Ticket #12849: colorize_unicode.diff

django/utils/termcolors.py

@@ -2 +2 @@
 termcolors.py
 """
 
+import locale
+from django.utils.encoding import smart_str
+
+
 color_names = ('black', 'red', 'green', 'yellow', 'blue', 'magenta', 'cyan', 'white')
 foreground = dict([(color_names[x], '3%s' % x) for x in range(8)])
 background = dict([(color_names[x], '4%s' % x) for x in range(8)])
@@ -9 +13 @@
 RESET = '0'
 opt_dict = {'bold': '1', 'underscore': '4', 'blink': '5', 'reverse': '7', 'conceal': '8'}
 
-def colorize(text='', opts=(), **kwargs):
+def colorize(text=u'', opts=(), encoding=None, **kwargs):
     """
     Returns your text, enclosed in ANSI graphics codes.
 
-    Depends on the keyword arguments 'fg' and 'bg', and the contents of
-    the opts tuple/list.
+    The returned string depends on the keyword arguments 'fg' and 'bg', and
+    the contents of the opts tuple/list. The text will be encoded using the
+    current locale's preferred encoding unless otherwise specified in the
+    'encoding' keyword argument.
 
     Returns the RESET code if no parameters are given.
 
@@ -38 +44 @@
         print colorize('and so should this')
         print 'this should not be red'
     """
-    text = str(text)
+    if encoding is None:
+        encoding = locale.getpreferredencoding()
+    text = smart_str(text, encoding=encoding)
     code_list = []
     if text == '' and len(opts) == 1 and opts[0] == 'reset':
         return '\x1b[%sm' % RESET

tests/regressiontests/utils/termcolors.py

@@ -1 +1 @@
 from unittest import TestCase
 
-from django.utils.termcolors import parse_color_setting, PALETTES, DEFAULT_PALETTE, LIGHT_PALETTE, DARK_PALETTE, NOCOLOR_PALETTE
+from django.utils.termcolors import colorize, parse_color_setting, PALETTES, DEFAULT_PALETTE, LIGHT_PALETTE, DARK_PALETTE, NOCOLOR_PALETTE
 
 class TermColorTests(TestCase):
 
+    def test_encoding(self):
+        """Verify that unicode strings are accepted and propertly encoded."""
+        test_data = (
+            ('\xc2\xa3', 'utf-8', '\xc2\xa3'),
+            ('\xc2\xa3', 'latin-1', '\xa3'),
+            (u'\u00a3', 'utf-8', '\xc2\xa3'),
+            (u'\u00a3', 'latin-1', '\xa3'),
+            )
+        for input, encoding, encoded in test_data:
+            output = colorize(input, fg='red', encoding=encoding)
+            self.assertEquals(output, '\x1b[31m%s\x1b[0m' % encoded)
+
     def test_empty_string(self):
         self.assertEquals(parse_color_setting(''), PALETTES[DEFAULT_PALETTE])