Fix encoding logic/order

vmuriart · vmuriart · commit 66b36af84fbe · 2017-01-10T19:13:32.000-07:00
- If user provides an encoding value, use it instead of trying to _guess_ first.
- If no value is provided, then decode with default of utf-8, otherwise
  try with unicode-escape
diff --git a/sqlparse/lexer.py b/sqlparse/lexer.py
@@ -43,12 +43,13 @@ def get_tokens(text, encoding=None):
         if isinstance(text, text_type):
             pass
         elif isinstance(text, bytes_type):
-            try:
-                text = text.decode('utf-8')
-            except UnicodeDecodeError:
-                if not encoding:
-                    encoding = 'unicode-escape'
+            if encoding:
                 text = text.decode(encoding)
+            else:
+                try:
+                    text = text.decode('utf-8')
+                except UnicodeDecodeError:
+                    text = text.decode('unicode-escape')
         else:
             raise TypeError(u"Expected text or file-like object, got {!r}".
                             format(type(text)))
