Run 2to3 on config files which contain Python 2.x unicode literals.

1 files changed, 62 insertions, 5 deletions

diff --git a/sphinx/util/pycompat.py b/sphinx/util/pycompat.py
index 229b54b4..2ec71e72 100644
--- a/sphinx/util/pycompat.py
+++ b/sphinx/util/pycompat.py
@@ -12,6 +12,7 @@
 import sys
 import codecs
 import encodings
+import re
 
 try:
     from types import ClassType
@@ -20,11 +21,6 @@ except ImportError:
     # Python 3
     class_types = (type,)
 
-try:
-    base_exception = BaseException
-except NameError:
-    base_exception = Exception
-
 
 # the ubiquitous "bytes" helper function
 if sys.version_info >= (3, 0):
@@ -34,6 +30,66 @@ else:
     b = str
 
 
+encoding_re = re.compile(b(r'coding[=:]\s*([-\w.]+)'))
+unicode_literal_re = re.compile(ur"""
+(?:
+    "(?:[^"\]]*(?:\\.[^"\\]*)*)"|
+    '(?:[^'\]]*(?:\\.[^'\\]*)*)'
+)
+""", re.VERBOSE)
+
+
+try:
+    from lib2to3.refactor import RefactoringTool, get_fixers_from_package
+except ImportError:
+    _run_2to3 = None
+    def should_run_2to3(filepath):
+        return False
+else:
+    def should_run_2to3(filepath):
+        # th default source code encoding for python 2.x
+        encoding = 'ascii'
+        # only the first match of the encoding cookie counts
+        encoding_set = False
+        f = open(filepath, 'rb')
+        try:
+            for i, line in enumerate(f):
+                if line.startswith(b('#')):
+                    if i == 0 and b('python3') in line:
+                        return False
+                    if not encoding_set:
+                        encoding_match = encoding_re.match(line)
+                        if encoding_match:
+                            encoding = encoding_match.group(1)
+                            encodin_set = True
+                elif line.strip():
+                    try:
+                        line = line.decode(encoding)
+                    except UnicodeDecodeError:
+                        # I'm not sure this will work but let's try it anyway
+                        return True
+                    if unicode_literal_re.search(line) is not None:
+                        return True
+        finally:
+            f.close()
+        return False
+
+    def run_2to3(filepath):
+        sys.path.append('..')
+        fixers = get_fixers_from_package('lib2to3.fixes')
+        fixers.extend(get_fixers_from_package('custom_fixers'))
+        refactoring_tool = RefactoringTool(fixers)
+        source = refactoring_tool._read_python_source(filepath)[0]
+        ast = refactoring_tool.refactor_string(source, 'conf.py')
+        return unicode(ast)
+
+
+try:
+    base_exception = BaseException
+except NameError:
+    base_exception = Exception
+
+
 try:
     next = next
 except NameError:
@@ -41,6 +97,7 @@ except NameError:
     def next(iterator):
         return iterator.next()
 
+
 try:
     bytes = bytes
 except NameError: