sort of merge cache branch but don't do anything automatically

allow manual caching of story text using diffbot, if an api key is
configured and Story#fetch_story_cache! is called

app/models/story.rb

@@ -132,7 +132,8 @@ class Story < ActiveRecord::Base
   end
 
   def log_moderation
-    if self.new_record? || self.editor_user_id == self.user_id
+    if self.new_record? || !self.editor_user_id ||
+    self.editor_user_id == self.user_id
       return
     end
 
@@ -234,6 +235,12 @@ class Story < ActiveRecord::Base
     @fetched_content
   end
 
+  def fetch_story_cache!
+    if self.url.present?
+      self.story_cache = StoryCacher.get_story_text(self.url)
+    end
+  end
+
   def calculated_hotness
     order = Math.log([ score.abs, 1 ].max, 10)
     if score > 0

db/migrate/20121004153529_add_story_text.rb

@@ -0,0 +1,9 @@
+class AddStoryText < ActiveRecord::Migration
+  def up
+    add_column :stories, :story_cache, :text
+  end
+
+  def down
+    remove_column :stories, :story_cache
+  end
+end

db/schema.rb

@@ -89,6 +89,7 @@ ActiveRecord::Schema.define(:version => 20130526164230) do
     t.integer  "is_moderated",           :limit => 1,                                        :default => 0,   :null => false
     t.decimal  "hotness",                                    :precision => 20, :scale => 10, :default => 0.0, :null => false
     t.text     "markeddown_description", :limit => 16777215
+    t.text     "story_cache"
   end
 
   add_index "stories", ["hotness"], :name => "hotness_idx"

extras/story_cacher.rb

@@ -0,0 +1,35 @@
+class StoryCacher
+  cattr_accessor :DIFFBOT_API_KEY
+
+  # this needs to be overridden in config/initializers/production.rb
+  @@DIFFBOT_API_KEY = nil
+
+  DIFFBOT_API_URL = "http://www.diffbot.com/api/article"
+
+  def self.get_story_text(url)
+    if !@@DIFFBOT_API_KEY
+      return
+    end
+
+    db_url = "#{DIFFBOT_API_URL}?token=#{@@DIFFBOT_API_KEY}&url=" <<
+      CGI.escape(url)
+
+    begin
+      s = Sponge.new
+      # we're not doing this interactively, so take a while
+      s.timeout = 20
+      res = s.fetch(db_url)
+      if res.present?
+        j = JSON.parse(res)
+
+        # turn newlines into double newlines, so they become paragraphs
+        return j["text"].gsub("\n", "\n\n")
+      end
+
+    rescue => e
+      Rails.logger.error "error fetching #{db_url}: #{e.message}"
+    end
+
+    nil
+  end
+end