generate_xml

#!/usr/bin/env ruby
require 'date'
require 'rubygems'
require 'activesupport'
require 'htmlentities'

# are there maybe dependency problems in the tweetparser gem?
# Why do I have to do this?
gem 'polyglot','=0.2.9'
gem 'treetop','=1.4.2'
gem 'tweetparser'
require 'tweetparser'

class String
  def to_xml
    unpack('U*').map {|n| 
      case n
        when 38
          "&"
        when 60
          "<"
        when 62
          ">"
        when 1..127
          n.chr
        else
          "&##{n};"
      end
    }.join
  end
  def from_html
    coder = HTMLEntities.new
    coder.decode(self)
  end
end

def translate(text)
  parsed = TweetParser.parse(text)
  parsed.map { |sexp|
    type, token = *sexp
    token = token.from_html.to_xml
    case type
      when :url
        "<link>"+token+"</link>"
      when :username
        "<name>"+token+"</name>"
      when :hashtag
        "<hashtag>"+token+"</hashtag>"
      else
        token
      end
  }.join("")
end

cur_month = ""
cur_day = ""

puts "<tweets>"
while (tweet = gets) 
    id, date,text = tweet.split("\t")
    date = DateTime.strptime(date,"%Y-%m-%dT%H:%M:%S%z")
    month = date.strftime("%B %Y")
    day = date.strftime("%A #{date.day.ordinalize} %B")
    time = date.strftime("%I:%M%p").downcase
    
    new_month = (cur_month != month)
    new_day = (cur_day != day)
    
    if new_day
       puts "</day>" unless cur_day == ""
    end
    
    if new_month
        puts "</month>" unless cur_month == ""
        puts "<month name=\"#{month}\">"
        cur_month = month
    end
    
    if new_day
      puts "<day name=\"#{day}\">"
      cur_day = day
    end

    puts "<tweet>#{time}&#160;&#160;#{translate(text)}</tweet>"
end

puts "</day>"
puts "</month>"
puts "</tweets>"