Search + delete for Solr

search_and_delete.rb

require 'rubygems'
require 'rest_client'
require 'nokogiri'

ROOT = 'http:/SOLRHOST:8983'



# Colorful strings

module Colourize
colour_codes = {
  :black => 30,   :white => 37,   :red => 31,     :green => 32,
  :yellow => 33,  :blue => 34,    :magenta => 35,  :cyan => 36
}
  colour_codes.each do |colour, colour_code|
    send :define_method, colour do
      ansi = "\e[#{colour_code}m"; reset = "\e[0m"
      colourized_string = ansi + self + reset
      colourized_string
    end
  end
end

class String
  include Colourize
end



# World's tiniest, toughest Solr client

class Solr
  require 'open-uri'
  require 'ostruct'
  require 'rubygems'
  require 'nokogiri'

  def initialize
    @xpaths = {
      :pid => '//str[@name="id"]',
      :title => '//arr[@name="title_t"]/str[1]',
      :url => '//arr[@name="url"]/*',
    }
  end

  # Search, returns array of ostructs
  def search
    query = "#{ROOT}/solr/select?q=#{ARGV[0]}&rows=50000&qt=tki"
    begin
      raw_results = Nokogiri::XML(open(URI.encode(query)))
    rescue SystemCallError, EOFError
      retry
    end
    results = []
    raw_results.xpath('//doc').each do |doc|
      doc = Nokogiri::XML.parse(doc.to_s) # confused... inefficient...
      item = OpenStruct.new
      # Populate struct objects with name/matches from @xpaths
      @xpaths.each do |k,v|
        item.send("#{k}=", doc.xpath(v).text)
      end
      results << item
    end
    results
  end

  # Update solr
  def update(record, uri="#{ROOT}/solr/update/?commit=true")
    begin
      response = RestClient.post uri, record, :content_type => 'text/xml; charset=utf-8'
      puts response.blue
    rescue
      warn "SolrHelpers.update failed to update #{uri}".red
    end
  end

  # Delete record from solr
  def delete(id)
    puts "Deleting #{id}...".red
    delete_reference = "<delete><id>#{id}</id></delete>"
    update(delete_reference)
  end
end





# Find records
solr = Solr.new
results = solr.search

# Print results
results.each do |x|
  puts "#{x.pid.red} - #{x.url.green}"
end


# Offer to remove them from Solr
puts; puts "Delete these records from #{ROOT}?"
puts "Y/N"

decision = $stdin.gets.chomp


# Delete from solr
if decision == 'Y'
  results.each do |x|
    solr.delete(x.pid)
  end
else
  # Or just exit
  puts 'Exiting...'.yellow
end