Write CSS out in the same encoding it claims to be.

3 files changed, 37 insertions, 18 deletions

diff --git a/scss/__init__.py b/scss/__init__.py
index 24ac7d8..865e996 100644
--- a/scss/__init__.py
+++ b/scss/__init__.py
@@ -48,7 +48,6 @@ __license__ = LICENSE
 
 from collections import defaultdict
 import glob
-import io
 from itertools import product
 import logging
 import warnings
@@ -160,8 +159,8 @@ class SourceFile(object):
 
     @classmethod
     def from_file(cls, f, filename, **kwargs):
-        encoding = determine_encoding(f)
         contents = f.read()
+        encoding = determine_encoding(contents)
         if isinstance(contents, six.binary_type):
             contents = contents.decode(encoding)
 
@@ -172,9 +171,9 @@ class SourceFile(object):
         if isinstance(string, six.text_type):
             # Already decoded; we don't know what encoding to use for output,
             # though, so still check for a @charset.
-            encoding = determine_encoding(io.StringIO(string))
+            encoding = determine_encoding(string)
         elif isinstance(string, six.binary_type):
-            encoding = determine_encoding(io.BytesIO(string))
+            encoding = determine_encoding(string)
             string = string.decode(encoding)
         else:
             raise TypeError("Expected a string, got {0!r}".format(string))
@@ -393,13 +392,17 @@ class Scss(object):
 
     #@profile
     #@print_timing(2)
-    def Compilation(self, scss_string=None, scss_file=None, super_selector=None, filename=None, is_sass=None, line_numbers=True):
+    def Compilation(self, scss_string=None, scss_file=None, source_files=None, super_selector=None, filename=None, is_sass=None, line_numbers=True):
+        # TODO this signature is totally wacky; it should just take a list of
+        # source files
         if super_selector:
             self.super_selector = super_selector + ' '
         self.reset()
 
         source_file = None
-        if scss_string is not None:
+        if source_files is not None:
+            self.source_files = source_files
+        elif scss_string is not None:
             source_file = SourceFile.from_string(scss_string, filename, is_sass, line_numbers)
         elif scss_file is not None:
             source_file = SourceFile.from_filename(scss_file, filename, is_sass, line_numbers)
@@ -1264,6 +1267,7 @@ class Scss(object):
         """
         Implements @-blocks
         """
+        # TODO handle @charset, probably?
         # Interpolate the current block
         # TODO this seems like it should be done in the block header.  and more
         # generally?
diff --git a/scss/cssdefs.py b/scss/cssdefs.py
index 398a4cc..e312ca5 100644
--- a/scss/cssdefs.py
+++ b/scss/cssdefs.py
@@ -347,16 +347,12 @@ def is_builtin_css_function(name):
 # CSS character set determination
 # Based upon: http://www.w3.org/TR/CSS2/syndata.html#charset
 
-def determine_encoding(f):
-    """Return the appropriate encoding for the given file, according to the CSS
-    charset rules.
+def determine_encoding(buf):
+    """Return the appropriate encoding for the given CSS source, according to
+    the CSS charset rules.
 
-    `f` should be a file-like object with the cursor at the beginning.
+    `buf` may be either a string or bytes.
     """
-    # 200 bytes should be enough for anyone
-    buf = f.read(200)
-    f.seek(0)
-
     # The ultimate default is utf8; bravo, W3C
     bom_encoding = 'UTF-8'
 
diff --git a/scss/tool.py b/scss/tool.py
index d765c8f..6b15a37 100644
--- a/scss/tool.py
+++ b/scss/tool.py
@@ -163,19 +163,38 @@ def run_tests():
 
 def do_build(options, args):
     if options.output is not None:
-        output = open(options.output, 'wt')
+        out = open(options.output, 'wb')
     else:
-        output = sys.stdout
+        out = sys.stdout
+        # Get the unencoded stream on Python 3
+        out = getattr(out, 'buffer', out)
 
     css = Scss(scss_opts={
         'style': options.style,
         'debug_info': options.debug_info,
     })
     if args:
+        source_files = [
+            SourceFile.from_filename(path, is_sass=options.is_sass)
+            for path in args
+        ]
         for path in args:
-            output.write(css.compile(scss_file=path, is_sass=options.is_sass))
+            out.write(css.compile(scss_file=path, is_sass=options.is_sass))
     else:
-        output.write(css.compile(sys.stdin.read(), is_sass=options.is_sass))
+        source_files = [
+            SourceFile.from_file(sys.stdin, "<stdin>", is_sass=options.is_sass)]
+
+    encodings = set(source.encoding for source in source_files)
+    if len(encodings) > 1:
+        sys.stderr.write(
+            "Can't combine these files!  "
+            "They have different encodings: {0}\n"
+            .format(', '.join(encodings))
+        )
+        sys.exit(3)
+
+    output = css.compile(source_files=source_files)
+    out.write(output.encode(source_files[0].encoding))
 
     for f, t in profiling.items():
         sys.stderr.write("%s took %03fs" % (f, t))