Fix sphinx-quickstart raises UnicodeEncodeError if "Project version" includes non-ASCII characters. Closes #1188

CHANGES

@@ -26,6 +26,8 @@ Bugs fixed
   Marcin Wojdyr.
 * PR#145: In parallel builds, sphinx drops second document file to write.
   Thanks to tychoish.
+* #1188: sphinx-quickstart raises UnicodeEncodeError if "Project version"
+  includes non-ASCII characters.
 
 
 Release 1.2 (beta1 released Mar 31, 2013)

sphinx/quickstart.py

@@ -866,9 +866,24 @@ def ok(x):
 def do_prompt(d, key, text, default=None, validator=nonempty):
     while True:
         if default:
-            prompt = purple(PROMPT_PREFIX + '%s [%s]: ' % (text, default))
+            prompt = PROMPT_PREFIX + '%s [%s]: ' % (text, default)
         else:
-            prompt = purple(PROMPT_PREFIX + text + ': ')
+            prompt = PROMPT_PREFIX + text + ': '
+        if sys.version_info < (3, 0):
+            # for Python 2.x, try to get a Unicode string out of it
+            if prompt.encode('ascii', 'replace').decode('ascii', 'replace') \
+                    != prompt:
+                if TERM_ENCODING:
+                    prompt = prompt.encode(TERM_ENCODING)
+                else:
+                    print turquoise('* Note: non-ASCII default value provided '
+                                    'and terminal encoding unknown -- assuming '
+                                    'UTF-8 or Latin-1.')
+                    try:
+                        prompt = prompt.encode('utf-8')
+                    except UnicodeEncodeError:
+                        prompt = prompt.encode('latin1')
+        prompt = purple(prompt)
         x = term_input(prompt).strip()
         if default and not x:
             x = default

tests/test_quickstart.py

@@ -28,6 +28,12 @@ def mock_raw_input(answers, needanswer=False):
             raise AssertionError('answer for %r missing and no default '
                                  'present' % prompt)
         called.add(prompt)
+        if sys.version_info < (3, 0):
+            prompt = str(prompt)  # Python2.x raw_input emulation
+            # `raw_input` encode `prompt` by default encoding to print.
+        else:
+            prompt = unicode(prompt)  # Python3.x input emulation
+            # `input` decode prompt by default encoding before print.
         for question in answers:
             if prompt.startswith(qs.PROMPT_PREFIX + question):
                 return answers[question]
@@ -95,6 +101,16 @@ def test_do_prompt():
     raises(AssertionError, qs.do_prompt, d, 'k6', 'Q6', validator=qs.boolean)
 
 
+def test_do_prompt_with_multibyte():
+    d = {}
+    answers = {
+        'Q1': ur'\u30c9\u30a4\u30c4',
+    }
+    qs.term_input = mock_raw_input(answers)
+    qs.do_prompt(d, 'k1', 'Q1', default=ur'\u65e5\u672c')
+    assert d['k1'] == ur'\u30c9\u30a4\u30c4'
+
+
 @with_tempdir
 def test_quickstart_defaults(tempdir):
     answers = {