Add string utilities to convert string to unicode and vice versa.

monty/string.py

@@ -12,6 +12,9 @@
 __date__ = "2/28/14"
 
 
+import sys
+
+
 def remove_non_ascii(s):
     """
     Remove non-ascii characters in a file. Needed when support for non-ASCII
@@ -26,6 +29,33 @@ def remove_non_ascii(s):
     return "".join(i for i in s if ord(i) < 128)
 
 
+def unicode2str(s):
+    """
+    Forces a unicode to a string in Python 2, but transparently handles
+    Python 3.
+
+    Args:
+        s (str/unicode): Input string / unicode.
+
+    Returns:
+        str in Python 2. Unchanged otherwise.
+    """
+    return s.encode('utf-8') if sys.version_info.major < 3 else s
+
+
+def str2unicode(s):
+    """
+    Converts strings to unicode in python 2. Ignores Python 3 strings.
+
+    Args:
+        s (str/unicode): Input string / unicode.
+
+    Returns:
+        Unicode.
+    """
+    return unicode(s) if sys.version_info.major < 3 else s
+
+
 def marquee(text="", width=78, mark='*'):
     """
     Return the input string centered in a 'marquee'.

tests/test_string.py

@@ -13,8 +13,9 @@
 
 import unittest
 import random
+import sys
 
-from monty.string import remove_non_ascii
+from monty.string import remove_non_ascii, str2unicode, unicode2str
 
 
 class FuncTest(unittest.TestCase):
@@ -25,6 +26,17 @@ def test_remove_non_ascii(self):
         clean = remove_non_ascii(s)
         self.assertEqual(len(clean), 10)
 
+    def test_str2unicode(self):
+        if sys.version_info.major < 3:
+            self.assertEqual(type(str2unicode("a")), unicode)
+        else:
+            self.assertEqual(type(str2unicode("a")), str)
+
+    def test_unicode2str(self):
+        if sys.version_info.major < 3:
+            self.assertEqual(type(unicode2str("a")), str)
+        else:
+            self.assertEqual(type(unicode2str("a")), str)
 
 if __name__ == '__main__':
     unittest.main()