lint: Do not allow non-printable characters in configuration

powerline/lint/__init__.py

@@ -41,7 +41,7 @@ def generate_json_config_loader(lhadproblem):
 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 @@ segment_spec = Spec(
 	display=Spec().type(bool).optional(),
 	module=segment_module_spec(),
 	priority=Spec().type(int, float, type(None)).optional(),
-	after=Spec().type(unicode).optional(),
+	after=Spec().printable().optional(),
-	before=Spec().type(unicode).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 @@ divside_spec = Spec(
 	soft=divider_spec(),
 ).copy
 segment_data_value_spec = Spec(
-	after=Spec().type(unicode).optional(),
+	after=Spec().printable().optional(),
-	before=Spec().type(unicode).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(),

powerline/lint/markedjson/error.py

@@ -7,7 +7,7 @@ import re
 from powerline.lib.unicode import unichr
 
 
-NON_PRINTABLE = re.compile(
+NON_PRINTABLE_STR = (
 	'[^'
 	# ASCII control characters: 0x00-0x19
 	+ '\t\n'           # Tab, newline: allowed ASCII control characters
@@ -26,6 +26,7 @@ NON_PRINTABLE = re.compile(
 		''
 	))
 )
+NON_PRINTABLE_RE = re.compile(NON_PRINTABLE_STR)
 
 
 def repl(s):
@@ -33,7 +34,7 @@ def repl(s):
 
 
 def strtrans(s):
-	return NON_PRINTABLE.sub(repl, s.replace('\t', '>---'))
+	return NON_PRINTABLE_RE.sub(repl, s.replace('\t', '>---'))
 
 
 class Mark:
@@ -73,6 +74,13 @@ class Mark:
 			+ ' ' * (indent + len(head) + len(snippet[0])) + '^'
 		)
 
+	def advance_string(self, diff):
+		ret = self.copy()
+		# FIXME Currently does not work properly with escaped strings.
+		ret.column += diff
+		ret.pointer += diff
+		return ret
+
 	def __str__(self):
 		snippet = self.get_snippet()
 		where = ('  in "%s", line %d, column %d' % (

powerline/lint/markedjson/markedvalue.py

@@ -33,12 +33,7 @@ class MarkedUnicode(unicode):
 		pointdiff = 1
 		r = []
 		for s in part_result:
-			mark = self.mark.copy()
+			r.append(MarkedUnicode(s, self.mark.advance_string(pointdiff)))
-			# XXX Does not work properly with escaped strings, but this requires 
-			# saving much more information in mark.
-			mark.column += pointdiff
-			mark.pointer += pointdiff
-			r.append(MarkedUnicode(s, mark))
 			pointdiff += len(s)
 		return tuple(r)
 

powerline/lint/markedjson/reader.py

@@ -3,7 +3,7 @@ from __future__ import (unicode_literals, division, absolute_import, print_funct
 
 import codecs
 
-from powerline.lint.markedjson.error import MarkedError, Mark, NON_PRINTABLE
+from powerline.lint.markedjson.error import MarkedError, Mark, NON_PRINTABLE_RE
 from powerline.lib.unicode import unicode
 
 
@@ -84,7 +84,7 @@ class Reader(object):
 		return Mark(self.name, self.line, self.column, self.full_buffer, self.full_pointer)
 
 	def check_printable(self, data):
-		match = NON_PRINTABLE.search(data)
+		match = NON_PRINTABLE_RE.search(data)
 		if match:
 			self.update_pointer(match.start())
 			raise ReaderError(

powerline/lint/spec.py

@@ -7,10 +7,19 @@ import re
 from copy import copy
 
 from powerline.lib.unicode import unicode
-from powerline.lint.markedjson.error import echoerr, DelayedEchoErr
+from powerline.lint.markedjson.error import echoerr, DelayedEchoErr, NON_PRINTABLE_STR
 from powerline.lint.selfcheck import havemarks
 
 
+NON_PRINTABLE_RE = re.compile(
+	NON_PRINTABLE_STR.translate({
+		ord('\t'): None,
+		ord('\n'): None,
+		0x0085: None,
+	})
+)
+
+
 class Spec(object):
 	'''Class that describes some JSON value
 
@@ -342,6 +351,26 @@ class Spec(object):
 				return False, hadproblem
 		return True, hadproblem
 
+	def check_printable(self, value, context_mark, data, context, echoerr, _):
+		'''Check that given unicode string contains only printable characters
+		'''
+		hadproblem = False
+		for match in NON_PRINTABLE_RE.finditer(value):
+			hadproblem = True
+			echoerr(
+				context=self.cmsg.format(key=context.key),
+				context_mark=value.mark,
+				problem='found not printable character U+{0:04x} in a configuration string'.format(
+					ord(match.group(0))),
+				problem_mark=value.mark.advance_string(match.start() + 1)
+			)
+		return True, hadproblem
+
+	def printable(self, *args):
+		self.type(unicode)
+		self.checks.append(('check_printable', args))
+		return self
+
 	def type(self, *args):
 		'''Describe value that has one of the types given in arguments