def cleanup(input_string):
"""
Removes and strip separators from input_string (but keep ',;' characters)
:param input_string:
:type input_string:
:return:
:rtype:
"""
for char in clean_chars:
input_string = input_string.replace(char, ' ')
return re.sub(' +', ' ', strip(input_string))
def cleanup(input_string):
"""
Removes and strip separators from input_string (but keep ',;' characters)
:param input_string:
:type input_string:
:return:
:rtype:
"""
for char in clean_chars:
input_string = input_string.replace(char, ' ')
return re.sub(' +', ' ', strip(input_string))
def cleanup(input_string):
"""
Removes and strip separators from input_string (but keep ',;' characters)
It also keep separators for single characters (Mavels Agents of S.H.I.E.L.D.)
:param input_string:
:type input_string: str
:return:
:rtype:
"""
clean_string = input_string
for char in clean_chars:
clean_string = clean_string.replace(char, ' ')
# Restore input separator if they separate single characters.
# Useful for Mavels Agents of S.H.I.E.L.D.
# https://github.com/guessit-io/guessit/issues/278
indices = [i for i, letter in enumerate(clean_string) if letter in seps]
dots = set()
if indices:
clean_list = list(clean_string)
potential_indices = []
for i in indices:
if _potential_before(i, input_string) and _potential_after(
i, input_string):
potential_indices.append(i)
replace_indices = []
for potential_index in potential_indices:
if potential_index - 2 in potential_indices or potential_index + 2 in potential_indices:
replace_indices.append(potential_index)
if replace_indices:
for replace_index in replace_indices:
dots.add(input_string[replace_index])
clean_list[replace_index] = input_string[replace_index]
clean_string = ''.join(clean_list)
clean_string = strip(clean_string,
''.join([c for c in seps if c not in dots]))
clean_string = re.sub(' +', ' ', clean_string)
return clean_string
def cleanup(input_string):
"""
Removes and strip separators from input_string (but keep ',;' characters)
It also keep separators for single characters (Mavels Agents of S.H.I.E.L.D.)
:param input_string:
:type input_string: str
:return:
:rtype:
"""
clean_string = input_string
for char in clean_chars:
clean_string = clean_string.replace(char, ' ')
# Restore input separator if they separate single characters.
# Useful for Mavels Agents of S.H.I.E.L.D.
# https://github.com/guessit-io/guessit/issues/278
indices = [i for i, letter in enumerate(clean_string) if letter in seps]
dots = set()
if indices:
clean_list = list(clean_string)
potential_indices = []
for i in indices:
if _potential_before(i, input_string) and _potential_after(i, input_string):
potential_indices.append(i)
replace_indices = []
for potential_index in potential_indices:
if potential_index - 2 in potential_indices or potential_index + 2 in potential_indices:
replace_indices.append(potential_index)
if replace_indices:
for replace_index in replace_indices:
dots.add(input_string[replace_index])
clean_list[replace_index] = input_string[replace_index]
clean_string = ''.join(clean_list)
clean_string = strip(clean_string, ''.join([c for c in seps if c not in dots]))
clean_string = re.sub(' +', ' ', clean_string)
return clean_string
