Does python's print function handle unicode differently now than when Dive Into Python was written?

Question

I'm trying to work my way through some frustrating encoding issues by going back to basics. In Dive Into Python example 9.14 (here) we have this:

>>> s = u'La Pe\xf1a'
>>> print s
Traceback (innermost last): File "<interactive input>", line 1, in ?
UnicodeError: ASCII encoding error: ordinal not in range(128)
>>> print s.encode('latin-1')
La Peña

But on my machine, this happens:

>>> sys.getdefaultencoding()
'ascii'
>>> s = u'La Pe\xf1a'
>>> print s
La Peña

I don't understand why these are different. Anybody?

Answer 1

The default encoding for print doesn't depend on sys.getdefaultencoding(), but on sys.stdout.encoding. If you launch python with e.g. LANG=C or redirect a python script to a file, the encoding for stdout will be ANSI_X3.4-1968. On the other hand, if sys.stdout is a terminal, it will use the terminal's encoding.

To explain what sys.getdefaultencoding() does -- it's used when implicitly converting strings from/to unicode. In this example, str(u'La Pe\xf1a') with the default ASCII encoding would fail, but with modified default encoding it would encode the string to Latin-1. However setting the default encoding is a horrible idea, you should always use explicit encoding when you want to go from unicode to str.