Ruby elasticsearch-model - 配置多个映射

时间:2015-06-19 21:58:42

标签: ruby-on-rails ruby elasticsearch spree

module Spree
    Product.class_eval do
       include Elasticsearch::Model

index_name Spree::ElasticsearchSettings.index
document_type 'spree_product'

  mapping _all: {"index_analyzer" => "nGram_analyzer", "search_analyzer" => "whitespace_analyzer"} do
    indexes :name, type: 'multi_field' do
      indexes :name, type: 'string', analyzer:'nGram_analyzer', boost: 100
      indexes :untouched, type: 'string', include_in_all: false, index: 'not_analyzed'
    end

    indexes :name_whole, type: 'multi_field' do
      indexes :name, type: 'string', index_analyzer:'simple'
    end

    indexes :name_completion, type: 'multi_field' do
      indexes :untouched, type: 'string', include_in_all: false, index: 'not_analyzed'
    end

    indexes :taxon_ids, type: 'multi_field' do
      indexes :taxon_ids, type:'string', analyzer: 'simple'
      indexes :taxon_ids_ngram, type:'string', analyzer:'nGram_analyzer'
    end
    indexes :description, analyzer: 'snowball', include_in_all:false
    indexes :available_on, type: 'date', format: 'dateOptionalTime', include_in_all: false
    indexes :price, type: 'double', include_in_all:false
    indexes :sku, type: 'string', index: 'not_analyzed', include_in_all:false
    indexes :properties, type: 'string', index: 'not_analyzed', include_in_all:false
 end

 mapping _suggest: {"index_analyzer" => "simple", "search_analyzer" => "whitespace_analyzer"} do
    indexes :name_completion, type: 'multi_field' do
      indexes :name, type: 'completion', index_analyzer:'simple', search_analyzer: 'simple', payloads: true
    end


    indexes :taxon_ids, type: 'multi_field' do
      indexes :taxon_ids, type:'string', analyzer: 'simple'
      indexes :taxon_ids_ngram, type:'string', analyzer:'nGram_analyzer'
    end
    indexes :description, analyzer: 'snowball', include_in_all:false
    indexes :available_on, type: 'date', format: 'dateOptionalTime', include_in_all: false
    indexes :price, type: 'double', include_in_all:false
    indexes :sku, type: 'string', index: 'not_analyzed', include_in_all:false
    indexes :properties, type: 'string', index: 'not_analyzed', include_in_all:false
 end

def as_indexed_json(options={})
  result = as_json({
    methods: [:price, :sku],
    only: [:available_on, :description, :name],
    include: {
      variants: {
        only: [:sku],
        include: {
          option_values: {
            only: [:name, :presentation]
          }
        }
      }
    }
  })
  result[:properties] = property_list unless property_list.empty?
  result[:taxon_ids] = taxons.map(&:self_and_ancestors).flatten.uniq.map(&:id) unless taxons.empty?
  result
end

# Inner class used to query elasticsearch. The idea is that the query is dynamically build based on the parameters.
class Product::ElasticsearchQuery
  include ::Virtus.model

  attribute :from, Integer, default: 0
  attribute :price_min, Float
  attribute :price_max, Float
  attribute :properties, Hash
  attribute :query, String
  attribute :taxons, Array
  attribute :browse_mode, Boolean
  attribute :sorting, String

  # When browse_mode is enabled, the taxon filter is placed at top level. This causes the results to be limited, but facetting is done on the complete dataset.
  # When browse_mode is disabled, the taxon filter is placed inside the filtered query. This causes the facets to be limited to the resulting set.

  # Method that creates the actual query based on the current attributes.
  # The idea is to always to use the following schema and fill in the blanks.
  # {
  #   query: {
  #     filtered: {
  #       query: {
  #         query_string: { query: , fields: [] }
  #       }
  #       filter: {
  #         and: [
  #           { terms: { taxons: [] } },
  #           { terms: { properties: [] } }
  #         ]
  #       }
  #     }
  #   }
  #   filter: { range: { price: { lte: , gte: } } },
  #   sort: [],
  #   from: ,
  #   facets:
  # }
  def to_hash
    q = { match_all: {} }
    unless query.blank? # nil or empty
      q = { query_string: { query: query, fields: ['name^5','description','sku'], default_operator: 'AND', use_dis_max: true } }
    end
    query = q

