embulk-input-mixpanel 0.5.2 → 0.5.3.alpha.1

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: 8534316e5eae7127b70afc10d9b247e872643c32
-  data.tar.gz: e629adeb4d42e6386564bd13cfba3bce107ecd9c
+  metadata.gz: 28a4798aa8352ee4fbf06a23a57ff44aeb4cb902
+  data.tar.gz: 433333137ed26c3a2d62b822b767621885095cf4
 SHA512:
-  metadata.gz: 70287307438223546321af68362fc63fd56f592eb09d17d543d53986990109bb53b285a8467500617f7a8b99ab4991b6850a7b53c0df241282ee2259f5a8808d
-  data.tar.gz: f8e370285d393fada617ccc2d4dcb9d5cd54396cf47757a9af6f425bdfb83ac77c6354127171a6859939367698cd9b23e0915b8c954e557533140dcc1ea1525c
+  metadata.gz: 2d5fbba923cf5e52c169ed94001633001a2fcd8bb83cf0138541581d5a7768e4fd51bf7d690f7105c3fb31cf480a27a1af4fa9633636c8e4a237ba5786c99722
+  data.tar.gz: a147657608530a5f0ef513fdd150e93cf62bee74abeeadc3046ad30df42d19c1b94c4f42606260c827f1785dfc3c5c9fef6cf231efbc55ef4351d7be38cd4002

data/embulk-input-mixpanel.gemspec

@@ -1,7 +1,6 @@
-
 Gem::Specification.new do |spec|
   spec.name          = "embulk-input-mixpanel"
-  spec.version       = "0.5.2"
+  spec.version       = "0.5.3.alpha.1"
   spec.authors       = ["yoshihara", "uu59"]
   spec.summary       = "Mixpanel input plugin for Embulk"
   spec.description   = "Loads records from Mixpanel."

data/lib/embulk/input/mixpanel.rb

@@ -29,17 +29,32 @@ module Embulk
       # between each 7 (SLICE_DAYS_COUNT) days.
       SLICE_DAYS_COUNT = 7
 
+      DEFAULT_TIME_COLUMN = 'time'
+
       def self.transaction(config, &control)
         timezone = config.param(:timezone, :string)
         TimezoneValidator.new(timezone).validate
 
         from_date = config.param(:from_date, :string, default: (Date.today - 2).to_s)
         fetch_days = config.param(:fetch_days, :integer, default: nil)
-        range = RangeGenerator.new(from_date, fetch_days).generate_range
-        Embulk.logger.info "Try to fetch data from #{range.first} to #{range.last}"
+
 
         fetch_unknown_columns = config.param(:fetch_unknown_columns, :bool, default: false)
 
+        incremental_column = config.param(:incremental_column, :string, default: nil)
+        incremental = config.param(:incremental, :bool, default: true)
+        latest_fetched_time = config.param(:latest_fetched_time, :integer, default: 0)
+
+        # Backfill from date if incremental and an incremental field is set and we are in incremental run
+        if incremental && !incremental_column.nil? && latest_fetched_time !=0
+          back_fill_days = config.param(:back_fill_days, :integer, default: 5)
+          puts "Backfill days #{back_fill_days}"
+          from_date = (Date.parse(from_date) - back_fill_days).to_s
+          fetch_days = fetch_days.nil? ? nil : fetch_days + back_fill_days
+        end
+
+        range = RangeGenerator.new(from_date, fetch_days).generate_range
+        Embulk.logger.info "Try to fetch data from #{range.first} to #{range.last}"
         task = {
           params: export_params(config),
           dates: range,
@@ -50,8 +65,10 @@ module Embulk
           fetch_unknown_columns: fetch_unknown_columns,
           fetch_custom_properties: config.param(:fetch_custom_properties, :bool, default: true),
           retry_initial_wait_sec: config.param(:retry_initial_wait_sec, :integer, default: 1),
+          incremental_column: incremental_column,
           retry_limit: config.param(:retry_limit, :integer, default: 5),
-          latest_fetched_time: config.param(:latest_fetched_time, :integer, default: 0),
+          latest_fetched_time: latest_fetched_time,
+          incremental: incremental
         }
 
         if task[:fetch_unknown_columns] && task[:fetch_custom_properties]
@@ -82,14 +99,16 @@ module Embulk
 
