embulk-output-elasticsearch_ruby 0.1.4 → 0.1.5

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: be42f1b9b4d5953efa82af4f11eff2d1871e28d7
-  data.tar.gz: 994268d29c57b5f1145e1f4ac79922da18282c9c
+  metadata.gz: 6824687f4de2bdcc12467725ea585462b9b0aa21
+  data.tar.gz: 337d8df78354516360a8c922b7dda2fe49abb12f
 SHA512:
-  metadata.gz: ffd1752f6ae7f52fb1bc8a07d90ff59633f05799c31cf05f207f1607a31ab298aab48c031e7c1b6634dd9751c757a449b4ad94cce930f36195f0c22c415aaa52
-  data.tar.gz: 8ea0f41e0de01e06f3916b4b1c2377260097536d9f9b68abc31a8962655ef0b74ee750d4de1b11bcd601c8c38b3b25f062b9a66aff5409d4ea67debf99f53d2b
+  metadata.gz: ab542815971e42add2330522e9ea867c3e8542c5003155ec9df69ea76d55b35f8289853326eee52e855b72ff8fa2e171cb2a4ee8a1f8bc9dcc572987a341de22
+  data.tar.gz: a734d8594b15f0be35b12cdf64d3b51d3c87427e7e2d5fe0865b55e281cf77d63d588b1f3b16c4a1ceaee38a2a7e0620f3f7bd85d6f618a71f0b041f3cbcb9ff

data/README.md

@@ -1,4 +1,4 @@
-# Elasticsearch Ruby output plugin for Embulk
+# Elasticsearch Ruby output plugin for Embulk [![Gem Version](https://badge.fury.io/rb/embulk-output-elasticsearch_ruby.svg)](http://badge.fury.io/rb/embulk-output-elasticsearch_ruby)
 
 Dumps records to Elasticsearch Ruby. Elasticsearch 1.X AND 2.X AND 5.X compatible.
 
@@ -10,26 +10,28 @@ Dumps records to Elasticsearch Ruby. Elasticsearch 1.X AND 2.X AND 5.X compatibl
 * **Cleanup supported**: yes
 
 ## Configuration
-  - **nodes**: nodes (array, default: [{ 'host' => 'localhost', 'port' => 9200 }])
-    - **host**: index (string)
-    - **port**: index (integer)
-  - **request_timeout**: request_timeout (integer, default: 60)
-  - **index**: index (string, , default: 'logstash-%Y.%m.%d')
-  - **mode**: mode, normal or update or replace (string, default: normal)
-  - **reload_connections**: reload_connections (bool, default: true)
-  - **reload_on_failure**: reload_on_failure (bool, default: false)
-  - **delete_old_index**: delete_old_index (bool, default: false)
-  - **index_type**: index_type (string)
-  - **id_keys**: id_keys (array, default: nil)
-  - **id_format**: id_format (string, default: nil)
-  - **array_columns**: array_columns (array, default: nil)
-    - **name**: Array convert column. (string)
-    - **delimiter**: delimiter for split. (string)
-    - **is_integer**: to integer. (bool)
-  - **bulk_actions**: bulk_actions (integer, default: 1000)
-  - **retry_on_failure**: retry_on_failure (integer, default: 5)
+- **nodes** nodes (array, default: [{ 'host' => 'localhost', 'port' => 9200 }])
+    - **host** host (string)
+    - **port** port (string)
+- **request_timeout** request timeout (integer, default: 60)
+- **index_type** index type (string)
+- **mode** mode (string, default: 'normal')
+- **reload_connections** reload connections (bool, default: true)
+- **reload_on_failure** reload on failure (bool, default: false)
+- **delete_old_index** delete old index (bool, default: false)
+- **delete_old_alias** delete old alias (bool, default: true)
+- **id_keys** id keys (array, default: nil)
+- **id_format** id format (string, default: nil)
+- **array_columns** array columns (array, default: nil)
+- **bulk_actions** bulk actions (integer, default: 1000)
+- **retry_on_failure** retry on failure (integer, default: 5)
+- **current_index_name** current index name (string, default: nil)
+- **index** index (string, default: 'logstash-%Y.%m.%d')
+- **before_delete_index** before delete index (bool, default: false)
+- **before_template_name** before template name (string, default: nil)
+- **before_template** before template (hash, default: nil)
 
-## Example
+## Example(minimum settings)
 
 ```yaml
 out:
@@ -39,7 +41,7 @@ out:
   index_type: page
 ```
 
-## Example(update)
+## Example(update mode)
 
 ```yaml
 out:
@@ -56,6 +58,23 @@ out:
     - _id
 ```
 
