If a colored message were wrapped just right (e.g., a colored number ending
the chunk), FormatParser would gobble up the color format code and the number
in the message, causing a KeyError when trying to look up the color in
mircColors.
Signed-off-by: James McCoy <jamessan@users.sourceforge.net>
These module-level variables had historically been created such that reloading
the module wouldn't redefine the variable. However, none of our code reloads
the modules and the guard to prevent redefining the variable was broken so it
would've been redefined anyway.
Signed-off-by: James McCoy <jamessan@users.sourceforge.net>
unbans did send the repr() of the ban list, and IrcMsg.__hash__ did try to hash a list.
Conflicts:
src/version.py
Signed-off-by: James McCoy <jamessan@users.sourceforge.net>
This also fixes a long-standing failing Misc test since it was relying on the
receivedAt tag.
Signed-off-by: James McCoy <jamessan@users.sourceforge.net>
At a minimum, the message gives us the server name, ircd version, supported
umodes, and supported channel modes. Add the umodes and channel modes to
self.supported.
Some IRCds (e.g., hybrid and ircd-seven) have an extra arg which seems to be
the channel modes that require arguments.
Signed-off-by: James McCoy <vega.james@gmail.com>
Define utils.python.glob2re based on the Python version being used.
Use glob2re in Todo and Note plugins.
Closes: Sf#3059292
Signed-off-by: James Vega <jamessan@users.sourceforge.net>
All ircmsgs that takeMsg will return should be processed each time
checkIrcForMsgs is called since there may be multiple available in the
fastqueue.
Reduced the time between calls of checkIrcForMsgs so the delay between
normally queued ircmsgs stays close to the configured throttleTime.
Closes: Sf#3018148
(cherry picked from commit adc5d62bbf6b89631ab25e9b5f5707bde0b06709)
Users were occasionally hitting a situation where the socket had errored,
causing a reconnect, but the socket wasn't closed nor the driver marked as
disconnected. This resulted in run() continuing to try and use the driver,
which would cause another error, schedule another reconnect, log an error, ad
infinitum.
Closes: Sf#2965530
Signed-off-by: James Vega <jamessan@users.sourceforge.net>
(cherry picked from commit a278d17f2bec3697987cbd403594393a4fb30021)
Regexp.error can't directly call Value.error because it's providing extra
information, so it needs to build the InvalidRegistryValue exception itself
and raise it.
Closes: Sf#2985241
Signed-off-by: James Vega <jamessan@users.sourceforge.net>
(cherry picked from commit ef8bd817e8b62ede76aa7501a9a8d69af7408efc)
Also update commands.urlSnarfer to do the same, which allows us to revert
"Don't bother snarfing URLs from non-Action CTCP messages."
This reverts commit 3282e3407ede364acbc92b4e9a6319800d50d46a.
Signed-off-by: James Vega <jamessan@users.sourceforge.net>
(cherry picked from commit 288d7c6e0216baa6445cbd90bb819d1fd0f113c4)
It's better to force people to use an RFC nick and change it after they
connect than to let non-RFC nicks get used and not be able to connect to the
network.
Signed-off-by: James Vega <jamessan@users.sourceforge.net>
(cherry picked from commit 07e283f4502db87cac55217f4567ad4feb418d44)
This gives back the ability to generate a banmask based on the global banmask
settings instead of per-channel settings.
Signed-off-by: James Vega <jamessan@users.sourceforge.net>
(cherry picked from commit 8a98653d3bc9cc638a1739e3d134ebdea8aa0112)
Using the compiled regexps for a PluginRegexp method's __doc__ doesn't work.
Closes Sourceforge #2879862
Signed-off-by: James Vega <jamessan@users.sourceforge.net>
(cherry picked from commit 25fc2de6430ca76b93054fc440f41e3cd93e67dd)
The code expects a float between 0 and 1 inclusive but was simply using
registry.Float. registry.Probability matches the behavior we want.
Signed-off-by: James Vega <jamessan@users.sourceforge.net>
(cherry picked from commit e9a896c736e8eac8402d6d94bce0c3962950ef62)
