def remove_free_digits(text):
"""
Removes free standing digits (digits not part of a word)
:param text: String to format (f.ex. "Only 90s kids will get this 1337 m8")
:return: The formatted String (f.ex. "Only 90s kids will get this m8")
"""
return RegexFilters.replace_free_digits(text, " ")
def remove_repeated_whitespace(text):
"""
Removes repeated whitespace
:param text: String to format (f.ex. "A string with maany spaces ")
:return: The formatted String (f.ex. "A string with many spaces ")
"""
return RegexFilters.replace_whitespace(text, " ")
def remove_non_alphabetic_text(text):
"""
Removes non alphabetic characters
:param text: String to format (f.ex "Hey, m8!")
:return: The formatted String (f.ex. "Hey m")
"""
return RegexFilters.replace_non_alphabetic_text(text, "")
def remove_inner_word_characters(text):
"""
Removes characters which are often part of a word (mostly apostrophes)
:param text: String to format (f.ex. "Here's a sentence!")
:return: The formatted String (f.ex. "Heres a sentence!")
"""
return RegexFilters.replace_inner_word_characters(text, "")
def remove_non_alphanumerical_text(text):
"""
Removes non-alphanumerical characters
:param text: String to format (f.ex "It's very nice!")
:return: The formatted String (f.ex "It s very nice ")
"""
return RegexFilters.replace_non_alphanumerical_text(text, " ")
def remove_non_syntactical_text(text):
"""
Removes all non-alphabetic or basic punctuation characters (!?,. )
:param text: String to format (f.ex. "This is' a #crazy tæst")
:return: The formatted String (f.ex. "This is a crazy tst")
"""
return RegexFilters.replace_non_syntactical_text(text, " ")
def remove_repeating_characters(text):
"""
Replaces repeating characters in String
:param text: String to format (f.ex. "Today is a greeeeeeaaaaaaat dayy!")
:return: The formatted String (f.ex. "Today is a great day!")
"""
return RegexFilters.replace_repeating_characters(text, "$1")
def placeholder_rt_tag(text):
return RegexFilters.replace_rt_tag(text, Filters.RTTAG_PLACEHOLDER)
def remove_non_syntactical_text_plus(text):
return RegexFilters.replace_non_syntactical_text_plus(text, " ")
def placeholder_url(text):
return RegexFilters.replace_url(text, Filters.URL_PLACEHOLDER)
def remove_emoticons(text):
return RegexFilters.replace_emoticons(text, "")
def remove_rt_tag(text):
return RegexFilters.replace_rt_tag(text, "")
def protect_hashtag(text):
return RegexFilters.replace_hashtag(text, " ||#$1|| ")
def hashtag_to_word(text):
return RegexFilters.replace_hashtag(text, "$1")
def placeholder_hashtag(text):
return RegexFilters.replace_hashtag(text, Filters.HASHTAG_PLACEHOLDER)
def placeholder_username(text):
return RegexFilters.replace_username(text,
Filters.USERNAME_PLACEHOLDER)
def remove_username(text):
return RegexFilters.replace_username(text, "")
def remove_url(text):
return RegexFilters.replace_url(text, "")
def parse_emoticons(text):
return RegexFilters.replace_emoticons(text, " ||$1|| ")