+## Example(replace mode)
+
+```yaml
+out:
+  type: elasticsearch_ruby
+  nodes:
+    - {host: localhost, port: 9200}
+  index: test_alias
+  index_type: crawl_companies
+  mode: replace
+  delete_old_index: true
+  before_delete_index: true
+  bulk_actions: 1000
+  request_timeout: 60
+```
+
+* create alias 
 
 ## Build
 

data/embulk-output-elasticsearch_ruby.gemspec

@@ -1,7 +1,7 @@
 
 Gem::Specification.new do |spec|
   spec.name          = "embulk-output-elasticsearch_ruby"
-  spec.version       = "0.1.4"
+  spec.version       = "0.1.5"
   spec.authors       = ["toyama0919"]
   spec.summary       = "Elasticsearch Ruby output plugin for Embulk. Elasticsearch 1.X AND 2.X AND 5.X compatible."
   spec.description   = "Dumps records to Elasticsearch Ruby. Elasticsearch 1.X AND 2.X AND 5.X compatible."

data/lib/embulk/output/elasticsearch/connection.rb

@@ -0,0 +1,129 @@
+require 'excon'
+require 'elasticsearch'
+
+module Embulk
+  module Output
+    class Elasticsearch < OutputPlugin
+      class Connection
+        def initialize(task)
+          @nodes = task["nodes"]
+          @index_type = task["index_type"]
+          @id_keys = task["id_keys"]
+          @id_format = task["id_format"]
+          @array_columns = task["array_columns"]
+          @retry_on_failure = task["retry_on_failure"]
+          @mode = task["mode"]
+          @delete_old_index = task['delete_old_index']
+          @delete_old_alias = task['delete_old_alias']
+          @index = task['index']
+          @alias = task['alias']
+          @action = (@mode == 'update') ? :update : :index
+
+          @client = create_client(
+            nodes: task['nodes'],
+            reload_connections: task['reload_connections'],
+            reload_on_failure: task['reload_on_failure'],
+            retry_on_failure: task['retry_on_failure'],
+            request_timeout: task['request_timeout']
+          )
+        end
+
+        def create_client(nodes: ,reload_connections: ,reload_on_failure: ,retry_on_failure: ,request_timeout:)
+          transport = ::Elasticsearch::Transport::Transport::HTTP::Faraday.new(
+            {
+              hosts: nodes.map{ |node| Hash[node.map{ |k, v| [k.to_sym, v] }] },
+              options: {
+                reload_connections: reload_connections,
+                reload_on_failure: reload_on_failure,
+                retry_on_failure: retry_on_failure,
+                transport_options: {
+                  request: { timeout: request_timeout }
+                }
+              }
+            }
+          )
+          ::Elasticsearch::Client.new transport: transport
+        end
+
+        def put_template(before_template_name, before_template)
+          Embulk.logger.info("put template => #{before_template_name}")
+          @client.indices.put_template name: before_template_name, body: before_template
+        end
+
+        def create_aliases
+          @client.indices.update_aliases body: {
+            actions: [{ add: { index: @index, alias: @alias } }]
+          }
+          Embulk.logger.info "created alias: #{@alias}, index: #{@index}"
+        end
+
+        def delete_aliases
+          indices = @client.indices.get_alias(name: @alias).keys
+          indices.each do |index|
+            if index != @index
+              if @delete_old_alias
+                @client.indices.delete_alias index: index, name: @alias
+                Embulk.logger.info "deleted alias: #{@alias}, index: #{index}"
+              end
+              if @delete_old_index
+                delete_index(index)
+              end
+            end
+          end
+        end
+
+        def delete_index(index)
+          indices = @client.cat.indices(format: 'json')
+          if indices.any? { |i| i['index'] == index }
+            @client.indices.delete index: index
+            Embulk.logger.info "deleted index: #{index}"
+          end
+        end
+
+        def send(bulk_message)
+          retries = 0
+          begin
+            @client.bulk body: bulk_message
+            Embulk.logger.info "bulk: #{bulk_message.size/2} success."
+          rescue => e
+            if retries < @retry_on_failure
+              retries += 1
+              Embulk.logger.warn "Could not push logs to Elasticsearch, resetting connection and trying again. #{e.message}"
+              sleep 2**retries
+              retry
+            end
+            raise "Could not push logs to Elasticsearch after #{retries} retries. #{e.message}"
+          end
+        end
+
+        def generate_source(record)
+          result = {}
+
+          record.each { |key, value|
+            result[key] = value
+            next if (value.nil? || !@array_columns)
+            @array_columns.each do |array_column|
+              if array_column['name'] == key
+                array_value = value.split(array_column['delimiter']).reject(&:empty?)
+                array_value = array_value.map(&:to_i) if array_column['is_integer']
+                result[key] = array_value
+              end
+            end
+          }
+          (@mode == 'update') ? {doc: result} : result
+        end
+
+        def generate_id(template, record, id_keys)
+          template % id_keys.map { |key| record[key] }
+        end
+
+        def generate_meta(record)
+          meta = {}
+          meta[@action] = { _index: @index, _type: @index_type }
+          meta[@action][:_id] = generate_id(@id_format, record, @id_keys) unless @id_keys.nil?
+          meta
+        end
+      end
+    end
+  end
+end

