aws-sdk-s3 1.199.1 → 1.202.0

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

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

@@ -11,7 +11,6 @@ module Aws
     class MultipartStreamUploader
 
       DEFAULT_PART_SIZE = 5 * 1024 * 1024 # 5MB
-      DEFAULT_THREAD_COUNT = 10
       CREATE_OPTIONS = Set.new(Client.api.operation(:create_multipart_upload).input.shape.member_names)
       UPLOAD_PART_OPTIONS = Set.new(Client.api.operation(:upload_part).input.shape.member_names)
       COMPLETE_UPLOAD_OPTIONS = Set.new(Client.api.operation(:complete_multipart_upload).input.shape.member_names)
@@ -19,9 +18,9 @@ module Aws
       # @option options [Client] :client
       def initialize(options = {})
         @client = options[:client] || Client.new
+        @executor = options[:executor]
         @tempfile = options[:tempfile]
         @part_size = options[:part_size] || DEFAULT_PART_SIZE
-        @thread_count = options[:thread_count] || DEFAULT_THREAD_COUNT
       end
 
       # @return [Client]
@@ -29,7 +28,6 @@ module Aws
 
       # @option options [required,String] :bucket
       # @option options [required,String] :key
-      # @option options [Integer] :thread_count (DEFAULT_THREAD_COUNT)
       # @return [Seahorse::Client::Response] - the CompleteMultipartUploadResponse
       def upload(options = {}, &block)
         Aws::Plugins::UserAgent.metric('S3_TRANSFER') do
@@ -54,28 +52,30 @@ module Aws
       end
 
       def upload_parts(upload_id, options, &block)
-        completed = Queue.new
-        thread_errors = []
-        errors = begin
+        completed_parts = Queue.new
+        errors = []
+
+        begin
           IO.pipe do |read_pipe, write_pipe|
-            threads = upload_in_threads(
-              read_pipe,
-              completed,
-              upload_part_opts(options).merge(upload_id: upload_id),
-              thread_errors
-            )
-            begin
-              block.call(write_pipe)
-            ensure
-              # Ensure the pipe is closed to avoid https://github.com/jruby/jruby/issues/6111
-              write_pipe.close
+            upload_thread = Thread.new do
+              upload_with_executor(
+                read_pipe,
+                completed_parts,
+                errors,
+                upload_part_opts(options).merge(upload_id: upload_id)
+              )
             end
-            threads.map(&:value).compact
+
+            block.call(write_pipe)
+          ensure
+            # Ensure the pipe is closed to avoid https://github.com/jruby/jruby/issues/6111
+            write_pipe.close
+            upload_thread.join
           end
         rescue StandardError => e
-          thread_errors + [e]
+          errors << e
         end
-        return ordered_parts(completed) if errors.empty?
+        return ordered_parts(completed_parts) if errors.empty?
 
         abort_upload(upload_id, options, errors)
       end
@@ -128,37 +128,34 @@ module Aws
         end
       end
 
-      def upload_in_threads(read_pipe, completed, options, thread_errors)
-        mutex = Mutex.new
+      def upload_with_executor(read_pipe, completed, errors, options)
+        completion_queue = Queue.new
+        queued_parts = 0
         part_number = 0
-        options.fetch(:thread_count, @thread_count).times.map do
-          thread = Thread.new do
-            loop do
-              body, thread_part_number = mutex.synchronize do
-                [read_to_part_body(read_pipe), part_number += 1]
-              end
-              break unless body || thread_part_number == 1
-
-              begin
-                part = options.merge(body: body, part_number: thread_part_number)
-                resp = @client.upload_part(part)
-                completed_part = create_completed_part(resp, part)
-                completed.push(completed_part)
-              ensure
-                clear_body(body)
-              end
-            end
-            nil
+        mutex = Mutex.new
+        loop do
+          part_body, current_part_num = mutex.synchronize do
+            [read_to_part_body(read_pipe), part_number += 1]
+          end
+          break unless part_body || current_part_num == 1
+
+          queued_parts += 1
+          @executor.post(part_body, current_part_num, options) do |body, num, opts|
+            part = opts.merge(body: body, part_number: num)
+            resp = @client.upload_part(part)
+            completed_part = create_completed_part(resp, part)
+            completed.push(completed_part)
           rescue StandardError => e
-            # keep other threads from uploading other parts
             mutex.synchronize do
-              thread_errors.push(e)
+              errors.push(e)
               read_pipe.close_read unless read_pipe.closed?
             end
-            e
+          ensure
+            clear_body(body)
+            completion_queue << :done
           end
-          thread
         end
+        queued_parts.times { completion_queue.pop }
       end
 
       def create_completed_part(resp, part)

