module SearchFlip # The SearchFlip::Criteria class serves the purpose of chaining various # filtering and aggregation methods. Each chainable method creates a new # criteria object until a method is called that finally sends the respective # request to Elasticsearch and returns the result. # # @example # CommentIndex.where(public: true).sort(id: "desc").limit(1_000).records # CommentIndex.range(:created_at, lt: Time.parse("2014-01-01").delete # CommentIndex.search("hello world").total_entries # CommentIndex.query(more_like_this: { "...", fields: ["description"] })] # CommentIndex.exists(:user_id).paginate(page: 1, per_page: 100) # CommentIndex.sort("_doc").find_each { |comment| "..." } class Criteria include Sortable include Sourceable include Highlightable include Explainable include Paginatable include Customable include Filterable include PostFilterable include Aggregatable extend Forwardable attr_accessor :target, :profile_value, :source_value, :suggest_values, :includes_values, :eager_load_values, :preload_values, :failsafe_value, :scroll_args, :terminate_after_value, :timeout_value, :preference_value, :search_type_value, :routing_value, :track_total_hits_value # Creates a new criteria while merging the attributes (constraints, # settings, etc) of the current criteria with the attributes of another one # passed as argument. For multi-value contstraints the resulting criteria # will include constraints of both criterias. For single-value constraints, # the values of the criteria passed as an argument are used. # # @example # CommentIndex.where(approved: true).merge(CommentIndex.range(:created_at, gt: Time.parse("2015-01-01"))) # CommentIndex.aggregate(:user_id).merge(CommentIndex.where(admin: true)) # # @return [SearchFlip::Criteria] A newly created extended criteria def merge(other) other = other.criteria fresh.tap do |criteria| [ :profile_value, :failsafe_value, :terminate_after_value, :timeout_value, :offset_value, :limit_value, :scroll_args, :source_value, :preference_value, :search_type_value, :routing_value, :track_total_hits_value, :explain_value ].each do |name| criteria.send(:"#{name}=", other.send(name)) unless other.send(name).nil? end [ :sort_values, :includes_values, :preload_values, :eager_load_values, :must_values, :must_not_values, :filter_values, :post_must_values, :post_must_not_values, :post_filter_values ].each do |name| criteria.send(:"#{name}=", (criteria.send(name) || []) + other.send(name)) if other.send(name) end [:highlight_values, :suggest_values, :custom_value, :aggregation_values].each do |name| criteria.send(:"#{name}=", (criteria.send(name) || {}).merge(other.send(name))) if other.send(name) end end end # Specifies if or how many hits should be counted/tracked. Check out the # elasticsearch docs for futher details. # # @example # CommentIndex.track_total_hits(true) # CommentIndex.track_total_hits(10_000) # # @param value The value for track_total_hits # # @return [SearchFlip::Criteria] A newly created extended criteria def track_total_hits(value) fresh.tap do |criteria| criteria.track_total_hits_value = value end end # Specifies a preference value for the request. Check out the elasticsearch # docs for further details. # # @example # CommentIndex.preference("_primary") # # @param value The preference value # # @return [SearchFlip::Criteria] A newly created extended criteria def preference(value) fresh.tap do |criteria| criteria.preference_value = value end end # Specifies the search type value for the request. Check out the elasticsearch # docs for further details. # # @example # CommentIndex.search_type("dfs_query_then_fetch") # # @param value The search type value # # @return [SearchFlip::Criteria] A newly created extended criteria def search_type(value) fresh.tap do |criteria| criteria.search_type_value = value end end # Specifies the routing value for the request. Check out the elasticsearch # docs for further details. # # @example # CommentIndex.routing("user_id") # # @param value The search type value # # @return [SearchFlip::Criteria] A newly created extended criteria def routing(value) fresh.tap do |criteria| criteria.routing_value = value end end # Specifies a query timeout, such that the processing will be stopped after # that timeout and only the results calculated up to that point will be # processed and returned. # # @example # ProductIndex.timeout("3s").search("hello world") # # @param value [String] The timeout value # # @return [SearchFlip::Criteria] A newly created extended criteria def timeout(value) fresh.tap do |criteria| criteria.timeout_value = value end end # Specifies early query termination, such that the processing will