rocketjob 5.3.0 → 5.4.0.beta2

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: fe39da29017eca601c104ca9d7c4fa90f903c32cbf3822e064cbd4a7f0780272
-  data.tar.gz: ce81cfb0d3a0ac4dbfe4ebe7c549077e56ca52d19ca41a2fda150561ecfeedd7
+  metadata.gz: df527427b1abf142761ba3be11df6f146708763a371c345c1a2c4a2d5d160f73
+  data.tar.gz: 2cf7a11dc3b82fd5fceb147991eb3dfa448d454c40264a96a2e9318cd2783e10
 SHA512:
-  metadata.gz: fa49046d738a5064da99363e0d854253d17c375ab1e75b095c118139d9081781b9e4177bc03cf70abd4edb906b43cdc033b96234bd5864b5f9959e81f1421074
-  data.tar.gz: 5e67742618a4ec874a1c43649d8619cec007380f8db84a3a81f584ec6f9896be0e14e062e69955158732b32e419a8ac7bfab9eef394c983a9f484a9c5c04c934
+  metadata.gz: 8af8995b77274fad9d791a9598191ca0dc1f28dcc13f1391d91accc39f193bb9a15a1447526cc550312d2160925276a271c89a9eb2be5927199d5ba8b96cfa1d
+  data.tar.gz: e0f48e2b522eae0a470ce1fa0af9daa02900a446990e52bbe472880818fde220b15575e365135cae2843c4aceeac8680a8f316f70a7961cdf788eb856f6771d3

data/lib/rocket_job/batch/io.rb

@@ -18,7 +18,7 @@ module RocketJob
           raise "Category #{category.inspect}, must be registered in input_categories: #{input_categories.inspect}"
         end
 
-        (@inputs ||= {})[category] ||= RocketJob::Sliced::Input.new(**rocket_job_io_slice_arguments("inputs", category))
+        (@inputs ||= {})[category] ||= RocketJob::Sliced.factory(:input, category, self)
       end
 
       # Returns [RocketJob::Sliced::Output] output collection for holding output slices
@@ -34,7 +34,7 @@ module RocketJob
           raise "Category #{category.inspect}, must be registered in output_categories: #{output_categories.inspect}"
         end
 
-        (@outputs ||= {})[category] ||= RocketJob::Sliced::Output.new(**rocket_job_io_slice_arguments("outputs", category))
+        (@outputs ||= {})[category] ||= RocketJob::Sliced.factory(:output, category, self)
       end
 
       # Upload the supplied file, io, IOStreams::Path, or IOStreams::Stream.
@@ -355,8 +355,18 @@ module RocketJob
 
         return output(category).download(header_line: header_line, &block) if block
 
-        IOStreams.new(stream).writer(:line, **args) do |io|
-          output(category).download(header_line: header_line) { |record| io << record }
+        output_collection = output(category)
+
+        if output_collection.binary?
+          IOStreams.new(stream).stream(:none).writer(**args) do |io|
+            raise(ArgumenError, "A `header_line` is not supported with binary output collections") if header_line
+
+            output_collection.download { |record| io << record[:binary] }
+          end
+        else
+          IOStreams.new(stream).writer(:line, **args) do |io|
+            output_collection.download(header_line: header_line) { |record| io << record }
+          end
         end
       end
 
@@ -393,21 +403,6 @@ module RocketJob
           RocketJob::Sliced::Writer::Output.collect(self, input_slice) { |writer| writer << result }
         end
       end
-
-      private
-
-      def rocket_job_io_slice_arguments(collection_type, category)
-        collection_name = "rocket_job.#{collection_type}.#{id}"
-        collection_name << ".#{category}" unless category == :main
-
-        args = {collection_name: collection_name, slice_size: slice_size}
-        if encrypt
-          args[:slice_class] = Sliced::EncryptedSlice
-        elsif compress
-          args[:slice_class] = Sliced::CompressedSlice
-        end
-        args
-      end
     end
   end
 end

data/lib/rocket_job/batch/model.rb

@@ -44,12 +44,12 @@ module RocketJob
         # Compress uploaded records.
         # The fields are not affected in any way, only the data stored in the
         # records and results collections will compressed
-        field :compress, type: Boolean, default: false, class_attribute: true
+        field :compress, type: Object, default: false, class_attribute: true
 