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

@@ -136,10 +136,10 @@ module Aws::S3
     end
 
     # The Base64 encoded, 32-bit `CRC32 checksum` of the object. This
-    # checksum is only be present if the checksum was uploaded with the
-    # object. When you use an API operation on an object that was uploaded
-    # using multipart uploads, this value may not be a direct checksum value
-    # of the full object. Instead, it's a calculation based on the checksum
+    # checksum is only present if the checksum was uploaded with the object.
+    # When you use an API operation on an object that was uploaded using
+    # multipart uploads, this value may not be a direct checksum value of
+    # the full object. Instead, it's a calculation based on the checksum
     # values of each individual part. For more information about how
     # checksums are calculated with multipart uploads, see [ Checking object
     # integrity][1] in the *Amazon S3 User Guide*.
@@ -181,8 +181,8 @@ module Aws::S3
       data[:checksum_crc64nvme]
     end
 
-    # The Base64 encoded, 160-bit `SHA1` digest of the object. This will
-    # only be present if the object was uploaded with the object. When you
+    # The Base64 encoded, 160-bit `SHA1` digest of the object. This checksum
+    # is only present if the checksum was uploaded with the object. When you
     # use the API operation on an object that was uploaded using multipart
     # uploads, this value may not be a direct checksum value of the full
     # object. Instead, it's a calculation based on the checksum values of
@@ -198,14 +198,14 @@ module Aws::S3
       data[:checksum_sha1]
     end
 
-    # The Base64 encoded, 256-bit `SHA256` digest of the object. This will
-    # only be present if the object was uploaded with the object. When you
-    # use an API operation on an object that was uploaded using multipart
-    # uploads, this value may not be a direct checksum value of the full
-    # object. Instead, it's a calculation based on the checksum values of
-    # each individual part. For more information about how checksums are
-    # calculated with multipart uploads, see [ Checking object integrity][1]
-    # in the *Amazon S3 User Guide*.
+    # The Base64 encoded, 256-bit `SHA256` digest of the object. This
+    # checksum is only present if the checksum was uploaded with the object.
+    # When you use an API operation on an object that was uploaded using
+    # multipart uploads, this value may not be a direct checksum value of
+    # the full object. Instead, it's a calculation based on the checksum
+    # values of each individual part. For more information about how
+    # checksums are calculated with multipart uploads, see [ Checking object
+    # integrity][1] in the *Amazon S3 User Guide*.
     #
     #
     #
@@ -757,6 +757,8 @@ module Aws::S3
     #     grant_read: "GrantRead",
     #     grant_read_acp: "GrantReadACP",
     #     grant_write_acp: "GrantWriteACP",
+    #     if_match: "IfMatch",
+    #     if_none_match: "IfNoneMatch",
     #     metadata: {
     #       "MetadataKey" => "MetadataValue",
     #     },
@@ -1013,6 +1015,35 @@ module Aws::S3
     #   * This functionality is not supported for Amazon S3 on Outposts.
     #
     #    </note>
+    # @option options [String] :if_match
+    #   Copies the object if the entity tag (ETag) of the destination object
+    #   matches the specified tag. If the ETag values do not match, the
+    #   operation returns a `412 Precondition Failed` error. If a concurrent
+    #   operation occurs during the upload S3 returns a `409
+    #   ConditionalRequestConflict` response. On a 409 failure you should
+    #   fetch the object's ETag and retry the upload.
+    #
+    #   Expects the ETag value as a string.
+    #
+    #   For more information about conditional requests, see [RFC 7232][1].
+    #
+    #
+    #
+    #   [1]: https://tools.ietf.org/html/rfc7232
+    # @option options [String] :if_none_match
+    #   Copies the object only if the object key name at the destination does
+    #   not already exist in the bucket specified. Otherwise, Amazon S3
+    #   returns a `412 Precondition Failed` error. If a concurrent operation
+    #   occurs during the upload S3 returns a `409 ConditionalRequestConflict`
+    #   response. On a 409 failure you should retry the upload.
+    #
+    #   Expects the '*' (asterisk) character.
+    #
+    #   For more information about conditional requests, see [RFC 7232][1].
+    #
+    #
+    #
+    #   [1]: https://tools.ietf.org/html/rfc7232
     # @option options [Hash<String,String>] :metadata
     #   A map of metadata to store with the object in S3.
     # @option options [String] :metadata_directive
@@ -1535,17 +1566,15 @@ module Aws::S3
     #   you provide does not match the actual owner of the bucket, the request
     #   fails with the HTTP status code `403 Forbidden` (access denied).
     # @option options [String] :if_match
-    #   The `If-Match` header field makes the request method conditional on
-    #   ETags. If the ETag value does not match, the operation returns a `412
-    #   Precondition Failed` error. If the ETag matches or if the object
-    #   doesn't exist, the operation will return a `204 Success (No Content)
-    #   response`.
-    #
-    #   For more information about conditional requests, see [RFC 7232][1].
+    #   Deletes the object if the ETag (entity tag) value provided during the
+    #   delete operation matches the ETag of the object in S3. If the ETag
+    #   values do not match, the operation returns a `412 Precondition Failed`
+    #   error.
     #
-    #   <note markdown="1"> This functionality is only supported for directory buckets.
+    #   Expects the ETag value as a string. `If-Match` does accept a string
+    #   value of an '*' (asterisk) character to denote a match of any ETag.
     #
-    #    </note>
+    #   For more information about conditional requests, see [RFC 7232][1].
     #
     #
     #