be # stopped after the specified number of results has been accumulated. # # @example # ProductIndex.terminate_after(10_000).search("hello world") # # @param value [Fixnum] The number of records to terminate after # # @return [SearchFlip::Criteria] A newly created extended criteria def terminate_after(value) fresh.tap do |criteria| criteria.terminate_after_value = value end end # @api private # # Convenience method to have a unified conversion api. # # @return [SearchFlip::Criteria] Simply returns self def criteria self end alias_method :all, :criteria # Creates a new SearchFlip::Criteria. # # @param attributes [Hash] Attributes to initialize the Criteria with def initialize(attributes = {}) attributes.each do |key, value| send "#{key}=", value end end # Allows to set query specific settings like e.g. connection and index # name. Please note, however, that this should only be used for special # cases and the subsequent query can not be serialized. Checkout # SearchFlip::Index.with_settings for more details. # # @example # UserIndex.where("...").with_settings(connection: ProxyConnection) # # @return [SearchFlip::Criteria] Simply returns self ruby2_keywords def with_settings(*args) fresh.tap do |criteria| criteria.target = target.with_settings(*args) end end # Generates the request object from the attributes specified via chaining, # like eg offset, limit, query, filters, aggregations, etc and returns a # Hash that later gets serialized as JSON. # # @return [Hash] The generated request object def request @request ||= begin res = {} if must_values || must_not_values || filter_values res[:query] = { bool: { must: must_values.to_a, must_not: must_not_values.to_a, filter: filter_values.to_a }.reject { |_, value| value.empty? } } end res.update(from: offset_value_with_default, size: limit_value_with_default) res[:track_total_hits] = track_total_hits_value unless track_total_hits_value.nil? res[:explain] = explain_value unless explain_value.nil? res[:timeout] = timeout_value if timeout_value res[:terminate_after] = terminate_after_value if terminate_after_value res[:highlight] = highlight_values if highlight_values res[:suggest] = suggest_values if suggest_values res[:sort] = sort_values if sort_values res[:aggregations] = aggregation_values if aggregation_values if post_must_values || post_must_not_values || post_filter_values res[:post_filter] = { bool: { must: post_must_values.to_a, must_not: post_must_not_values.to_a, filter: post_filter_values.to_a }.reject { |_, value| value.empty? } } end res[:_source] = source_value unless source_value.nil? res[:profile] = true if profile_value res.update(custom_value) if custom_value res end end # Adds a suggestion section with the given name to the request. # # @example # query = CommentIndex.suggest(:suggestion, text: "helo", term: { field: "message" }) # query.suggestions(:suggestion).first["text"] # => "hello" # # @param name [String, Symbol] The name of the suggestion section # # @param options [Hash] Additional suggestion options. Check out the Elasticsearch # docs for further details. # # @return [SearchFlip::Criteria] A new criteria including the suggestion section def suggest(name, options = {}) fresh.tap do |criteria| criteria.suggest_values = (criteria.suggest_values || {}).merge(name => options) end end # Sets whether or not query profiling should be enabled. # # @example # query = CommentIndex.profile(true) # query.raw_response["profile"] # => { "shards" => ... } # # @param value [Boolean] Whether query profiling should be enabled or not # # @return [SearchFlip::Criteria] A newly created extended criteria def profile(value) fresh.tap do |criteria| criteria.profile_value = value end end # Adds scrolling to the request with or without an already existing scroll # id and using the specified timeout. # # @example # query = CommentIndex.scroll(timeout: "5m") # # until query.records.empty? # # ... # # query = query.scroll(id: query.scroll_id, timeout: "5m") # end # # @param id [String, nil] The scroll id of the last request returned by # SearchFlip or nil # # @param timeout [String] The timeout of the scroll request, ie. how long # SearchFlip should keep the scroll handle open # # @return [SearchFlip::Criteria] A newly created extended criteria def scroll(id: nil, timeout: "1m") fresh.tap do |criteria| criteria.scroll_args = { id: id, timeout: timeout } end end # Sends a delete by query request to Elasticsearch, such that all documents # matching the query get deleted. Please note, for certain Elasticsearch # versions you need to install the delete-by-query plugin to get support # for this feature. Refreshes the index if the auto_refresh is