         # Encrypt uploaded records.
         # The fields are not affected in any way, only the data stored in the
         # records and results collections will be encrypted
-        field :encrypt, type: Boolean, default: false, class_attribute: true
+        field :encrypt, type: Object, default: false, class_attribute: true
 
         #
         # Values that jobs can also update during processing

data/lib/rocket_job/batch/tabular/input.rb

@@ -12,6 +12,7 @@ module RocketJob
         included do
           field :tabular_input_header, type: Array, class_attribute: true, user_editable: true
           field :tabular_input_format, type: Symbol, default: :csv, class_attribute: true, user_editable: true
+          field :tabular_input_options, type: Hash, class_attribute: true
 
           # tabular_input_mode: [:line | :array | :hash]
           #   :line
@@ -53,7 +54,9 @@ module RocketJob
           input_stream = stream.nil? ? nil : IOStreams.new(stream)
 
           if stream && (tabular_input_type == :text)
-            input_stream.option_or_stream(:encode, encoding: "UTF-8", cleaner: :printable, replace: "")
+            # Cannot change the length of fixed width lines
+            replace = tabular_input_format == :fixed ? " " : ""
+            input_stream.option_or_stream(:encode, encoding: "UTF-8", cleaner: :printable, replace: replace)
           end
 
           # If an input header is not required, then we don't extract it'
@@ -96,14 +99,15 @@ module RocketJob
             allowed_columns:  tabular_input_white_list,
             required_columns: tabular_input_required,
             skip_unknown:     tabular_input_skip_unknown,
-            format:           tabular_input_format
+            format:           tabular_input_format,
+            format_options:   tabular_input_options&.deep_symbolize_keys
           )
         end
 
         def tabular_input_render
-          unless tabular_input_header.blank? && tabular_input.header?
-            @rocket_job_input = tabular_input.record_parse(@rocket_job_input)
-          end
+          return if tabular_input_header.blank? && tabular_input.header?
+
+          @rocket_job_input = tabular_input.record_parse(@rocket_job_input)
         end
 
         # Cleanse custom input header if supplied.

data/lib/rocket_job/batch/tabular/output.rb

@@ -12,6 +12,7 @@ module RocketJob
         included do
           field :tabular_output_header, type: Array, class_attribute: true, user_editable: true, copy_on_restart: true
           field :tabular_output_format, type: Symbol, default: :csv, class_attribute: true, user_editable: true, copy_on_restart: true
+          field :tabular_output_options, type: Hash, class_attribute: true
 
           validates_inclusion_of :tabular_output_format, in: IOStreams::Tabular.registered_formats
 
@@ -31,8 +32,9 @@ module RocketJob
 
         # Overrides: `RocketJob::Batch::IO#download` to add the `tabular_output_header`.
         def download(file_name_or_io = nil, category: :main, **args, &block)
-          # No header required
-          return super(file_name_or_io, category: category, **args, &block) unless tabular_output.requires_header?(category)
+          unless tabular_output.requires_header?(category)
+            return super(file_name_or_io, category: category, **args, &block)
+          end
 
           header = tabular_output.render_header(category)
           super(file_name_or_io, header_line: header, category: category, **args, &block)
@@ -43,7 +45,11 @@ module RocketJob
         # Delimited instance used for this slice, by a single worker (thread)
         def tabular_output
           @tabular_output ||= Tabular.new(
-            main: IOStreams::Tabular.new(columns: tabular_output_header, format: tabular_output_format)
+            main: IOStreams::Tabular.new(
+              columns:        tabular_output_header,
+              format:         tabular_output_format,
+              format_options: tabular_output_options&.deep_symbolize_keys
+            )
           )
         end
 

data/lib/rocket_job/batch/throttle_windows.rb

@@ -1,4 +1,5 @@
 require "active_support/concern"
+require "fugit"
 