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

@@ -354,6 +354,8 @@ module Aws::S3
     #     grant_read: "GrantRead",
     #     grant_read_acp: "GrantReadACP",
     #     grant_write_acp: "GrantWriteACP",
+    #     if_match: "IfMatch",
+    #     if_none_match: "IfNoneMatch",
     #     metadata: {
     #       "MetadataKey" => "MetadataValue",
     #     },
@@ -610,6 +612,35 @@ module Aws::S3
     #   * This functionality is not supported for Amazon S3 on Outposts.
     #
     #    </note>
+    # @option options [String] :if_match
+    #   Copies the object if the entity tag (ETag) of the destination object
+    #   matches the specified tag. If the ETag values do not match, the
+    #   operation returns a `412 Precondition Failed` error. If a concurrent
+    #   operation occurs during the upload S3 returns a `409
+    #   ConditionalRequestConflict` response. On a 409 failure you should
+    #   fetch the object's ETag and retry the upload.
+    #
+    #   Expects the ETag value as a string.
+    #
+    #   For more information about conditional requests, see [RFC 7232][1].
+    #
+    #
+    #
+    #   [1]: https://tools.ietf.org/html/rfc7232
+    # @option options [String] :if_none_match
+    #   Copies the object only if the object key name at the destination does
+    #   not already exist in the bucket specified. Otherwise, Amazon S3
+    #   returns a `412 Precondition Failed` error. If a concurrent operation
+    #   occurs during the upload S3 returns a `409 ConditionalRequestConflict`
+    #   response. On a 409 failure you should retry the upload.
+    #
+    #   Expects the '*' (asterisk) character.
+    #
+    #   For more information about conditional requests, see [RFC 7232][1].
+    #
+    #
+    #
+    #   [1]: https://tools.ietf.org/html/rfc7232
     # @option options [Hash<String,String>] :metadata
     #   A map of metadata to store with the object in S3.
     # @option options [String] :metadata_directive
@@ -1132,17 +1163,15 @@ module Aws::S3
     #   you provide does not match the actual owner of the bucket, the request
     #   fails with the HTTP status code `403 Forbidden` (access denied).
     # @option options [String] :if_match
-    #   The `If-Match` header field makes the request method conditional on
-    #   ETags. If the ETag value does not match, the operation returns a `412
-    #   Precondition Failed` error. If the ETag matches or if the object
-    #   doesn't exist, the operation will return a `204 Success (No Content)
-    #   response`.
+    #   Deletes the object if the ETag (entity tag) value provided during the
+    #   delete operation matches the ETag of the object in S3. If the ETag
+    #   values do not match, the operation returns a `412 Precondition Failed`
+    #   error.
     #
-    #   For more information about conditional requests, see [RFC 7232][1].
+    #   Expects the ETag value as a string. `If-Match` does accept a string
+    #   value of an '*' (asterisk) character to denote a match of any ETag.
     #
-    #   <note markdown="1"> This functionality is only supported for directory buckets.
-    #
-    #    </note>
+    #   For more information about conditional requests, see [RFC 7232][1].
     #
     #
     #

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

@@ -312,17 +312,15 @@ module Aws::S3
     #   you provide does not match the actual owner of the bucket, the request
     #   fails with the HTTP status code `403 Forbidden` (access denied).
     # @option options [String] :if_match
-    #   The `If-Match` header field makes the request method conditional on
-    #   ETags. If the ETag value does not match, the operation returns a `412
-    #   Precondition Failed` error. If the ETag matches or if the object
-    #   doesn't exist, the operation will return a `204 Success (No Content)
-    #   response`.
-    #
-    #   For more information about conditional requests, see [RFC 7232][1].
+    #   Deletes the object if the ETag (entity tag) value provided during the
+    #   delete operation matches the ETag of the object in S3. If the ETag
+    #   values do not match, the operation returns a `412 Precondition Failed`
+    #   error.
     #
-    #   <note markdown="1"> This functionality is only supported for directory buckets.
+    #   Expects the ETag value as a string. `If-Match` does accept a string
+    #   value of an '*' (asterisk) character to denote a match of any ETag.
     #
-    #    </note>
+    #   For more information about conditional requests, see [RFC 7232][1].
     #
     #
     #

data/lib/aws-sdk-s3/plugins/endpoints.rb

@@ -25,7 +25,7 @@ The endpoint provider used to resolve endpoints. Any object that responds to
 
       option(
         :disable_s3_express_session_auth,
-        doc_type: 'Boolean',
+        doc_type: 'boolean',
         docstring: <<~DOCS) do |cfg|
 Parameter to indicate whether S3Express session auth should be disabled
         DOCS