string.decode() vs. unicode(string)

Question

myString = 'éíěřáé'

I need to decode this string to unicode. Is there any difference between folowing usages and between these two methods in general?

myString.decode(encoding='UTF-8', errors='ignore')

and

unicode(myString, encoding='UTF-8', errors='ignore')

Answer 1

The unicode constructor can take other types apart from strings:

>>> unicode(10)
u'10'

For the bytestring case, however, the two forms are mostly equivalent. Some encoding options are not valid for the unicode constructor as they do not result in unicode output, but are valid for the .decode method of bytestrings, such as 'hex':

>>> unicode('10', encoding='hex')
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
TypeError: decoder did not return an unicode object (type=str)

Answer 2

They're essentially the same, but with some minor performance shortcuts in either case; str.decode knows that its argument is a string, so it can shortcut type checking of its argument, while unicode.__new__ has shortcuts for some common encodings including UTF-8.

Both methods call into PyCodec_Decode in the general case.