aws-sdk-s3 1.198.0 → 1.202.0

data/lib/aws-sdk-s3/file_downloader.rb

@@ -8,184 +8,245 @@ module Aws
   module S3
     # @api private
     class FileDownloader
-
       MIN_CHUNK_SIZE = 5 * 1024 * 1024
       MAX_PARTS = 10_000
+      HEAD_OPTIONS = Set.new(Client.api.operation(:head_object).input.shape.member_names)
+      GET_OPTIONS = Set.new(Client.api.operation(:get_object).input.shape.member_names)
 
       def initialize(options = {})
         @client = options[:client] || Client.new
+        @executor = options[:executor]
       end
 
       # @return [Client]
       attr_reader :client
 
       def download(destination, options = {})
-        valid_types = [String, Pathname, File, Tempfile]
-        unless valid_types.include?(destination.class)
-          raise ArgumentError, "Invalid destination, expected #{valid_types.join(', ')} but got: #{destination.class}"
-        end
-
-        @destination = destination
-        @mode = options.delete(:mode) || 'auto'
-        @thread_count = options.delete(:thread_count) || 10
-        @chunk_size = options.delete(:chunk_size)
-        @on_checksum_validated = options.delete(:on_checksum_validated)
-        @progress_callback = options.delete(:progress_callback)
-        @params = options
-        validate!
+        validate_destination!(destination)
+        opts = build_download_opts(destination, options)
+        validate_opts!(opts)
 
         Aws::Plugins::UserAgent.metric('S3_TRANSFER') do
-          case @mode
-          when 'auto' then multipart_download
-          when 'single_request' then single_request
-          when 'get_range'
-            raise ArgumentError, 'In get_range mode, :chunk_size must be provided' unless @chunk_size
-
-            resp = @client.head_object(@params)
-            multithreaded_get_by_ranges(resp.content_length, resp.etag)
-          else
-            raise ArgumentError, "Invalid mode #{@mode} provided, :mode should be single_request, get_range or auto"
+          case opts[:mode]
+          when 'auto' then multipart_download(opts)
+          when 'single_request' then single_request(opts)
+          when 'get_range' then range_request(opts)
           end
         end
-        File.rename(@temp_path, @destination) if @temp_path
+        File.rename(opts[:temp_path], destination) if opts[:temp_path]
       ensure
-        File.delete(@temp_path) if @temp_path && File.exist?(@temp_path)
+        cleanup_temp_file(opts)
       end
 
       private
 
-      def validate!
-        return unless @on_checksum_validated && !@on_checksum_validated.respond_to?(:call)
+      def build_download_opts(destination, opts)
+        {
+          destination: destination,
+          mode: opts.delete(:mode) || 'auto',
+          chunk_size: opts.delete(:chunk_size),
+          on_checksum_validated: opts.delete(:on_checksum_validated),
+          progress_callback: opts.delete(:progress_callback),
+          params: opts,
+          temp_path: nil
+        }
+      end
+
+      def cleanup_temp_file(opts)
+        return unless opts
+
+        temp_file = opts[:temp_path]
+        File.delete(temp_file) if temp_file && File.exist?(temp_file)
+      end
+
+      def download_with_executor(part_list, total_size, opts)
+        download_attempts = 0
+        completion_queue = Queue.new
+        abort_download = false
+        error = nil
+        progress = MultipartProgress.new(part_list, total_size, opts[:progress_callback])
+
+        while (part = part_list.shift)
+          break if abort_download
+
+          download_attempts += 1
+          @executor.post(part) do |p|
+            update_progress(progress, p)
+            resp = @client.get_object(p.params)
+            range = extract_range(resp.content_range)
+            validate_range(range, p.params[:range]) if p.params[:range]
+            write(resp.body, range, opts)
+
+            execute_checksum_callback(resp, opts)
+          rescue StandardError => e
+            abort_download = true
+            error = e
+          ensure
+            completion_queue << :done
+          end
+        end
+
+        download_attempts.times { completion_queue.pop }
+        raise error unless error.nil?
+      end
+
+      def get_opts(opts)
+        GET_OPTIONS.each_with_object({}) { |k, h| h[k] = opts[k] if opts.key?(k) }
+      end
+
+      def head_opts(opts)
+        HEAD_OPTIONS.each_with_object({}) { |k, h| h[k] = opts[k] if opts.key?(k) }
+      end
+
+      def compute_chunk(chunk_size, file_size)
+        raise ArgumentError, ":chunk_size shouldn't exceed total file size." if chunk_size && chunk_size > file_size
 
