ichat.rb

puts "*** Loading Objective-C dependencies, this will take a moment"

require 'rubygems'
require 'htmlentities'
require 'json'

framework 'Foundation'
framework 'AppKit'

class Person
  def initWithCoder decoder
  end
end

class Presentity
  def initWithCoder decoder
    if decoder.allowsKeyedCoding
      @service, @sender = decoder.decodeObjectForKey(:ServiceName), decoder.decodeObjectForKey(:ID)
    else
      @service, @sender = decoder.decodeObject, decoder.decodeObject
    end
    [@service, @sender]
  end
end

class InstantMessage
  def initWithCoder decoder
    if decoder.allowsKeyedCoding
      @sender = decoder.decodeObjectForKey(:Sender)
      @when = decoder.decodeObjectForKey(:Time)
      @text = decoder.decodeObjectForKey(:MessageText)
      @flags = decoder.decodeObjectForKey(:Flags)
    else
      @sender = decoder.decodeObject
      @when = decoder.decodeObject
      @text = decoder.decodeObject
      @flags = Pointer.new_with_type('I')
      decoder.decodeValueOfObjCType('I', at: @flags)
    end
    attributes = {NSDocumentTypeDocumentAttribute => NSHTMLTextDocumentType, NSExcludedElementsDocumentAttribute => ['doctype', 'xml', 'html', 'head', 'body']}
    mhtml = @text.nil? ? '' : @text.dataFromRange(NSMakeRange(0, @text.length), documentAttributes: attributes, error: nil)
    [@sender, @when, NSString.stringWithUTF8String(mhtml)]
  end
end

class GroupchatMessage < InstantMessage
end

if __FILE__ == $0
  start = Time.now
  $stderr.puts "*** Scanning directories"
  logs = ARGV.collect{|d| Dir.glob("#{File.join d, '**', '*.ichat'}")}.flatten.sort{|a, b| File.basename(File.dirname(a)) <=> File.basename(File.dirname(b))}
  $stderr.puts "*** Preparing to parse #{logs.count} chat logs"
  logs.collect! do |f|
    $stderr.puts "*** Reading: #{File.basename f, File.extname(f)}"
    contents = File.open(f, 'rb').read
    o = (contents[0,8] == 'bplist00' ? NSKeyedUnarchiver : NSUnarchiver).unarchiveObjectWithData(contents.to_data)
    {
      service: o.first,
      participants: (o[3].is_a?(Array) ? o[3] : [o[3]]).collect {|(service, user)| {service: service, user: user}},
      date: File.basename(File.dirname(f)),
      messages: o[2].collect do |presentity, time, msg|
        h = {
          type: (presentity.nil? ? :system : :user),
          time: time,
          msg:  HTMLEntities.new.decode(msg.gsub(%r{</?[^>]+?>}, '').strip)
        }

        # Get /me messages
        me = h[:msg].match(/\A\/me\s*(.+)\z/)
        h[:type], h[:msg] = :action, me[1] if !me.nil? && !me[1].nil?
        h[:service], h[:user] = presentity.first, presentity.last unless presentity.nil?

        h
      end
    }
  end

  $stderr.puts "*** Dumping to stdout, wait a second"
  $stdout.write JSON.pretty_generate(logs)
  $stderr.puts "*** Done in #{Time.now - start} seconds"
end