enabled. # Raises SearchFlip::ResponseError in case any errors occur. # # @see SearchFlip::Config See SearchFlip::Config for auto_refresh # # @example # CommentIndex.range(lt: Time.parse("2014-01-01")).delete # CommentIndex.where(public: false).delete def delete(params = {}) dupped_request = request.dup dupped_request.delete(:from) dupped_request.delete(:size) if connection.version.to_i >= 5 connection.http_client.post("#{target.type_url}/_delete_by_query", params: request_params.merge(params), json: dupped_request) else connection.http_client.delete("#{target.type_url}/_query", params: request_params.merge(params), json: dupped_request) end target.refresh if SearchFlip::Config[:auto_refresh] true end # Specify associations of the target model you want to include via # ActiveRecord's or other ORM's mechanisms when records get fetched from # the database. # # @example # CommentIndex.includes(:user, :post).records # PostIndex.includes(:comments => :user).records # # @param args The args that get passed to the includes method of # ActiveRecord or other ORMs # # @return [SearchFlip::Criteria] A newly created extended criteria def includes(*args) fresh.tap do |criteria| criteria.includes_values = (includes_values || []) + args end end # Specify associations of the target model you want to eager load via # ActiveRecord's or other ORM's mechanisms when records get fetched from # the database. # # @example # CommentIndex.eager_load(:user, :post).records # PostIndex.eager_load(:comments => :user).records # # @param args The args that get passed to the eager load method of # ActiveRecord or other ORMs # # @return [SearchFlip::Criteria] A newly created extended criteria def eager_load(*args) fresh.tap do |criteria| criteria.eager_load_values = (eager_load_values || []) + args end end # Specify associations of the target model you want to preload via # ActiveRecord's or other ORM's mechanisms when records get fetched from # the database. # # @example # CommentIndex.preload(:user, :post).records # PostIndex.includes(:comments => :user).records # # @param args The args that get passed to the preload method of # ActiveRecord or other ORMs # # @return [SearchFlip::Criteria] A newly created extended criteria def preload(*args) fresh.tap do |criteria| criteria.preload_values = (preload_values || []) + args end end # Fetches the records specified by the criteria in batches using the # ElasicSearch scroll API and yields each batch. The batch size and scroll # API timeout can be specified. Check out the Elasticsearch docs for # further details. # # @example # CommentIndex.search("hello world").find_in_batches(batch_size: 100) do |batch| # # ... # end # # @param options [Hash] The options to control the fetching of batches # @option options batch_size [Fixnum] The number of records to fetch per # batch. Uses #limit to control the batch size. # @option options timeout [String] The timeout per scroll request, ie how # long Elasticsearch will keep the request handle open. def find_in_batches(options = {}) return enum_for(:find_in_batches, options) unless block_given? yield_in_batches(options) do |criteria| yield(criteria.records) if criteria.records.size > 0 end end # Fetches the results specified by the criteria in batches using the # Elasticsearch scroll API and yields each batch. The batch size and scroll # API timeout can be specified. Checkout out the Elasticsearch docs for # further details. # # @example # CommentIndex.search("hello world").find_results_in_batches(batch_size: 100) do |batch| # # ... # end # # @param options [Hash] The options to control the fetching of batches # @option options batch_size [Fixnum] The number of records to fetch per # batch. Uses #limit to control the batch size. # @option options timeout [String] The timeout per scroll request, ie how # long Elasticsearch will keep the request handle open. def find_results_in_batches(options = {}) return enum_for(:find_results_in_batches, options) unless block_given? yield_in_batches(options) do |criteria| yield criteria.results end end # Fetches the records specified by the relatin in batches using the # Elasticsearch scroll API and yields each record. The batch size and # scroll API timeout can be specified. Check out the Elasticsearch docs for # further details. # # @example # CommentIndex.search("hello world").find_each(batch_size: 100) do |record| # # ... # end # # @param options [Hash] The options to control the fetching of batches # @option options batch_size [Fixnum] The number of records to fetch per # batch. Uses #limit to control the batch size. # @option options timeout [String] The timeout per scroll request, ie how # long