Provide correct default encoding.

language/google/cloud/language/document.py

@@ -18,6 +18,7 @@
 """
 
 import collections
+import sys
 
 from google.cloud.language import api_responses
 from google.cloud.language.entity import Entity
@@ -64,6 +65,17 @@ class Encoding(object):
     UTF32 = 'UTF32'
     """UTF-32 encoding type."""
 
+    @classmethod
+    def get_default(cls):
+        """Return the appropriate default encoding on this system.
+
+        :rtype: str
+        :returns: The correct default encoding on this system.
+        """
+        if sys.maxunicode == 65535:
+            return cls.UTF16
+        return cls.UTF32
+
 
 class Document(object):
     """Document to send to Google Cloud Natural Language API.
@@ -115,7 +127,7 @@ class Document(object):
     """HTML document type."""
 
     def __init__(self, client, content=None, gcs_url=None, doc_type=PLAIN_TEXT,
-                 language=None, encoding=Encoding.UTF8):
+                 language=None, encoding=Encoding.get_default()):
         if content is not None and gcs_url is not None:
             raise ValueError('A Document cannot contain both local text and '
                              'a link to text in a Google Cloud Storage object')

language/unit_tests/test_document.py

@@ -105,6 +105,19 @@ def make_mock_client(response):
     return mock.Mock(_connection=connection, spec=Client)
 
 
+class TestEncoding(unittest.TestCase):
+    def test_default_low_maxunicode(self):
+        import sys
+        import mock
+
+        from google.cloud.language.document import Encoding
+
+        with mock.patch.dict(sys.__dict__, maxunicode=65535):
+            self.assertEqual(Encoding.get_default(), Encoding.UTF16)
+        with mock.patch.dict(sys.__dict__, maxunicode=1114111):
+            self.assertEqual(Encoding.get_default(), Encoding.UTF32)
+
+
 class TestDocument(unittest.TestCase):
 
     @staticmethod
@@ -127,7 +140,7 @@ def test_constructor_defaults(self):
         self.assertIsNone(document.gcs_url)
         self.assertIsNone(document.language)
         self.assertEqual(document.doc_type, MUT.Document.PLAIN_TEXT)
-        self.assertEqual(document.encoding, MUT.Encoding.UTF8)
+        self.assertEqual(document.encoding, MUT.Encoding.get_default())
 
     def test_constructor_explicit(self):
         import google.cloud.language.document as MUT
@@ -287,7 +300,7 @@ def test_analyze_entities(self):
 
         # Verify the request.
         expected = self._expected_data(
-            content, encoding_type=Encoding.UTF8)
+            content, encoding_type=Encoding.get_default())
         client._connection.api_request.assert_called_once_with(
             path='analyzeEntities', method='POST', data=expected)
 
@@ -428,7 +441,7 @@ def test_analyze_syntax(self):
 
         # Verify the request.
         expected = self._expected_data(
-            content, encoding_type=Encoding.UTF8)
+            content, encoding_type=Encoding.get_default())
         client._connection.api_request.assert_called_once_with(
             path='analyzeSyntax', method='POST', data=expected)
 
@@ -506,7 +519,7 @@ def _annotate_text_helper(self, include_sentiment,
 
         # Verify the request.
         expected = self._expected_data(
-            ANNOTATE_CONTENT, encoding_type=Encoding.UTF8,
+            ANNOTATE_CONTENT, encoding_type=Encoding.get_default(),
             extract_sentiment=include_sentiment,
             extract_entities=include_entities,
             extract_syntax=include_syntax)

system_tests/language.py

@@ -76,15 +76,15 @@ def _check_analyze_entities_result(self, entities):
         self.assertGreater(entity1.salience, 0.0)
         # Other mentions may occur, e.g. "painter".
         self.assertIn(entity1.name, entity1.mentions)
-        self.assertEqual(entity1.wikipedia_url,
+        self.assertEqual(entity1.metadata['wikipedia_url'],
                          'http://en.wikipedia.org/wiki/Caravaggio')
         self.assertIsInstance(entity1.metadata, dict)
         # Verify entity 2.
         self.assertEqual(entity2.name, self.NAME2)
         self.assertEqual(entity2.entity_type, EntityType.LOCATION)
         self.assertGreater(entity2.salience, 0.0)
         self.assertEqual(entity2.mentions, [entity2.name])
-        self.assertEqual(entity2.wikipedia_url,
+        self.assertEqual(entity2.metadata['wikipedia_url'],
                          'http://en.wikipedia.org/wiki/Italy')
         self.assertIsInstance(entity2.metadata, dict)
         # Verify entity 3.
@@ -95,7 +95,7 @@ def _check_analyze_entities_result(self, entities):
         self.assertEqual(entity3.mentions, [entity3.name])
         wiki_url = ('http://en.wikipedia.org/wiki/'
                     'The_Calling_of_St_Matthew_(Caravaggio)')
-        self.assertEqual(entity3.wikipedia_url, wiki_url)
+        self.assertEqual(entity3.metadata['wikipedia_url'], wiki_url)
         self.assertIsInstance(entity3.metadata, dict)
 
     def test_analyze_entities(self):