python_regex_german_phone_number_extractor.py

import sys, re

if sys.version_info.major != 3:
	raise ValueError("You must use Python 3.")
if sys.version_info.minor < 4:
	raise ValueError("You must use at least Python 3.4")
if sys.version_info.minor < 6:
	print("Recommended Python Version is 3.6")

test_phone_numbers = [
	"+49174321324",
	"0164883423",
	"0049(0)16483311724",
	"+49-8332-3010",
	"00498513994",
	"0164 5554454",
	"0851 509",
	"0851 509-0",
	"(0851) 3394"
]

gold_phone_numbers = [
	"+49 174 321324",
	"+49 164 83311724",
	"+49 83323010",
	"+49 8513994",
	"+49 164 883423",
	"+49 164 5554454",
	"+49 851509",
	"+49 8515090",
	"+49 8513394"
]


def validate_phone(parsed_numbers):
	for parsed, test, gold in zip(parsed_numbers, test_phone_numbers, gold_phone_numbers):
		if parsed != gold:
			print("WRONG   ? \n\t Input : {}\n\t Gold  : {}\n\t Parsed: {}".format(test, gold, parsed))
		else:
			print("CORRECT ? \n\t Input : {}\n\t Gold  : {}\n\t Parsed: {}".format(test, gold, parsed))


parsed_list = []
cleaned_list = []
telephone_list = []
non_telephone_list = []
country_coded_list = []
non_country_coded_list = []
valid_list = []

for i in test_phone_numbers:
	a = i.replace("(0)", "")
	a = a.replace("-", "")
	a = a.replace(" ", "")
	a = a.replace("(", "")
	a = a.replace(")", "")
	cleaned_list.append(a)


for i in cleaned_list:
	try:
		country_check1 = re.match(r"^[+]49$", i[:3])
	except:
		country_check1 = None

	if not country_check1:
		try:
			country_check2 = re.match(r"^0+49", i[:5])
		except:
			country_check2 = None

		if country_check2:
			country_coded_list.append(i.replace("0049", "+49 "))
		else:
			non_country_coded_list.append(i)
	else:
		country_coded_list.append(i.replace("+49", "+49 "))

for i in country_coded_list:
	try:
		zero_check = re.match(r"^[0]$", i[4])
	except:
		zero_check = None

	if not zero_check:
		valid_list.append(i)

for i in valid_list:
	try:
		one_check = re.match(r"^[1]$", i[4])
	except:
		one_check = None

	if one_check:
		telephone_list.append(i)
	else:
		non_telephone_list.append(i)

for i in telephone_list:
	tel_code = i[4:7]
	total = "+49 " + tel_code + " " + i[7:]
	parsed_list.append(total)

for i in non_telephone_list:
	total = "+49 " + i[4:]
	parsed_list.append(total)

for i in non_country_coded_list:
	if len(i) < 10:
		a = "+49 " + i[1:]
		parsed_list.append(a)
	else:
		a = "+49 " + i[1:4] + " " + i[4:]
		parsed_list.append(a)


validate_phone(parsed_list)