aws-sdk-s3 1.10.0 → 1.208.0

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

@@ -1,138 +1,317 @@
+# frozen_string_literal: true
+
 require 'pathname'
-require 'thread'
+require 'securerandom'
 require 'set'
-require 'tmpdir'
 
 module Aws
   module S3
     # @api private
     class FileDownloader
-
       MIN_CHUNK_SIZE = 5 * 1024 * 1024
       MAX_PARTS = 10_000
-      THREAD_COUNT = 10
+      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 = {})
-        @path = destination
-        @mode = options[:mode] || "auto"
-        @thread_count = options[:thread_count] || THREAD_COUNT
-        @chunk_size = options[:chunk_size]
-        @bucket = options[:bucket]
-        @key = options[:key]
-
-        case @mode
-        when "auto" then multipart_download
-        when "single_request" then single_request
-        when "get_range"
-          if @chunk_size
-            resp = @client.head_object(bucket: @bucket, key: @key)
-            multithreaded_get_by_ranges(construct_chunks(resp.content_length))
-          else
-            msg = "In :get_range mode, :chunk_size must be provided"
-            raise ArgumentError, msg
+        validate_destination!(destination)
+        opts = build_download_opts(destination, options)
+        validate_opts!(opts)
+
+        Aws::Plugins::UserAgent.metric('S3_TRANSFER') do
+          case opts[:mode]
+          when 'auto' then multipart_download(opts)
+          when 'single_request' then single_request(opts)
+          when 'get_range' then range_request(opts)
           end
-        else
-          msg = "Invalid mode #{@mode} provided, "\
-            "mode should be :single_request, :get_range or :auto"
-          raise ArgumentError, msg
         end
+        File.rename(opts[:temp_path], destination) if opts[:temp_path]
+      ensure
+        cleanup_temp_file(opts)
       end
 
       private
 
-      def multipart_download
-        resp = @client.head_object(bucket: @bucket, key: @key, part_number: 1)
+      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 handle_checksum_mode_option(option_key, opts)
+        return false unless option_key == :checksum_mode && opts[:checksum_mode] == 'DISABLED'
+
+        msg = ':checksum_mode option is deprecated. Checksums will be validated by default. ' \
+          'To disable checksum validation, set :response_checksum_validation to "when_required" on your S3 client.'
+        warn(msg)
+        true
+      end
+
+      def get_opts(opts)
+        GET_OPTIONS.each_with_object({}) do |k, h|
+          next if k == :checksum_mode
+
+          h[k] = opts[k] if opts.key?(k)
+        end
+      end
+
+      def head_opts(opts)
+        HEAD_OPTIONS.each_with_object({}) do |k, h|
+          next if handle_checksum_mode_option(k, opts)
+
+          h[k] = opts[k] if opts.key?(k)
+        end
+      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
+
+        chunk_size || [(file_size.to_f / MAX_PARTS).ceil, MIN_CHUNK_SIZE].max.to_i
+      end
+
+      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
-          resp.content_length < MIN_CHUNK_SIZE ?
-            single_request :
-            multithreaded_get_by_ranges(construct_chunks(resp.content_length))
+          if resp.content_length <= MIN_CHUNK_SIZE
+            single_request(opts)
+          else
+            resolve_temp_path(opts)
+            multithreaded_get_by_ranges(resp.content_length, resp.etag, opts)
+          end
         else
-          # partNumber is an option
-          resp = @client.head_object(bucket: @bucket, key: @key)
-          resp.content_length < MIN_CHUNK_SIZE ?
-            single_request :
-            compute_mode(resp.content_length, count)
+          # covers cases when given object is not uploaded via UploadPart API
+          resp = @client.head_object(head_opts(opts[:params])) # partNumber is an option
+          if resp.content_length <= MIN_CHUNK_SIZE
+            single_request(opts)
+          else
+            compute_mode(resp.content_length, count, resp.etag, opts)
+          end
         end
       end
 
-      def compute_mode(file_size, count)
-        chunk_size = compute_chunk(file_size)
-        part_size = (file_size.to_f / count.to_f).ceil
-        if chunk_size < part_size
-          multithreaded_get_by_ranges(construct_chunks(file_size))
-        else
-          multithreaded_get_by_parts(count)
+      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 construct_chunks(file_size)
+      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 = []
-        while offset <= file_size
+        part_number = 1 # parts start at 1
+        while offset < file_size
           progress = offset + default_chunk_size
-          chunks << "bytes=#{offset}-#{progress < file_size ? progress : file_size}"
-          offset = progress + 1
+          progress = file_size if progress > file_size
+          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
-        chunks
+        download_with_executor(PartList.new(chunks), file_size, opts)
       end
 
-      def compute_chunk(file_size)
-        if @chunk_size && @chunk_size > file_size
-          raise ArgumentError, ":chunk_size shouldn't exceed total file size."
-        else
-          @chunk_size || [(file_size.to_f / MAX_PARTS).ceil, MIN_CHUNK_SIZE].max.to_i
-        end
+      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 batches(chunks, mode)
-        chunks = (1..chunks) if mode.eql? 'part_number'
-        chunks.each_slice(@thread_count).to_a
+      def resolve_temp_path(opts)
+        return if [File, Tempfile].include?(opts[:destination].class)
+
+        opts[:temp_path] ||= "#{opts[:destination]}.s3tmp.#{SecureRandom.alphanumeric(8)}"
       end
 