 module RocketJob
   module Batch
@@ -43,6 +44,10 @@ module RocketJob
         field :secondary_duration, type: Integer, class_attribute: true, user_editable: true, copy_on_restart: true
 
         define_batch_throttle :throttle_windows_exceeded?, filter: :throttle_filter_id
+
+        validates_each :primary_schedule, :secondary_schedule do |record, attr, value|
+          record.errors.add(attr, "Invalid #{attr}: #{value.inspect}") if value && !Fugit::Cron.new(value)
+        end
       end
 
       private
@@ -56,10 +61,10 @@ module RocketJob
       end
 
       def throttle_outside_window?(schedule, duration)
-        cron = Plugins::Rufus::CronLine.new(schedule)
-        time = Time.now + 1
+        cron = Fugit::Cron.new(schedule)
+        time = Time.now.utc + 1
         # Add 1 second since right now could be the very beginning of the processing window.
-        previous_time = cron.previous_time(time).to_time
+        previous_time = cron.previous_time(time).to_utc_time
         previous_time + duration < time
       end
     end

data/lib/rocket_job/batch/worker.rb

@@ -46,7 +46,7 @@ module RocketJob
               next if slice.failed?
 
               slice.fail_on_exception!(re_raise_exceptions) { rocket_job_process_slice(slice) }
-            elsif record_count && rocket_job_batch_complete?(worker.name)
+            elsif record_count && fail_on_exception!(re_raise_exceptions) { rocket_job_batch_complete?(worker.name) }
               return false
             else
               logger.debug "No more work available for this job"
@@ -110,8 +110,6 @@ module RocketJob
         servers
       end
 
-      private
-
       def rocket_job_batch_throttled?(slice, worker)
         filter = self.class.rocket_job_batch_throttles.matching_filter(self, slice)
         return false unless filter
@@ -150,7 +148,7 @@ module RocketJob
           records = slice.records
 
           # Skip records already processed, if any.
-          #slice.processing_record_number ||= 0
+          # slice.processing_record_number ||= 0
           # TODO: Must append to existing output slices before this can be enabled.
           # if !collect_output && (slice.processing_record_number > 1)
           #   records = records[slice.processing_record_number - 1..-1]

data/lib/rocket_job/event.rb

@@ -104,8 +104,6 @@ module RocketJob
       end
     end
 
-    private
-
     @load_time   = Time.now.utc
     @subscribers = Concurrent::Map.new { Concurrent::Array.new }
 

data/lib/rocket_job/extensions/mongoid/clients/options.rb

@@ -22,8 +22,6 @@ module RocketJob
         @collection_name = collection_name&.to_sym
       end
 
-      private
-
       module ClassMethods
         def with_collection(collection_name)
           all.with_collection(collection_name)

data/lib/rocket_job/extensions/mongoid/remove_warnings.rb

@@ -0,0 +1,12 @@
+require "mongoid/fields/validators/macro"
+require "semantic_logger"
+module RocketJob
+  module RemoveMongoidWarnings
+    # Remove annoying warnings about Symbols type being deprecated.
+    def validate_options(*params)
+      SemanticLogger.silence(:error) { super(*params) }
+    end
+  end
+end
+
+::Mongoid::Fields::Validators::Macro.extend(RocketJob::RemoveMongoidWarnings)

data/lib/rocket_job/jobs/copy_file_job.rb

@@ -25,7 +25,7 @@ module RocketJob
 
       self.destroy_on_complete = false
       # Number of times to automatically retry the copy. Set to `0` for no retry attempts.
-      self.retry_limit = 5
+      self.retry_limit = 10
 
       # File names in IOStreams URL format.
       field :source_url, type: String, user_editable: true

data/lib/rocket_job/jobs/re_encrypt/relational_job.rb

@@ -1,8 +1,5 @@
-begin
-  require "active_record"
-rescue LoadError
-  raise 'RocketJob::Jobs::ReEncrypt::RelationalJob uses ActiveRecord to obtain the database connection, please install the gem "activerecord".'
-end
+require "active_record"
+require "sync_attr"
 
