We might be able to add support for territorial zones that are not countries or subdivisions per-se, like the European Union, or an

This requires #6.

Use case: implement territory matching for business rules.

Better yet: let the user creates its own zones. Think of arbitrary availability zones for datacenters, business regions, supranational European bodies, ...

We have been forced to avoid version 0.8.4 of Faker in this commit : 3c9cedf

But we should try to make postal-address works with the latest faker package.

Just discovered there is a whole industry around postal address normalization in the French market.

See:

• https://fr.wikipedia.org/wiki/Restructuration,_normalisation_et_validation_postale
• https://www.normad1.fr/en
• http://www.laposte.fr/entreprise/produits-et-services/sna-normalisation-des-adresses

There might be an opportunity here for the postal_address project to cover most, if not all, of these local specifications.

Package up all territory utilities into a proper Territory class.

The goal of this class will be to reconcile and normalize territory taxonomies (ISO countries, subdivisions and european legacy country codes).

Once #1 is done, we might eventually include support for NUTS subdivisions, if it make sense.

I would like to use this module in Italy

Postal codes requires their own parsing and normalization library. The idea is to render them under the {country_code}-{postal_code} scheme.

The main motivation behind this feature is support for territorial exceptions, as defined in https://github.com/kdeldycke/vat-rates/blob/master/vat_rates.csv

Where can i find documentation about this package API ?

https://postal-address.readthedocs.io/

0.3.0 release moved some code around and thus breaks components using the symbols where they were located resulting in this sample backtrace:

[...]
   from postal_address.address import supported_territory_codes
ImportError: cannot import name supported_territory_codes

This specific break occurs when upgrading from 0.2.1 to 0.3.0 and the CHANGES.rst did not make any mention of that.

There is no such thing as private symbols in python, except for well documented _ or __ class or module members so please write a little something about the stability or lack thereof of the module.

To help solve territory hierarchy, implement an utility to find the lowest common ancestor of several territory codes.

Address rendering is currently stupid. We need to implement proper localized address rendering.

For localization, we can reuse address metadata from Google's I18n Services Address Data Service. These data seems to be published under no particular license.

Hi @kdeldycke and @scaleway,

Do you think we could use this library to create a Django PostalAddressField?
Could you write some documentation, snippet or maybe a library for this?

By default, we rely on the ISO 3166-1 classification for country codes. This taxonomy is not up to date and some inaccuracies are currently hidden because we use common names instead of the official one (see commit 8d1753c ).

CLDR unicode data seems like a good candidate:

• https://www.drupal.org/node/1938892
• http://www.unicode.org/repos/cldr/trunk/common/main/en.xml

Territory code aliases implementation is quite a mess, and to make it easier to maintain, use a directed acyclic graph to store the subdivision / country relationship as well as aliases.

With such a structure, we can easily solve #5 and check for stupid cycles in unit tests.

I think I'll leverage the NetworkX package to do so.

Hi, the lib intro sounds great, but I couldn't find any doc examples on how to actually use it.

There's the API documentation at https://postal-address.readthedocs.io/en/develop/ , which is quite detailed and contains a myriad of classes and methods, but not helpful in figuring out what to do.

Is there a quick start example or something? How do I detect addresses in text using postal-address?

Apologies if I missed something obvious.

When validating an address using address.validate(), on error the method always raises a ValueError.
We can also use the address.valid() method, but we do not have the information about what field is invalid.

It would be great if you could expose a new property like address.errors that returns a dict with something like {'field': 'error'}.

And maybe also embed in the exception body this dict.

Non standard, addresses' country code are not normalized to their good values, even if defined as alias in territory.py:

>>> import postal_address
>>> postal_address.__version__
'0.7.1'
>>> from postal_address import Address
>>> Address(country_code='EL')
  File "(...)/postal_address-0.7.1-py2.7.egg/postal_address/address.py", line 152, in __repr__
  File "(...)/postal_address-0.7.1-py2.7.egg/postal_address/address.py", line 456, in country_name
  File "(...)/postal_address-0.7.1-py2.7.egg/postal_address/address.py", line 445, in country
  File "(...)/pycountry-1.10-py2.7.egg/pycountry/db.py", line 114, in get
    return self.indices[field][value]
KeyError: 'EL'

Do you have any particular reason to use the GPL ? We'd like to use your library in a LGPL projet, it would be great if you could change to something more liberal.

Country and/or state derived from subdivision have some edge-cases that need to be addressed.

See be12158 for unittests highlighting the case.