Elasticsearch will keep the request handle open. def find_each(options = {}) return enum_for(:find_each, options) unless block_given? find_in_batches options do |batch| batch.each do |record| yield record end end end alias_method :each, :find_each # Fetches the results specified by the criteria in batches using the # Elasticsearch scroll API and yields each result. The batch size and scroll # API timeout can be specified. Checkout out the Elasticsearch docs for # further details. # # @example # CommentIndex.search("hello world").find_each_result(batch_size: 100) do |result| # # ... # end # # @param options [Hash] The options to control the fetching of batches # @option options batch_size [Fixnum] The number of records to fetch per # batch. Uses #limit to control the batch size. # @option options timeout [String] The timeout per scroll request, ie how # long Elasticsearch will keep the request handle open. def find_each_result(options = {}) return enum_for(:find_each_result, options) unless block_given? find_results_in_batches options do |batch| batch.each do |result| yield result end end end # Executes the search request for the current criteria, ie sends the # request to Elasticsearch and returns the response. Connection and # response errors will be rescued if you specify the criteria to be # #failsafe, such that an empty response is returned instead. # # @example # response = CommentIndex.search("hello world").execute # # @return [SearchFlip::Response] The response object def execute @response ||= begin Config[:instrumenter].instrument("request.search_flip", index: target, request: request) do |payload| response = execute! payload[:response] = response response end end end alias_method :response, :execute # Marks the criteria to be failsafe, ie certain exceptions raised due to # invalid queries, inavailability of Elasticsearch, etc get rescued and an # empty criteria is returned instead. # # @see #execute See #execute for further details # # @example # CommentIndex.search("invalid/request").execute # # raises SearchFlip::ResponseError # # # ... # # CommentIndex.search("invalid/request").failsafe(true).execute # # => # # # @param value [Boolean] Whether or not the criteria should be failsafe # # @return [SearchFlip::Response] A newly created extended criteria def failsafe(value) fresh.tap do |criteria| criteria.failsafe_value = value end end # Returns a fresh, ie dupped, criteria with the response cache being # cleared. # # @example # CommentIndex.search("hello world").fresh # # @return [SearchFlip::Response] A dupped criteria with the response # cache being cleared def fresh dup.tap do |criteria| criteria.instance_variable_set(:@request, nil) criteria.instance_variable_set(:@response, nil) end end def respond_to_missing?(name, *args) target.respond_to?(name, *args) || super end def method_missing(name, *args, &block) if target.respond_to?(name) merge(target.send(name, *args, &block)) else super end end ruby2_keywords :method_missing def_delegators :response, :total_entries, :total_count, :current_page, :previous_page, :prev_page, :next_page, :first_page?, :last_page?, :out_of_range?, :total_pages, :hits, :ids, :count, :size, :length, :took, :aggregations, :suggestions, :scope, :results, :records, :scroll_id, :raw_response def_delegators :target, :connection private def execute! http_request = connection.http_client.headers(accept: "application/json") http_response = if scroll_args && scroll_args[:id] http_request.post( "#{connection.base_url}/_search/scroll", params: request_params, json: { scroll: scroll_args[:timeout], scroll_id: scroll_args[:id] } ) elsif scroll_args http_request.post( "#{target.type_url}/_search", params: request_params.merge(scroll: scroll_args[:timeout]), json: request ) else http_request.post("#{target.type_url}/_search", params: request_params, json: request) end SearchFlip::Response.new(self, http_response.parse) rescue SearchFlip::ConnectionError, SearchFlip::ResponseError => e raise e unless failsafe_value SearchFlip::Response.new(self, "took" => 0, "hits" => { "total" => 0, "hits" => [] }) end def yield_in_batches(options = {}) return enum_for(:yield_in_batches, options) unless block_given? batch_size = options[:batch_size] || 1_000 timeout = options[:timeout] || "1m" criteria = limit(batch_size).scroll(timeout: timeout) until criteria.ids.empty? yield criteria.response criteria = criteria.scroll(id: criteria.scroll_id, timeout: timeout) end end def request_params res = {} res[:preference] = preference_value if preference_value res[:search_type] = search_type_value if search_type_value res[:routing] = routing_value if routing_value res end end end