-        raise ArgumentError, ':on_checksum_validated must be callable'
+        chunk_size || [(file_size.to_f / MAX_PARTS).ceil, MIN_CHUNK_SIZE].max.to_i
       end
 
-      def multipart_download
-        resp = @client.head_object(@params.merge(part_number: 1))
+      def compute_mode(file_size, total_parts, etag, opts)
+        chunk_size = compute_chunk(opts[:chunk_size], file_size)
+        part_size = (file_size.to_f / total_parts).ceil
+
+        resolve_temp_path(opts)
+        if chunk_size < part_size
+          multithreaded_get_by_ranges(file_size, etag, opts)
+        else
+          multithreaded_get_by_parts(total_parts, file_size, etag, opts)
+        end
+      end
+
+      def extract_range(value)
+        value.match(%r{bytes (?<range>\d+-\d+)/\d+})[:range]
+      end
+
+      def multipart_download(opts)
+        resp = @client.head_object(head_opts(opts[:params].merge(part_number: 1)))
         count = resp.parts_count
 
         if count.nil? || count <= 1
           if resp.content_length <= MIN_CHUNK_SIZE
-            single_request
+            single_request(opts)
           else
-            multithreaded_get_by_ranges(resp.content_length, resp.etag)
+            resolve_temp_path(opts)
+            multithreaded_get_by_ranges(resp.content_length, resp.etag, opts)
           end
         else
           # covers cases when given object is not uploaded via UploadPart API
-          resp = @client.head_object(@params) # partNumber is an option
+          resp = @client.head_object(head_opts(opts[:params])) # partNumber is an option
           if resp.content_length <= MIN_CHUNK_SIZE
-            single_request
+            single_request(opts)
           else
-            compute_mode(resp.content_length, count, resp.etag)
+            compute_mode(resp.content_length, count, resp.etag, opts)
           end
         end
       end
 
-      def compute_mode(file_size, count, etag)
-        chunk_size = compute_chunk(file_size)
-        part_size = (file_size.to_f / count).ceil
-        if chunk_size < part_size
-          multithreaded_get_by_ranges(file_size, etag)
-        else
-          multithreaded_get_by_parts(count, file_size, etag)
+      def multithreaded_get_by_parts(total_parts, file_size, etag, opts)
+        parts = (1..total_parts).map do |part|
+          params = get_opts(opts[:params].merge(part_number: part, if_match: etag))
+          Part.new(part_number: part, params: params)
         end
+        download_with_executor(PartList.new(parts), file_size, opts)
       end
 
