support multibyte filename handling.

https://bitbucket.org/birkenfeld/sphinx/issue/703

2 files changed, 26 insertions, 6 deletions

diff --git a/tests/path.py b/tests/path.py
index 8e9afeaa8..bd0552d70 100644
--- a/tests/path.py
+++ b/tests/path.py
@@ -16,16 +16,16 @@ from codecs import open
 FILESYSTEMENCODING = sys.getfilesystemencoding() or sys.getdefaultencoding()
 
 
-class path(str):
+class path(unicode):
     """
     Represents a path which behaves like a string.
     """
     if sys.version_info < (3, 0):
         def __new__(cls, s, encoding=FILESYSTEMENCODING, errors='strict'):
-            if isinstance(s, unicode):
-                s = s.encode(encoding, errors=errors)
-                return str.__new__(cls, s)
-            return str.__new__(cls, s)
+            if isinstance(s, str):
+                s = s.decode(encoding, errors)
+                return unicode.__new__(cls, s)
+            return unicode.__new__(cls, s)
 
     @property
     def parent(self):
@@ -193,4 +193,4 @@ class path(str):
     __div__ = __truediv__ = joinpath
 
     def __repr__(self):
-        return '%s(%s)' % (self.__class__.__name__, str.__repr__(self))
+        return '%s(%s)' % (self.__class__.__name__, unicode.__repr__(self))
diff --git a/tests/test_build.py b/tests/test_build.py
index 5f24f89ac..ae3f29211 100644
--- a/tests/test_build.py
+++ b/tests/test_build.py
@@ -10,6 +10,7 @@
 """
 
 from util import *
+from textwrap import dedent
 
 
 def teardown_module():
@@ -61,3 +62,22 @@ else:
 @with_app(buildername='singlehtml', cleanenv=True)
 def test_singlehtml(app):
     app.builder.build_all()
+
+@with_app(buildername='html', srcdir='(temp)')
+def test_multibyte_path(app):
+    srcdir = path(app.srcdir)
+    mb_name = u'\u65e5\u672c\u8a9e'
+    (srcdir / mb_name).makedirs()
+    (srcdir / mb_name / (mb_name + '.txt')).write_text(dedent("""
+        multi byte file name page
+        ==========================
+        """))
+
+    master_doc = srcdir / 'contents.txt'
+    master_doc.write_bytes((master_doc.text() + dedent("""
+            .. toctree::
+            
+               %(mb_name)s/%(mb_name)s
+            """ % locals())
+    ).encode('utf-8'))
+    app.builder.build_all()