From c201b07a60ee1a3e669a74db942370186d0939d5 Mon Sep 17 00:00:00 2001
From: Jacob Harris <jharris@nytimes.com>
Date: Tue, 9 Jul 2013 21:54:44 -0400
Subject: [PATCH] Fixes for launch

---
 ebook.rb     | 51 +++++++++++++++++++++++++++++++++++++++++++++++
 ebook.worker |  9 +++++++++
 markov.rb    | 56 ++++++++++++++++++++++++++++++++++++++++++++++++++++
 3 files changed, 116 insertions(+)
 create mode 100644 ebook.rb
 create mode 100644 ebook.worker
 create mode 100644 markov.rb

diff --git a/ebook.rb b/ebook.rb
new file mode 100644
index 0000000..7f92980
--- /dev/null
+++ b/ebook.rb
@@ -0,0 +1,51 @@
+# Worker code can be anything you want.
+require 'rubygems'
+require 'twitter'
+
+require 'twitter_init'
+require 'markov'
+
+source_tweets = []
+
+$rand_limit ||= 10
+
+# randomly running only about 1 in $rand_limit times
+unless rand($rand_limit) == 0
+  puts "Not running this time"
+else
+  # Fetch a thousand tweets
+  begin
+    user_tweets = Twitter.user_timeline($source_account, :count => 200, :trim_user => true, :exclude_replies => false, :include_replies => true)
+    max_id = user_tweets.last.id
+    source_tweets += user_tweets.reject {|t| t.text =~ /(http:\/\/)|(\bRT\b)|(\bMT\b)|@/ }
+  
+    25.times do
+      user_tweets = Twitter.user_timeline($source_account, :count => 200, :trim_user => true, :max_id => max_id - 1, :exclude_replies => false, :include_replies => true)
+      max_id = user_tweets.last.id
+      source_tweets += user_tweets.reject {|t| t.text =~ /(http:\/\/)|(\bRT\b)|(\bMT\b)|@/ }
+    end
+  rescue
+  end
+  
+  puts "#{source_tweets.length} tweets found"
+  
+  markov = MarkovChainer.new(2)
+  
+  source_tweets.each do |twt|
+    markov.add_text(twt.text)
+  end
+  
+  tweet = nil
+  
+  5.times do
+    tweet = markov.generate_sentence
+    break if !source_tweets.any? {|t| t.text == tweet }
+  end
+  
+  puts "TWEET: #{tweet}"
+  
+  unless tweet.nil? || tweet == ''
+    Twitter.update(tweet)
+  end
+end
+
diff --git a/ebook.worker b/ebook.worker
new file mode 100644
index 0000000..33ff648
--- /dev/null
+++ b/ebook.worker
@@ -0,0 +1,9 @@
+# define the runtime language
+runtime "ruby"
+# exec is the file that will be executed:
+exec "ebook.rb"
+
+file "twitter_init.rb"
+file "markov.rb"
+
+gem 'twitter'
diff --git a/markov.rb b/markov.rb
new file mode 100644
index 0000000..f527b4a
--- /dev/null
+++ b/markov.rb
@@ -0,0 +1,56 @@
+class MarkovChainer
+   attr_reader :order
+   def initialize(order)
+     @order = order
+     @beginnings = []
+     @freq = {}
+   end
+
+   def add_text(text)
+     # make sure each paragraph ends with some sentence terminator
+     text.gsub!(/\n\s*\n/m, ".")
+     text << "."
+     seps = /([.!?;])/
+     sentence = ""
+     text.split(seps).each { |p|
+       if seps =~ p
+         add_sentence(sentence, p)
+         sentence = ""
+       else
+         sentence = p
+       end
+     }
+   end
+
+   def generate_sentence
+     res = @beginnings[rand(@beginnings.size)]
+     return nil if res.nil?
+     loop {
+       unless nw = next_word_for(res[-order, order])
+         return res[0..-2].join(" ") + res.last
+       end
+       res << nw
+     }
+   end
+
+private
+   def add_sentence(str, terminator)
+     words = str.scan(/[\w'\-]+/)
+     return unless words.size > order # ignore short sentences
+     words << terminator
+     buf = []
+     words.each { |w|
+       buf << w
+       if buf.size == order + 1
+         (@freq[buf[0..-2]] ||= []) << buf[-1]
+         buf.shift
+       end
+     }
+     @beginnings << words[0, order]
+   end
+
+   def next_word_for(words)
+     arr = @freq[words]
+     arr && arr[rand(arr.size)]
+   end
+end