About:

The project PyAfipWs is a library and tools for government web-services.

I'm one of its creators & maintainer, being a python enthusiast since long time ago.

Also I'm a software developer, tech leader and now architect, and studied a Master in Free and Open Source Software (CV).

We formed a micro startup (Sistemas Ágiles: Agile Systems) to support the project and complement the community effort.

Summary:

Technology: 

• Multiplatform library & end user tools (windows executables, interoperability & legacy interfaces, spin-offs: SOAP WS / PDF projects
• Python 3 challenges, crypto/ssl issues
• Useful GitHub actions for windows automation and test against external webservices.

• Tips and lessons learned to sustain a non trivial user base and hundreds of customers.
• Building a 2k user group & GSoC mentorship / contributions

• Commercial dual-license funding / tech support / maintenance
• Alternatives to create sustainable open source projects for local or regional needs.

ChatGPT intro experiment:

As an alternative presentation, when asked the AI to give some information to the topic:

Q: What are the Argentina major government webservices (AFIP, ANMAT and ARBA)? Q: And what Python open source library can I use to interact with them?

The major government webservices in Argentina are: AFIP (Administración Federal de Ingresos Públicos) - This is the federal tax agency in Argentina responsible for collecting taxes and ensuring compliance with tax laws. ANMAT (Administración Nacional de Medicamentos, Alimentos y Tecnología Médica) - This is the national regulatory agency in Argentina responsible for ensuring the safety, efficacy, and quality of drugs, medical devices, and food products. ARBA (Agencia de Recaudación de la Provincia de Buenos Aires) - This is the tax agency for the province of Buenos Aires in Argentina, responsible for collecting taxes and ensuring compliance with tax laws in the province. To interact with these webservices, you can use the Python Requests library to make HTTP requests and retrieve data from their APIs. Additionally, you may need to use Python libraries specific to each webservice to handle authentication and data parsing. For AFIP, you can use the PyAFIP library, which provides a Python interface for interacting with AFIP's SOAP web services. 😀 For ANMAT, there is no specific Python library available, but you can use the Requests library to interact with their RESTful API. 😕 For ARBA, you can use the PyARBA library, which provides a Python interface for interacting with ARBA's SOAP web services. 😱

Warning with AI hallucinations! 

PyARBA project doesn’t seem to exist (as far I could search eg. PyPI nor Google nor GitHub). 

In another question the AI also mentioned WSSRv1, a government webservice that seems to not exist either.

In fact, our library (which its complete name is PyAfipWS) supports several webservices including  AFIP, ANMAT and ARBA.

Probably it would be better to divide this project per gov agency PyARBA / PyANMAT etc. but that could be harder to maintain.

Most of them are SOAP (Simple Object Access Protocol), a mixture of HTTP + XML.

It is not really simple, except ARBA that uses something like a REST API with CSV request and XML response (so ChatGPT mixed a bit that or is referencing other webservices).

It seems the AI is translating some concepts wrongly (method names?).

Naming convention debate aside… Spanish is used in official webservice methods documentation, and some words doesn’t have a direct English translation.

but after that there are some de

The code snippet is not that bad, the first line written (from pyafipws.wsfev1 import WSFEv1) looked promising as it mention our library and correct webservice class, but after that there are some problems. It confuses the request and response, uses the version 0 fields instead version 1, misses some tax structures as we have different VAT tax percentages, and it doesn’t point out the major result (CAE: authorization code).

Maybe ChatGPT is mixing other libraries/examples?

The correct code could be found in the project examples directory: factura_electronica.py (and there are also programs and code fragments for other programming languages and webservices).

A short fragment to illustrate and compare:

import datetime from pyafipws.wsaa import WSAA from pyafipws.wsfev1 import WSFEv1

# electronic invoice webservice setup for testing: wsfev1 = WSFEv1() wsfev1.Token = wsaa.Token wsfev1.Sign = wsaa.Sign wsfev1.Cuit = "20267565393" ok = wsfev1.Conectar(wsdl="https://wswhomo.afip.gov.ar/wsfev1/service.asmx?WSDL") # get last invoice number: cbte_nro = int(wsfev1.CompUltimoAutorizado(1, 3)) or 0 # create the electronic invoice (header fields, simple case): wsfev1.CrearFactura( concepto=1, tipo_doc=80, nro_doc="30500010912", tipo_cbte=1, punto_vta=3, cbt_desde=cbte_nro+1, cbt_hasta=cbte_nro+1, imp_total=122.00, imp_tot_conc=0.00, imp_neto=100.00, imp_iva=21.00, imp_trib=1.00, imp_op_ex=0.00, fecha_cbte=datetime.datetime.now().strftime("%Y%m%d"), moneda_id="PES", moneda_ctz=1.000, ) # add subtotals per each VAT tax percentage (21% case): wsfev1.AgregarIva(iva_id=5, base_imp=100.00, importe=21.00) # add other taxes: wsfev1.AgregarTributo( tributo_id=99, desc = 'Impuesto Municipal', base_imp=None, alic=None, importe=1.00, ) # authorize the invoice: wsfev1.CAESolicitar() # Electronic Authorization Code (result): print("CAE", wsfev1.CAE)

Anyway it is very interesting as it give a overall help on this topic (if you know the details).

Probably we should use the Stack Overflow question/response style and provide more examples, so the information corpus about these topics increases and AI responses get more accurate.

Quick Intro (real):

After the ChatGPT experiment, that was interesting but probably there is no enought information for more correct responses (as documentation is in Spanish and very limited).

Let's do a traditional quick introduction to these topics.

AFIP is the country's main tax agency, like the "I.R.S." in the US (Internal revenue service)..

Note that every invoice in Argentina must be sent to the government to get approval (including total amount, taxes plus item description, quantities and unit prices in some cases), or use a special printer with government software (fiscal controller).

There are special regulations to use invoices for tax returns (industrial fiscal bonus), foreign trade commerce (exports), tourism or using credit invoices as checking-account to pay other invoices.

There is a tax-payers registry to validate customers' VAT numbers and status. 

And received invoices should be verified using a webservices to validate the amounts for tax porpouses.

Also the primary sector is important in Argentina, so sales and shipments of cereal crops, flour, cattle, meat or milk production and derivatives should be informed to  Afip too, using several Agriculture or Livestock webservices.

Customs and foreign exchange are also complicated, for example in the past needing authorization to buy or sell US dollars (via web service for companies/banks or online government website for taxpayers).

Other related web services supported by the library are: 

ARBA: Buenos Aires province tax agency

• Freight transport Shipments code (COT), also for Córdoba & Santa Fe provinces.
• Gross income tax withholdings (IIBB).

National Traceability System web services: to track movements of

• Medicines and medical equipment (ANMAT TrazaMed/ProdMed)
• Controlled substances (SEDRONAR/RENPRE TrazaRenpre)
• National Health and Food Quality (SENASA: TrazaVet/TrazaFito)