Search code examples
ruby-on-railsruby-on-rails-3sunspotsunspot-rails

How to use sunspot_rails gem to search for related articles


I have a mini blog app and i would like user to view articles that relates to what they are reading in the article show page. without the sunspot_rails gem i would do something like this

in my model

  def self.related_search(query, join = "AND")
    find(:all, :conditions => related_search_conditions(query, join))
  end

  def self.related_search_conditions(query, join)
    query.split(/\s+/).map do |word|
      '(' + %w[name description notes].map { |col| "#{col} LIKE #{sanitize('%' + word.to_s + '%')}" }.join(' OR ') + ')'
    end.join(" #{join} ")
  end

then in my view it would be like this

@article.related_search

but i want to use the sunspot_rails gem to make this way easy. Any help. Thanks


Solution

  • As RocketR mentions, this is a trivial use case for Sunspot.

    First, use Sunspot to specify that you have three fields to be indexed as text.

    class Article < ActiveRecord::Base
      searchable do
        text :name
        text :description
        text :notes
      end
    end
    

    Then issue a search, likely from within a controller action. The @search object below contains metadata about the search response, including the matching objects under its results method.

    @search = Article.search do
      keywords query
    end
    @results = @search.results
    

    To find other documents that are similar to an object you already have loaded, say in a show action, you can call the more_like_this instance method. This is a special kind of search, which uses Solr's "More Like This" functionality, and which returns a search object similar to the above full-text search. You can use its results method to render the results of that search.

    <%= render @article.more_like_this.results %>
    

    The more_like_this method also accepts a block with similar options to the search block, so you can have more control over how you're judging similarity.

    Hope that helps!