In the unicodeData.txt file the old name(field 11) is only used for the character name if the name is "".

So rather than printing it instead of name when it exists I should be printing it alongside names that are surrounded by <>.

See http://ftp.unicode.org/Public/3.0-Update/UnicodeData-3.0.0.html

Provide a unicode search command that uses the latest unicode consortium data.

Interface

Search by codepoint

./u <code>

Where code can have any of the following formats:

• uFFFF
• u+FFFF
• \uFFFF
• 0xFFFF
• 0xF0 0x9F 0x98 0xBC where the provided numbers form a single valid codepoint
• 240 159 152 188 where the provided numbers form a single valid codepoint
• 128572 which is treated as a decimal unicode identifier

Search by string:

./u <regex>...

Search against the entity name field(s) using the specified regex. I will probably end up doing this with the awk regex engine.

Seach by char:

./u <char>

Will search through the character representation fields. Not sure if/how I want to handle normalization with that. This search is performed instead of string search when the argument is exactly one unicode character.

Update database:

./u -u

Print current version and retrieval date:

./u -v

Output

The queries should all output the following information:

U+[ID], [codepoints array], [description] ([char])

Command to add a line from the recent chat history to a file of user quotes.

Command Syntax

Adding a quote:

./quote <nick> <quote text...>

Getting a quote, with optional target user:

./quote [nick]

Getting all quotes, with optional target user:

./quote -a [nick]

The logs are full of \e[7m^M when viewed with most editors. The debug output should only go through showm when it goes to a terminal.

As is the timestamps in the logs are local unix time. This means that if you read them back after changing the timezone it will get the time wrong.

They should be written out as UTC and converted to local when read back in.

There are probably many similar time handling issues that should be fixed the same way.

I would like Shellby to provide completely invisible karma tracking.

Thus it should:

• report karma only when asked
• populate initial karma from logs
• attempt to determine user accounts and combine karma for nicks that share an account(maybe)
• recognize the following forms:
  • /^nickname[:,]? \+\+([^+]|$)/
  • /\bnickname\+\+([^+]|$)/
• only accept one inline increment, because seriously

Command to make using python with eval.in easier.

Simply send the following as the code to execute:

import sys;from code import InteractiveInterpreter;
InteractiveInterpreter().runsource(sys.stdin.read())

Then pass the user submitted code to stdin.

Command to provide a reminder for a user next time they speak.

Syntax

./tell <nickname> <message>

When the user talks shellby will send them a message in the following format:

<nickname>: <sender> asked me to tell you <message>

I am not sure whether it is desirable to add so much overhead. The max message length would be, in the worst case, SERVER_MAX - 25 - NICK_MAX * 2. Though anything over 400 seems sufficient for almost all non annoying messages.

It may also be desirable to provide ./ask that uses asked me to ask you in the reminder.

A user should be able to view logs by channel and date in text, html, and json.

netcat is not part of a base linux, you should make it use /dev/tcp.

More information can be found here:

http://www.linuxjournal.com/content/more-using-bashs-built-devtcp-file-tcpip

@zellio

Strictly speaking case-invariant compare is insufficient for irc nicks because of how {|} should equal [/] respectively.

There are many places in the code that need to be changed to use normalized nicks.

It may be wise to attempt to normalize them when they are read in with a preprocessor.