solr-brute-force-index-sync.rb

#
# Experimenting with a brute force check to see if your Solr index is in sync
# with your database. This example uses Sunspot for the #solr_index method,
# but not for the searching. We query Solr and the local database directly
# to page through this stuff as quickly as possible.
#
# Got a better approach? Ping @websolr on Twitter :)
#

start = 0
rows  = 100

while start < Post.count

  search_url = "http://index.websolr.com/solr/a1b2c3d4e5f/select?wt=ruby&q=*:*&rows=#{rows}&start=#{start}"
  results = eval(open(search_url).string)
  ids = results["response"]["docs"].collect{|d|d["id"][/[0-9]+/].to_i}

  unless Post.count(:conditions => { :id => ids }) == rows
    Post.find(:all, :conditions => [ "id not in (?)", ids ]).each do |post|
      puts "Didn't find Post #{post.id}, re-indexing..."
      post.solr_index
    end
  end

  start = start + rows

end