Release 1.3.1

- Fixed Command-T finder labels: now `CommandT::` and  `Finder::` parts are
  removed because all finders were moved to `CommandT::Finder` module in place
  of just `CommandT`.
- Fixed dbus players support: it is now OK not to have album title.
- Characters that form the surrogate pair that represents some character above
  U+10000 in UCS-2 Python builds are no longer considered non-printable.
- Added support for calculating lengths of the characters above U+10000 in UCS-2
  Python builds.
- Made linter support characters above U+10000 if they are represented as
  a `\uXXXX\uXXXX` surrogate pair in JSON.
- Made linter validate that all characters in some strings are printable.
- Made `powerline-daemon` work in FreeBSD: only use abstract socket namespace on
  linux systems.
- Fixed `string` and `safe_unicode` functions in Python-3.

client/powerline.c

@@ -42,12 +42,12 @@ void do_write(int sd, const char *raw, size_t len) {
 	}
 }
 
-#ifdef __APPLE__
-# define ADDRESS_TEMPLATE "/tmp/powerline-ipc-%d"
-# define A
-#else
+#ifdef __linux__
 # define ADDRESS_TEMPLATE "powerline-ipc-%d"
 # define A +1
+#else
+# define ADDRESS_TEMPLATE "/tmp/powerline-ipc-%d"
+# define A
 #endif
 
 #define ADDRESS_SIZE sizeof(ADDRESS_TEMPLATE) + (sizeof(uid_t) * 4)

client/powerline.py

@@ -26,9 +26,7 @@
 	print('Must provide at least one argument.', file=sys.stderr)
 	raise SystemExit(1)
 
-platform = sys.platform.lower()
-use_filesystem = 'darwin' in platform
-del platform
+use_filesystem = not sys.platform.lower().startswith('linux')
 
 if sys.argv[1] == '--socket':
 	address = sys.argv[2]

client/powerline.sh

@@ -1,20 +1,39 @@
 #!/bin/sh
 
-test "${OSTYPE#darwin}" = "${OSTYPE}" && darwin=n || darwin=y
+use_filesystem=1
+darwin=
+if test -n "$OSTYPE" ; then
+	# OSTYPE variable is a shell feature. supported by bash and zsh, but not 
+	# dash, busybox or (m)ksh.
+	if test "${OSTYPE#linux}" '!=' "${OSTYPE}" ; then
+		use_filesystem=
+	elif test "${OSTYPE#darwin}" ; then
+		darwin=1
+	fi
+elif which uname >/dev/null ; then
+	if uname -o | grep -iqF linux ; then
+		use_filesystem=
+	elif uname -o | grep -iqF darwin ; then
+		darwin=1
+	fi
+fi
 
 if test "$1" = "--socket" ; then
 	shift
 	ADDRESS="$1"
 	shift
 else
 	ADDRESS="powerline-ipc-${UID:-`id -u`}"
-	test "$darwin" = y && ADDRESS="/tmp/$ADDRESS"
+	test -n "$use_filesystem" && ADDRESS="/tmp/$ADDRESS"
 fi
 
-if test "$darwin" = y; then
+if test -n "$darwin" ; then
 	ENV=genv
 else
 	ENV=env
+fi
+
+if test -z "$use_filesystem" ; then
 	ADDRESS="abstract-client:$ADDRESS"
 fi
 

docs/source/installation.rst

@@ -7,6 +7,15 @@ Generic requirements
 
 * Python 2.6 or later, 3.2 or later, PyPy 2.0 or later. It is the only 
   non-optional requirement.
+
+  .. warning:
+     It is highly advised to use UCS-4 version of Python because UCS-2 version 
+     uses significantly slower text processing (length determination and 
+     non-printable character replacement) functions due to the need of 
+     supporting unicode characters above U+FFFF which are represented as 
+     surrogate pairs. This price will be paid even if configuration has no such 
+     characters.
+
 * C compiler. Required to build powerline client on linux. If it is not present 
   then powerline will fall back to shell script or python client.
 * ``socat`` program. Required for shell variant of client which runs a bit 
@@ -43,7 +52,7 @@ powerline with ``pip``:
 
 .. code-block:: sh
 
-    pip install -e --user {path_to_powerline}
+    pip install --user --editable={path_to_powerline}
 
