Improved tests

3 files changed, 149 insertions, 29 deletions

diff --git a/tests/test_feedgenerator.py b/tests/test_feedgenerator.py
index 5b8bc42..dbc90b3 100644
--- a/tests/test_feedgenerator.py
+++ b/tests/test_feedgenerator.py
@@ -27,19 +27,6 @@ FIXT_ITEM = dict(
     description="Testing."
 )
 
-# Cannot use unicode chars here. six.b() encodes string to latin-1, which would fail if
-# we have unicode chars in these literals.
-FIXT_FEED_BYTES = dict(
-    title=six.b("Poynter E-Media Tidbits"),
-    link=six.b("http://www.poynter.org/column.asp?id=31"),
-    description=six.b("A group Weblog by the sharpest minds in online media/journalism/publishing."),
-    language=six.b("en")
-)
-FIXT_ITEM_BYTES = dict(
-    title=six.b("Hello"),
-    link=six.b("http://www.holovaty.com/test/"),
-    description=six.b("Testing.")
-)
 
 EXPECTED_RESULT = """<?xml version="1.0" encoding="utf-8"?>
 <rss xmlns:atom="http://www.w3.org/2005/Atom" version="2.0"><channel><title>Poynter E-Media Tidbits</title><link>http://www.poynter.org/column.asp?id=31</link><description>A group Weblog by the sharpest minds in online media/journalism/publishing.
@@ -48,31 +35,57 @@ EXPECTED_RESULT = """<?xml version="1.0" encoding="utf-8"?>
     Finnish: Mustan kissan paksut posket. (ah, no special chars) Kärpänen sanoi kärpäselle: tuu kattoon kattoon ku kaveri tapettiin tapettiin.
     </description><language>en</language><lastBuildDate>%DATE%</lastBuildDate><item><title>Hello</title><link>http://www.holovaty.com/test/</link><description>Testing.</description></item></channel></rss>"""
 
-EXPECTED_RESULT_BYTES = six.b("""<?xml version="1.0" encoding="utf-8"?>
-<rss xmlns:atom="http://www.w3.org/2005/Atom" version="2.0"><channel><title>Poynter E-Media Tidbits</title><link>http://www.poynter.org/column.asp?id=31</link><description>A group Weblog by the sharpest minds in online media/journalism/publishing.</description><language>en</language><lastBuildDate>%DATE%</lastBuildDate><item><title>Hello</title><link>http://www.holovaty.com/test/</link><description>Testing.</description></item></channel></rss>""")
-
 
 ENCODING = 'utf-8'
 
-
 def build_expected_result(feed, expected_result, encoding):
+    # Result's date is of course different from the date in the fixture.
+    # So make them equal!
     d = feedgenerator.rfc2822_date(feed.latest_post_date())
-    return expected_result.replace('%DATE%', d)
+    s = expected_result.replace('%DATE%', d)
+    if encoding:
+        return s.encode(encoding)
+    else:
+        return s
 
 
 class TestFeedGenerator(unittest.TestCase):
 
-    def test_001_default_encoding(self):
-        # Default encoding is unicode since we have imported unicode_literals!
+    def test_000_types(self):
+        ty = str if six.PY3 else unicode
+        for k, v in FIXT_FEED.items():
+            self.assertEqual(type(v), ty)
+        for k, v in FIXT_ITEM.items():
+            self.assertEqual(type(v), ty)
+        self.assertEqual(type(EXPECTED_RESULT), ty)
+
+    def test_001_string_results(self):
         feed = feedgenerator.Rss201rev2Feed(**FIXT_FEED)
         feed.add_item(**FIXT_ITEM)
         result = feed.writeString(ENCODING)
-        expected_result = build_expected_result(feed, EXPECTED_RESULT, ENCODING)
+        if six.PY3:
+            # On Python 3, result of feedgenerator is a unicode string!
+            # So do not encode our expected_result.
+            expected_result = build_expected_result(feed, EXPECTED_RESULT, None)
+        else:
+            # On Python 2, result of feedgenerator is a str string!
+            # Expected_result must therefore encoded likewise.
+            expected_result = build_expected_result(feed, EXPECTED_RESULT, ENCODING)
+        # XXX Is the following resoning true? XXX
+        # The different result types of Python 2 (str=bytes) and Python 3
+        # (str=text=unicode) stems from a different implementation of StringIO.
+        # As I understand it, getvalue() in Python 2 returns the type you
+        # originally wrote into the buffer. In Python 3 getvalue() always
+        # returns a str (=text=unicode).
+        # See other test: test_stringio.py
+        #print type(result), type(expected_result)
+        self.assertEqual(type(result), type(expected_result))
         self.assertEqual(result, expected_result)
 