         # NOTE: If this plugin supports to run by multi threads, this
         # implementation is terrible.
-        task_report = task_reports.first
-        next_to_date = Date.parse(task_report[:to_date])
-
-        next_config_diff = {
-          from_date: next_to_date.to_s,
-          latest_fetched_time: task_report[:latest_fetched_time],
-        }
-        return next_config_diff
+        if task[:incremental]
+          task_report = task_reports.first
+          next_to_date = Date.parse(task_report[:to_date])
+          next_config_diff = {
+            from_date: next_to_date.to_s,
+            latest_fetched_time: task_report[:latest_fetched_time],
+          }
+          return next_config_diff
+        end
+        return {}
       end
 
       def self.guess(config)
@@ -109,7 +128,6 @@ module Embulk
           "from_date" => range.first,
           "to_date" => range.last,
         )
-
         columns = guess_from_records(client.export_for_small_dataset(params))
         return {"columns" => columns}
       end
@@ -133,6 +151,8 @@ module Embulk
         @schema = task[:schema]
         @dates = task[:dates]
         @fetch_unknown_columns = task[:fetch_unknown_columns]
+        @incremental_column = task[:incremental_column]
+        @incremental = task[:incremental]
       end
 
       def run
@@ -146,19 +166,25 @@ module Embulk
           unless preview?
             Embulk.logger.info "Fetching data from #{dates.first} to #{dates.last} ..."
           end
-
-          fetch(dates).each do |record|
-            record_time = record["properties"]["time"]
-            if record_time <= prev_latest_fetched_time
-              ignored_record_count += 1
-              next
-            end
-
-            current_latest_fetched_time= [
-              current_latest_fetched_time,
-              record_time,
-            ].max
-
+          record_time_column=@incremental_column || DEFAULT_TIME_COLUMN
+          fetch(dates,prev_latest_fetched_time).each do |record|
+            if @incremental
+              if !record["properties"].include?(record_time_column)
+                raise Embulk::ConfigError.new("Incremental column not exists in fetched data #{record_time_column}")
+              end
+              record_time = record["properties"][record_time_column]
+             if @incremental_column.nil?
+               if record_time <= prev_latest_fetched_time
+                 ignored_record_count += 1
+                 next
+               end
+             end
+
+             current_latest_fetched_time= [
+               current_latest_fetched_time,
+               record_time,
+             ].max
+           end
             values = extract_values(record)
             if @fetch_unknown_columns
               unknown_values = extract_unknown_values(record)
@@ -175,14 +201,12 @@ module Embulk
           end
           break if preview?
         end
-
         page_builder.finish
-
         task_report = {
           latest_fetched_time: current_latest_fetched_time,
           to_date: @dates.last || Date.today - 1,
         }
-        return task_report
+        task_report
       end
 
       private
@@ -236,13 +260,19 @@ module Embulk
         end
       end
 
-      def fetch(dates, &block)
+      def fetch(dates,last_fetch_time, &block)
         from_date = dates.first
         to_date = dates.last
         params = @params.merge(
           "from_date" => from_date,
-          "to_date" => to_date,
+          "to_date" => to_date
         )
+        if !@incremental_column.nil? && !last_fetch_time.nil? && last_fetch_time!=0 # can't do filter on time column, time column need to be filter manually.
+          params = params.merge(
+            "where" => "#{params['where'].nil? ? '' : "(#{params['where']}) and " }properties[\"#{@incremental_column}\"] > #{last_fetch_time}"
+          )
+        end
+        puts "Where params is #{params["where"]}"
         client = MixpanelApi::Client.new(@api_key, @api_secret, self.class.perfect_retry(task))
 
