[python-debian/master 26/36] Fix up debian.changelog for string handling changes in Python 3.

[Colin Watson]

---
 lib/debian/changelog.py |   43 +++++++++++++++++++++++++++++++----------
 tests/test_changelog.py |   48 +++++++++++++++++++++++++++++-----------------
 2 files changed, 62 insertions(+), 29 deletions(-)

diff --git a/lib/debian/changelog.py b/lib/debian/changelog.py
index df0d3df..e99c06a 100644
--- a/lib/debian/changelog.py
+++ b/lib/debian/changelog.py
@@ -30,6 +30,9 @@ import pwd
 import re
 import socket
 import warnings
+import sys
+
+import six
 
 from debian import debian_support
 
@@ -137,7 +140,7 @@ class ChangeBlock(object):
                 changes.append(change)
             self._changes = changes
 
-    def __unicode__(self):
+    def _format(self):
         # TODO(jsw): Switch to StringIO or a list to join at the end.
         block = ""
         if self.package is None:
@@ -170,8 +173,16 @@ class ChangeBlock(object):
             block += line + "\n"
         return block
 
-    def __str__(self):
-        return unicode(self).encode(self._encoding)
+    if sys.version >= '3':
+        __str__ = _format
+
+        def __bytes__(self):
+            return str(self).encode(self._encoding)
+    else:
+        __unicode__ = _format
+
+        def __str__(self):
+            return unicode(self).encode(self._encoding)
 
 topline = re.compile(r'^(\w%(name_chars)s*) \(([^\(\) \t]+)\)'
                      '((\s+%(name_chars)s+)+)\;'
@@ -268,7 +279,9 @@ class Changelog(object):
         
         state = first_heading
         old_state = None
-        if isinstance(file, basestring):
+        if isinstance(file, bytes):
+            file = file.decode(encoding)
+        if isinstance(file, six.string_types):
             # Make sure the changelog file is not empty.
             if len(file.strip()) == 0:
                 self._parse_error('Empty changelog file.', strict)
@@ -276,7 +289,7 @@ class Changelog(object):
 
             file = file.splitlines()
         for line in file:
-            if not isinstance(line, unicode):
+            if not isinstance(line, six.text_type):
                 line = line.decode(encoding)
             # Support both lists of lines without the trailing newline and
             # those with trailing newlines (e.g. when given a file object
@@ -468,15 +481,23 @@ class Changelog(object):
     def _raw_versions(self):
         return [block._raw_version for block in self._blocks]
 
-    def __unicode__(self):
+    def _format(self):
         pieces = []
-        pieces.append(u'\n'.join(self.initial_blank_lines))
+        pieces.append(six.u('\n').join(self.initial_blank_lines))
         for block in self._blocks:
-            pieces.append(unicode(block))
-        return u''.join(pieces)
+            pieces.append(six.text_type(block))
+        return six.u('').join(pieces)
 
-    def __str__(self):
-        return unicode(self).encode(self._encoding)
+    if sys.version >= '3':
+        __str__ = _format
+
+        def __bytes__(self):
+            return str(self).encode(self._encoding)
+    else:
+        __unicode__ = _format
+
+        def __str__(self):
+            return unicode(self).encode(self._encoding)
 
     def __iter__(self):
         return iter(self._blocks)
diff --git a/tests/test_changelog.py b/tests/test_changelog.py
index 2ba0676..2b2ba78 100755
--- a/tests/test_changelog.py
+++ b/tests/test_changelog.py
@@ -29,10 +29,21 @@ from __future__ import absolute_import
 import sys
 import unittest
 
+import six
+
 sys.path.insert(0, '../lib/')
 
 from debian import changelog
 
+
+def open_utf8(filename, mode='r'):
+    """Open a UTF-8 text file in text mode."""
+    if sys.version < '3':
+        return open(filename, mode=mode)
+    else:
+        return open(filename, mode=mode, encoding='UTF-8')
+
+
 class ChangelogTests(unittest.TestCase):
 
     def test_create_changelog(self):
@@ -186,45 +197,46 @@ class ChangelogTests(unittest.TestCase):
             self.assertEqual(str(c), cl_data)
 
     def test_utf8_encoded_file_input(self):
-        f = open('test_changelog_unicode')
+        f = open_utf8('test_changelog_unicode')
         c = changelog.Changelog(f)
         f.close()
-        u = unicode(c)
-        expected_u = u"""haskell-src-exts (1.8.2-3) unstable; urgency=low
+        u = six.text_type(c)
+        expected_u = six.u("""haskell-src-exts (1.8.2-3) unstable; urgency=low
 
   * control: Use versioned Replaces: and Conflicts:
 
- -- Marco Túlio Gontijo e Silva <marcot at debian.org>  Wed, 05 May 2010 18:01:53 -0300
+ -- Marco T\xfalio Gontijo e Silva <marcot at debian.org>  Wed, 05 May 2010 18:01:53 -0300
 
 haskell-src-exts (1.8.2-2) unstable; urgency=low
 
   * debian/control: Rename -doc package.
 
- -- Marco Túlio Gontijo e Silva <marcot at debian.org>  Tue, 16 Mar 2010 10:59:48 -0300
-"""
+ -- Marco T\xfalio Gontijo e Silva <marcot at debian.org>  Tue, 16 Mar 2010 10:59:48 -0300
+""")
         self.assertEqual(u, expected_u)
-        self.assertEqual(str(c), u.encode('utf-8'))
+        self.assertEqual(bytes(c), u.encode('utf-8'))
 
     def test_unicode_object_input(self):
-        f = open('test_changelog_unicode')
-        c_str = f.read()
+        f = open('test_changelog_unicode', 'rb')
+        c_bytes = f.read()
         f.close()
-        c_unicode = c_str.decode('utf-8')
+        c_unicode = c_bytes.decode('utf-8')
         c = changelog.Changelog(c_unicode)
-        self.assertEqual(unicode(c), c_unicode)
-        self.assertEqual(str(c), c_str)
+        self.assertEqual(six.text_type(c), c_unicode)
+        self.assertEqual(bytes(c), c_bytes)
 
     def test_non_utf8_encoding(self):
-        f = open('test_changelog_unicode')
-        c_str = f.read()
+        f = open('test_changelog_unicode', 'rb')
+        c_bytes = f.read()
         f.close()
-        c_unicode = c_str.decode('utf-8')
+        c_unicode = c_bytes.decode('utf-8')
         c_latin1_str = c_unicode.encode('latin1')
         c = changelog.Changelog(c_latin1_str, encoding='latin1')
-        self.assertEqual(unicode(c), c_unicode)
-        self.assertEqual(str(c), c_latin1_str)
+        self.assertEqual(six.text_type(c), c_unicode)
+        self.assertEqual(bytes(c), c_latin1_str)
         for block in c:
-            self.assertEqual(str(block), unicode(block).encode('latin1'))
+            self.assertEqual(bytes(block),
+                             six.text_type(block).encode('latin1'))
 
     def test_block_iterator(self):
         f = open('test_changelog')
-- 
1.7.2.5