tire 0.4.0 → 0.4.1

data/lib/tire.rb

@@ -6,6 +6,7 @@ require 'cgi'
 
 require 'active_support/core_ext/object/to_param'
 require 'active_support/core_ext/object/to_query'
+require 'active_support/core_ext/hash/except.rb'
 
 # Ruby 1.8 compatibility
 require 'tire/rubyext/ruby_1_8' if defined?(RUBY_VERSION) && RUBY_VERSION < '1.9'
@@ -23,6 +24,7 @@ require 'tire/search/sort'
 require 'tire/search/facet'
 require 'tire/search/filter'
 require 'tire/search/highlight'
+require 'tire/search/scan'
 require 'tire/results/pagination'
 require 'tire/results/collection'
 require 'tire/results/item'

data/lib/tire/dsl.rb

@@ -31,5 +31,19 @@ module Tire
       Index.new(name, &block)
     end
 
+    def scan(names, options={}, &block)
+      Search::Scan.new(names, options, &block)
+    end
+
+    def aliases
+      @response = Configuration.client.get "#{Configuration.url}/_aliases"
+      MultiJson.decode(@response.body).inject({}) do |acc, (index, value)|
+        next acc if value['aliases'].empty?
+
+        acc[index] = value['aliases'].keys
+        acc
+      end
+    end
+
   end
 end

data/lib/tire/index.rb

@@ -8,39 +8,81 @@ module Tire
       instance_eval(&block) if block_given?
     end
 
+    def url
+      "#{Configuration.url}/#{@name}"
+    end
+
     def exists?
-      @response = Configuration.client.head("#{Configuration.url}/#{@name}")
+      @response = Configuration.client.head("#{url}")
       @response.success?
 
     ensure
-      curl = %Q|curl -I "#{Configuration.url}/#{@name}"|
+      curl = %Q|curl -I "#{url}"|
       logged('HEAD', curl)
     end
 
     def delete
-      @response = Configuration.client.delete "#{Configuration.url}/#{@name}"
+      @response = Configuration.client.delete url
       @response.success?
 
     ensure
-      curl = %Q|curl -X DELETE "#{Configuration.url}/#{@name}"|
+      curl = %Q|curl -X DELETE #{url}|
       logged('DELETE', curl)
     end
 
     def create(options={})
       @options = options
-      @response = Configuration.client.post "#{Configuration.url}/#{@name}", MultiJson.encode(options)
+      @response = Configuration.client.post url, MultiJson.encode(options)
       @response.success? ? @response : false
 
     ensure
-      curl = %Q|curl -X POST "#{Configuration.url}/#{@name}" -d '#{MultiJson.encode(options)}'|
+      curl = %Q|curl -X POST #{url} -d '#{MultiJson.encode(options)}'|
       logged('CREATE', curl)
     end
 
+    def add_alias(alias_name, configuration={})
+      payload = {'actions' => [ {'add' => {'index' => @name, 'alias' => alias_name}.merge(configuration) } ]}
+      @response = Configuration.client.post "#{Configuration.url}/_aliases", MultiJson.encode(payload)
+      @response.success?
+
+    ensure
+      curl = %Q|curl -X POST "#{Configuration.url}/_aliases" -d '#{MultiJson.encode(payload)}'|
+      logged('POST', curl)
+    end
+
+    def remove_alias(alias_name)
+      payload = {'actions' => [{'remove' => {'index' => @name, 'alias' => alias_name}}]}
+      @response = Configuration.client.post "#{Configuration.url}/_aliases", MultiJson.encode(payload)
+      @response.success?
+
+    ensure
+      curl = %Q|curl -X POST "#{Configuration.url}/_aliases" -d '#{MultiJson.encode(payload)}'|
+      logged('POST', curl)
+    end
+
+    def aliases(alias_name = nil)
+      @response = Configuration.client.get "#{url}/_aliases"
+      if alias_name
+        MultiJson.decode(@response.body)[@name]['aliases'][alias_name]
+      else
+        MultiJson.decode(@response.body)[@name]['aliases'].keys
+      end
+
+    ensure
+      curl = %Q|curl "#{url}/_aliases?pretty"|
+      logged('GET', curl)
+    end
+
     def mapping
-      @response = Configuration.client.get("#{Configuration.url}/#{@name}/_mapping")
+      @response = Configuration.client.get("#{url}/_mapping")
       MultiJson.decode(@response.body)[@name]
     end
 
