Refactored get_view_description, moved appropriate tests to test_description.py

rest_framework/tests/test_description.py

@@ -3,8 +3,10 @@
 from __future__ import unicode_literals
 from django.test import TestCase
 from rest_framework.views import APIView
-from rest_framework.compat import apply_markdown
+from rest_framework.compat import apply_markdown, smart_text
 from rest_framework.utils.formatting import get_view_name, get_view_description
+from rest_framework.tests.views import (
+    ViewWithNonASCIICharactersInDocstring, UTF8_TEST_DOCSTRING)
 
 # We check that docstrings get nicely un-indented.
 DESCRIPTION = """an example docstring
@@ -83,11 +85,10 @@ class TestViewNamesAndDescriptions(TestCase):
         Unicode in docstrings should be respected.
         """
 
-        class MockView(APIView):
-            """Проверка"""
-            pass
-
-        self.assertEqual(get_view_description(MockView), "Проверка")
+        self.assertEqual(
+            get_view_description(ViewWithNonASCIICharactersInDocstring),
+            smart_text(UTF8_TEST_DOCSTRING)
+        )
 
     def test_view_description_can_be_empty(self):
         """

rest_framework/tests/test_utils.py

-# -*- coding: utf-8 -*-
-
-from django.test import TestCase
-from rest_framework.utils import formatting
-import sys
-
-
-class FormattingUnitTests(TestCase):
-    def setUp(self):
-        # test strings snatched from http://www.columbia.edu/~fdc/utf8/,
-        # http://winrus.com/utf8-jap.htm and memory
-        self.utf8_test_string = (
-            'zażółć gęślą jaźń'
-            'Sîne klâwen durh die wolken sint geslagen'
-            'Τη γλώσσα μου έδωσαν ελληνική'
-            'யாமறிந்த மொழிகளிலே தமிழ்மொழி'
-            'На берегу пустынных волн'
-            ' てすと'
-            'ｱｲｳｴｵｶｷｸｹｺｻｼｽｾｿﾀﾁﾂﾃ'
-        )
-        self.non_utf8_test_string = ('The quick brown fox jumps over the lazy '
-                                     'dog')
-
-    def test_for_ascii_support_in_remove_leading_indent(self):
-        if sys.version_info < (3, 0):
-            # only Python 2.x is affected, so we skip the test entirely
-            # if on Python 3.x
-            self.assertEqual(formatting._remove_leading_indent(
-                self.non_utf8_test_string), self.non_utf8_test_string)
-
-    def test_for_utf8_support_in_remove_leading_indent(self):
-        if sys.version_info < (3, 0):
-            # only Python 2.x is affected, so we skip the test entirely
-            # if on Python 3.x
-            self.assertEqual(formatting._remove_leading_indent(
-                self.utf8_test_string), self.utf8_test_string.decode('utf-8'))

rest_framework/tests/views.py

+# -*- coding: utf-8 -*-
+
+from rest_framework.views import APIView
+
+
+# test strings snatched from http://www.columbia.edu/~fdc/utf8/,
+# http://winrus.com/utf8-jap.htm and memory
+UTF8_TEST_DOCSTRING = (
+    'zażółć gęślą jaźń'
+    'Sîne klâwen durh die wolken sint geslagen'
+    'Τη γλώσσα μου έδωσαν ελληνική'
+    'யாமறிந்த மொழிகளிலே தமிழ்மொழி'
+    'На берегу пустынных волн'
+    'てすと'
+    'ｱｲｳｴｵｶｷｸｹｺｻｼｽｾｿﾀﾁﾂﾃ'
+)
+
+
+# Apparently there is an issue where docstrings of imported view classes
+# do not retain their encoding information even if a module has a proper
+# encoding declaration at the top of its source file. Therefore for tests
+# to catch unicode related errors, a mock view has to be declared in a separate
+# module.
+class ViewWithNonASCIICharactersInDocstring(APIView):
+    __doc__ = UTF8_TEST_DOCSTRING

rest_framework/utils/formatting.py

@@ -5,7 +5,7 @@ from __future__ import unicode_literals
 
 from django.utils.html import escape
 from django.utils.safestring import mark_safe
-from rest_framework.compat import apply_markdown
+from rest_framework.compat import apply_markdown, smart_text
 import re
 
 
@@ -24,11 +24,6 @@ def _remove_leading_indent(content):
     Remove leading indent from a block of text.
     Used when generating descriptions from docstrings.
     """
-    try:
-        content = content.decode('utf-8')
-    except (AttributeError, UnicodeEncodeError):
-        pass  # the string should keep the default 'ascii' encoding in
-              # Python 2.x or stay a unicode string in Python 3.x
     whitespace_counts = [len(line) - len(line.lstrip(' '))
                          for line in content.splitlines()[1:] if line.lstrip()]
 
@@ -68,7 +63,7 @@ def get_view_description(cls, html=False):
     Return a description for an `APIView` class or `@api_view` function.
     """
     description = cls.__doc__ or ''
-    description = _remove_leading_indent(description)
+    description = _remove_leading_indent(smart_text(description))
     if html:
         return markup_description(description)
     return description