-      def multithreaded_get_by_ranges(chunks)
-        thread_batches(chunks, 'range')
+      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 opts[:on_checksum_validated]
+
+        opts[:on_checksum_validated].call(resp.checksum_validated, resp) if resp.checksum_validated
+        resp
       end
 
-      def multithreaded_get_by_parts(parts)
-        thread_batches(parts, 'part_number')
+      def single_part_progress(opts)
+        proc do |_chunk, bytes_read, total_size|
+          opts[:progress_callback].call([bytes_read], [total_size], total_size)
+        end
       end
 
-      def thread_batches(chunks, param)
-        batches(chunks, param).each do |batch|
-          threads = []
-          batch.each do |chunk|
-            threads << Thread.new do
-              resp = @client.get_object(
-                :bucket => @bucket,
-                :key => @key,
-                param.to_sym => chunk
-              )
-              write(resp)
-            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
-          threads.each(&:join)
+      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(resp)
-        range, _ = resp.content_range.split(" ").last.split("/")
-        head, _ = range.split("-").map {|s| s.to_i}
-        IO.write(@path, resp.body.read, head)
+      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
+      end
+
+      # @api private
+      class PartList
+        include Enumerable
+        def initialize(parts = [])
+          @parts = parts
+          @mutex = Mutex.new
+        end
+
+        def shift
+          @mutex.synchronize { @parts.shift }
+        end
+
+        def size
+          @mutex.synchronize { @parts.size }
+        end
+
+        def clear!
+          @mutex.synchronize { @parts.clear }
+        end
+
+        def each(&block)
+          @mutex.synchronize { @parts.each(&block) }
+        end
       end
 
-      def single_request
-        @client.get_object(
-          bucket: @bucket, key: @key, response_target: @path
-        )
+      # @api private
+      class MultipartProgress
+        def initialize(parts, total_size, progress_callback)
+          @bytes_received = Array.new(parts.size, 0)
+          @part_sizes = parts.map(&:size)
+          @total_size = total_size
+          @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
+          # part size may not be known until we get the first response
+          @part_sizes[part_number - 1] ||= total
+          @progress_callback.call(@bytes_received, @part_sizes, @total_size)
+        end
       end
     end
   end

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

@@ -1,15 +1,20 @@
+# frozen_string_literal: true
+
 module Aws
   module S3
 
-    # A utility class that provides an IO-like interface to a portion of
-    # a file on disk.
+    # A utility class that provides an IO-like interface to a portion of a file
+    # on disk.
     # @api private
     class FilePart
 
-      # @option options [required,String,Pathname,File,Tempfile] :source
-      # @option options [required,Integer] :offset The file part will read
+      # @option options [required, String, Pathname, File, Tempfile] :source
+      #   The file to upload.
+      #
+      # @option options [required, Integer] :offset The file part will read
       #   starting at this byte offset.
-      # @option options [required,Integer] :size The maximum number of bytes to
+      #
+      # @option options [required, Integer] :size The maximum number of bytes to
       #   read from the `:offset`.
       def initialize(options = {})
         @source = options[:source]
@@ -19,7 +24,7 @@ module Aws
         @file = nil
       end
 
-      # @return [String,Pathname,File,Tempfile]
+      # @return [String, Pathname, File, Tempfile]
       attr_reader :source
 
       # @return [Integer]
@@ -56,14 +61,12 @@ module Aws
       end
 
       def read_from_file(bytes, output_buffer)
-        if bytes
-          data = @file.read([remaining_bytes, bytes].min)
-          data = nil if data == ''
-        else
-          data = @file.read(remaining_bytes)
-        end
+        length = [remaining_bytes, *bytes].min
+        data   = @file.read(length, output_buffer)
+
         @position += data ? data.bytesize : 0
-        output_buffer ? output_buffer.replace(data || '') : data
+
+        data.to_s unless bytes && (data.nil? || data.empty?)
       end
 
       def remaining_bytes

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

@@ -1,3 +1,5 @@
+# frozen_string_literal: true
+
 require 'pathname'
 
 module Aws
@@ -5,54 +7,67 @@ 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 Files greater than
-      #   `:multipart_threshold` bytes are uploaded using S3 multipart APIs.
+      # @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]
+        @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
-      # @option options [required,String] :bucket
-      # @option options [required,String] :key
+      # @param [String, Pathname, File, Tempfile] source The file to upload.
+      # @option options [required, String] :bucket The bucket to upload to.
+      # @option options [required, String] :key The key for the object.
+      # @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).upload(source, options)
+          else
+            put_object(source, options)
+          end
         end
       end
 
       private
 
+      def open_file(source, &block)
+        if source.is_a?(String) || source.is_a?(Pathname)
+          File.open(source, 'rb', &block)
+        else
+          yield(source)
+        end
+      end
+
       def put_object(source, options)
+        if (callback = options.delete(:progress_callback))
+          options[:on_chunk_sent] = single_part_progress(callback)
+        end
         open_file(source) do |file|
           @client.put_object(options.merge(body: file))
         end
       end
 
-      def open_file(source)
-        if String === source || Pathname === source
-          file = File.open(source, 'rb')
-          yield(file)
-          file.close
-        else
-          yield(source)
+      def single_part_progress(progress_callback)
+        proc do |_chunk, bytes_read, total_size|
+          progress_callback.call([bytes_read], [total_size])
         end
       end
-
     end
   end
 end

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

@@ -1,8 +1,10 @@
+# frozen_string_literal: true
+
 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
@@ -155,33 +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