Street Address Parser

This is a command line utility using Python standard libraries to parse street addresses in the country schemes DE, ES, FR, and US.

Prerequisites

• A working Python installation, minimum version 3.6.
• Add your Python command line interpreter to the PATH

Usage

Example for parsing an address (simple address profile) - open a command line prompt, change to the project directory and execute:

$ python street_address_parser.py "foobar 23"
{"foobar", "23"}

Help

From the project directory, you can get command line help by running this:

$ python street_address_parser.py --help

Test Execution

Open a command line prompt, change to the project directory and run the following command:

$ python -m unittest discover
....
----------------------------------------------------------------------
Ran 4 tests in 0.001s

OK

Implementation Notes

Assumptions

• Addresses don't contain funny unicode characters. I only considered the test cases provided in the coding challenge.
• Functional correctness is primary for this task, performance is secondary. If performance would be primary (e.g., for quickly parsing large numbers of addresses), I would reconsider the use of regular expressions and possibly also the use of Python.

Home-made vs off-the-shelf solutions

I first tried several open-source libraries (Python) for address parsing, but they all failed one or more of the addresses. The most promising one was https://github.com/datamade/usaddress, but it failed parse ES addresses correctly. Apart from that, the library looks really good and they even offer a API for bulk processing (https://parserator.datamade.us/api-docs). With these insights, I opted for a home-made solution. However, if I were to require a fully fledged parser for all sorts of address formats as utility for professional use, I would try not to build this myself and rather focus my efforts on the core business functionality.