+    def settings
+      @response = Configuration.client.get("#{url}/_settings")
+      MultiJson.decode(@response.body)[@name]['settings']
+    end
+
     def store(*args)
       document, options = args
       type = get_type_from_document(document)
@@ -53,7 +95,7 @@ module Tire
       id       = get_id_from_document(document)
       document = convert_document_to_json(document)
 
-      url  = id ? "#{Configuration.url}/#{@name}/#{type}/#{id}" : "#{Configuration.url}/#{@name}/#{type}/"
+      url  = id ? "#{self.url}/#{type}/#{id}" : "#{self.url}/#{type}/"
       url += "?percolate=#{percolate}" if percolate
 
       @response = Configuration.client.post url, document
@@ -82,7 +124,7 @@ module Tire
       count = 0
 
       begin
-        response = Configuration.client.post("#{Configuration.url}/_bulk", payload.join("\n"))
+        response = Configuration.client.post("#{url}/_bulk", payload.join("\n"))
         raise RuntimeError, "#{response.code} > #{response.body}" if response.failure?
         response
       rescue StandardError => error
@@ -96,7 +138,7 @@ module Tire
         end
 
       ensure
-        curl = %Q|curl -X POST "#{Configuration.url}/_bulk" -d '{... data omitted ...}'|
+        curl = %Q|curl -X POST "#{url}/_bulk" -d '{... data omitted ...}'|
         logged('BULK', curl)
       end
     end
@@ -124,6 +166,17 @@ module Tire
       end
     end
 
+    def reindex(name, options={}, &block)
+      new_index = Index.new(name)
+      new_index.create(options) unless new_index.exists?
+
+      Search::Scan.new(self.name, &block).each do |results|
+        new_index.bulk_store results.map do |document|
+          document.to_hash.except(:type, :_index, :_explanation, :_score, :_version, :highlight, :sort)
+        end
+      end
+    end
+
     def remove(*args)
       if args.size > 1
         type, document = args
@@ -136,7 +189,7 @@ module Tire
       end
       raise ArgumentError, "Please pass a document ID" unless id
 
-      url    = "#{Configuration.url}/#{@name}/#{type}/#{id}"
+      url    = "#{self.url}/#{type}/#{id}"
       result = Configuration.client.delete url
       MultiJson.decode(result.body) if result.success?
 
@@ -149,7 +202,7 @@ module Tire
       raise ArgumentError, "Please pass a document ID" unless id
 
       type      = Utils.escape(type)
-      url       = "#{Configuration.url}/#{@name}/#{type}/#{id}"
+      url       = "#{self.url}/#{type}/#{id}"
       @response = Configuration.client.get url
 
       h = MultiJson.decode(@response.body)
@@ -167,40 +220,40 @@ module Tire
     end
 
     def refresh
-      @response = Configuration.client.post "#{Configuration.url}/#{@name}/_refresh", ''
+      @response = Configuration.client.post "#{url}/_refresh", ''
 
     ensure
-      curl = %Q|curl -X POST "#{Configuration.url}/#{@name}/_refresh"|
+      curl = %Q|curl -X POST "#{url}/_refresh"|
       logged('_refresh', curl)
     end
 
     def open(options={})
       # TODO: Remove the duplication in the execute > rescue > ensure chain
-      @response = Configuration.client.post "#{Configuration.url}/#{@name}/_open", MultiJson.encode(options)
+      @response = Configuration.client.post "#{url}/_open", MultiJson.encode(options)
       MultiJson.decode(@response.body)['ok']
 
     ensure
-      curl = %Q|curl -X POST "#{Configuration.url}/#{@name}/_open"|
+      curl = %Q|curl -X POST "#{url}/_open"|
       logged('_open', curl)
     end
 
     def close(options={})
-      @response = Configuration.client.post "#{Configuration.url}/#{@name}/_close", MultiJson.encode(options)
+      @response = Configuration.client.post "#{url}/_close", MultiJson.encode(options)
       MultiJson.decode(@response.body)['ok']
 
     ensure
-      curl = %Q|curl -X POST "#{Configuration.url}/#{@name}/_close"|
+      curl = %Q|curl -X POST "#{url}/_close"|
       logged('_close', curl)
     end
 
     def analyze(text, options={})
       options = {:pretty => true}.update(options)
       params  = options.to_param
-      @response = Configuration.client.get "#{Configuration.url}/#{@name}/_analyze?#{params}", text
+      @response = Configuration.client.get "#{url}/_analyze?#{params}", text
       @response.success? ? MultiJson.decode(@response.body) : false
 
     ensure
-      curl = %Q|curl -X GET "#{Configuration.url}/#{@name}/_analyze?#{params}" -d '#{text}'|
+      curl = %Q|curl -X GET "#{url}/_analyze?#{params}" -d '#{text}'|
       logged('_analyze', curl)
     end
 
