# Run with: merb -a runner -r lib/workers/scorer.rb

DataMapper.read_from!(:job)

class Scorer < NfsLockable
  def transaction
    Merb.logger.info "Calculating popularity scores"

    # Vars
    scored_cutoff = 1.months.ago.to_date #Older than this is not scored, set to 0
    cutoff = 1.week.ago.to_date #Events older than this do not apply to their score
    new_profile_cutoff = 1.day.ago.to_date #Profiles older than this do not get a boost
    base_popularity = 180

    # Zero out everyone who hasn't logged in in a month
    Merb.logger.info "Zeroing out dormant profiles"
    Profile.all(:last_login.lte => scored_cutoff).update!(:popularity => base_popularity, :paid_popularity => base_popularity )

    #Construct scoring query
    log_repo_name = repository(:logging).adapter.uri.path.split("/").last
    sql = <<SQL
SELECT profile_id,
  new_profile_score + paid_score + activated_score + logged_in_score + recent_messages_received_score + recent_replies_score +
  ifnull( reply_rate_score, 0)  + ifnull( spam_score, 0) +
  recent_profile_views_score + public_photos_score + private_photos_score + recent_photo_upload_score + blocks_score + base_score as popularity,
  paid_modifier FROM
  ( SELECT profile_id,
  is_new_profile * 50 as new_profile_score,
  is_paid * 0 as paid_score,
  is_activated * 10 as activated_score,
  -- If has logged in within the last week, rank higher than those within last month. Older than that are zeroed.
  has_logged_in * 15 as logged_in_score,
  -- Recent inbound messages indicate good profiles
  LEAST( messages_received * 2, 20 ) as recent_messages_received_score,
  -- Recent outbound replies is good for women
  LEAST( messages_replied * 20, 50 ) * gender as recent_replies_score,
  -- Rate of reply to incoming messages for all time is important for women
  LEAST( messages_replied / messages_received * messages_replied * 2, 50 ) * gender as reply_rate_score,
  -- To punish spammers, rate of reply to your messages is important
  -- If less than 16 messages sent for every 1 received penalize
  - LEAST( GREATEST( messages_sent / responses_to_messages_sent - 15, 0 ) * 10, 70 ) as spam_score,
  -- Recent profile views
  LEAST( profile_views * 1, 30 ) as recent_profile_views_score,
  -- Current Public Photo Count
  LEAST( public_photos * 5, 15 ) as public_photos_score,
  -- Current Private Photo Count
  LEAST( private_photos * 10, 10 ) as private_photos_score,
  -- Photos uploaded in last week
  LEAST( photos_uploaded * 10, 20 ) as recent_photo_upload_score,
  -- Profiles with more than one block indicate poor profile quality and/or spammers
  - LEAST( GREATEST( blocks - 1, 0 ) * 30, 60) as blocks_score,
  -- Sphinx doesn't allow negative numbers so adjust all up to avoid
  -- Spam penalty (70) + Block Penalty (60) + Paid Modifier (50)
  180 as base_score,
  -- If searched by a paid member, demote bot results
  is_bot * -50 AS paid_modifier
  FROM
  -- Raw data for use in scoring
  (SELECT
   profiles.id as profile_id,
   profiles.gender = 'f' as gender,
   profiles.created_at > '#{new_profile_cutoff}' AS is_new_profile,
   profiles.subscription_status IN ('reactivated_subscription', 'currently_subscribed') AS is_paid,
   users.user_type = 'decoy' AS is_bot,
   profiles.last_login > '#{cutoff}' AS has_logged_in,
   profiles.activated_at IS NOT NULL as is_activated,
   site_id as site,
   -- Messages received in last week from distinct senders (reward freshness)
   ( SELECT count( DISTINCT messages.sender_id) FROM messages WHERE recipient_id = profiles.id AND messages.created_at > '#{cutoff}') AS messages_received,
   -- Messages replied to in last week with distinct reciever (reward freshness)
   ( SELECT count( DISTINCT messages.recipient_id ) FROM messages WHERE sender_id = profiles.id AND messages.created_at > '#{cutoff}' AND in_reply_to_id IS NOT NULL) AS messages_replied,
   -- People who have viewed their profile in the last week (reward freshness)
   ( SELECT count( DISTINCT profile_interactions.profile_id) FROM #{log_repo_name}.profile_interactions WHERE with_id = profiles.id AND action = 'viewed_profile_of' AND
 profile_interactions.at > '#{cutoff}' ) as profile_views,
   -- Responses (for calculating response rate for messages sent to them)
   ( SELECT count( DISTINCT recipient_id ) FROM messages WHERE sender_id = profiles.id AND in_reply_to_id IS NOT NULL AND thread_id IN ( SELECT messages.id FROM messages WHERE recipient_id = profiles.id AND in_reply_to_id IS NULL AND type != 'AccessGrantedMessage' ) ) as responses_to_inbound_contacts,
   -- Initial Contacts (for calculating response rate for messages sent to them)
   ( SELECT count( DISTINCT messages.sender_id ) FROM messages WHERE recipient_id = profiles.id AND in_reply_to_id IS NULL AND type != 'AccessGrantedMessage' ) as inbound_contacts,
   -- Messages sent (for calculating response rate for messages they sent)
    ( SELECT count(messages.id) FROM messages WHERE sender_id = profiles.id AND in_reply_to_id IS NULL) as messages_sent,
   -- Messages sent that are responded to (for calculating response rate for messages they sent)
    ( SELECT count(DISTINCT inbound.thread_id) FROM messages outbound, messages inbound WHERE outbound.sender_id = profiles.id AND outbound.in_reply_to_id IS NULL AND outbound.id = inbound.thread_id AND inbound.thread_id != inbound.id ) as responses_to_messages_sent,
   -- Approved Public Photos
   ( SELECT count(photos.id) FROM photos WHERE profile_id = profiles.id AND private = 0 AND moderation != 'd') as public_photos,
   -- Approved Private Photos
   ( SELECT count(photos.id) FROM photos WHERE profile_id = profiles.id AND private = 1 AND moderation != 'd') as private_photos,
   -- Photos uploaded in last week (reward freshness)
   ( SELECT count(photos.id) FROM photos WHERE profile_id = profiles.id AND moderation != 'd' AND photos.created_at > '#{cutoff}') as photos_uploaded,
   -- Blocks (penalize)
   ( SELECT count(profile_blocks.block_id) FROM profile_blocks WHERE block_id = profiles.id ) as blocks
   FROM profiles
   JOIN users ON profiles.user_id = users.id
   WHERE profiles.last_login > '#{scored_cutoff}' AND deleted_at IS NULL AND profiles.status != 'fraud') as scores
   ) as popularity_scores
SQL

    # Get all scores from slave
    Merb.logger.info "Calculating Scores"
    scores = Profile.repository.adapter(true).query(sql)

    # Update master
    Merb.logger.info "Updating Scores"
    scores.in_groups_of(1000) do |score_group|
      score_group.compact.each do |score|
        Profile.all(:id => score.profile_id).update!(:popularity => score.popularity, :paid_popularity => score.popularity + score.paid_modifier)
      end
      Merb.logger.info "... updating scores ..."
      sleep(5)
    end
    Merb.logger.info "Score Update Complete"
  end
end

Scorer.new().execute