 , but note that in this case ``pip`` will not install ``powerline`` executable 
 and you will have to do something like

powerline/bindings/config.py

@@ -145,6 +145,8 @@ def get_highlighting(group):
 	left_dividers = powerline.renderer.theme.dividers['left']
 	set_tmux_environment('_POWERLINE_LEFT_HARD_DIVIDER', left_dividers['hard'])
 	set_tmux_environment('_POWERLINE_LEFT_SOFT_DIVIDER', left_dividers['soft'])
+	set_tmux_environment('_POWERLINE_LEFT_HARD_DIVIDER_SPACES', (
+		' ' * powerline.renderer.strwidth(left_dividers['hard'])))
 
 
 def get_main_config(args):

powerline/bindings/tmux/powerline-base.conf

@@ -4,7 +4,7 @@ set -g status-interval 2
 set -g status-left-length 20
 set -g status-right '#(eval $POWERLINE_COMMAND tmux right -R pane_id=`tmux display -p "#D"`)'
 set -g status-right-length 150
-set -g window-status-format "#[$_POWERLINE_WINDOW_COLOR]  #I #[$_POWERLINE_WINDOW_DIVIDER_COLOR]$_POWERLINE_LEFT_SOFT_DIVIDER#[default]#W "
+set -g window-status-format "#[$_POWERLINE_WINDOW_COLOR]$_POWERLINE_LEFT_HARD_DIVIDER_SPACES#I #[$_POWERLINE_WINDOW_DIVIDER_COLOR]$_POWERLINE_LEFT_SOFT_DIVIDER#[default]#W $_POWERLINE_LEFT_HARD_DIVIDER_SPACES"
 set -g window-status-current-format "#[$_POWERLINE_WINDOW_CURRENT_HARD_DIVIDER_COLOR]$_POWERLINE_LEFT_HARD_DIVIDER#[$_POWERLINE_WINDOW_CURRENT_COLOR]#I $_POWERLINE_LEFT_SOFT_DIVIDER#[$_POWERLINE_WINDOW_NAME_COLOR]#W #[$_POWERLINE_WINDOW_CURRENT_HARD_DIVIDER_NEXT_COLOR]$_POWERLINE_LEFT_HARD_DIVIDER"
 
 # Legacy status-left definition to be overwritten for tmux Versions 1.8+

powerline/config_files/colorschemes/tmux/default.json

@@ -4,8 +4,8 @@
 		"window_status":        {"fg": "gray70",     "bg": "gray0",    "attr": []},
 		"activity_status":      {"fg": "yellow",     "bg": "gray0",    "attr": []},
 		"bell_status":          {"fg": "red",        "bg": "gray0",    "attr": []},
-		"window":               {"fg": "gray6",      "bg": "gray11",   "attr": []},
-		"window:divider":       {"fg": "gray4",      "bg": "gray11",   "attr": []},
+		"window":               {"fg": "gray6",      "bg": "gray0",    "attr": []},
+		"window:divider":       {"fg": "gray4",      "bg": "gray0",    "attr": []},
 		"window:current":       {"fg": "mediumcyan", "bg": "darkblue", "attr": []},
 		"window_name":          {"fg": "white",      "bg": "darkblue", "attr": ["bold"]},
 		"session":              {"fg": "black",      "bg": "gray90",   "attr": ["bold"]},

powerline/lib/debug.py

@@ -1,4 +1,3 @@
-#!/usr/bin/env python
 # vim:fileencoding=utf-8:noet
 from __future__ import (unicode_literals, division, absolute_import, print_function)
 

powerline/lib/unicode.py

@@ -4,6 +4,8 @@
 import sys
 import codecs
 
+from unicodedata import east_asian_width, combining
+
 from powerline.lib.encoding import get_preferred_output_encoding
 
 
@@ -19,6 +21,17 @@
 	unichr = chr
 
 
+if sys.maxunicode < 0x10FFFF:
+	_unichr = unichr
+
+	def unichr(ch):
+		if ch <= sys.maxunicode:
+			return _unichr(ch)
+		else:
+			ch -= 0x10000
+			return _unichr((ch >> 10) + 0xD800) + _unichr((ch & ((1 << 10) - 1)) + 0xDC00)
+
+
 def u(s):
 	'''Return unicode instance assuming UTF-8 encoded string.
 	'''
@@ -56,6 +69,27 @@ def powerline_decode_error(e):
 