         if preview?
@@ -299,7 +329,7 @@ module Embulk
       end
 
       def self.guess_from_records(records)
-        sample_props = records.first(GUESS_RECORDS_COUNT).map{|r| r["properties"]}
+        sample_props = records.first(GUESS_RECORDS_COUNT).map {|r| r["properties"]}
         schema = Guess::SchemaGuess.from_hash_records(sample_props)
         columns = schema.map do |col|
           next if col.name == "time"
@@ -311,6 +341,7 @@ module Embulk
           result
         end.compact
         columns.unshift(name: NOT_PROPERTY_COLUMN, type: :string)
+        # Shift incremental column to top
         columns.unshift(name: "time", type: :long)
       end
     end

data/lib/range_generator.rb

@@ -47,7 +47,7 @@ class RangeGenerator
     end
 
     if fetch_days
-      from_date..(from_date + fetch_days - 1)
+      from_date..(from_date + (fetch_days > 1? fetch_days - 1 : fetch_days))
     else
       from_date..today
     end

data/test/embulk/input/test_mixpanel.rb

@@ -116,7 +116,7 @@ module Embulk
 
         def test_json_type
           sample_records = records.map do |r|
-            r.merge("properties" => {"array" => [1,2], "hash" => {foo: "FOO"}})
+            r.merge("properties" => {"time" => 1, "array" => [1, 2], "hash" => {foo: "FOO"}})
           end
           actual = Mixpanel.guess_from_records(sample_records)
           assert actual.include?(name: "array", type: :json)
@@ -193,6 +193,14 @@ module Embulk
           end
         end
 
+        def test_default_configuration
+          stub(Mixpanel).resume {|task|
+            assert_nil(task[:incremental_column])
+            assert_true(task[:incremental])
+          }
+          Mixpanel.transaction(transaction_config(Date.today))
+        end
+
         private
 
         def transaction_config(from_date)
@@ -282,6 +290,8 @@ module Embulk
               dates: DATES.map {|date| date.to_s},
               api_key: API_KEY,
               api_secret: API_SECRET,
+              incremental: true,
+              incremental_column: nil,
               timezone: timezone,
               schema: schema
             )
@@ -304,6 +314,25 @@ module Embulk
             Mixpanel.transaction(transaction_config(days), &control)
           end
 
+          def test_valid_days_with_backfill
+            days = 5
+
+            stub(Mixpanel).resume() do |task|
+              assert_equal(["2015-02-17", "2015-02-18", "2015-02-19", "2015-02-20", "2015-02-21", "2015-02-22", "2015-02-23", "2015-02-24", "2015-02-25", "2015-02-26"], task[:dates])
+            end
+            config=transaction_config(days).merge("back_fill_days" => 5, "incremental_column" => "test_column", "latest_fetched_time" => 1501599491000)
+            Mixpanel.transaction(config, &control)
+          end
+
+          def test_valid_days_with_backfill_first_run
+            days = 5
+            stub(Mixpanel).resume() do |task|
+              assert_equal(transaction_task(days)[:dates], task[:dates])
+            end
+            config=transaction_config(days).merge("back_fill_days" => 5, "incremental_column" => "test_column")
+            Mixpanel.transaction(config, &control)
+          end
+
           def test_invalid_days
             days = 0
 