-    def OFF_test_002_bytes(self):
-        feed = feedgenerator.Rss201rev2Feed(**FIXT_FEED_BYTES)
-        feed.add_item(**FIXT_ITEM_BYTES)
-        result = feed.writeString(ENCODING)
-        expected_result = build_expected_result(feed, EXPECTED_RESULT_BYTES, ENCODING)
-        self.assertEqual(result, expected_result)
+    def OFF_test_002_file_results(self):
+        pass
+        # DO-IT_YOURSELF: Run usage_example with python2 and python3.
+        #                 Each will create a feed file.
+        #                 Compare the files, they must be equal!
+        #                 XXX  Argh -- No, the lastBuildDate will differ.
+        #                              But this is allowed.
diff --git a/tests/test_stringio.py b/tests/test_stringio.py
new file mode 100644
index 0000000..5ba56a8
--- /dev/null
+++ b/tests/test_stringio.py
@@ -0,0 +1,89 @@
+# -*- encoding: utf-8 -*-
+
+from __future__ import unicode_literals, print_function
+
+try:
+    import unittest2 as unittest
+except ImportError:
+    import unittest
+
+import six
+from six import StringIO
+
+ENCODING = 'utf-8'
+
+S0 = 'hello world, Umlauts: äöüßÄÖÜ, Chinese: 四是四，十是十，十四是十四，四十是四十，四十四隻不識字之石獅子是死的'
+S0_BYTES = 'fe fi foe fam'.encode(ENCODING)
+
+print("###", StringIO, "###")
+
+class TestStringIO(unittest.TestCase):
+
+    def test_001_text(self):
+        # If we throw unicode into the StringIO buffer, we'll
+        # get unicode out of it.
+        # Thank god this is the same in Python 2 and 3.
+        if six.PY3:
+            self.assertEqual(type(S0), str)
+        else:
+            self.assertEqual(type(S0), unicode)
+        buf = StringIO()
+        print(S0, file=buf, end="")
+        s1 = buf.getvalue()
+        self.assertEqual(type(S0), type(s1))
+        self.assertEqual(S0, s1)
+        if six.PY3:
+            self.assertEqual(type(s1), str)
+        else:
+            self.assertEqual(type(s1), unicode)
+
+    def test_002_bytes(self):
+        buf = StringIO()
+        print(S0_BYTES, file=buf, end="")
+        s1 = buf.getvalue()
+        
+        if six.PY3:
+            # In Python 3 StringIO *ALWAYS* returns str (=text=unicode) !
+            # Even if we originally write bytes into the buffer, the value
+            # we get out of it has type str!
+
+            # Input is bytes
+            self.assertEqual(type(S0_BYTES), bytes)
+            # Output is NOT bytes...
+            self.assertNotEqual(type(S0_BYTES), type(s1))
+            self.assertNotEqual(type(s1), bytes)
+            # ...but str!
+            self.assertEqual(type(s1), str)
+            # So the contents are not equal!
+            self.assertNotEqual(S0_BYTES, s1)
+            # StringIO coerced bytes into str:
+            # b'xyz' ---> "b'xyz'"
+            self.assertEqual(str(S0_BYTES), s1)
+            # See, the type info is literally present in the output str!
+            self.assertEqual("b'" + str(S0_BYTES, encoding=ENCODING) + "'", s1)
+            # Coercion is NOT decoding!
+            self.assertNotEqual(S0_BYTES.decode(ENCODING), s1)
+            self.assertNotEqual(str(S0_BYTES, encoding=ENCODING), s1)
+            # These are the same
+            self.assertEqual(S0_BYTES.decode(ENCODING), 
+                str(S0_BYTES, encoding=ENCODING))
+            # Additional note:
+            # If we do not specify an encoding when we create a StringIO
+            # buffer, Python 3 automatically uses the locale's preferred
+            # encoding: locale.getpreferredencoding()
+            # Cf. http://docs.python.org/release/3.0.1/library/io.html#io.TextIOWrapper
+            # In my case this is the same encoding as the encoding of this source file,
+            # namely UTF-8. If on your system both encodings are different, you may 
+            # encounter other results than the above.
+            #
+            # In Python 3.2 the signature of StringIO() has changed. It is no more
+            # possible to specify an encoding here.
+        else:
+            # In Python 2 StringIO returns the type that we originally
+            # wrote into the buffer.
+            # Here we see that if we write bytes into the buffer, we'll get
+            # bytes out of it.
+            self.assertEqual(type(S0_BYTES), str)
+            self.assertEqual(type(S0_BYTES), type(s1))
+            self.assertEqual(type(s1), str)
+            self.assertEqual(S0_BYTES, s1)
diff --git a/tests/usage_example.py b/tests/usage_example.py
index e074148..69f258b 100644
--- a/tests/usage_example.py
+++ b/tests/usage_example.py
@@ -1,5 +1,12 @@
 # -*- encoding: utf-8 -*-
+
+from __future__ import unicode_literals
+
+import os
+import tempfile
+import six
 import feedgenerator
+
 feed = feedgenerator.Rss201rev2Feed(
     title="Poynter E-Media Tidbits",
     link="http://www.poynter.org/column.asp?id=31",
@@ -15,6 +22,17 @@ feed.add_item(
     link="http://www.holovaty.com/test/",
     description="Testing."
 )
-with open('test.rss', 'w') as fp:
-    feed.write(fp, 'utf-8')
 
+if six.PY3:
+    FN_PREFIX = 'feed_py3-'
+else:
+    FN_PREFIX = 'feed_py2-'
+
+# Usage example in feedgenerator docs opens the file in text mode, not binary.
+# So we do this here likewise.
+fd, filename = tempfile.mkstemp(prefix=FN_PREFIX, suffix='.txt', text=True)
+try:
+    fh = os.fdopen(fd, 'w')
+    feed.write(fh, 'utf-8')
+finally:
+    fh.close()