@@ -235,11 +288,11 @@ module Tire
       payload = { :doc => document }
       payload.update( :query => query ) if query
 
-      @response = Configuration.client.get "#{Configuration.url}/#{@name}/#{type}/_percolate", MultiJson.encode(payload)
+      @response = Configuration.client.get "#{url}/#{type}/_percolate", MultiJson.encode(payload)
       MultiJson.decode(@response.body)['matches']
 
     ensure
-      curl = %Q|curl -X GET "#{Configuration.url}/#{@name}/#{type}/_percolate?pretty=1" -d '#{payload.to_json}'|
+      curl = %Q|curl -X GET "#{url}/#{type}/_percolate?pretty=1" -d '#{payload.to_json}'|
       logged('_percolate', curl)
     end
 
@@ -304,7 +357,8 @@ module Tire
                      "please pass an object which responds to `to_indexed_json` or a plain Hash."
           document
         when document.respond_to?(:to_indexed_json) then document.to_indexed_json
-        else raise ArgumentError, "Please pass a JSON string or object with a 'to_indexed_json' method"
+        else raise ArgumentError, "Please pass a JSON string or object with a 'to_indexed_json' method," +
+                                  "'#{document.class}' given."
       end
     end
 

data/lib/tire/results/item.rb

@@ -32,7 +32,11 @@ module Tire
       end
 
       def id
-        @attributes[:_id] || @attributes[:id]
+        @attributes[:_id]   || @attributes[:id]
+      end
+
+      def type
+        @attributes[:_type] || @attributes[:type]
       end
 
       def persisted?

data/lib/tire/rubyext/hash.rb

@@ -1,6 +1,6 @@
 class Hash
 
-  def to_json
+  def to_json(options=nil)
     MultiJson.encode(self)
   end unless respond_to?(:to_json)
 

data/lib/tire/search/scan.rb

@@ -0,0 +1,114 @@
+module Tire
+  module Search
+
+
+    # Performs a "scan/scroll" search request, which obtains a `scroll_id`
+    # and keeps returning documents matching the passed query (or all documents) in batches.
+    #
+    # You may want to iterate over the batches being returned:
+    #
+    #     search = Tire::Search::Scan.new('articles')
+    #     search.each do |results|
+    #       puts results.map(&:title)
+    #     end
+    #
+    # The scan object has a fully Enumerable-compatible interface, so you may
+    # call methods like `map` or `each_with_index` on it.
+    #
+    # To iterate over individual documents, use the `each_document` method:
+    #
+    #     search.each_document do |document|
+    #       puts document.title
+    #     end
+    #
+    # You may limit the result set being returned by a regular Tire DSL query
+    # (or a hash, if you prefer), passed as a second argument:
+    #
+    #     search = Tire::Search::Scan.new('articles') do
+    #       query { term 'author.exact', 'John Smith' }
+    #     end
+    #
+    # The feature is also exposed in the Tire top-level DSL:
+    #
+    #     search = Tire.scan 'articles' do
+    #       query { term 'author.exact', 'John Smith' }
+    #     end
+    #
+    # See ElasticSearch documentation for further reference:
+    #
+    # * http://www.elasticsearch.org/guide/reference/api/search/search-type.html
+    # * http://www.elasticsearch.org/guide/reference/api/search/scroll.html
+    #
+    class Scan
+      include Enumerable
+
+      attr_reader :indices, :options, :search
+
+      def initialize(indices=nil, options={}, &block)
+        @indices = Array(indices)
+        @options = options.update(:search_type => 'scan', :scroll => '10m')
+        @seen    = 0
+        @search  = Search.new(@indices, @options, &block)
+      end
+
+      def url;                Configuration.url + "/_search/scroll";                           end
+      def params;             @options.empty? ? '' : '?' + @options.to_param;                  end
+      def results;            @results  || (__perform; @results);                              end
+      def response;           @response || (__perform; @response);                             end
+      def json;               @json     || (__perform; @json);                                 end
+      def total;              @total    || (__perform; @total);                                end
+      def seen;               @seen     || (__perform; @seen);                                 end
+
+      def scroll_id
+        @scroll_id ||= @search.perform.json['_scroll_id']
+      end
+
+      def each
+        until results.empty?
+          yield results.results
+          __perform
+        end
+      end
+
+      def each_document
+        until results.empty?
+          results.each { |item| yield item }
+          __perform
+        end
+      end
+
+      def size
+        results.size
+      end
+
+      def __perform
+        @response  = Configuration.client.get [url, params].join, scroll_id
+        @json      = MultiJson.decode @response.body
+        @results   = Results::Collection.new @json, @options
+        @total     = @json['hits']['total'].to_i
+        @seen     += @results.size
+        @scroll_id = @json['_scroll_id']
+        return self
+      ensure
+        __logged
+      end
+
+      def to_a;        results; end; alias :to_ary :to_a
+      def to_curl;     %Q|curl -X GET "#{url}?pretty=true" -d '#{@scroll_id}'|; end
+
+      def __logged(error=nil)
+        if Configuration.logger
+          Configuration.logger.log_request 'scroll', nil, to_curl
+
+          took = @json['took']        rescue nil
+          code = @response.code       rescue nil
+          body = "#{@seen}/#{@total} (#{@seen/@total.to_f*100}%)" rescue nil
+
+          Configuration.logger.log_response code || 'N/A', took || 'N/A', body
+        end
+      end
+
+    end
+
+  end
+end

