Handle fonts with missing ulCodePageRange1 entries. (#906)

I think this happens for either very old fonts or fonts which aren't fully
compliant with the open type specification.  Not much either we or the user
can do about it.

kiva/fonttools/_util.py

@@ -155,10 +155,14 @@ def get_ttf_prop_dict(font):
                 if bit in _ot_unicode_range_bits:
                     languages.add(_ot_unicode_range_bits[bit])
         # Check the codepage range bits
-        cp_bits = table.ulCodePageRange1
-        for lang, mask in _ot_code_page_masks.items():
-            if cp_bits & mask:
-                languages.add(lang)
+        try:
+            cp_bits = table.ulCodePageRange1
+            for lang, mask in _ot_code_page_masks.items():
+                if cp_bits & mask:
+                    languages.add(lang)
+        except AttributeError:
+            # some fonts don't have ulCodePageRange1
+            pass
         # Lock the set so that it's hashable
         propdict["languages"] = frozenset(languages)
     except KeyError: