Replace u() function with out_u for safer unicode conversion

powerline/lib/unicode.py

@@ -1,6 +1,9 @@
 # vim:fileencoding=utf-8:noet
 from __future__ import (unicode_literals, division, absolute_import, print_function)
 
+import sys
+import codecs
+
 from locale import getpreferredencoding
 
 
@@ -25,6 +28,45 @@ def u(s):
 		return unicode(s, 'utf-8')
 
 
+if sys.version_info < (3,):
+	def tointiter(s):
+		'''Convert a byte string to the sequence of integers
+		'''
+		return (ord(c) for c in s)
+else:
+	def tointiter(s):
+		'''Convert a byte string to the sequence of integers
+		'''
+		return iter(s)
+
+
+def powerline_decode_error(e):
+	if not isinstance(e, UnicodeDecodeError):
+		raise NotImplementedError
+	return (''.join((
+		'<{0:02X}>'.format(c)
+		for c in tointiter(e.object[e.start:e.end])
+	)), e.end)
+
+
+codecs.register_error('powerline_decode_error', powerline_decode_error)
+
+
+def out_u(s):
+	'''Return unicode string suitable for displaying
+
+	Unlike other functions assumes getpreferredencoding() first. Unlike u() does 
+	not throw exceptions for invalid unicode strings. Unlike safe_unicode() does 
+	throw an exception if object is not a string.
+	'''
+	if isinstance(s, unicode):
+		return s
+	elif isinstance(s, bytes):
+		return unicode(s, getpreferredencoding(), 'powerline_decode_error')
+	else:
+		raise TypeError('Expected unicode or bytes instance, got {0}'.format(repr(type(s))))
+
+
 def safe_unicode(s):
 	'''Return unicode instance without raising an exception.
 

powerline/segments/common.py

@@ -17,7 +17,7 @@ from powerline.lib.vcs import guess, tree_status
 from powerline.lib.threaded import ThreadedSegment, KwThreadedSegment
 from powerline.lib.monotonic import monotonic
 from powerline.lib.humanize_bytes import humanize_bytes
-from powerline.lib.unicode import u
+from powerline.lib.unicode import out_u
 from powerline.theme import requires_segment_info, requires_filesystem_watcher
 from powerline.segments import Segment, with_docstring
 
@@ -94,7 +94,7 @@ class CwdSegment(Segment):
 
 	def get_shortened_path(self, pl, segment_info, shorten_home=True, **kwargs):
 		try:
-			path = u(segment_info['getcwd']())
+			path = out_u(segment_info['getcwd']())
 		except OSError as e:
 			if e.errno == 2:
 				# user most probably deleted the directory
@@ -106,7 +106,7 @@ class CwdSegment(Segment):
 		if shorten_home:
 			home = segment_info['home']
 			if home:
-				home = u(home)
+				home = out_u(home)
 				if path.startswith(home):
 					path = '~' + path[len(home):]
 		return path
@@ -1123,12 +1123,12 @@ class NowPlayingSegment(Segment):
 			return
 		if not info:
 			return
-		album = u(info.get('xesam:album'))
+		album = out_u(info.get('xesam:album'))
-		title = u(info.get('xesam:title'))
+		title = out_u(info.get('xesam:title'))
 		artist = info.get('xesam:artist')
 		state = self._convert_state(status)
 		if artist:
-			artist = u(artist[0])
+			artist = out_u(artist[0])
 		return {
 			'state': state,
 			'album': album,