UnicodeDecodeError in mimetypes.guess_type on Windows

[vldmit]

BPO	10113
Nosy	@bitdancer
Superseder	bpo-9291: mimetypes initialization fails on Windows because of non-Latin characters in registry
Files	mime content types.reg: non ASCII content-type keys are at the bottom of the reg file

^{Note: these values reflect the state of the issue at the time it was migrated and might not reflect the current state.}

Show more details

GitHub fields:

assignee = None
closed_at = <Date 2010-10-15.12:52:38.685>
created_at = <Date 2010-10-15.10:57:52.435>
labels = ['type-bug', 'library', 'OS-windows']
title = 'UnicodeDecodeError in mimetypes.guess_type on Windows'
updated_at = <Date 2010-10-15.12:52:38.683>
user = 'https://bugs.python.org/vldmit'

bugs.python.org fields:

activity = <Date 2010-10-15.12:52:38.683>
actor = 'r.david.murray'
assignee = 'none'
closed = True
closed_date = <Date 2010-10-15.12:52:38.685>
closer = 'r.david.murray'
components = ['Library (Lib)', 'Windows']
creation = <Date 2010-10-15.10:57:52.435>
creator = 'vldmit'
dependencies = []
files = ['19242']
hgrepos = []
issue_num = 10113
keywords = []
message_count = 2.0
messages = ['118758', '118764']
nosy_count = 2.0
nosy_names = ['r.david.murray', 'vldmit']
pr_nums = []
priority = 'normal'
resolution = 'duplicate'
stage = 'resolved'
status = 'closed'
superseder = '9291'
type = 'behavior'
url = 'https://bugs.python.org/issue10113'
versions = ['Python 2.7']

[vldmit]

Windows 7, Python 2.7

Some windows applications (QuickTime) add content-types to Windows registry with non-ascii names. mimetypes in unaware of that and fails with UnicodeDecodeError:

>>> mimetypes.guess_type('test.js')
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "c:\Python27\lib\mimetypes.py", line 294, in guess_type
    init()
  File "c:\Python27\lib\mimetypes.py", line 355, in init
    db.read_windows_registry()
  File "c:\Python27\lib\mimetypes.py", line 260, in read_windows_registry
    for ctype in enum_types(mimedb):
  File "c:\Python27\lib\mimetypes.py", line 250, in enum_types
    ctype = ctype.encode(default_encoding) # omit in 3.x!
UnicodeDecodeError: 'ascii' codec can't decode byte 0xe0 in position 0: ordinal not in range(128)

Example registry leaf is attached to previous message.

I believe the correct behavior would be either to wrap UnicodeDecodeError exception and skip those content-typer or use .decode() method for registry keys and get encoding using locale.getdefaultlocale()

[bitdancer]

This is a duplicate of bpo-9291.