Use UTF-8 in IRC

lib/ircbot.py

@@ -45,7 +45,8 @@ class SingleServerIRCBot(SimpleIRCClient):
     have operator or voice modes.  The "database" is kept in the
     self.channels attribute, which is an IRCDict of Channels.
     """
-    def __init__(self, server_list, nickname, realname, reconnection_interval=60):
+    def __init__(self, server_list, nickname, realname,
+                 reconnection_interval=60, encoding="utf-8"):
         """Constructor for SingleServerIRCBot objects.
 
         Arguments:
@@ -61,6 +62,8 @@ class SingleServerIRCBot(SimpleIRCClient):
             reconnection_interval -- How long the bot should wait
                                      before trying to reconnect.
 
+            encoding -- An encoding to use.
+
             dcc_connections -- A list of initiated/accepted DCC
             connections.
         """
@@ -74,6 +77,7 @@ class SingleServerIRCBot(SimpleIRCClient):
 
         self._nickname = nickname
         self._realname = realname
+        self._encoding = encoding
         for i in ["disconnect", "join", "kick", "mode",
                   "namreply", "nick", "part", "quit"]:
             self.connection.add_global_handler(i,
@@ -96,7 +100,8 @@ class SingleServerIRCBot(SimpleIRCClient):
                          self.server_list[0][1],
                          self._nickname,
                          password,
-                         ircname=self._realname)
+                         ircname=self._realname,
+                         encoding=self._encoding)
         except ServerConnectionError:
             pass
 

lib/irclib.py

@@ -1,3 +1,5 @@
+# -*- coding: utf-8 -*-
+
 # Copyright (C) 1999--2002  Joel Rosdahl
 #
 # This library is free software; you can redistribute it and/or
@@ -380,7 +382,7 @@ class ServerConnection(Connection):
         self.socket = None
 
     def connect(self, server, port, nickname, password=None, username=None,
-                ircname=None, localaddress="", localport=0):
+                ircname=None, localaddress="", localport=0, encoding="utf-8"):
         """Connect/reconnect to a server.
 
         Arguments:
@@ -401,6 +403,8 @@ class ServerConnection(Connection):
 
             localport -- Bind the connection to a specific local port.
 
+            encoding -- The encoding to use.
+
         This function can be called to reconnect a closed connection.
 
         Returns the ServerConnection object.
@@ -418,6 +422,7 @@ class ServerConnection(Connection):
         self.username = username or nickname
         self.ircname = ircname or nickname
         self.password = password
+        self.encoding = encoding
         self.localaddress = localaddress
         self.localport = localport
         self.localhost = socket.gethostname()
@@ -489,7 +494,12 @@ class ServerConnection(Connection):
             self.disconnect("Connection reset by peer")
             return
 
-        lines = _linesep_regexp.split(self.previous_buffer + new_data.decode("latin-1", "replace"))
+        try:
+            new_data = new_data.decode(self.encoding, "strict")
+        except (UnicodeDecodeError, UnicodeEncodeError):
+            new_data = new_data.decode("latin-1", "replace")
+
+        lines = _linesep_regexp.split(self.previous_buffer + new_data)
 
         # Save the last, unfinished line.
         self.previous_buffer = lines[-1]
@@ -777,10 +787,17 @@ class ServerConnection(Connection):
 
         The string will be padded with appropriate CR LF.
         """
+        if isinstance(string, bytes):
+            data = string
+        else:
+            try:
+                data = string.encode(self.encoding, "strict")
+            except (UnicodeEncodeError, UnicodeDecodeError):
+                data = string.encode("latin-1", "replace")
         if self.socket is None:
             raise ServerNotConnectedError("Not connected.")
         try:
-            self.socket.send(string.encode("latin-1", "replace") + b"\r\n")
+            self.socket.send(data + b"\r\n")
             if DEBUG:
                 print("TO SERVER: %s\n" % string)
         except socket.error as x:
@@ -1046,7 +1063,7 @@ class SimpleIRCClient:
         self.dcc_connections.remove(c)
 
     def connect(self, server, port, nickname, password=None, username=None,
-                ircname=None, localaddress="", localport=0):
+                ircname=None, localaddress="", localport=0, encoding="utf-8"):
         """Connect/reconnect to a server.
 
         Arguments:
@@ -1067,11 +1084,13 @@ class SimpleIRCClient:
 
             localport -- Bind the connection to a specific local port.
 
+            encoding -- The encoding to use.
+
         This function can be called to reconnect a closed connection.
         """
         self.connection.connect(server, port, nickname,
                                 password, username, ircname,
-                                localaddress, localport)
+                                localaddress, localport, encoding)
 
     def dcc_connect(self, address, port, dcctype="chat"):
         """Connect to a DCC peer.