data/lib/embulk/output/elasticsearch_ruby.rb

@@ -1,5 +1,4 @@
-require 'excon'
-require 'elasticsearch'
+require_relative 'elasticsearch/connection'
 
 module Embulk
   module Output
@@ -12,40 +11,47 @@ module Embulk
         task = {
           "nodes" => config.param("nodes", :array, default: [{ 'host' => 'localhost', 'port' => 9200 }]),
           "request_timeout" => config.param("request_timeout", :integer, default: 60),
-          "index" => config.param("index", :string, default: 'logstash-%Y.%m.%d'),
+          "index_type" => config.param("index_type", :string),
           "mode" => config.param("mode", :string, default: 'normal'),
           "reload_connections" => config.param("reload_connections", :bool, default: true),
           "reload_on_failure" => config.param("reload_on_failure", :bool, default: false),
           "delete_old_index" => config.param("delete_old_index", :bool, default: false),
           "delete_old_alias" => config.param("delete_old_alias", :bool, default: true),
-          "index_type" => config.param("index_type", :string),
           "id_keys" => config.param("id_keys", :array, default: nil),
           "id_format" => config.param("id_format", :string, default: nil),
           "array_columns" => config.param("array_columns", :array, default: nil),
           "bulk_actions" => config.param("bulk_actions", :integer, default: 1000),
           "retry_on_failure" => config.param("retry_on_failure", :integer, default: 5),
-          "before_template_name" => config.param("before_template_name", :string, default: nil),
-          "before_template" => config.param("before_template", :hash, default: nil),
-          "current_index_name" => config.param("current_index_name", :string, default: nil),
         }
-        task['time_value'] = Time.now.strftime('%Y.%m.%d.%H.%M.%S')
-        task['index'] = Time.now.strftime(task['index'])
-
-        task['current_index_name'] = if task['current_index_name']
-          task['current_index_name']
-        else
-          "#{task['index']}-#{task['index_type']}-#{task['time_value']}"
-        end
 
         unless ENABLE_MODE.include?(task['mode'])
           raise ConfigError.new "`mode` must be one of #{ENABLE_MODE.join(', ')}"
         end
         Embulk.logger.info("mode => #{task['mode']}")
 
-        if task['before_template_name'] && task['before_template']
-          client = create_client(task)
-          Embulk.logger.info("put template => #{task['before_template_name']}")
-          client.indices.put_template name: task['before_template_name'], body: task['before_template']
+        current_index_name = config.param("current_index_name", :string, default: nil)
+        index = config.param("index", :string, default: 'logstash-%Y.%m.%d')
+        if task['mode'] == 'replace'
+          task['alias'] = index
+          task['index'] = if current_index_name
+            current_index_name
+          else
+            "#{index}-#{task['index_type']}-#{Time.now.strftime('%Y.%m.%d.%H.%M.%S')}"
+          end
+        else
+          task['index'] = Time.now.strftime(index)
+        end
+
+        connection = Connection.new(task)
+        before_delete_index = config.param("before_delete_index", :bool, default: false)
+        if before_delete_index
+          connection.delete_index(task['index'])
+        end
+
+        before_template_name = config.param("before_template_name", :string, default: nil)
+        before_template = config.param("before_template", :hash, default: nil)
+        if before_template_name && before_template
+          connection.put_template(before_template_name, before_template)
         end
 
         task_reports = yield(task)
@@ -55,57 +61,12 @@ module Embulk
 
       def self.cleanup(task, schema, count, task_reports)
         if task['mode'] == 'replace'
-          client = create_client(task)
-          create_aliases(client, task['index'], get_index(task))
-          delete_aliases(client, task)
+          connection = Connection.new(task)
+          connection.create_aliases
+          connection.delete_aliases
         end
       end
 