-      def compute_chunk(file_size)
-        raise ArgumentError, ":chunk_size shouldn't exceed total file size." if @chunk_size && @chunk_size > file_size
-
-        @chunk_size || [(file_size.to_f / MAX_PARTS).ceil, MIN_CHUNK_SIZE].max.to_i
-      end
-
-      def multithreaded_get_by_ranges(file_size, etag)
+      def multithreaded_get_by_ranges(file_size, etag, opts)
         offset = 0
-        default_chunk_size = compute_chunk(file_size)
+        default_chunk_size = compute_chunk(opts[:chunk_size], file_size)
         chunks = []
         part_number = 1 # parts start at 1
         while offset < file_size
           progress = offset + default_chunk_size
           progress = file_size if progress > file_size
-          params = @params.merge(range: "bytes=#{offset}-#{progress - 1}", if_match: etag)
+          params = get_opts(opts[:params].merge(range: "bytes=#{offset}-#{progress - 1}", if_match: etag))
           chunks << Part.new(part_number: part_number, size: (progress - offset), params: params)
           part_number += 1
           offset = progress
         end
-        download_in_threads(PartList.new(chunks), file_size)
-      end
-
-      def multithreaded_get_by_parts(n_parts, total_size, etag)
-        parts = (1..n_parts).map do |part|
-          Part.new(part_number: part, params: @params.merge(part_number: part, if_match: etag))
-        end
-        download_in_threads(PartList.new(parts), total_size)
-      end
-
-      def download_in_threads(pending, total_size)
-        threads = []
-        progress = MultipartProgress.new(pending, total_size, @progress_callback) if @progress_callback
-        unless [File, Tempfile].include?(@destination.class)
-          @temp_path = "#{@destination}.s3tmp.#{SecureRandom.alphanumeric(8)}"
-        end
-        @thread_count.times do
-          thread = Thread.new do
-            begin
-              while (part = pending.shift)
-                if progress
-                  part.params[:on_chunk_received] =
-                    proc do |_chunk, bytes, total|
-                      progress.call(part.part_number, bytes, total)
-                    end
-                end
-                resp = @client.get_object(part.params)
-                range = extract_range(resp.content_range)
-                validate_range(range, part.params[:range]) if part.params[:range]
-                write(resp.body, range)
-                if @on_checksum_validated && resp.checksum_validated
-                  @on_checksum_validated.call(resp.checksum_validated, resp)
-                end
-              end
-              nil
-            rescue StandardError => e
-              pending.clear! # keep other threads from downloading other parts
-              raise e
-            end
-          end
-          threads << thread
-        end
-        threads.map(&:value).compact
+        download_with_executor(PartList.new(chunks), file_size, opts)
       end
 
-      def extract_range(value)
-        value.match(%r{bytes (?<range>\d+-\d+)/\d+})[:range]
+      def range_request(opts)
+        resp = @client.head_object(head_opts(opts[:params]))
+        resolve_temp_path(opts)
+        multithreaded_get_by_ranges(resp.content_length, resp.etag, opts)
       end
 
-      def validate_range(actual, expected)
-        return if actual == expected.match(/bytes=(?<range>\d+-\d+)/)[:range]
-
-        raise MultipartDownloadError, "multipart download failed: expected range of #{expected} but got #{actual}"
-      end
+      def resolve_temp_path(opts)
+        return if [File, Tempfile].include?(opts[:destination].class)
 
-      def write(body, range)
-        path = @temp_path || @destination
-        File.write(path, body.read, range.split('-').first.to_i)
+        opts[:temp_path] ||= "#{opts[:destination]}.s3tmp.#{SecureRandom.alphanumeric(8)}"
       end
 
-      def single_request
-        params = @params.merge(response_target: @destination)
-        params[:on_chunk_received] = single_part_progress if @progress_callback
+      def single_request(opts)
+        params = get_opts(opts[:params]).merge(response_target: opts[:destination])
+        params[:on_chunk_received] = single_part_progress(opts) if opts[:progress_callback]
         resp = @client.get_object(params)
-        return resp unless @on_checksum_validated
+        return resp unless opts[:on_checksum_validated]
 
-        @on_checksum_validated.call(resp.checksum_validated, resp) if resp.checksum_validated
+        opts[:on_checksum_validated].call(resp.checksum_validated, resp) if resp.checksum_validated
         resp
       end
 
-      def single_part_progress
+      def single_part_progress(opts)
         proc do |_chunk, bytes_read, total_size|
-          @progress_callback.call([bytes_read], [total_size], total_size)
+          opts[:progress_callback].call([bytes_read], [total_size], total_size)
         end
       end
 
