def filename_to_unicode(opsysstring):
"""Converts a str representing a filename from GLib to unicode.
:param str opsysstring: a string in the (GLib) encoding for filenames
:returns: the converted filename
:rtype: unicode
>>> from gi.repository import GLib
>>> filename_to_unicode('/ascii/only/path')
u'/ascii/only/path'
>>> filename_to_unicode(None) is None
True
This is just a more Pythonic wrapper around g_filename_to_utf8() for
now. If there are compatibility reasons to change it, fallbacks
involving sys.getfilesystemencoding exist.
"""
if opsysstring is None:
return None
# On Windows, they're always UTF-8 regardless.
if sys.platform == "win32":
return opsysstring.decode("utf-8")
# Other systems are dependent in opaque ways on the environment.
if not isinstance(opsysstring, str):
raise TypeError("Argument must be bytes")
ustring = GLib.filename_to_utf8(opsysstring, -1, 0, 0)
if ustring is None:
raise UnicodeDecodeError(
"GLib failed to convert %r to a UTF-8 string. "
"Consider setting G_FILENAME_ENCODING if your file system's "
"filename encoding scheme is not UTF-8." % (opsysstring, ))
return ustring.decode("utf-8")
def filename_to_unicode(opsysstring):
"""Converts a str representing a filename from GLib to unicode.
:param str opsysstring: a string in the (GLib) encoding for filenames
:returns: the converted filename
:rtype: unicode
>>> from gi.repository import GLib
>>> filename_to_unicode('/ascii/only/path')
u'/ascii/only/path'
>>> filename_to_unicode(None) is None
True
This is just a more Pythonic wrapper around g_filename_to_utf8() for
now. If there are compatibility reasons to change it, fallbacks
involving sys.getfilesystemencoding exist.
"""
if opsysstring is None:
return None
# On Windows, they're always UTF-8 regardless.
if sys.platform == "win32":
return opsysstring.decode("utf-8")
# Other systems are dependent in opaque ways on the environment.
if not isinstance(opsysstring, str):
raise TypeError("Argument must be bytes")
ustring = GLib.filename_to_utf8(opsysstring, -1, 0, 0)
if ustring is None:
raise UnicodeDecodeError(
"GLib failed to convert %r to a UTF-8 string. "
"Consider setting G_FILENAME_ENCODING if your file system's "
"filename encoding scheme is not UTF-8."
% (opsysstring,)
)
return ustring.decode("utf-8")
def filename_to_unicode(opsysstring):
"""Converts a str representing a filename from GLib to unicode.
:param bytes opsysstring: a string in the (GLib) encoding for filenames
:returns: the converted filename
:rtype: str
>>> filename_to_unicode(b'/ascii/only/path') == u'/ascii/only/path'
True
>>> filename_to_unicode(None) is None
True
This is just a more Pythonic wrapper around g_filename_to_utf8() for
now. If there are compatibility reasons to change it, fallbacks
involving sys.getfilesystemencoding exist.
"""
if opsysstring is None:
return None
# Let's assume that if the string is already unicode under Python 3,
# then it's already correct.
if PY3 and isinstance(opsysstring, str):
return opsysstring
# On Windows, they're always UTF-8 regardless.
if sys.platform == "win32":
return opsysstring.decode("utf-8")
# Other systems are dependent in opaque ways on the environment.
if not isinstance(opsysstring, bytes):
raise TypeError("Argument must be bytes")
# This function's annotation seems to vary quite a bit.
# See https://github.com/mypaint/mypaint/issues/634
try:
ustring, _, _ = GLib.filename_to_utf8(opsysstring, -1)
except TypeError:
ustring = GLib.filename_to_utf8(opsysstring, -1, 0, 0)
if ustring is None:
raise UnicodeDecodeError(
"GLib failed to convert %r to a UTF-8 string. "
"Consider setting G_FILENAME_ENCODING if your file system's "
"filename encoding scheme is not UTF-8." % (opsysstring, ))
if PY2:
ustring = ustring.decode("utf-8")
return ustring
def filename_to_unicode(opsysstring):
"""Converts a str representing a filename from GLib to unicode.