data/lib/tire/version.rb

@@ -1,27 +1,14 @@
 module Tire
-  VERSION   = "0.4.0"
+  VERSION   = "0.4.1"
 
   CHANGELOG =<<-END
     IMPORTANT CHANGES LATELY:
 
-    * Persistence supports property defaults and casting model properties as Ruby objects
-    * Added Hashr (http://rubygems.org/gems/hashr) as dependency
-    * Search in persistence models returns model instances, not Items
-    * Fixed errors in the Curb client
-    * Re-raise the RestClient::RequestTimeout and RestClient::ServerBrokeConnection exceptions
-    * Index#bulk_store and Index#import support the `:raise` option to re-raise exceptions
-    * Prefer ELASTICSEARCH_URL environment variable as the default URL, if present
-    * Added the "text" search query
-    * Deprecated the support for passing JSON strings to `Index#store`
-    * ActiveModel mapping has the `:as` option dynamically set property value for serialization
-    * ActiveModel supports any level of mappings in `mapping`
-    * ActiveModel search can eagerly load records of multiple types/classes
-    * ActiveModel integration now properly supports namespaced models
-    * Added support for passing search params (`search_type`, `timeout`, etc.) to search requests
-    * Added the "tire:index:drop" Rake task
-    * Added the "Filter" facet type
-    * Added the "Fuzzy" search query type
-    * Various test suite refactorings and changes
-    * Relaxed gem dependencies
+    * Added a Index#settings method to retrieve index settings as a Hash
+    * Added support for the "scan" search in the Ruby API
+    * Added support for reindexing the index documents into new index
+    * Added basic support for index aliases
+    * Changed, that Index#bulk_store runs against an index endpoint, not against `/_bulk`
+    * Refactorings, fixes, Ruby 1.8 compatibility
   END
 end

data/test/integration/highlight_test.rb

@@ -21,6 +21,19 @@ module Tire
         assert doc.highlight.title.to_s.include?('<em>'), "Highlight does not include default highlight tag"
       end
 
+      should "highlight multiple fields with custom highlight tag" do
+        s = Tire.search('articles-test') do
+          query { string 'Two OR ruby' }
+          highlight :tags, :title, :options => { :tag => '<strong>' }
+        end
+
+        doc = s.results.first
+
+        assert_equal 1, doc.highlight.title.size
+        assert_equal "<strong>Two</strong>", doc.highlight.title.first, "Highlight does not include highlight tag"
+        assert_equal "<strong>ruby</strong>", doc.highlight.tags.first, "Highlight does not include highlight tag"
+      end
+
       should "return entire content with highlighted fragments" do
         # Tire::Configuration.logger STDERR, :level => 'debug'