@@ -549,6 +578,96 @@ module Embulk
           @plugin.run
         end
 
+        class NonIncrementalRunTest < self
+
+          def test_non_incremental_run
+
+            mock(@page_builder).add(anything).times(records.length * 2)
+            mock(@page_builder).finish
+            task_report = @plugin.run
+            assert_equal(0, task_report[:latest_fetched_time])
+          end
+
+          def task
+            super.merge(incremental: false)
+          end
+
+        end
+
+        class IncrementalRunTest < self
+
+          def test_incremental_run
+            dont_allow(mock(@page_builder)).add(anything)
+            mock(@page_builder).finish
+            task_report = @plugin.run
+            assert_equal(record_epoch+1, task_report[:latest_fetched_time])
+          end
+
+          def task
+            super.merge(incremental: true, latest_fetched_time: record_epoch+1)
+          end
+
+        end
+
+        class IncrementalColumnTest < self
+
+          def setup
+          end
+
+          def test_incremental_column_with_where
+            page_builder = Object.new
+            plugin = Mixpanel.new(task.merge(params: task[:params].merge("where" => "abc==def")), nil, nil, page_builder)
+            stub(plugin).preview? {false}
+            adjusted = record_epoch - timezone_offset_seconds
+            mock(page_builder).add(["FOO", adjusted, "event"]).times(records.length * 2)
+            mock(page_builder).finish
+            any_instance_of(MixpanelApi::Client) do |klass|
+              stub(klass).export() do |params, block|
+                assert_equal('(abc==def) and properties["mp_processing_time_ms"] > 0',params["where"])
+                records.each{|record| block.call(record) }
+              end
+            end
+            task_report = plugin.run
+            assert_equal(1234567919, task_report[:latest_fetched_time])
+          end
+
+          def test_incremental_column
+            page_builder = Object.new
+            plugin = Mixpanel.new(task, nil, nil, page_builder)
+            stub(plugin).preview? {false}
+            adjusted = record_epoch - timezone_offset_seconds
+            mock(page_builder).add(["FOO", adjusted, "event"]).times(records.length * 2)
+            mock(page_builder).finish
+            any_instance_of(MixpanelApi::Client) do |klass|
+              stub(klass).export() do |params, block|
+                assert_equal('properties["mp_processing_time_ms"] > 0',params["where"])
+                records.each{|record| block.call(record) }
+              end
+            end
+            task_report = plugin.run
+            assert_equal(1234567919, task_report[:latest_fetched_time])
+          end
+
+          def records
+            super.each_with_index.map {|record, i|
+              record['properties']['mp_processing_time_ms'] = record_epoch+i
+              record
+            }
+          end
+
+          def schema
+            [
+              {"name" => "foo", "type" => "string"},
+              {"name" => "time", "type" => "integer"},
+              {"name" => "event", "type" => "string"},
+            ]
+          end
+
+          def task
+            super.merge(incremental_column: 'mp_processing_time_ms')
+          end
+        end
+
         class CustomPropertiesTest < self
           def setup
             super
@@ -671,6 +790,8 @@ module Embulk
           api_key: API_KEY,
           api_secret: API_SECRET,
           timezone: TIMEZONE,
+          incremental: true,
+          incremental_column: nil,
           schema: schema,
           dates: DATES.to_a.map(&:to_s),
           params: Mixpanel.export_params(embulk_config),

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: embulk-input-mixpanel
 version: !ruby/object:Gem::Version
-  version: 0.5.2
+  version: 0.5.3.alpha.1
 platform: ruby
 authors:
 - yoshihara
@@ -9,7 +9,7 @@ authors:
 autorequire:
 bindir: bin
 cert_chain: []
-date: 2017-07-26 00:00:00.000000000 Z
+date: 2017-08-02 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   requirement: !ruby/object:Gem::Requirement
@@ -219,12 +219,12 @@ required_ruby_version: !ruby/object:Gem::Requirement
       version: '0'
 required_rubygems_version: !ruby/object:Gem::Requirement
   requirements:
-  - - ">="
+  - - ">"
     - !ruby/object:Gem::Version
-      version: '0'
+      version: 1.3.1
 requirements: []
 rubyforge_project:
-rubygems_version: 2.4.8
+rubygems_version: 2.6.12
 signing_key:
 specification_version: 4
 summary: Mixpanel input plugin for Embulk