aws-sdk-s3 1.79.1 → 1.212.0

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

@@ -7,23 +7,22 @@ module Aws
     # @api private
     class FileUploader
 
-      FIFTEEN_MEGABYTES = 15 * 1024 * 1024
+      DEFAULT_MULTIPART_THRESHOLD = 100 * 1024 * 1024
 
       # @param [Hash] options
       # @option options [Client] :client
-      # @option options [Integer] :multipart_threshold (15728640)
+      # @option options [Integer] :multipart_threshold (104857600)
       def initialize(options = {})
-        @options = options
         @client = options[:client] || Client.new
-        @multipart_threshold = options[:multipart_threshold] ||
-                               FIFTEEN_MEGABYTES
+        @executor = options[:executor]
+        @http_chunk_size = options[:http_chunk_size]
+        @multipart_threshold = options[:multipart_threshold] || DEFAULT_MULTIPART_THRESHOLD
       end
 
       # @return [Client]
       attr_reader :client
 
-      # @return [Integer] Files larger than this in bytes are uploaded
-      #   using a {MultipartFileUploader}.
+      # @return [Integer] Files larger than or equal to this in bytes are uploaded using a {MultipartFileUploader}.
       attr_reader :multipart_threshold
 
       # @param [String, Pathname, File, Tempfile] source The file to upload.
@@ -32,20 +31,29 @@ module Aws
       # @option options [Proc] :progress_callback
       #   A Proc that will be called when each chunk of the upload is sent.
       #   It will be invoked with [bytes_read], [total_sizes]
+      # @option options [Integer] :thread_count
+      #   The thread count to use for multipart uploads. Ignored for
+      #   objects smaller than the multipart threshold.
       # @return [void]
       def upload(source, options = {})
-        if File.size(source) >= multipart_threshold
-          MultipartFileUploader.new(@options).upload(source, options)
-        else
-          put_object(source, options)
+        Aws::Plugins::UserAgent.metric('S3_TRANSFER') do
+          if File.size(source) >= @multipart_threshold
+            MultipartFileUploader.new(
+              client: @client,
+              executor: @executor,
+              http_chunk_size: @http_chunk_size
+            ).upload(source, options)
+          else
+            put_object(source, options)
+          end
         end
       end
 
       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
@@ -56,7 +64,10 @@ module Aws
           options[:on_chunk_sent] = single_part_progress(callback)
         end
         open_file(source) do |file|
+          Thread.current[:net_http_override_body_stream_chunk] = @http_chunk_size if @http_chunk_size
           @client.put_object(options.merge(body: file))
+        ensure
+          Thread.current[:net_http_override_body_stream_chunk] = nil
         end
       end
 

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

@@ -3,8 +3,8 @@
 require 'set'
 require 'time'
 require 'openssl'
-require 'cgi'
-require 'webrick/httputils'
+require "cgi/escape"
+require "cgi/util" if RUBY_VERSION < "3.5"
 require 'aws-sdk-core/query'
 
 module Aws
@@ -157,33 +157,24 @@ module Aws
       end
 
       def uri_escape(s)
-
         #URI.escape(s)
 
-        # URI.escape is deprecated, replacing it with escape from webrick
-        # to squelch the massive number of warnings generated from Ruby.
-        # The following script was used to determine the differences
-        # between the various escape methods available. The webrick
-        # escape only had two differences and it is available in the
-        # standard lib.
-        #
-        #     (0..255).each {|c|
-        #       s = [c].pack("C")
-        #       e = [
-        #         CGI.escape(s),
-        #         ERB::Util.url_encode(s),
-        #         URI.encode_www_form_component(s),
-        #         WEBrick::HTTPUtils.escape_form(s),
-        #         WEBrick::HTTPUtils.escape(s),
-        #         URI.escape(s),
-        #       ]
-        #       next if e.uniq.length == 1
-        #       puts("%5s %5s %5s %5s %5s %5s %5s" % ([s.inspect] + e))
-        #     }
-        #
-        WEBrick::HTTPUtils.escape(s).gsub('%5B', '[').gsub('%5D', ']')
+        # (0..255).each {|c|
+        #   s = [c].pack("C")
+        #   e = [
+        #     CGI.escape(s),
+        #     ERB::Util.url_encode(s),
+        #     URI.encode_www_form_component(s),
+        #     WEBrick::HTTPUtils.escape_form(s),
+        #     WEBrick::HTTPUtils.escape(s),
+        #     URI.escape(s),
+        #     URI::DEFAULT_PARSER.escape(s)
+        #   ]
+        #   next if e.uniq.length == 1
+        #   puts("%5s %5s %5s %5s %5s %5s %5s %5s" % ([s.inspect] + e))
+        # }
+        URI::DEFAULT_PARSER.escape(s)
       end
-
     end
   end
 end

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

@@ -0,0 +1,8 @@
+# frozen_string_literal: true
+
+module Aws
+  module S3
+    # Raised when multipart download fails to complete.
+    class MultipartDownloadError < StandardError; end
+  end
+end