:param bytes opsysstring: a string in the (GLib) encoding for filenames
:returns: the converted filename
:rtype: unicode/str
>>> filename_to_unicode(b'/ascii/only/path') == u'/ascii/only/path'
True
>>> filename_to_unicode(None) is None
True
This is just a more Pythonic wrapper around g_filename_to_utf8() for
now, which works around a ton of weird bugs and corner cases with
the typelib annotations for it. It is intended for cleaning up the
output of certain GLib functions.
Currently, if you're using Python 3 and the input is already unicode
then this function assumes that GLib+GI have already done the work,
and that the unicode string was correct. You get the same string
back.
For Python 2, this accepts only "bytes" string input. If we find a
corner case where GLib functions return degenerate unicode, we can
adapt it for that case (those funcs need their own wrappers though).
"""
if opsysstring is None:
return None
# Let's assume that if the string is already unicode under Python 3,
# then it's already correct.
if PY3 and isinstance(opsysstring, unicode):
return opsysstring
# On Windows, they're always UTF-8 regardless.
# That's what the docs say.
if sys.platform == "win32":
return opsysstring.decode("utf-8")
# Other systems are dependent in opaque ways on the environment.
if not isinstance(opsysstring, bytes):
raise TypeError("Argument must be bytes")
opsysstring_degenerate_unicode = opsysstring.decode("latin_1")
# This function's annotation seems to vary quite a bit.
# See https://github.com/mypaint/mypaint/issues/634
ustring = None
# The sensible, modern case! Byte strings in, unicode strings
# out hopefully, and the C func's arguments are correctly
# [out]-annotated. It works like this as of...
#
# - Python 2.7.14 OR Python 3.6.4
# - gobject-introspection 1.54.1
# - glib2 2.54.3
# - Debian buster/sid amd64 OR MSYS2 MINGW64 on Windows 7 64-bit.
if ustring is None:
for s in [opsysstring, opsysstring_degenerate_unicode]:
try:
ustring, _bytes_read, _bytes_written \
= GLib.filename_to_utf8(s, -1)
break
except TypeError:
pass
# Try an older, bad typelib's form.
# This is the case for Ubuntu 14.04 LTS "trusty" (which is ancient,
# but that's what our current Travis CI solution uses). For the
# record, this weirdness is applicable to the following combination:
#
# - Python 2.7.6 OR Python 3.4.3
# - gobject-introspection 1.40.0
# - glib2 2.40.2
# - Ubuntu 14.04.5 LTS amd64.
#
# Of note: the Py3 wrappings are weird in Trusty. Other GLib funcs
# return bytes, but GLib.filename_to_utf8() expects those degenerate
# unicode strings. Byte strings will not do. Unusual tastes.
if ustring is None:
for s in [opsysstring, opsysstring_degenerate_unicode]:
try:
ustring = GLib.filename_to_utf8(s, -1, 0, 0)
break
except TypeError:
pass
# Congratulations! You found a new bug.
if ustring is None:
raise UnicodeDecodeError(
"New or unknown bugs in g_filename_to_utf8()'s typelib. "
"Failed to convert %r. Please tell the developers about this."
% (opsysstring,)
)
# Python2's wrappers tended to do this.
# I suspect it's reasonable to convert for all, now that we're
# reasonably sure that the data would be utf-8.
if isinstance(ustring, bytes):
ustring = ustring.decode("utf-8")
return ustring
from gi.repository import GLib
+logger = logging.getLogger(__name__)
+
## File path getter functions
@@ -35,7 +36,6 @@ def filename_to_unicode(opsysstring):
:returns: the converted filename
:rtype: unicode
- >>> from gi.repository import GLib
>>> filename_to_unicode('/ascii/only/path')
u'/ascii/only/path'
>>> filename_to_unicode(None) is None
@@ -54,7 +54,12 @@ def filename_to_unicode(opsysstring):
# Other systems are dependent in opaque ways on the environment.
if not isinstance(opsysstring, str):
raise TypeError("Argument must be bytes")
- ustring = GLib.filename_to_utf8(opsysstring, -1, 0, 0)
+ # This function's annotation seems to vary quite a bit.