 # Batch Worker to Re-encrypt all encrypted fields in MySQL that start with `encrytped_`.
 #

data/lib/rocket_job/jobs/upload_file_job.rb

@@ -40,7 +40,7 @@ module RocketJob
         job.id = job_id if job_id
         upload_file(job)
         job.save!
-      rescue StandardError => e
+      rescue Exception => e
         # Prevent partial uploads
         job&.cleanup! if job.respond_to?(:cleanup!)
         raise(e)

data/lib/rocket_job/plugins/cron.rb

@@ -1,4 +1,5 @@
 require "active_support/concern"
+require "fugit"
 
 module RocketJob
   module Plugins
@@ -17,7 +18,9 @@ module RocketJob
 
         field :cron_schedule, type: String, class_attribute: true, user_editable: true, copy_on_restart: true
 
-        validate :rocket_job_cron_valid
+        validates_each :cron_schedule do |record, attr, value|
+          record.errors.add(attr, "Invalid cron_schedule: #{value.inspect}") if value && !Fugit::Cron.new(value)
+        end
         before_save :rocket_job_cron_set_run_at
 
         private
@@ -42,30 +45,10 @@ module RocketJob
         end
       end
 
-      # Returns [Time] the next time this job will be scheduled to run at.
-      #
-      # Parameters
-      #   time: [Time]
-      #     The next time as of this time.
-      #     Default: Time.now
-      def rocket_job_cron_next_time(time = Time.now)
-        RocketJob::Plugins::Rufus::CronLine.new(cron_schedule).next_time(time)
-      end
-
-      private
-
       def rocket_job_cron_set_run_at
-        return unless cron_schedule
-
-        self.run_at = rocket_job_cron_next_time if cron_schedule_changed? && !run_at_changed?
-      end
-
-      def rocket_job_cron_valid
-        return unless cron_schedule
+        return if cron_schedule.nil? || !(cron_schedule_changed? && !run_at_changed?)
 
-        RocketJob::Plugins::Rufus::CronLine.new(cron_schedule)
-      rescue ArgumentError => e
-        errors.add(:cron_schedule, e.message)
+        self.run_at = Fugit::Cron.new(cron_schedule).next_time.to_utc_time
       end
     end
   end

data/lib/rocket_job/plugins/job/worker.rb

@@ -73,12 +73,13 @@ module RocketJob
             if failed? || !may_fail?
               self.exception        = JobException.from_exception(e)
               exception.worker_name = worker_name
-              save! unless new_record? || destroyed?
-            elsif new_record? || destroyed?
-              fail(worker_name, e)
             else
-              fail!(worker_name, e)
+              fail(worker_name, e)
             end
+
+            # Prevent validation failures from failing the job
+            save(validate: false) unless new_record? || destroyed?
+
             raise e if re_raise_exceptions
           end
         end

data/lib/rocket_job/plugins/processing_window.rb

@@ -1,4 +1,5 @@
 require "active_support/concern"
+require "fugit"
 
 module RocketJob
   module Plugins
@@ -47,18 +48,14 @@ module RocketJob
 
         validates_presence_of :processing_schedule, :processing_duration
         validates_each :processing_schedule do |record, attr, value|
-          begin
-            RocketJob::Plugins::Rufus::CronLine.new(value)
-          rescue ArgumentError => e
-            record.errors.add(attr, e.message)
-          end
+          record.errors.add(attr, "Invalid schedule: #{value.inspect}") unless Fugit::Cron.new(value)
         end
       end
 
       # Returns [true|false] whether this job is currently inside its processing window
       def rocket_job_processing_window_active?
-        time          = Time.now
-        previous_time = rocket_job_processing_schedule.previous_time(time)
+        time          = Time.now.utc
+        previous_time = Fugit::Cron.new(processing_schedule).previous_time(time).to_utc_time
         # Inside previous processing window?
         previous_time + processing_duration > time
       end
@@ -69,17 +66,14 @@ module RocketJob
       def rocket_job_processing_window_check
         return if rocket_job_processing_window_active?
 