-      def self.create_client(task)
-        transport = ::Elasticsearch::Transport::Transport::HTTP::Faraday.new(
-          {
-            hosts: task['nodes'].map{ |node| Hash[node.map{ |k, v| [k.to_sym, v] }] },
-            options: {
-              reload_connections: task['reload_connections'],
-              reload_on_failure: task['reload_on_failure'],
-              retry_on_failure: task['retry_on_failure'],
-              transport_options: {
-                request: { timeout: task['request_timeout'] }
-              }
-            }
-          }
-        )
-
-        ::Elasticsearch::Client.new transport: transport
-      end
-
-      def self.create_aliases(client, als, index)
-        client.indices.update_aliases body: {
-          actions: [{ add: { index: index, alias: als } }]
-        }
-        Embulk.logger.info "created alias: #{als}, index: #{index}"
-      end
-
-      def self.delete_aliases(client, task)
-        indices = client.indices.get_alias(name: task['index']).keys
-        indices.each { |index|
-          if index != get_index(task)
-            if task['delete_old_alias']
-              client.indices.delete_alias index: index, name: task['index']
-              Embulk.logger.info "deleted alias: #{task['index']}, index: #{index}"
-            end
-            if task['delete_old_index']
-              client.indices.delete index: index
-              Embulk.logger.info "deleted index: #{index}"
-            end
-          end
-        }
-      end
-
-      def self.get_index(task)
-        task['mode'] == 'replace' ? task['current_index_name'] : task['index']
-      end
-
       #def self.resume(task, schema, count, &control)
       #  task_reports = yield(task)
       #
@@ -114,17 +75,8 @@ module Embulk
       #end
 
       def init
-        @nodes = task["nodes"]
-        @index_type = task["index_type"]
-        @id_keys = task["id_keys"]
-        @id_format = task["id_format"]
+        @connection = Connection.new(task)
         @bulk_actions = task["bulk_actions"]
-        @array_columns = task["array_columns"]
-        @retry_on_failure = task["retry_on_failure"]
-        @mode = task["mode"]
-        @index = self.class.get_index(task)
-
-        @client = self.class.create_client(task)
         @bulk_message = []
       end
 
@@ -134,22 +86,24 @@ module Embulk
       def add(page)
         page.each do |record|
           hash = Hash[schema.names.zip(record)]
-          action = (@mode == 'update') ? :update : :index
-          meta = {}
-          meta[action] = { _index: @index, _type: @index_type }
-          meta[action][:_id] = generate_id(@id_format, hash, @id_keys) unless @id_keys.nil?
-          source = generate_array(hash)
+          meta = @connection.generate_meta(hash)
+          source = @connection.generate_source(hash)
+
+          Embulk.logger.debug("meta => #{meta}")
+          Embulk.logger.debug("source => #{source}")
+
           @bulk_message << meta
           @bulk_message << source
           if @bulk_actions * 2 <= @bulk_message.size
-            send
+            @connection.send(@bulk_message)
+            @bulk_message.clear
           end
         end
       end
 
       def finish
         if @bulk_message.size > 0
-          send
+          @connection.send(@bulk_message)
         end
       end
 
@@ -160,46 +114,6 @@ module Embulk
         task_report = {}
         return task_report
       end
-
-      private
-
-      def generate_array(record)
-        result = {}
-
-        record.each { |key, value|
-          result[key] = value
-          next if (value.nil? || !@array_columns)
-          @array_columns.each do |array_column|
-            if array_column['name'] == key
-              array_value = value.split(array_column['delimiter']).reject(&:empty?)
-              array_value = array_value.map(&:to_i) if array_column['is_integer']
-              result[key] = array_value
-            end
-          end
-        }
-        (@mode == 'update') ? {doc: result} : result
-      end
-
-      def generate_id(template, record, id_keys)
-        template % id_keys.map { |key| record[key] }
-      end
-
-      def send
-        retries = 0
-        begin
-          @client.bulk body: @bulk_message
-          Embulk.logger.info "bulk: #{@bulk_message.size/2} success."
-        rescue => e
-          if retries < @retry_on_failure
-            retries += 1
-            Embulk.logger.warn "Could not push logs to Elasticsearch, resetting connection and trying again. #{e.message}"
-            sleep 2**retries
-            retry
-          end
-          raise "Could not push logs to Elasticsearch after #{retries} retries. #{e.message}"
-        end
-        @bulk_message.clear
-      end
     end
   end
 end

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: embulk-output-elasticsearch_ruby
 version: !ruby/object:Gem::Version
-  version: 0.1.4
+  version: 0.1.5
 platform: ruby
 authors:
 - toyama0919
 autorequire:
 bindir: bin
 cert_chain: []
-date: 2017-02-27 00:00:00.000000000 Z
+date: 2017-03-27 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   requirement: !ruby/object:Gem::Requirement
@@ -121,6 +121,7 @@ files:
 - README.md
 - Rakefile
 - embulk-output-elasticsearch_ruby.gemspec
+- lib/embulk/output/elasticsearch/connection.rb
 - lib/embulk/output/elasticsearch_ruby.rb
 - test/helper.rb
 - test/test_transaction.rb