lib/ej/core.rb in ej-0.0.10 vs lib/ej/core.rb in ej-0.0.12

@@ -6,20 +6,27 @@
 require 'hashie'
 require 'pp'
 
 module Ej
   class Core
-    DEFAULT_PER = 30000
+    DEFAULT_PER = 1000
     def initialize(host, index, debug)
       @logger =  debug ? Logger.new($stderr) : nil
       @index = index
       @client = Elasticsearch::Client.new hosts: host, logger: @logger, index: @index
     end
 
-    def search(type, query, size, from, source_only, routing = nil, fields = nil)
+    def search(type, query, size, from, source_only, routing = nil, fields = nil, sort = nil)
       body = { from: from }
       body[:size] = size unless size.nil?
+      if sort
+        sorts = []
+        sort.each do |k, v|
+          sorts << { k => v }
+        end
+        body[:sort] = sorts
+      end
       body[:query] = { query_string: { query: query } } unless query.nil?
       search_option = { index: @index, type: type, body: body }
       search_option[:routing] = routing unless routing.nil?
       search_option[:_source] = fields.nil? ? nil : fields.join(',')
       results = Hashie::Mash.new(@client.search(search_option))
@@ -30,12 +37,12 @@
       body = { size: 0, "aggs"=>{ term + "_count"=>{"cardinality"=>{"field"=>term}}}}
       body[:query] = { query_string: { query: query } } unless query.nil?
       @client.search index: @index, type: type, body: body
     end
 
-    def copy(source, dest, query)
-      per = DEFAULT_PER
+    def copy(source, dest, query, per_size)
+      per = per_size || DEFAULT_PER
       num = 0
       logger = Logger.new($stdout)
       source_client = Elasticsearch::Client.new hosts: source, index: @index
       dest_client = Elasticsearch::Client.new hosts: dest
       while true
@@ -57,15 +64,15 @@
         logger.info("copy complete #{from + docs.size}/#{total}")
         num += 1
       end
     end
 
-    def dump(query)
-      per = DEFAULT_PER
+    def dump(query, per_size)
+      per = per_size || DEFAULT_PER
       num = 0
-      bulk_message = []
       while true
+        bulk_message = []
         from = num * per
         body = { size: per, from: from }
         body[:query] = { query_string: { query: query } } unless query.nil?
         data = Hashie::Mash.new(@client.search index: @index, body: body)
         docs = data.hits.hits
@@ -75,12 +82,12 @@
           doc.delete('_score')
           bulk_message << Yajl::Encoder.encode({ 'index' => doc.to_h })
           bulk_message << Yajl::Encoder.encode(source)
         end
         num += 1
+        puts bulk_message.join("\n")
       end
-      puts bulk_message.join("\n")
     end
 
     def facet(term, size, query)
       body = {"facets"=>
         {"terms"=>
@@ -104,10 +111,11 @@
                   {"query"=>{"query_string"=>{"query"=> query}},
                    "filter"=>{"bool"=>{"must"=>[], "must_not"=>[]}}}},
                "aggs"=>
                 {"agg_" + term =>
                   {"terms"=>{"field"=>term, "size"=>size, "order"=>{"_count"=>"desc"}}}}}
-      @client.search index: @index, body: body
+      results = @client.search index: @index, body: body
+      results['aggregations']["agg_" + term]['buckets']
     end
 
     def min(term)
       body = {
         aggs: {