Python 3: Initial support for Python 3 (CLI)

This changeset makes gPodder's codebase convertable
to Python 3 using the "2to3" utility. Right now, only
the CLI module (bin/gpo) has been tested.

See the README file for instructions and remarks.

README

@@ -85,6 +85,26 @@
     To install gPodder system-wide, use "make install".
 
 
+ [ PYTHON 3 SUPPORT ]
+
+    The CLI version of gPodder (bin/gpo) is compatible with Python 3
+    after converting the codebase with the 2to3 utility:
+
+      2to3 -w bin/* src/gpodder/
+
+    You will also need a copy of "mygpoclient" converted using 2to3 and
+    a copy of "feedparser" converted using 2to3 (see the feedparser README
+    for details on how to get it set up on Python 3, including sgmllib).
+
+    Please note that the Gtk UI is not compatible with Python 3 (it will
+    be once we migrate the codebase to Gtk3/GObject Introspection), and
+    the QML UI - while theoretically compatible - has not been tested
+    with Python 3 yet due to the Python 3 support status in PySide.
+
+    As of January 2012, Python 3 support is still experimental. Please
+    report any bugs that you find to the gPodder bug tracker (see below).
+
+
  [ PORTABLE MODE / ROAMING PROFILES ]
 
     The run-time environment variable GPODDER_HOME is used to set
@@ -104,5 +124,5 @@
     - IRC channel                      #gpodder on irc.freenode.net
 
   ............................................................................
-             Last updated: 2012-01-09 by Thomas Perl <thp.io/about>
+             Last updated: 2012-01-10 by Thomas Perl <thp.io/about>
 

bin/gpo

@@ -283,7 +283,7 @@ class gPodderCli(object):
 
         episodes = (u'%3d. %s %s' % (i+1, status_str(e), e.title)
                 for i, e in enumerate(podcast.get_episodes()))
-	return episodes
+        return episodes
 
     @FirstArgumentIsPodcastURL
     def info(self, url):

makefile

@@ -136,6 +136,7 @@ clean:
 	$(PYTHON) setup.py clean
 	find src/ -name '*.pyc' -exec rm '{}' \;
 	find src/ -name '*.pyo' -exec rm '{}' \;
+	find src/ -type d -name '__pycache__' -exec rm -r '{}' \;
 	find data/ui/ -name '*.ui.h' -exec rm '{}' \;
 	rm -f MANIFEST PKG-INFO data/messages.pot~ $(DESKTOPFILE_H)
 	rm -f data/gpodder-??x??.png .coverage

src/gpodder/__init__.py

@@ -88,8 +88,16 @@ osx = (platform.system() == 'Darwin')
 textdomain = 'gpodder'
 locale_dir = gettext.bindtextdomain(textdomain)
 t = gettext.translation(textdomain, locale_dir, fallback=True)
-gettext = t.ugettext
-ngettext = t.ungettext
+
+try:
+    # Python 2
+    gettext = t.ugettext
+    ngettext = t.ungettext
+except AttributeError:
+    # Python 3
+    gettext = t.gettext
+    ngettext = t.ngettext
+
 if win32:
     try:
         # Workaround for bug 650

src/gpodder/dbsqlite.py

@@ -37,6 +37,7 @@ import logging
 logger = logging.getLogger(__name__)
 
 from gpodder import schema
+from gpodder import util
 
 import threading
 import re
@@ -202,11 +203,8 @@ class Database(object):
         with self.lock:
             try:
                 cur = self.cursor()
-                def convert(x):
-                    if isinstance(x, str):
-                        x = x.decode('utf-8', 'ignore')
-                    return x
-                values = [convert(getattr(o, name)) for name in columns]
+                values = [util.convert_bytes(getattr(o, name))
+                        for name in columns]
 
                 if o.id is None:
                     qmarks = ', '.join('?'*len(columns))
@@ -248,20 +246,20 @@ class Database(object):
         Returns True if a foldername for a channel exists.
         False otherwise.
         """
-        if not isinstance(foldername, unicode):
-            foldername = foldername.decode('utf-8', 'ignore')
+        foldername = util.convert_bytes(foldername)
 
-        return self.get("SELECT id FROM %s WHERE download_folder = ?" % self.TABLE_PODCAST, (foldername,)) is not None
+        return self.get("SELECT id FROM %s WHERE download_folder = ?" %
+                self.TABLE_PODCAST, (foldername,)) is not None
 
     def episode_filename_exists(self, podcast_id, filename):
         """
         Returns True if a filename for an episode exists.
         False otherwise.
         """
-        if not isinstance(filename, unicode):
-            filename = filename.decode('utf-8', 'ignore')
+        filename = util.convert_bytes(filename)
 
-        return self.get("SELECT id FROM %s WHERE podcast_id = ? AND download_filename = ?" % self.TABLE_EPISODE, (podcast_id, filename,)) is not None
+        return self.get("SELECT id FROM %s WHERE podcast_id = ? AND download_filename = ?" %
+                self.TABLE_EPISODE, (podcast_id, filename,)) is not None
 
     def get_last_published(self, podcast):
         """
@@ -275,11 +273,10 @@ class Database(object):
         a given channel. Used after feed updates for
         episodes that have disappeared from the feed.
         """
-        if not isinstance(guid, unicode):
-            guid = guid.decode('utf-8', 'ignore')
+        guid = util.convert_bytes(guid)
 