 
 def register_strwidth_error(strwidth):
+	'''Create new encode errors handling method similar to ``replace``
+
+	Like ``replace`` this method uses question marks in place of the characters 
+	that cannot be represented in the requested encoding. Unlike ``replace`` the 
+	amount of question marks is identical to the amount of display cells 
+	offending character occupies. Thus encoding ``…`` (U+2026, HORIZONTAL 
+	ELLIPSIS) to ``latin1`` will emit one question mark, but encoding ``Ａ`` 
+	(U+FF21, FULLWIDTH LATIN CAPITAL LETTER A) will emit two question marks.
+
+	Since width of some characters depends on the terminal settings and 
+	powerline knows how to respect them a single error handling method cannot be 
+	used. Instead of it the generator function is used which takes ``strwidth`` 
+	function (function that knows how to compute string width respecting all 
+	needed settings) and emits new error handling method name.
+
+	:param function strwidth:
+		Function that computs string width measured in display cells the string 
+		occupies when displayed.
+
+	:return: New error handling method name.
+	'''
 	global last_swe_idx
 	last_swe_idx += 1
 
@@ -98,7 +132,10 @@ def safe_unicode(s):
 	'''
 	try:
 		try:
-			return unicode(s)
+			if type(s) is bytes:
+				return unicode(s, 'ascii')
+			else:
+				return unicode(s)
 		except UnicodeDecodeError:
 			try:
 				return unicode(s, 'utf-8')
@@ -111,8 +148,7 @@ def safe_unicode(s):
 
 
 class FailedUnicode(unicode):
-	'''Builtin ``unicode`` (``str`` in python 3) subclass indicating fatal 
-	error.
+	'''Builtin ``unicode`` subclass indicating fatal error
 
 	If your code for some reason wants to determine whether `.render()` method 
 	failed it should check returned string for being a FailedUnicode instance. 
@@ -123,8 +159,125 @@ class FailedUnicode(unicode):
 	pass
 
 
-def string(s):
-	if type(s) is not str:
-		return s.encode('utf-8')
-	else:
-		return s
+if sys.version_info < (3,):
+	def string(s):
+		if type(s) is not str:
+			return s.encode('utf-8')
+		else:
+			return s
+else:
+	def string(s):
+		if type(s) is not str:
+			return s.decode('utf-8')
+		else:
+			return s
+
+
+string.__doc__ = (
+	'''Transform ``unicode`` or ``bytes`` object into ``str`` object
+
+	On Python-2 this encodes ``unicode`` to ``bytes`` (which is ``str``) using 
+	UTF-8 encoding; on Python-3 this decodes ``bytes`` to ``unicode`` (which is 
+	``str``) using UTF-8 encoding.
+
+	Useful for functions that expect an ``str`` object in both unicode versions, 
+	not caring about the semantic differences between them in Python-2 and 
+	Python-3.
+	'''
+)
+
+
+def surrogate_pair_to_character(high, low):
+	'''Transform a pair of surrogate codepoints to one codepoint
+	'''
+	return 0x10000 + ((high - 0xD800) << 10) + (low - 0xDC00)
+
+
+_strwidth_documentation = (
+	'''Compute string width in display cells
+
+	{0}
+
+	:param dict width_data:
+		Dictionary which maps east_asian_width property values to strings 
+		lengths. It is expected to contain the following keys and values (from 
+		`East Asian Width annex <http://www.unicode.org/reports/tr11/>`_):
+
+		===  ======  ===========================================================
+		Key  Value   Description
+		===  ======  ===========================================================
+		F    2       Fullwidth: all characters that are defined as Fullwidth in 
+		             the Unicode Standard [Unicode] by having a compatibility 
+		             decomposition of type <wide> to characters elsewhere in the 
+		             Unicode Standard that are implicitly narrow but unmarked.
+		H    1       Halfwidth: all characters that are explicitly defined as 
+		             Halfwidth in the Unicode Standard by having a compatibility 
+		             decomposition of type <narrow> to characters elsewhere in 
+		             the Unicode Standard that are implicitly wide but unmarked, 
+		             plus U+20A9 ₩ WON SIGN.
+		W    2       Wide: all other characters that are always wide. These 
+		             characters occur only in the context of East Asian 
+		             typography where they are wide characters (such as the 
+		             Unified Han Ideographs or Squared Katakana Symbols). This 
+		             category includes characters that have explicit halfwidth 
+		             counterparts.
+		Na   1       Narrow: characters that are always narrow and have explicit 
+		             fullwidth or wide counterparts. These characters are 
+		             implicitly narrow in East Asian typography and legacy 
+		             character sets because they have explicit fullwidth or wide 
+		             counterparts. All of ASCII is an example of East Asian 
+		             Narrow characters.
+		A    1 or 2  Ambigious: characters that may sometimes be wide and 
+		             sometimes narrow. Ambiguous characters require additional 
+		             information not contained in the character code to further 
+		             resolve their width. This information is usually defined in 
+		             terminal setting that should in turn respect glyphs widths 
+		             in used fonts. Also see :ref:`ambiwidth configuration 
+		             option <config-common-ambiwidth>`.
+		N    1       Neutral characters: character that does not occur in legacy 
+		             East Asian character sets.
+		===  ======  ===========================================================
+
+	:param unicode string:
+		String whose width will be calculated.
+
+	:return: unsigned integer.''')
+
+
+def strwidth_ucs_4(width_data, string):
+	return sum(((
+		(
+			0
+		) if combining(symbol) else (
+			width_data[east_asian_width(symbol)]
+		)
+	) for symbol in string))
+
+
+strwidth_ucs_4.__doc__ = _strwidth_documentation.format(
+	'''This version of function expects that characters above 0xFFFF are 
+	represented using one symbol. This is only the case in UCS-4 Python builds.
+
+	.. note:
+		Even in UCS-4 Python builds it is possible to represent characters above 
+		0xFFFF using surrogate pairs. Characters represented this way are not 
+		supported.''')
+
+
+def strwidth_ucs_2(width_data, string):
+	return sum(((
+		(
+			width_data[east_asian_width(string[i - 1] + symbol)]
+		) if 0xDC00 <= ord(symbol) <= 0xDFFF else (
+			0
+		) if combining(symbol) or 0xD800 <= ord(symbol) <= 0xDBFF else (
+			width_data[east_asian_width(symbol)]
+		)
+	) for i, symbol in enumerate(string)))
+
+
+strwidth_ucs_2.__doc__ = _strwidth_documentation.format(
+	'''This version of function expects that characters above 0xFFFF are 
+	represented using two symbols forming a surrogate pair, which is the only 
+	option in UCS-2 Python builds. It still works correctly in UCS-4 Python 
+	builds, but is slower then its UCS-4 counterpart.''')