-        logger.warn("Processing window closed before job was processed. Job is re-scheduled to run at: #{rocket_job_processing_schedule.next_time}")
+        next_time = Fugit::Cron.new(processing_schedule).next_time.to_utc_time
+        logger.warn("Processing window closed before job was processed. Job is re-scheduled to run at: #{next_time}")
         self.worker_name ||= "inline"
         requeue!(worker_name)
       end
 
       def rocket_job_processing_window_set_run_at
-        self.run_at = rocket_job_processing_schedule.next_time unless rocket_job_processing_window_active?
-      end
-
-      def rocket_job_processing_schedule
-        RocketJob::Plugins::Rufus::CronLine.new(processing_schedule)
+        self.run_at = Fugit::Cron.new(processing_schedule).next_time.to_utc_time unless rocket_job_processing_window_active?
       end
     end
   end

data/lib/rocket_job/sliced.rb

@@ -0,0 +1,91 @@
+module RocketJob
+  module Sliced
+    autoload :BZip2OutputSlice, "rocket_job/sliced/bzip2_output_slice"
+    autoload :CompressedSlice, "rocket_job/sliced/compressed_slice"
+    autoload :EncryptedSlice, "rocket_job/sliced/encrypted_slice"
+    autoload :Input, "rocket_job/sliced/input"
+    autoload :Output, "rocket_job/sliced/output"
+    autoload :Slice, "rocket_job/sliced/slice"
+    autoload :Slices, "rocket_job/sliced/slices"
+    autoload :Store, "rocket_job/sliced/store"
+
+    module Writer
+      autoload :Input, "rocket_job/sliced/writer/input"
+      autoload :Output, "rocket_job/sliced/writer/output"
+    end
+
+    # Returns [RocketJob::Sliced::Slices] for the relevant type and category.
+    #
+    # Supports compress and encrypt with [true|false|Hash] values.
+    # When [Hash] they must specify whether the apply to the input or output collection types.
+    #
+    # Example, compress both input and output collections:
+    #   class MyJob < RocketJob::Job
+    #     include RocketJob::Batch
+    #     self.compress = true
+    #   end
+    #
+    # Example, compress just the output collections:
+    #   class MyJob < RocketJob::Job
+    #     include RocketJob::Batch
+    #     self.compress = {output: true}
+    #   end
+    #
+    # To use the specialized BZip output compressor, and the regular compressor for the input collections:
+    #   class MyJob < RocketJob::Job
+    #     include RocketJob::Batch
+    #     self.compress = {output: :bzip2, input: true}
+    #   end
+    def self.factory(type, category, job)
+      raise(ArgumentError, "Unknown type: #{type.inspect}") unless %i[input output].include?(type)
+
+      collection_name = "rocket_job.#{type}s.#{job.id}"
+      collection_name << ".#{category}" unless category == :main
+
+      args               = {collection_name: collection_name, slice_size: job.slice_size}
+      klass              = slice_class(type, job)
+      args[:slice_class] = klass if klass
+
+      if type == :input
+        RocketJob::Sliced::Input.new(args)
+      else
+        RocketJob::Sliced::Output.new(args)
+      end
+    end
+
+    private
+
+    # Parses the encrypt and compress options to determine which slice serializer to use.
+    # `encrypt` takes priority over any `compress` option.
+    def self.slice_class(type, job)
+      encrypt  = extract_value(type, job.encrypt)
+      compress = extract_value(type, job.compress)
+
+      if encrypt
+        case encrypt
+        when true
+          EncryptedSlice
+        else
+          raise(ArgumentError, "Unknown job `encrypt` value: #{compress}") unless compress.is_a?(Slices)
+          # Returns the supplied class to use for encryption.
+          encrypt
+        end
+      elsif compress
+        case compress
+        when true
+          CompressedSlice
+        when :bzip2
+          BZip2OutputSlice
+        else
+          raise(ArgumentError, "Unknown job `compress` value: #{compress}") unless compress.is_a?(Slices)
+          # Returns the supplied class to use for compression.
+          compress
+        end
+      end
+    end
+
+    def self.extract_value(type, value)
+      value.is_a?(Hash) ? value[type] : value
+    end
+  end
+end