         with self.lock:
             cur = self.cursor()
-            cur.execute('DELETE FROM %s WHERE podcast_id = ? AND guid = ?' % self.TABLE_EPISODE, \
-                    (podcast_id, guid))
+            cur.execute('DELETE FROM %s WHERE podcast_id = ? AND guid = ?' %
+                    self.TABLE_EPISODE, (podcast_id, guid))
 

src/gpodder/download.py

@@ -46,7 +46,13 @@ import collections
 
 import mimetypes
 import email
-import email.Header
+
+try:
+    # Python 2
+    from email.Header import decode_header
+except ImportError:
+    # Python 3
+    from email.header import decode_header
 
 import cgi
 
@@ -69,7 +75,7 @@ def get_header_param(headers, param, header_name):
             value = msg.get_param(param, header=header_name)
             if value is None:
                 return None
-            decoded_list = email.Header.decode_header(value)
+            decoded_list = decode_header(value)
             value = []
             for part, encoding in decoded_list:
                 if encoding:

src/gpodder/minidb.py

@@ -88,6 +88,7 @@ class Store(object):
         if isinstance(v, unicode):
             return v
         elif isinstance(v, str):
+            # XXX: Rewrite ^^^ as "isinstance(v, bytes)" in Python 3
             return v.decode('utf-8')
         else:
             return str(v)

src/gpodder/model.py

@@ -40,7 +40,14 @@ import glob
 import shutil
 import time
 import datetime
-import rfc822
+
+try:
+    # Python 2
+    from rfc822 import mktime_tz
+except ImportError:
+    # Python 3
+    from email.utils import mktime_tz
+
 import hashlib
 import feedparser
 import collections
@@ -212,7 +219,7 @@ class PodcastEpisode(PodcastModelObject):
             episode.description = entry.get('subtitle', '')
 
         if entry.get('updated_parsed', None):
-            episode.published = rfc822.mktime_tz(entry.updated_parsed+(0,))
+            episode.published = mktime_tz(entry.updated_parsed+(0,))
 
         enclosures = entry.get('enclosures', [])
         media_rss_content = entry.get('media_content', [])
@@ -454,10 +461,7 @@ class PodcastEpisode(PodcastModelObject):
             return _('No description available')
         else:
             # Decode the description to avoid gPodder bug 1277
-            if isinstance(desc, str):
-                desc = desc.decode('utf-8', 'ignore')
-
-            desc = desc.strip()
+            desc = util.convert_bytes(desc).strip()
 
             if len(desc) > MAX_LINE_LENGTH:
                 return desc[:MAX_LINE_LENGTH] + '...'
@@ -873,9 +877,7 @@ class PodcastChannel(PodcastModelObject):
 
     @classmethod
     def sort_key(cls, podcast):
-        key = podcast.title.lower()
-        if not isinstance(key, unicode):
-            key = key.decode('utf-8', 'ignore')
+        key = util.convert_bytes(podcast.title.lower())
         return re.sub('^the ', '', key).translate(cls.UNICODE_TRANSLATE)
 
     @classmethod

src/gpodder/qmlui/helper.py

@@ -19,15 +19,15 @@
 
 import gpodder
 
+from gpodder import util
+
 from PySide import QtCore
 
 
 class Action(QtCore.QObject):
     def __init__(self, caption, action, target=None):
         QtCore.QObject.__init__(self)
-        if isinstance(caption, str):
-            caption = caption.decode('utf-8')
-        self._caption = caption
+        self._caption = util.convert_bytes(caption)
 
         self.action = action
         self.target = target

src/gpodder/qmlui/model.py

@@ -37,14 +37,7 @@ from gpodder import model
 import threading
 import os
 
-def convert(s):
-    if s is None:
-        return None
-
-    if isinstance(s, unicode):
-        return s
-
-    return s.decode('utf-8', 'ignore')
+convert = util.convert_bytes
 
 class QEpisode(QObject):
     def __init__(self, wrapper_manager, podcast, episode):

src/gpodder/util.py

@@ -1182,6 +1182,33 @@ def open_website(url):
     else:
         threading.Thread(target=webbrowser.open, args=(url,)).start()
 
+def convert_bytes(d):
+    """
+    Convert byte strings to unicode strings
+
+    This function will decode byte strings into unicode
+    strings. Any other data types will be left alone.
+
+    >>> convert_bytes(None)
+    >>> convert_bytes(1)
+    1
+    >>> convert_bytes(True)
+    True
+    >>> convert_bytes(3.1415)
+    3.1415
+    >>> convert_bytes('Hello')
+    u'Hello'
+    >>> convert_bytes(u'Hey')
+    u'Hey'
+    """
+    if d is None:
+        return d
+    if any(isinstance(d, t) for t in (int, bool, float)):
+        return d
+    elif not isinstance(d, unicode):
+        return d.decode('utf-8', 'ignore')
+    return d
+
 def sanitize_encoding(filename):
     r"""
     Generate a sanitized version of a string (i.e.
@@ -1290,12 +1317,16 @@ def find_mount_point(directory):
     >>> restore()
     """
     if isinstance(directory, unicode):
+        # XXX: This is only valid for Python 2 - misleading error in Python 3?
         # We do not accept unicode strings, because they could fail when
         # trying to be converted to some native encoding, so fail loudly
         # and leave it up to the callee to encode into the proper encoding.
         raise ValueError('Convert unicode objects to str first.')
 
     if not isinstance(directory, str):
+        # In Python 2, we assume it's a byte str; in Python 3, we assume
+        # that it's a unicode str. The abspath/ismount/split functions of
+        # os.path work with unicode str in Python 3, but not in Python 2.
         raise ValueError('Directory names should be of type str.')
 
     directory = os.path.abspath(directory)