nadoka 0.8.0 → 0.8.1

data/{README.org → README.md}

@@ -1,16 +1,11 @@
-#######################################################
 # Nadoka: IRC Client server program.
-#
-# Written by SASADA Koichi <ko1 at atdot.net>
-#
-#
-# $Id$
-#
 
-* What's this?
+Written by SASADA Koichi <ko1 at atdot.net>
+
+## What's this?
 
 Nadoka is IRC Client Server program.
-It's same concept as [[http://www.madoka.org/][madoka]].
+It's same concept as [madoka](http://www.madoka.org/).
 
 You can do with this software:
 
@@ -18,14 +13,10 @@ You can do with this software:
 - easy to make a bot with ruby
 
 
-* more about
+## more about
 
 See Web pages:
+
 - https://github.com/nadoka/nadoka
 - http://rubyforge.org/projects/nadoka/
 - http://www.atdot.net/nadoka/
-
-
-: Thank you,
-: --
-: SASADA Koichi at atdot dot net

data/nadoka.rb

@@ -16,14 +16,6 @@
 # Create : K.S. 03/07/10 20:29:07
 #
 
-unless "".respond_to?(:force_encoding)
-  class String
-    def force_encoding(enc)
-      self
-    end
-  end
-end
-
 $LOAD_PATH.unshift File.dirname(__FILE__)
 require 'ndk/version'
 

data/ndk/config.rb

@@ -58,7 +58,7 @@ module Nadoka
     Away_Message = 'away'
     Away_Nick    = nil
 
-    Quit_Message = "Quit Nadoka #{::Nadoka::NDK_Version} - http://www.atdot.net/nadoka/"
+    Quit_Message = "Quit Nadoka #{::Nadoka::NDK_Version}"
     
     #
     Channel_info = {}
@@ -418,7 +418,11 @@ module Nadoka
 
     def identical_channel_name ch
       # use 4 gsub() because of the compatibility of RFC2813(3.2)
-      ch.toeuc.downcase.tr('[]\\~', '{}|^').tojis.force_encoding('ASCII-8BIT')
+      ch = ch.toeuc.downcase.tr('[]\\~', '{}|^').tojis
+      if ch.respond_to?(:force_encoding)
+        ch.force_encoding('ASCII-8BIT')
+      end
+      ch
     end
 
     RName = {        # ('&','#','+','!')

data/ndk/version.rb

@@ -11,12 +11,21 @@
 #
 
 module Nadoka
-  VERSION  = '0.8.0'
+  VERSION  = '0.8.1'
   NDK_Version  = VERSION.dup
   NDK_Created  = Time.now
 
   if File.directory?(File.expand_path('../../.git', __FILE__))
-    NDK_Version.concat("+git")
+    git_describe = nil
+    Dir.chdir(File.expand_path('../..', __FILE__)) do
+      git_describe = `git describe --tags --long --dirty=-dt`
+      if $?.success?
+        git_describe = "(#{git_describe.strip})"
+      else
+        git_describe = nil
+      end
+    end
+    NDK_Version.concat("+git#{git_describe}")
   end
   if /trunk/ =~ '$HeadURL$'
     NDK_Version.concat('-trunk')

data/plugins/googlebot.nb

@@ -162,6 +162,7 @@ class GoogleBot < Nadoka::NDK_Bot
       e = result["responseData"]["results"][0]
       url   = e['unescapedUrl'] || e['url'] || e['postUrl']
       title = show_char_code_and_erase_tag(e['titleNoFormatting'])
+      url = shorten_url(url)
       "#{title} - #{url} (and #{count} hit#{(count.to_i > 1) ? 's' : ''})"
     else
       "no match"
@@ -342,4 +343,14 @@ class GoogleBot < Nadoka::NDK_Bot
       raise
     end
   end
+
+  def shorten_url(url)
+    case url
+    when %r!\Ahttp://www\.amazon\.co\.jp/.*(/dp/.+)\z!
+      "http://amazon.jp#{$1}"
+    else
+      # default: do nothing
+      url
+    end
+  end
 end

data/plugins/titlebot.nb

@@ -1,6 +1,7 @@
 # -*-ruby-*-
+# vim:set ft=ruby:
 #
-# Copyright (c) 2009, 2011 Kazuhiro NISHIYAMA
+# Copyright (c) 2009, 2011, 2012 Kazuhiro NISHIYAMA
 #
 # This program is free software with ABSOLUTELY NO WARRANTY.
 # You can re-distribute and/or modify this program under
@@ -19,10 +20,14 @@ Reply title of URL.
     :name => :TitleBot,
     :ch => //,
     :timeout => 10,
+    :headers => {
+      #"User-Agent" => "Ruby/#{RUBY_VERSION}",
+    }
   }
 
 =end
 
+require 'iconv'
 require 'nkf'
 require 'open-uri'
 require 'timeout'
@@ -37,7 +42,7 @@ end
 module URL2Title
   module_function
 
-  def get_title(url)
+  def get_title(url, headers)
     uri = URI(url)
     info = { :uri => uri }
     info[:errors] = []
@@ -46,7 +51,10 @@ module URL2Title
       info[:title] = "(ignored)"
       return info
     end
-    uri.open(:content_length_proc => proc{|x| raise Errno::EFBIG if x && x > 1048576}) do |f|
+    headers = {
+      :content_length_proc => proc{|x| raise Errno::EFBIG if x && x > 1048576},
+    }.merge!(headers)
+    uri.open(headers) do |f|
       info[:uri] = f.base_uri
       body = f.read
 
@@ -71,15 +79,52 @@ module URL2Title
           body = Zlib::GzipReader.new(StringIO.new(body)).read || ''
         end
 