+ # See https://github.com/mypaint/mypaint/issues/634
+ try:
+ ustring, _, _ = GLib.filename_to_utf8(opsysstring, -1)
+ except TypeError:
+ ustring = GLib.filename_to_utf8(opsysstring, -1, 0, 0)
if ustring is None:
raise UnicodeDecodeError(
"GLib failed to convert %r to a UTF-8 string. "
def filename_to_unicode(opsysstring):
"""Converts a str representing a filename from GLib to unicode.
:param bytes opsysstring: a string in the (GLib) encoding for filenames
:returns: the converted filename
:rtype: unicode/str
>>> filename_to_unicode(b'/ascii/only/path') == u'/ascii/only/path'
True
>>> filename_to_unicode(None) is None
True
This is just a more Pythonic wrapper around g_filename_to_utf8() for
now, which works around a ton of weird bugs and corner cases with
the typelib annotations for it. It is intended for cleaning up the
output of certain GLib functions.
Currently, if you're using Python 3 and the input is already unicode
then this function assumes that GLib+GI have already done the work,
and that the unicode string was correct. You get the same string
back.
For Python 2, this accepts only "bytes" string input. If we find a
corner case where GLib functions return degenerate unicode, we can
adapt it for that case (those funcs need their own wrappers though).
"""
if opsysstring is None:
return None
# Let's assume that if the string is already unicode under Python 3,
# then it's already correct.
if PY3 and isinstance(opsysstring, unicode):
return opsysstring
# On Windows, they're always UTF-8 regardless.
# That's what the docs say.
if sys.platform == "win32":
return opsysstring.decode("utf-8")
# Other systems are dependent in opaque ways on the environment.
if not isinstance(opsysstring, bytes):
raise TypeError("Argument must be bytes")
opsysstring_degenerate_unicode = opsysstring.decode("latin_1")
# This function's annotation seems to vary quite a bit.
# See https://github.com/mypaint/mypaint/issues/634
ustring = None
# The sensible, modern case! Byte strings in, unicode strings
# out hopefully, and the C func's arguments are correctly
# [out]-annotated. It works like this as of...
#
# - Python 2.7.14 OR Python 3.6.4
# - gobject-introspection 1.54.1
# - glib2 2.54.3
# - Debian buster/sid amd64 OR MSYS2 MINGW64 on Windows 7 64-bit.
if ustring is None:
for s in [opsysstring, opsysstring_degenerate_unicode]:
try:
ustring, _bytes_read, _bytes_written \
= GLib.filename_to_utf8(s, -1)
break
except TypeError:
pass
# Try an older, bad typelib's form.
# This is the case for Ubuntu 14.04 LTS "trusty" (which is ancient,
# but that's what our current Travis CI solution uses). For the
# record, this weirdness is applicable to the following combination:
#
# - Python 2.7.6 OR Python 3.4.3
# - gobject-introspection 1.40.0
# - glib2 2.40.2
# - Ubuntu 14.04.5 LTS amd64.
#
# Of note: the Py3 wrappings are weird in Trusty. Other GLib funcs
# return bytes, but GLib.filename_to_utf8() expects those degenerate
# unicode strings. Byte strings will not do. Unusual tastes.
if ustring is None:
for s in [opsysstring, opsysstring_degenerate_unicode]:
try:
ustring = GLib.filename_to_utf8(s, -1, 0, 0)
break
except TypeError:
pass
# Congratulations! You found a new bug.
if ustring is None:
raise UnicodeDecodeError(
"New or unknown bugs in g_filename_to_utf8()'s typelib. "
"Failed to convert %r. Please tell the developers about this." %
(opsysstring, ))
# Python2's wrappers tended to do this.
# I suspect it's reasonable to convert for all, now that we're
# reasonably sure that the data would be utf-8.
if isinstance(ustring, bytes):
ustring = ustring.decode("utf-8")
return ustring
