bot: Try to strip CouchDB

bot/bot.rb

@@ -7,7 +7,6 @@
 require File.expand_path('../command_patterns', __FILE__)
 require File.expand_path('../finish_notifier', __FILE__)
 require File.expand_path('../pipeline_notifier', __FILE__)
-require File.expand_path('../../lib/couchdb', __FILE__)
 
 opts = Trollop.options do
   opt :server, 'IRC server, expressed as a URI (irc://SERVER:PORT or ircs://SERVER:PORT for SSL)', :type => String
@@ -17,8 +16,6 @@
   opt :schemes, 'Comma-separated list of acceptable URI schemes', :default => 'http,https,ftp'
   opt :redis, 'URL of Redis server', :default => ENV['REDIS_URL'] || 'redis://localhost:6379/0'
   opt :password, 'IRC server password', :default => nil, :type => String
-  opt :db, 'URL of CouchDB database', :default => ENV['COUCHDB_URL'] || 'http://localhost:5984/archivebot'
-  opt :db_credentials, 'Credentials for CouchDB database (USERNAME:PASSWORD)', :type => String, :default => nil
 end
 
 redis = Redis.new(:url => opts[:redis])
@@ -54,8 +51,7 @@
     c.messages_per_second = 1.0
   end
 
-  couchdb = Couchdb.new(URI(opts[:db]), opts[:db_credentials])
-  brain = Brain.new(schemes, redis, couchdb)
+  brain = Brain.new(schemes, redis)
 
   on :message, CommandPatterns::ARCHIVE do |m, target, params|
     brain.request_archive(m, target, params)

bot/brain.rb

@@ -1,4 +1,5 @@
 require 'addressable/uri'
+require 'json'
 
 require File.expand_path('../../lib/job', __FILE__)
 require File.expand_path('../../lib/pipeline_info', __FILE__)
@@ -17,13 +18,11 @@ class Brain
   include AddIgnoreSets
   include PipelineOptions
 
-  attr_reader :couchdb
   attr_reader :redis
   attr_reader :schemes
   attr_reader :url_pattern
 
-  def initialize(schemes, redis, couchdb)
-    @couchdb = couchdb
+  def initialize(schemes, redis)
     @redis = redis
     @schemes = schemes
     @url_pattern ||= %r{(?:#{schemes.join('|')})://.+}
@@ -125,7 +124,20 @@ def request_archive(m, target, params, depth=:inf, url_file=false)
 
     if h[:user_agent_alias]
       ua_alias = h[:user_agent_alias]
-      user_agent = couchdb.user_agent_for_alias(ua_alias)
+
+      user_agents = {}
+
+      Dir.glob(File.join(File.expand_path('../../db/user_agents', __FILE__), "*.json")).each do |file|
+        data = JSON.parse(File.read(file))
+
+        data["agents"].each do |agent|
+          agent["aliases"].each do |alias_name|
+            user_agents[alias_name] = agent["name"]
+          end
+        end
+      end
+
+      user_agent = user_agents[ua_alias]
 
       if !user_agent
         reply m, %Q{Sorry, I don't know what the user agent "#{ua_alias}" is.}
@@ -260,7 +272,23 @@ def add_ignore_sets(m, job, names, need_authorization=true)
 
     return unless names && !names.empty?
 
-    ignore_pairs = couchdb.resolve_ignore_sets(names)
+    unknown = []
+    ignore_pairs = []
+
+    names.each do |name|
+      file = File.join(File.expand_path('../../db/ignore_patterns', __FILE__), "#{name}.json")
+
+      if File.file?(file)
+        data = JSON.parse(File.read(file))
+
+        data.fetch('patterns', []).map do |pattern|
+          ignore_pairs << [name, pattern]
+        end
+      else
+        unknown << name
+        []
+      end
+    end
 
     resolved = ignore_pairs.map(&:first).uniq
     patterns = ignore_pairs.map(&:last)
@@ -273,8 +301,6 @@ def add_ignore_sets(m, job, names, need_authorization=true)
       reply m, %Q{Added #{patterns.length} patterns from ignore set #{resolved.first} to job #{job.ident}.}
     end
 
-    unknown = names - resolved
-
     if !unknown.empty?
       reply m, "The following sets are unknown: #{unknown.join(', ')}"
     end

cogs/start.rb

@@ -2,8 +2,6 @@
 require 'trollop'
 require 'uri'
 
-require File.expand_path('../ignore_pattern_updater', __FILE__)
-require File.expand_path('../user_agent_updater', __FILE__)
 require File.expand_path('../../lib/job', __FILE__)
 require File.expand_path('../../lib/redis_subscriber', __FILE__)
 require File.expand_path('../../lib/shared_config', __FILE__)
@@ -35,16 +33,6 @@ def on_receive(ident)
 Reaper.supervise_as :reaper, opts[:redis]
 TwitterTweeter.supervise_as :twitter_tweeter, opts[:redis], opts[:twitter_config]
 
-ignore_patterns_path = File.expand_path('../../db/ignore_patterns', __FILE__)
-
-IgnorePatternUpdater.supervise_as :ignore_pattern_updater,
-  ignore_patterns_path, db_uri, opts[:db_credentials]
-
-user_agents_path = File.expand_path('../../db/user_agents', __FILE__)
-
-UserAgentUpdater.supervise_as :user_agent_updater,
-  user_agents_path, db_uri, opts[:db_credentials]
-
 # This should be started after all actors it references have started, which is
 # why it's the last actor to start up.
 if opts[:twitter_config]
@@ -55,8 +43,6 @@ def on_receive(ident)
   if opts[:twitter_config]
     Celluloid::Actor[:broadcaster].stop
   end
-  Celluloid::Actor[:ignore_pattern_updater].stop
-  Celluloid::Actor[:user_agent_updater].stop
 end
 
 trap('INT') do