+
         # encoding
-        if NKF.guess(body) == NKF::BINARY
-          info[:title] = "(binary)"
-          return info
+        begin
+          if charset
+            # ok
+          elsif /charset=[\'\"]?([^;\'\">]+)/ni =~ body
+            charset = $1
+          end
+          case charset
+          when /shift_jis/i
+            charset = "cp932"
+          when /euc-jp/i # euc-jp, x-euc-jp
+            charset = "eucjp-ms"
+          end
+          if /\Autf-8\z/i =~ charset
+            # avoid #<ArgumentError: invalid byte sequence in UTF-8>
+            # or Iconv::IllegalSequence
+            body = NKF.nkf("-Wwm0x", body)
+          elsif charset
+            charset.sub!(/\Ax-?/i, '')
+            body = Iconv.conv("utf-8", charset, body)
+          else
+            body = NKF.nkf("-wm0x", body)
+          end
+        rescue Iconv::IllegalSequence => e
+          info[:errors] << e
+        rescue Iconv::InvalidEncoding => e
+          info[:errors] << e
         end
-        body = NKF.nkf("-wm0x --numchar-input", body)
 
         title = nil
         case uri.host
+        when /\A(?:twitter\.com)\z/
+          if %r"<title\b(?>[^<>]*)>(.*?)</title(?>[^<>]*)>"miu =~ body
+            title = $1
+          end
+          if defined?(::Nokogiri)
+            doc = Nokogiri::HTML(body, uri.to_s, 'utf-8')
+            tweet = doc.css('.tweet-text')
+            unless tweet.empty?
+              tweet = tweet.inner_html
+              tweet.gsub!(/<.*?>/, '')
+              tweet.strip!
+              title = tweet unless tweet.empty?
+            end
+          end
         when /\A(?:www\.so-net\.ne\.jp)\z/
           if %r"<title\b(?>[^<>]*)>(.*?)</title(?>[^<>]*)>"miu =~ body
             title = $1
@@ -138,17 +183,23 @@ module URL2Title
     url.sub(/\/\#!\//, '/')
   end
 
-  def url2title(url)
+  def url2title(url, headers)
     url = prepare_url(url)
-    info = get_title(url)
+    info = get_title(url, headers)
     info[:title] = truncate(info[:title])
     info
   end
 end
 
 if __FILE__ == $0
+  require File.expand_path('../../ndk/version', __FILE__)
   def u2t(url)
-    URL2Title.url2title(url)
+    firefox = "Mozilla/5.0 (X11; Ubuntu; Linux x86_64; rv:12.0) Gecko/20100101 Firefox/12.0"
+    ua = "NadokaTitleBot/#{Nadoka::VERSION}"
+    headers = {
+      "User-Agent" => "#{firefox} Ruby/#{RUBY_VERSION} #{ua}",
+    }
+    URL2Title.url2title(url, headers)
   rescue
     $!.inspect
   end
@@ -179,9 +230,10 @@ class TitleBot < Nadoka::NDK_Bot
     end
 
     @same_bot = @bot_config.fetch(:same_bot, /(?!)/)
-    @nkf_options = @bot_config.fetch(:nkf, "--oc=CP50221 --numchar-input --fb-xml")
+    @nkf_options = @bot_config.fetch(:nkf, "--oc=CP50221 --ic=UTF-8 --numchar-input --fb-xml")
     @timeout = @bot_config.fetch(:timeout, 10)
-    @too_long_threshold = @bot_config.fetch(:too_long_threshold, 250)
+    @fragment_size_range = @bot_config.fetch(:fragment_size_range, 5..100)
+    @headers = @bot_config.fetch(:headers, {})
   end
 
   def send_notice(ch, msg)
@@ -195,12 +247,15 @@ class TitleBot < Nadoka::NDK_Bot
   def on_privmsg prefix, ch, msg
     return unless @available_channel === ch
 
-    if /https?:/ === msg
+    case msg
+    when /^\S+>/
+      # ignore messages to other bots
+    when /https?:/
       return if @state.channel_users(ccn(ch)).find{|x| @same_bot =~ x }
 
       url, = URI.extract(msg, ["http", "https"])
       info = Timeout.timeout(@timeout) do
-        url2title(url)
+        url2title(url, @headers)
       end
       return unless info[:title]
       if url != info[:uri].to_s
@@ -209,8 +264,8 @@ class TitleBot < Nadoka::NDK_Bot
         send_notice(ch, "title bot: #{info[:title]}")
       end
       if info[:fragment_text]
-        # ignore when fragment_text is too long
-        if info[:fragment_text].size < @too_long_threshold
+        # ignore when fragment_text is too long or too short
+        if @fragment_size_range === info[:fragment_text].size
           send_notice(ch, "title bot:: #{info[:fragment_text]}")
         end
       end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: nadoka
 version: !ruby/object:Gem::Version
-  version: 0.8.0
+  version: 0.8.1
   prerelease: 
 platform: ruby
 authors:
@@ -10,7 +10,7 @@ authors:
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2012-02-29 00:00:00.000000000 Z
+date: 2012-05-29 00:00:00.000000000 Z
 dependencies: []
 description: Nadoka is a tool for monitoring and logging IRC conversations and responding
   to specially formatted requests. You define and customize these responses in Ruby.
@@ -25,7 +25,7 @@ files:
 - .gitignore
 - ChangeLog.old
 - Gemfile
-- README.org
+- README.md
 - Rakefile
 - bin/nadoka
 - lib/rss_check.rb
@@ -95,7 +95,7 @@ required_rubygems_version: !ruby/object:Gem::Requirement
       version: '0'
 requirements: []
 rubyforge_project: nadoka
-rubygems_version: 1.8.11
+rubygems_version: 1.8.23
 signing_key: 
 specification_version: 3
 summary: IRC logger, monitor and proxy program ("bot")