+      def update_progress(progress, part)
+        return unless progress.progress_callback
+
+        part.params[:on_chunk_received] =
+          proc do |_chunk, bytes, total|
+            progress.call(part.part_number, bytes, total)
+          end
+      end
+
+      def execute_checksum_callback(resp, opts)
+        return unless opts[:on_checksum_validated] && resp.checksum_validated
+
+        opts[:on_checksum_validated].call(resp.checksum_validated, resp)
+      end
+
+      def validate_destination!(destination)
+        valid_types = [String, Pathname, File, Tempfile]
+        return if valid_types.include?(destination.class)
+
+        raise ArgumentError, "Invalid destination, expected #{valid_types.join(', ')} but got: #{destination.class}"
+      end
+
+      def validate_opts!(opts)
+        if opts[:on_checksum_validated] && !opts[:on_checksum_validated].respond_to?(:call)
+          raise ArgumentError, ':on_checksum_validated must be callable'
+        end
+
+        valid_modes = %w[auto get_range single_request]
+        unless valid_modes.include?(opts[:mode])
+          msg = "Invalid mode #{opts[:mode]} provided, :mode should be single_request, get_range or auto"
+          raise ArgumentError, msg
+        end
+
+        if opts[:mode] == 'get_range' && opts[:chunk_size].nil?
+          raise ArgumentError, 'In get_range mode, :chunk_size must be provided'
+        end
+
+        if opts[:chunk_size] && opts[:chunk_size] <= 0
+          raise ArgumentError, ':chunk_size must be positive'
+        end
+      end
+
+      def validate_range(actual, expected)
+        return if actual == expected.match(/bytes=(?<range>\d+-\d+)/)[:range]
+
+        raise MultipartDownloadError, "multipart download failed: expected range of #{expected} but got #{actual}"
+      end
+
+      def write(body, range, opts)
+        path = opts[:temp_path] || opts[:destination]
+        File.write(path, body.read, range.split('-').first.to_i)
+      end
+
       # @api private
       class Part < Struct.new(:part_number, :size, :params)
         include Aws::Structure
@@ -225,6 +286,8 @@ module Aws
           @progress_callback = progress_callback
         end
 
+        attr_reader :progress_callback
+
         def call(part_number, bytes_received, total)
           # part numbers start at 1
           @bytes_received[part_number - 1] = bytes_received

data/lib/aws-sdk-s3/file_uploader.rb

@@ -13,8 +13,8 @@ module Aws
       # @option options [Client] :client
       # @option options [Integer] :multipart_threshold (104857600)
       def initialize(options = {})
-        @options = options
         @client = options[:client] || Client.new
+        @executor = options[:executor]
         @multipart_threshold = options[:multipart_threshold] || DEFAULT_MULTIPART_THRESHOLD
       end
 
@@ -36,11 +36,9 @@ module Aws
       # @return [void]
       def upload(source, options = {})
         Aws::Plugins::UserAgent.metric('S3_TRANSFER') do
-          if File.size(source) >= multipart_threshold
-            MultipartFileUploader.new(@options).upload(source, options)
+          if File.size(source) >= @multipart_threshold
+            MultipartFileUploader.new(client: @client, executor: @executor).upload(source, options)
           else
-            # remove multipart parameters not supported by put_object
-            options.delete(:thread_count)
             put_object(source, options)
           end
         end
@@ -48,9 +46,9 @@ module Aws
 
       private
 
-      def open_file(source)
-        if String === source || Pathname === source
-          File.open(source, 'rb') { |file| yield(file) }
+      def open_file(source, &block)
+        if source.is_a?(String) || source.is_a?(Pathname)
+          File.open(source, 'rb', &block)
         else
           yield(source)
         end

data/lib/aws-sdk-s3/multipart_file_uploader.rb

@@ -7,10 +7,8 @@ module Aws
   module S3
     # @api private
     class MultipartFileUploader
-
       MIN_PART_SIZE = 5 * 1024 * 1024 # 5MB
       MAX_PARTS = 10_000
-      DEFAULT_THREAD_COUNT = 10
       CREATE_OPTIONS = Set.new(Client.api.operation(:create_multipart_upload).input.shape.member_names)
       COMPLETE_OPTIONS = Set.new(Client.api.operation(:complete_multipart_upload).input.shape.member_names)
       UPLOAD_PART_OPTIONS = Set.new(Client.api.operation(:upload_part).input.shape.member_names)
@@ -21,10 +19,9 @@ module Aws
       )
 
