pypostalwin

pypostalwin is the Un-Official Python wrapper to libpostal only for Windows, a fast statistical parser/normalizer for street addresses anywhere in the world.

About libpostal

libpostal is a C library for parsing/normalizing street addresses around the world using statistical NLP and open data. The goal of this project is to understand location-based strings in every language, everywhere.

Installation

1. Build the libpostal in Windows

• Before using the Python wrapper, you need to build the libpostal C library as a bundle which can be accessed by the python package. (still under development)

(OR)

• Download and Install MSYS2
• You can use the libpostal prebuilt zipped file Download here
• Unpack the zip to C:\Workbench\libpostal\
• If you don't have Workbench folder in C Drive, then create one.
• Copy the zip inside the Workbench and unzip using 7zip

2. Install the python wrapper

Install using pip. Offical Python Package Here!!

pip install pypostalwin

(OR)

Clone this Repository. Run this from the root directory to install

python setup.py install

Usage

Import Package

import pypostalwin

1. Initialize Address Parser Object

parser = pypostalwin.AddressParser()
parsed_address = parser.parse_address("The White House 1600 Pennsylvania Avenue NW, Washington, DC 20500, USA")
print(parsed_address)

output

[
  {'house': 'the white house'}, 
  {'house_number': '1600'},
  {'road': 'pennsylvania avenue nw'}, 
  {'city': 'washington'}, 
  {'state': 'dc'}, 
  {'postcode': '20500'}, 
  {'country': 'usa'}
]

Note: In a single runtime, the first-time parser.runParser() will take a few seconds to run since it is loading the models from libpostal in the backend process. Once it is loaded, the recurrent runs will be faster as usual. You need to use the same object instance to get the results faster.

for eg:

parser = pypostalwin.AddressParser()
parsed_address1 = parser.parse_address("The White House 1600") #only first time will take few seconds to load
parsed_address2 = parser.parse_address("Washington, DC 20500, USA") #will be faster as usual
parsed_address3 = parser.parse_address(" 20500, USA") #will be faster as usual
parsed_address4 = parser.parse_address("Pennsylvania Avenue NW, Washington,") #will be faster as usual

2. Expand the Address

expanded_address = parser.expand_address("District Science Cntr, Kokkirakulam Rd, Tirunelveli, TamilNadu 627009")
print(expanded_address)

output

['district science center kokkirakulam road tirunelveli tamilnadu 627009',
 'district science connector kokkirakulam road tirunelveli tamilnadu 627009']

3. Terminate Address Parser Object

parser.terminate_parser()