powerline/lint/__init__.py

@@ -41,7 +41,7 @@ def load_json_config(config_file_path, load=load, open_file=open_file):
 function_name_re = '^(\w+\.)*[a-zA-Z_]\w*$'
 
 
-divider_spec = Spec().type(unicode).len(
+divider_spec = Spec().printable().len(
 	'le', 3, (lambda value: 'Divider {0!r} is too large!'.format(value))).copy
 ext_theme_spec = Spec().type(unicode).func(lambda *args: check_config('themes', *args)).copy
 top_theme_spec = Spec().type(unicode).func(check_top_theme).copy
@@ -211,12 +211,12 @@ def load_json_config(config_file_path, load=load, open_file=open_file):
 	display=Spec().type(bool).optional(),
 	module=segment_module_spec(),
 	priority=Spec().type(int, float, type(None)).optional(),
-	after=Spec().type(unicode).optional(),
-	before=Spec().type(unicode).optional(),
+	after=Spec().printable().optional(),
+	before=Spec().printable().optional(),
 	width=Spec().either(Spec().unsigned(), Spec().cmp('eq', 'auto')).optional(),
 	align=Spec().oneof(set('lr')).optional(),
 	args=args_spec().func(lambda *args, **kwargs: check_args(get_one_segment_function, *args, **kwargs)),
-	contents=Spec().type(unicode).optional(),
+	contents=Spec().printable().optional(),
 	highlight_group=Spec().list(
 		highlight_group_spec().re(
 			'^(?:(?!:divider$).)+$',
@@ -243,11 +243,11 @@ def load_json_config(config_file_path, load=load, open_file=open_file):
 	soft=divider_spec(),
 ).copy
 segment_data_value_spec = Spec(
-	after=Spec().type(unicode).optional(),
-	before=Spec().type(unicode).optional(),
+	after=Spec().printable().optional(),
+	before=Spec().printable().optional(),
 	display=Spec().type(bool).optional(),
 	args=args_spec().func(lambda *args, **kwargs: check_args(get_all_possible_functions, *args, **kwargs)),
-	contents=Spec().type(unicode).optional(),
+	contents=Spec().printable().optional(),
 ).copy
 dividers_spec = Spec(
 	left=divside_spec(),