handle unicode better by decoding early and encoding late

jac/compat.py

@@ -11,7 +11,7 @@
 if is_py2:
 
     def _u(text):
-        if type(text) == str:
+        if isinstance(text, str):
             return text.decode('utf-8')
         return unicode(text)
     u = _u

jac/compilers/__init__.py

@@ -19,6 +19,9 @@ def __new__(meta, name, bases, attrs):
 def compile(what, mimetype, cwd=None, uri_cwd=None, debug=None):
     """
     Compile a given text based on mimetype.
+
+    The text to compile must be provided as a Unicode object and this
+    function must return the compiled text as a Unicode object.
     """
 
     try:

jac/compilers/less.py

@@ -1,7 +1,7 @@
 import subprocess
 from six import with_metaclass
 
-from jac.compat import file
+from jac.compat import file, u
 
 from . import CompilerMeta
 
@@ -35,7 +35,8 @@ def compile(cls, what, mimetype='text/less', cwd=None, uri_cwd=None,
 
         if isinstance(what, file):
             what = what.read()
-        (stdout, stderr) = handler.communicate(input=what)
+        (stdout, stderr) = handler.communicate(input=what.encode('utf-8'))
+        stdout = u(stdout)
 
         if handler.returncode == 0:
             return stdout

jac/compilers/sass.py

@@ -1,7 +1,7 @@
 import subprocess
 from six import with_metaclass
 
-from jac.compat import file
+from jac.compat import file, u
 
 from . import CompilerMeta
 
@@ -27,6 +27,7 @@ def compile(cls, what, mimetype='text/sass', cwd=None,
         if isinstance(what, file):
             what = what.read()
         (stdout, stderr) = handler.communicate(input=what)
+        stdout = u(stdout)
 
         if handler.returncode == 0:
             return stdout

jac/extension.py

@@ -88,9 +88,9 @@ def _make_hash(self, html, compilables):
 
     def _get_contents(self, src):
         if isinstance(src, file):
-            return src.read()
+            return u(src.read())
         else:
-            return src
+            return u(src)
 
     def _compile(self, compression_type, caller):
         html = caller()
@@ -143,7 +143,6 @@ def _compile(self, compression_type, caller):
                                uri_cwd=uri_cwd, debug=debug)
             else:
                 text = self._get_contents(src)
-            text = u(text)
 
             if not debug:
                 outfile = cached_file