panni
36 lines
971 B
Python
36 lines
971 B
Python
# coding=utf-8
|
|
|
|
import re
|
|
|
|
|
|
def sanitize(string, ignore_characters=None, default_characters={'-', ':', '(', ')', '.'}):
|
|
"""Sanitize a string to strip special characters.
|
|
|
|
:param str string: the string to sanitize.
|
|
:param set ignore_characters: characters to ignore.
|
|
:return: the sanitized string.
|
|
:rtype: str
|
|
|
|
"""
|
|
# only deal with strings
|
|
if string is None:
|
|
return
|
|
|
|
ignore_characters = ignore_characters or set()
|
|
|
|
# replace some characters with one space
|
|
characters = default_characters - ignore_characters
|
|
if characters:
|
|
string = re.sub(r'[%s]' % re.escape(''.join(characters)), ' ', string)
|
|
|
|
# remove some characters
|
|
characters = {'\''} - ignore_characters
|
|
if characters:
|
|
string = re.sub(r'[%s]' % re.escape(''.join(characters)), '', string)
|
|
|
|
# replace multiple spaces with one
|
|
string = re.sub(r'\s+', ' ', string)
|
|
|
|
# strip and lower case
|
|
return string.strip().lower()
|