Detect the declared encoding in Python 3; fixes lp:1299169

3 files changed, 21 insertions, 6 deletions

diff --git a/NEWS.txt b/NEWS.txt
index aee125c..776fb6f 100644
--- a/NEWS.txt
+++ b/NEWS.txt
@@ -1,3 +1,6 @@
+UNRELEASED:
+  - Detect the declared encoding in Python 3.
+
 0.8.0 (2014-03-22):
   - Adapt for the AST in Python 3.4.
   - Fix caret position on SyntaxError.
diff --git a/pyflakes/api.py b/pyflakes/api.py
index 03848c3..f32881e 100644
--- a/pyflakes/api.py
+++ b/pyflakes/api.py
@@ -13,8 +13,6 @@ from pyflakes import reporter as modReporter
 
 __all__ = ['check', 'checkPath', 'checkRecursive', 'iterSourceCode', 'main']
 
-universal_newline = ('U' if sys.version_info < (3, 0) else 'r')
-
 
 def check(codeString, filename, reporter=None):
     """
@@ -76,8 +74,10 @@ def checkPath(filename, reporter=None):
     if reporter is None:
         reporter = modReporter._makeDefaultReporter()
     try:
-        with open(filename, universal_newline) as f:
-            codestr = f.read() + '\n'
+        with open(filename, 'rb') as f:
+            codestr = f.read()
+        if sys.version_info < (2, 7):
+            codestr += '\n'     # Work around for Python <= 2.6
     except UnicodeError:
         reporter.unexpectedError(filename, 'problem decoding source')
         return 1
diff --git a/pyflakes/test/test_api.py b/pyflakes/test/test_api.py
index 1d58d80..82b92f6 100644
--- a/pyflakes/test/test_api.py
+++ b/pyflakes/test/test_api.py
@@ -15,7 +15,7 @@ from pyflakes.api import (
     checkRecursive,
     iterSourceCode,
 )
-from pyflakes.test.harness import TestCase, skipIf
+from pyflakes.test.harness import TestCase
 
 if sys.version_info < (3,):
     from cStringIO import StringIO
@@ -437,7 +437,19 @@ foo = '\\xyz'
         self.assertEqual(
             errors, [('flake', str(UnusedImport(sourcePath, Node(1), 'foo')))])
 
-    @skipIf(sys.version_info >= (3,), "not relevant")
+    def test_encodedFileUTF8(self):
+        """
+        If a source file contains bytes which cannot be decoded, this is
+        reported on stderr.
+        """
+        SNOWMAN = unichr(0x2603)
+        source = ("""\
+# coding: utf-8
+x = "%s"
+""" % SNOWMAN).encode('utf-8')
+        sourcePath = self.makeTempFile(source)
+        self.assertHasErrors(sourcePath, [])
+
     def test_misencodedFileUTF8(self):
         """
         If a source file contains bytes which cannot be decoded, this is