       # @option options [Client] :client
-      # @option options [Integer] :thread_count (DEFAULT_THREAD_COUNT)
       def initialize(options = {})
         @client = options[:client] || Client.new
-        @thread_count = options[:thread_count] || DEFAULT_THREAD_COUNT
+        @executor = options[:executor]
       end
 
       # @return [Client]
@@ -38,11 +35,12 @@ module Aws
       #   It will be invoked with [bytes_read], [total_sizes]
       # @return [Seahorse::Client::Response] - the CompleteMultipartUploadResponse
       def upload(source, options = {})
-        raise ArgumentError, 'unable to multipart upload files smaller than 5MB' if File.size(source) < MIN_PART_SIZE
+        file_size = File.size(source)
+        raise ArgumentError, 'unable to multipart upload files smaller than 5MB' if file_size < MIN_PART_SIZE
 
         upload_id = initiate_upload(options)
-        parts = upload_parts(upload_id, source, options)
-        complete_upload(upload_id, parts, source, options)
+        parts = upload_parts(upload_id, source, file_size, options)
+        complete_upload(upload_id, parts, file_size, options)
       end
 
       private
@@ -51,22 +49,21 @@ module Aws
         @client.create_multipart_upload(create_opts(options)).upload_id
       end
 
-      def complete_upload(upload_id, parts, source, options)
+      def complete_upload(upload_id, parts, file_size, options)
         @client.complete_multipart_upload(
-          **complete_opts(options).merge(
-            upload_id: upload_id,
-            multipart_upload: { parts: parts },
-            mpu_object_size: File.size(source)
-          )
+          **complete_opts(options),
+          upload_id: upload_id,
+          multipart_upload: { parts: parts },
+          mpu_object_size: file_size
         )
       rescue StandardError => e
         abort_upload(upload_id, options, [e])
       end
 
-      def upload_parts(upload_id, source, options)
+      def upload_parts(upload_id, source, file_size, options)
         completed = PartList.new
-        pending = PartList.new(compute_parts(upload_id, source, options))
-        errors = upload_in_threads(pending, completed, options)
+        pending = PartList.new(compute_parts(upload_id, source, file_size, options))
+        errors = upload_with_executor(pending, completed, options)
         if errors.empty?
           completed.to_a.sort_by { |part| part[:part_number] }
         else
@@ -81,22 +78,25 @@ module Aws
       rescue MultipartUploadError => e
         raise e
       rescue StandardError => e
-        msg = "failed to abort multipart upload: #{e.message}. "\
-          "Multipart upload failed: #{errors.map(&:message).join('; ')}"
+        msg = "failed to abort multipart upload: #{e.message}. " \
+              "Multipart upload failed: #{errors.map(&:message).join('; ')}"
         raise MultipartUploadError.new(msg, errors + [e])
       end
 
-      def compute_parts(upload_id, source, options)
-        size = File.size(source)
-        default_part_size = compute_default_part_size(size)
+      def compute_parts(upload_id, source, file_size, options)
+        default_part_size = compute_default_part_size(file_size)
         offset = 0
         part_number = 1
         parts = []
-        while offset < size
+        while offset < file_size
           parts << upload_part_opts(options).merge(
             upload_id: upload_id,
             part_number: part_number,
-            body: FilePart.new(source: source, offset: offset, size: part_size(size, default_part_size, offset))
+            body: FilePart.new(
+              source: source,
+              offset: offset,
+              size: part_size(file_size, default_part_size, offset)
+            )
           )
           part_number += 1
           offset += default_part_size
@@ -112,20 +112,23 @@ module Aws
         keys.any? { |key| checksum_key?(key) }
       end
 
