Added python 3 compatibility.

We're now using the six module and some helper functions to deal with
str, unicode and bytes. This is tested on python 2.7 and 3.6.

1 files changed, 36 insertions, 15 deletions

diff --git a/pastepy.py b/pastepy.py
index 8c715fa..32a2f74 100644
--- a/pastepy.py
+++ b/pastepy.py
@@ -21,9 +21,22 @@ import pygments
 from pygments import highlight
 from pygments.lexers import get_all_lexers, get_lexer_by_name
 from pygments.formatters import HtmlFormatter
+import six
 
 base62_alphabet = '0123456789abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ'
 
+def b2t(s):
+	'''Convert binary types to text types.'''
+	if isinstance(s, six.binary_type):
+		return s.decode('utf-8')
+	return s
+
+def t2b(s):
+	'''Convert text types to binary types.'''
+	if isinstance(s, six.text_type):
+		return s.encode('utf-8')
+	return s
+
 class PasteError(Exception): pass
 class UnknownSyntaxError(PasteError): pass
 
@@ -89,7 +102,7 @@ class Paste(object):
 			'title': settings.pastebin_name,
 			'header': settings.pastebin_name,
 			'lexers': self.lexers,
-			'nick': c['nick'].value if 'nick' in c else 'Anonymous',
+			'nick': b2t(c['nick'].value if 'nick' in c else 'Anonymous'),
 			'syntax': c['syntax'].value if 'syntax' in c else settings.default,
 			'remember_me': 'nick' in c,
 			'remember_syntax': 'syntax' in c,
@@ -97,27 +110,30 @@ class Paste(object):
 
 	def preview(self, mp):
 		try:
-			lexername, text = self.get_formatted(mp['syntax'].value.decode('utf-8'), mp['text'].value.decode('utf-8'))
+			lexername, text = self.get_formatted(mp['syntax'].value, mp['text'].value)
 		except UnknownSyntaxError:
 			return self.message('Could not find lexer "%s".' % mp['syntax'].value, 'Error')
 
+		# These values to be unicode in py2; in py3 they will be str which is OK.
+		nick = b2t(mp['nick'].value or 'Anonymous')
+		title = b2t(mp['title'].value or 'Untitled')
 		return self.render_template('view.html', {
 			'title': settings.pastebin_name,
 			'header': '%s – Preview' % settings.pastebin_name,
 			'hash': None,
-			'nick': mp['nick'].value.decode('utf-8') or 'Anoynmous',
+			'nick': nick,
 			'date': '%s UTC' % datetime.datetime.utcnow().ctime(),
 			'syntax': lexername,
-			'pastetitle': mp['title'].value.decode('utf-8') or 'Untitled',
+			'pastetitle': title,
 			'text': text,
 			'rendered': (lexername or '').startswith('Rendered '),
 			})
 
 	def add_paste(self, mp):
-		nick = mp['nick'].value.decode('utf-8') or None
-		syntax = mp['syntax'].value.decode('utf-8') or None
-		title = mp['title'].value.decode('utf-8') or None
-		text = mp['text'].value.decode('utf-8').replace('\r', '') or None
+		nick = b2t(mp['nick'].value or None)
+		syntax = b2t(mp['syntax'].value or None)
+		title = b2t(mp['title'].value or None)
+		text = b2t(mp['text'].value.replace('\r', '') or None)
 
 		hash = ''.join(random.choice(base62_alphabet) for x in range(5))
 
@@ -131,13 +147,18 @@ class Paste(object):
 
 		headers = [('Location', '/view/%s' % hash)]
 		c = SimpleCookie()
+		if six.PY2:
+			# In py2 we need nick to be a str, or non-ascii characters will fail with UnicodeDecodeError.
+			# In py3 it must be a str or we will get garbage in the cookie value.
+			nick = t2b(nick)
 		c['nick'] = nick
 		dt = (datetime.datetime.utcnow() + datetime.timedelta(days = 30)) if 'remember_me' in mp else datetime.datetime.utcfromtimestamp(0)
 		c['nick']['expires'] = dt.strftime('%a, %d-%b-%y %H:%M:%S GMT')
 		c['syntax'] = syntax
 		dt = (datetime.datetime.utcnow() + datetime.timedelta(days = 30)) if 'remember_syntax' in mp else datetime.datetime.utcfromtimestamp(0)
 		c['syntax']['expires'] = dt.strftime('%a, %d-%b-%y %H:%M:%S GMT')
-		headers.append(('Set-Cookie', c.output()))
+		headers.append(('Set-Cookie', c['nick'].OutputString()))
+		headers.append(('Set-Cookie', c['syntax'].OutputString()))
 		self.start_response('302 Found', headers)
 		return []
 
@@ -169,7 +190,7 @@ class Paste(object):
 					self.start_response('404 Not Found', [])
 					return []
 				try:
-					lexername, text = self.get_formatted(paste.syntax, paste.text if type(paste.text) == unicode else paste.text.decode('utf-8'))
+					lexername, text = self.get_formatted(paste.syntax, paste.text)
 				except UnknownSyntaxError:
 					return self.message('Could not find the lexer "%s".' % paste.syntax, 'Error')
 				cache = db.Cache(hash, lexername, text)
@@ -185,10 +206,10 @@ class Paste(object):
 			'header': '%s – View paste' % settings.pastebin_name,
 			'hash': hash,
 			'date': '%s UTC' % paste.date.ctime(),
-			'nick': paste.nick or 'Anonymous',
+			'nick': b2t(paste.nick or 'Anonymous'),
 			'syntax': cache.syntax_name,
-			'pastetitle': paste.title or 'Untitled',
-			'text': cache.text,
+			'pastetitle': b2t(paste.title or 'Untitled'),
+			'text': b2t(cache.text),
 			'rendered': (cache.syntax_name or '').startswith('Rendered '),
 			})
 
@@ -202,11 +223,11 @@ class Paste(object):
 			session.close()
 
 		self.start_response('200 OK', [('Content-Type', 'text/plain; charset=UTF-8')])
-		return [paste.text.encode('utf-8')]
+		return [t2b(paste.text)]
 
 	def highlight_stylesheet(self):
 		self.start_response('200 OK', [('Content-Type', 'text/css')])
-		return [self.formatter.get_style_defs()]
+		return [t2b(self.formatter.get_style_defs())]
 
 	def static(self):
 		filename = settings.static_root + os.path.sep + self.path[1]