def normalize_str(value):
import re
is_bytes = isinstance(value, bytes)
regex, sep = r'(\S+)\s*', ' '
if is_bytes:
regex, sep = bytes(regex), bytes(sep)
regex = re.compile(regex)
matches = regex.findall(value)
names = (m.capitalize() if len(m) >= 3 else m.lower() for m in matches)
return sep.join(names)
def safe_encode(u, encoding=None):
'''Similar to unicode `encode` method returning bytes.
Encodes `u` using the given `encoding`, or determining one from the system.
Returning type is always `bytes`; but in python 2.x is also `str`.
.. versionadded:: 1.1.3
'''
# TODO: This is not nice for Python 3, bytes is not valid string any more
# See :func:`json.encoder.py_encode_basestring_ascii` of Python 2.x
from xoutil.eight import string_types, text_type
if isinstance(u, bytes):
return u
else:
encoding = force_encoding(encoding)
try:
try:
if isinstance(u, string_types):
# In Python 2.x bytes does not allows an encoding argument.
return bytes(u)
else:
return text_type(u).encode(encoding, 'replace')
except:
return text_type(u).encode(encoding, 'replace')
except LookupError:
return safe_encode(u)