    and_filter = []
    unless @properties.nil? || @properties.empty?
      # transform properties from [{"key1" => ["value_a","value_b"]},{"key2" => ["value_a"]}
      # to { terms: { properties: ["key1||value_a","key1||value_b"] }
      #    { terms: { properties: ["key2||value_a"] }
      # This enforces "and" relation between different property values and "or" relation between same property values
      properties = @properties.map {|k,v| [k].product(v)}.map do |pair|
        and_filter << { terms: { properties: pair.map {|prop| prop.join("||")} } }
      end
    end

    sorting = case @sorting
    when "name_asc"
      [ {"name.untouched" => { order: "asc" }}, {"price" => { order: "asc" }}, "_score" ]
    when "name_desc"
      [ {"name.untouched" => { order: "desc" }}, {"price" => { order: "asc" }}, "_score" ]
    when "price_asc"
      [ {"price" => { order: "asc" }}, {"name.untouched" => { order: "asc" }}, "_score" ]
    when "price_desc"
      [ {"price" => { order: "desc" }}, {"name.untouched" => { order: "asc" }}, "_score" ]
    when "score"
      [ "_score", {"name.untouched" => { order: "asc" }}, {"price" => { order: "asc" }} ]
    else
      [ {"name.untouched" => { order: "asc" }}, {"price" => { order: "asc" }}, "_score" ]
    end

    # facets
    facets = {
      price: { statistical: { field: "price" } },
      properties: { terms: { field: "properties", order: "count", size: 1000000 } },
      taxon_ids: { terms: { field: "taxon_ids", size: 1000000 } }
    }

    # basic skeleton
    result = {
      min_score: 0.1,
      query: { filtered: {} },
      sort: sorting,
      from: from,
      facets: facets
    }

    # add query and filters to filtered
    result[:query][:filtered][:query] = query
    # taxon and property filters have an effect on the facets
    and_filter << { terms: { taxon_ids: taxons } } unless taxons.empty?
    # only return products that are available
    #and_filter << { range: { available_on: { lte: "now" } } }
    result[:query][:filtered][:filter] = { "and" => and_filter } unless and_filter.empty?

    # add price filter outside the query because it should have no effect on facets
    if price_min && price_max && (price_min < price_max)
      result[:filter] = { range: { price: { gte: price_min, lte: price_max } } }
    end
    result
  end
end

private

def property_list
  product_properties.map{|pp| "#{pp.property.name}||#{pp.value}"}
end
end
end

我有以下模型有两个映射,一个用于常规搜索,另一个用于搜索建议。根据ElasticSearch文档,后一个映射是必需的,以启用Completion Suggestor - https://www.elastic.co/guide/en/elasticsearch/reference/current/search-suggesters-completion.html

索引上述模型时,我收到以下错误: [400] {&#34;错误&#34;:&#34; MapperParsingException [mapping [spree_product]];嵌套:MapperParsingException [解析后根类型映射不为空!剩余字段:[_ abstract :: {search_analyzer = whitespace_analyzer,index_analyzer = simple}]]; &#34;&#34;状态&#34; 400} /usr/local/bundle/gems/elasticsearch-transport-1.0.12/lib/elasticsearch/transport/transport/base.rb:135:in __raise_transport_error' /usr/local/bundle/gems/elasticsearch-transport-1.0.12/lib/elasticsearch/transport/transport/base.rb:227:in perform_request&#39; /usr/local/bundle/gems/elasticsearch-transport-1.0.12/lib/elasticsearch/transport/transport/http/faraday.rb:20:in perform_request' /usr/local/bundle/gems/elasticsearch-transport-1.0.12/lib/elasticsearch/transport/client.rb:119:in perform_request&#39; /usr/local/bundle/gems/elasticsearch-api-1.0.12/lib/elasticsearch/api/namespace/common.rb:21:in perform_request' /usr/local/bundle/gems/elasticsearch-api-1.0.12/lib/elasticsearch/api/actions/indices/create.rb:77:in创建&#39; /tmp/spree_elasticsearch/lib/tasks/load_products.rake:5:in块(2级)&#39; 任务:TOP =&gt; spree_elasticsearch:load_products

1 个答案:

答案 0 :(得分:0)

发现了这个问题。将该字段嵌套完成可能不是一个好主意

设置你的映射:     映射做         索引:name_completion,type:&#39; completion&#39;,payloads:true     端

确保返回正确的JSON以进行索引:

[:name_completion] = {           输入:名称,           输出:名称,           有效载荷:&#34;&#34;    }

从这个例子中引用: https://github.com/elastic/elasticsearch-rails/commit/ded20356920802c35d258756113acfd95b25ade6