+      def checksum_not_required?(options)
+        @client.config.request_checksum_calculation == 'when_required' && !options[:checksum_algorithm]
+      end
+
       def create_opts(options)
-        opts = { checksum_algorithm: Aws::Plugins::ChecksumAlgorithm::DEFAULT_CHECKSUM }
-        opts[:checksum_type] = 'FULL_OBJECT' if has_checksum_key?(options.keys)
-        CREATE_OPTIONS.each_with_object(opts) do |key, hash|
-          hash[key] = options[key] if options.key?(key)
+        opts = {}
+        unless checksum_not_required?(options)
+          opts[:checksum_algorithm] = Aws::Plugins::ChecksumAlgorithm::DEFAULT_CHECKSUM
         end
+        opts[:checksum_type] = 'FULL_OBJECT' if has_checksum_key?(options.keys)
+        CREATE_OPTIONS.each_with_object(opts) { |k, h| h[k] = options[k] if options.key?(k) }
       end
 
       def complete_opts(options)
         opts = {}
         opts[:checksum_type] = 'FULL_OBJECT' if has_checksum_key?(options.keys)
-        COMPLETE_OPTIONS.each_with_object(opts) do |key, hash|
-          hash[key] = options[key] if options.key?(key)
-        end
+        COMPLETE_OPTIONS.each_with_object(opts) { |k, h| h[k] = options[k] if options.key?(k) }
       end
 
       def upload_part_opts(options)
@@ -135,43 +138,45 @@ module Aws
         end
       end
 
-      def upload_in_threads(pending, completed, options)
-        threads = []
-        if (callback = options[:progress_callback])
-          progress = MultipartProgress.new(pending, callback)
-        end
-        options.fetch(:thread_count, @thread_count).times do
-          thread = Thread.new do
-            begin
-              while (part = pending.shift)
-                if progress
-                  part[:on_chunk_sent] =
-                    proc do |_chunk, bytes, _total|
-                      progress.call(part[:part_number], bytes)
-                    end
-                end
-                resp = @client.upload_part(part)
-                part[:body].close
-                completed_part = { etag: resp.etag, part_number: part[:part_number] }
-                algorithm = resp.context.params[:checksum_algorithm]
-                k = "checksum_#{algorithm.downcase}".to_sym
-                completed_part[k] = resp.send(k)
-                completed.push(completed_part)
-              end
-              nil
-            rescue StandardError => e
-              # keep other threads from uploading other parts
-              pending.clear!
-              e
-            end
+      def upload_with_executor(pending, completed, options)
+        upload_attempts = 0
+        completion_queue = Queue.new
+        abort_upload = false
+        errors = []
+        progress = MultipartProgress.new(pending, options[:progress_callback])
+
+        while (part = pending.shift)
+          break if abort_upload
+
+          upload_attempts += 1
+          @executor.post(part) do |p|
+            update_progress(progress, p)
+            resp = @client.upload_part(p)
+            p[:body].close
+            completed_part = { etag: resp.etag, part_number: p[:part_number] }
+            apply_part_checksum(resp, completed_part)
+            completed.push(completed_part)
+          rescue StandardError => e
+            abort_upload = true
+            errors << e
+          ensure
+            completion_queue << :done
           end
-          threads << thread
         end
-        threads.map(&:value).compact
+
+        upload_attempts.times { completion_queue.pop }
+        errors
       end
 
-      def compute_default_part_size(source_size)
-        [(source_size.to_f / MAX_PARTS).ceil, MIN_PART_SIZE].max.to_i
+      def apply_part_checksum(resp, part)
+        return unless (checksum = resp.context.params[:checksum_algorithm])
+
+        k = :"checksum_#{checksum.downcase}"
+        part[k] = resp.send(k)
+      end
+
+      def compute_default_part_size(file_size)
+        [(file_size.to_f / MAX_PARTS).ceil, MIN_PART_SIZE].max.to_i
       end
 
       def part_size(total_size, part_size, offset)
@@ -182,6 +187,15 @@ module Aws
         end
       end
 
+      def update_progress(progress, part)
+        return unless progress.progress_callback
+
+        part[:on_chunk_sent] =
+          proc do |_chunk, bytes, _total|
+            progress.call(part[:part_number], bytes)
+          end
+      end
+
       # @api private
       class PartList
         def initialize(parts = [])
@@ -222,6 +236,8 @@ module Aws
           @progress_callback = progress_callback
         end
 
+        attr_reader :progress_callback
+
         def call(part_number, bytes_read)
           # part numbers start at 1
           @bytes_sent[part_number - 1] = bytes_read