Internationalises valid phone numbers given for a particular country, to E.164

phone_number_internationalisation.py

import re

# Dial-in prefixes
international_prefixes = {
    'FR': "33", # France
    'AU': "61", # Australia
    'NZ': "64", # New Zealand
    'UK': "44", # United Kingdon
    'US': "1", # United States
    'CA': "1", # Canada
    'CN': "86", # China
    'JP': "81", # Japan
    'PH': "63", # Philippines
    'SG': "65", # Singapore
    'ES': "34", # Spain
    'TH': "66", # Thailand
    'TR': "90", # Turkey
    'AE': "971", # UAE
}

# Dial-out prefixes
international_call_prefix_itu = "^(00).*$"

international_call_prefix_regexs = {
    'FR': international_call_prefix_itu,
    'AU': "^(001[14589]).*$",
    'NZ': international_call_prefix_itu,
    'UK': international_call_prefix_itu,
    'US': "^(0[1]{1,2}).*$",
    'CA': "^(0[1]{1,2}).*$",
    'CN': international_call_prefix_itu,
    'JP': "^(001).*$",
    'PH': international_call_prefix_itu,
    'SG': "^(^|)(000|001|002|008|011|013|018|019|020|021).*$",
    'ES': international_call_prefix_itu,
    'TH': "^(001).*$",
    'TR': international_call_prefix_itu,
    'AE': international_call_prefix_itu,
}

''' Accepts a domestic or international phone number and the country in which it is valid,
    and generalises that number to the international standard E.164 "+" format, compatible
    with various SMS gateways.
    
    Works in some countries. I only had to deal with a few in this instance, but this approach
    should be easily extendible.
'''
def internationalise_number(number, country):
	# Are we dealing with an international number, relative to the country?
	icp = re.compile(international_call_prefix_regexs[country])
	icp_code = icp.match(number)
	is_international = (icp_code != None)

	# Remove ICP if present.
	if is_international:
		icp_code = icp_code.group(1)
		number = number[len(icp_code):]
		
	# Remove trunk prefix if we're looking at a domestic number.
	else:
		if (country != "US" and country != "CA") and number[0] == '0':
			number = number[1:]
		elif (country == "US"  or country == "CA") and number[0] == '1':
			number = number[1:]
	
	# Remove separators
	number = number.replace('-','').replace(' ', '')
	
	# Add country code if necessary
	if number[0] == '+':
		return number
	if is_international:
		return "+" + number
	else:
		return "+" + international_prefixes[country] + number