formatting characters

System Dependencies and their installation

This package depends on qreader, which in turn depends on pyzbar, which in turn depends on the system library zbar ZBar

Check your OS documentation on what package to install to get ZBar working with pyzbar.

On Linux (Ubuntu 22.04):

sudo apt-get install libzbar0

On Mac OS X:

brew install zbar

Installation using pip

After installing system dependencies, you can install using the PyPI python package

pip install afipcaeqrdecode

First run notice

On first run qreader will download the weights to run its QR detector AI model, then it will resume program operation automatically.

WARNING

This is an experimental package, USE IN PRODUCTION AT YOUR OWN RISK.

It is barely even tested, i'm sharing it so I can actually import it as a PyPI package in another project that consumes it.

Credits

All the other library authors this package depends on. Facundo Mainere for helping with JWT decode.

Author: Emiliano Mesquita.

License

GNU GPLv3.