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

@@ -7,30 +7,22 @@ module Aws
   module S3
     # @api private
     class MultipartFileUploader
-
       MIN_PART_SIZE = 5 * 1024 * 1024 # 5MB
-
-      FILE_TOO_SMALL = "unable to multipart upload files smaller than 5MB"
-
       MAX_PARTS = 10_000
-
-      THREAD_COUNT = 10
-
-      # @api private
-      CREATE_OPTIONS = Set.new(
-        Client.api.operation(:create_multipart_upload).input.shape.member_names
-      )
-
-      # @api private
-      UPLOAD_PART_OPTIONS = Set.new(
-        Client.api.operation(:upload_part).input.shape.member_names
+      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)
+      CHECKSUM_KEYS = Set.new(
+        Client.api.operation(:upload_part).input.shape.members.map do |n, s|
+          n if s.location == 'header' && s.location_name.start_with?('x-amz-checksum-')
+        end.compact
       )
 
       # @option options [Client] :client
-      # @option options [Integer] :thread_count (THREAD_COUNT)
       def initialize(options = {})
         @client = options[:client] || Client.new
-        @thread_count = options[:thread_count] || THREAD_COUNT
+        @executor = options[:executor]
+        @http_chunk_size = options[:http_chunk_size]
       end
 
       # @return [Client]
@@ -42,15 +34,14 @@ module Aws
       # @option options [Proc] :progress_callback
       #   A Proc that will be called when each chunk of the upload is sent.
       #   It will be invoked with [bytes_read], [total_sizes]
-      # @return [void]
+      # @return [Seahorse::Client::Response] - the CompleteMultipartUploadResponse
       def upload(source, options = {})
-        if File.size(source) < MIN_PART_SIZE
-          raise ArgumentError, FILE_TOO_SMALL
-        else
-          upload_id = initiate_upload(options)
-          parts = upload_parts(upload_id, source, options)
-          complete_upload(upload_id, parts, options)
-        end
+        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, file_size, options)
+        complete_upload(upload_id, parts, file_size, options)
       end
 
       private
@@ -59,19 +50,21 @@ module Aws
         @client.create_multipart_upload(create_opts(options)).upload_id
       end
 
-      def complete_upload(upload_id, parts, options)
+      def complete_upload(upload_id, parts, file_size, options)
         @client.complete_multipart_upload(
-          bucket: options[:bucket],
-          key: options[:key],
+          **complete_opts(options),
           upload_id: upload_id,
-          multipart_upload: { parts: parts }
+          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)
-        pending = PartList.new(compute_parts(upload_id, source, options))
+      def upload_parts(upload_id, source, file_size, options)
         completed = PartList.new
-        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
@@ -80,34 +73,30 @@ module Aws
       end
 
       def abort_upload(upload_id, options, errors)
-        @client.abort_multipart_upload(
-          bucket: options[:bucket],
-          key: options[:key],
-          upload_id: upload_id
-        )
-        msg = "multipart upload failed: #{errors.map(&:message).join("; ")}"
+        @client.abort_multipart_upload(bucket: options[:bucket], key: options[:key], upload_id: upload_id)
+        msg = "multipart upload failed: #{errors.map(&:message).join('; ')}"
         raise MultipartUploadError.new(msg, errors)
-      rescue MultipartUploadError => error
-        raise error
-      rescue => error
-        msg = "failed to abort multipart upload: #{error.message}"
-        raise MultipartUploadError.new(msg, errors + [error])
+      rescue MultipartUploadError => e
+        raise e
+      rescue StandardError => e
+        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)
+              size: part_size(file_size, default_part_size, offset)
             )
           )
           part_number += 1
@@ -116,54 +105,81 @@ module Aws
         parts
       end
 
+      def checksum_key?(key)
+        CHECKSUM_KEYS.include?(key)
+      end
+
+      def has_checksum_key?(keys)
+        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)
-        CREATE_OPTIONS.inject({}) do |hash, key|
-          hash[key] = options[key] if options.key?(key)
-          hash
+        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) { |k, h| h[k] = options[k] if options.key?(k) }
       end
 
       def upload_part_opts(options)
-        UPLOAD_PART_OPTIONS.inject({}) do |hash, key|
-          hash[key] = options[key] if options.key?(key)
-          hash
+        UPLOAD_PART_OPTIONS.each_with_object({}) do |key, hash|
+          # don't pass through checksum calculations
+          hash[key] = options[key] if options.key?(key) && !checksum_key?(key)
         end
       end
 
-      def upload_in_threads(pending, completed, options)
-        threads = []
-        if (callback = options[:progress_callback])
-          progress = MultipartProgress.new(pending, callback)
-        end
-        @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.push(etag: resp.etag, part_number: part[:part_number])
-              end
-              nil
-            rescue => error
-              # keep other threads from uploading other parts
-              pending.clear!
-              error
-            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|
+            Thread.current[:net_http_override_body_stream_chunk] = @http_chunk_size if @http_chunk_size
+            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
+            Thread.current[:net_http_override_body_stream_chunk] = nil if @http_chunk_size
+            completion_queue << :done
           end
-          thread.abort_on_exception = true
-          threads << thread
         end
-        threads.map(&:value).compact
+
+        upload_attempts.times { completion_queue.pop }
+        errors
+      end
+
+      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(source_size)
-        [(source_size.to_f / MAX_PARTS).ceil, MIN_PART_SIZE].max.to_i
+      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)
@@ -174,9 +190,17 @@ 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 = [])
           @parts = parts
           @mutex = Mutex.new
@@ -205,7 +229,6 @@ module Aws
         def to_a
           @mutex.synchronize { @parts.dup }
         end
-
       end
 
       # @api private
@@ -216,6 +239,8 @@ 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