ext_backup 5.0.0.beta.2.1

data/lib/backup/cloud_io/s3.rb

@@ -0,0 +1,253 @@
+require "backup/cloud_io/base"
+require "ext_fog_aws"
+require "digest/md5"
+require "base64"
+require "stringio"
+
+module Backup
+  module CloudIO
+    class S3 < Base
+      class Error < Backup::Error; end
+
+      MAX_FILE_SIZE       = 1024**3 * 5   # 5 GiB
+      MAX_MULTIPART_SIZE  = 1024**4 * 5   # 5 TiB
+
+      attr_reader :access_key_id, :secret_access_key, :use_iam_profile,
+        :region, :bucket, :chunk_size, :encryption, :storage_class,
+        :fog_options
+
+      def initialize(options = {})
+        super
+
+        @access_key_id      = options[:access_key_id]
+        @secret_access_key  = options[:secret_access_key]
+        @use_iam_profile    = options[:use_iam_profile]
+        @region             = options[:region]
+        @bucket             = options[:bucket]
+        @chunk_size         = options[:chunk_size]
+        @encryption         = options[:encryption]
+        @storage_class      = options[:storage_class]
+        @fog_options        = options[:fog_options]
+      end
+
+      # The Syncer may call this method in multiple threads.
+      # However, #objects is always called prior to multithreading.
+      def upload(src, dest)
+        file_size = File.size(src)
+        chunk_bytes = chunk_size * 1024**2
+        if chunk_bytes > 0 && file_size > chunk_bytes
+          raise FileSizeError, <<-EOS if file_size > MAX_MULTIPART_SIZE
+            File Too Large
+            File: #{src}
+            Size: #{file_size}
+            Max Multipart Upload Size is #{MAX_MULTIPART_SIZE} (5 TiB)
+          EOS
+
+          chunk_bytes = adjusted_chunk_bytes(chunk_bytes, file_size)
+          upload_id = initiate_multipart(dest)
+          parts = upload_parts(src, dest, upload_id, chunk_bytes, file_size)
+          complete_multipart(dest, upload_id, parts)
+        else
+          raise FileSizeError, <<-EOS if file_size > MAX_FILE_SIZE
+            File Too Large
+            File: #{src}
+            Size: #{file_size}
+            Max File Size is #{MAX_FILE_SIZE} (5 GiB)
+          EOS
+
+          put_object(src, dest)
+        end
+      end
+
+      # Returns all objects in the bucket with the given prefix.
+      #
+      # - #get_bucket returns a max of 1000 objects per request.
+      # - Returns objects in alphabetical order.
+      # - If marker is given, only objects after the marker are in the response.
+      def objects(prefix)
+        objects = []
+        resp = nil
+        prefix = prefix.chomp("/")
+        opts = { "prefix" => prefix + "/" }
+
+        while resp.nil? || resp.body["IsTruncated"]
+          opts["marker"] = objects.last.key unless objects.empty?
+          with_retries("GET '#{bucket}/#{prefix}/*'") do
+            resp = connection.get_bucket(bucket, opts)
+          end
+          resp.body["Contents"].each do |obj_data|
+            objects << Object.new(self, obj_data)
+          end
+        end
+
+        objects
+      end
+
+      # Used by Object to fetch metadata if needed.
+      def head_object(object)
+        resp = nil
+        with_retries("HEAD '#{bucket}/#{object.key}'") do
+          resp = connection.head_object(bucket, object.key)
+        end
+        resp
+      end
+
+      # Delete object(s) from the bucket.
+      #
+      # - Called by the Storage (with objects) and the Syncer (with keys)
+      # - Deletes 1000 objects per request.
+      # - Missing objects will be ignored.
+      def delete(objects_or_keys)
+        keys = Array(objects_or_keys).dup
+        keys.map!(&:key) if keys.first.is_a?(Object)
+
+        opts = { quiet: true } # only report Errors in DeleteResult
+        until keys.empty?
+          keys_partial = keys.slice!(0, 1000)
+          with_retries("DELETE Multiple Objects") do
+            resp = connection.delete_multiple_objects(bucket, keys_partial, opts.dup)
+            unless resp.body["DeleteResult"].empty?
+              errors = resp.body["DeleteResult"].map do |result|
+                error = result["Error"]
+                "Failed to delete: #{error["Key"]}\n" \
+                  "Reason: #{error["Code"]}: #{error["Message"]}"
+              end.join("\n")
+              raise Error, "The server returned the following:\n#{errors}"
+            end
+          end
+        end
+      end
+
+      private
+
+      def connection
+        @connection ||=
+          begin
+            opts = { provider: "AWS", region: region }
+            if use_iam_profile
+              opts[:use_iam_profile] = true
+            else
+              opts[:aws_access_key_id] = access_key_id
+              opts[:aws_secret_access_key] = secret_access_key
+            end
+            opts.merge!(fog_options || {})
+            conn = Fog::Storage.new(opts)
+            conn.sync_clock
+            conn
+          end
+      end
+
+      def put_object(src, dest)
+        md5 = Base64.encode64(Digest::MD5.file(src).digest).chomp
+        options = headers.merge("Content-MD5" => md5)
+        with_retries("PUT '#{bucket}/#{dest}'") do
+          File.open(src, "r") do |file|
+            connection.put_object(bucket, dest, file, options)
+          end
+        end
+      end
+
+      def initiate_multipart(dest)
+        Logger.info "\s\sInitiate Multipart '#{bucket}/#{dest}'"
+
+        resp = nil
+        with_retries("POST '#{bucket}/#{dest}' (Initiate)") do
+          resp = connection.initiate_multipart_upload(bucket, dest, headers)
+        end
+        resp.body["UploadId"]
+      end
+
+      # Each part's MD5 is sent to verify the transfer.
+      # AWS will concatenate all parts into a single object
+      # once the multipart upload is completed.
+      def upload_parts(src, dest, upload_id, chunk_bytes, file_size)
+        total_parts = (file_size / chunk_bytes.to_f).ceil
+        progress = (0.1..0.9).step(0.1).map { |n| (total_parts * n).floor }
+        Logger.info "\s\sUploading #{total_parts} Parts..."
+
+        parts = []
+        File.open(src, "r") do |file|
+          part_number = 0
+          while data = file.read(chunk_bytes)
+            part_number += 1
+            md5 = Base64.encode64(Digest::MD5.digest(data)).chomp
+
+            with_retries("PUT '#{bucket}/#{dest}' Part ##{part_number}") do
+              resp = connection.upload_part(
+                bucket, dest, upload_id, part_number, StringIO.new(data),
+                "Content-MD5" => md5
+              )
+              parts << resp.headers["ETag"]
+            end
+
+            if i = progress.rindex(part_number)
+              Logger.info "\s\s...#{i + 1}0% Complete..."
+            end
+          end
+        end
+        parts
+      end
+
+      def complete_multipart(dest, upload_id, parts)
+        Logger.info "\s\sComplete Multipart '#{bucket}/#{dest}'"
+
+        with_retries("POST '#{bucket}/#{dest}' (Complete)") do
+          resp = connection.complete_multipart_upload(bucket, dest, upload_id, parts)
+          raise Error, <<-EOS if resp.body["Code"]
+            The server returned the following error:
+            #{resp.body["Code"]}: #{resp.body["Message"]}
+          EOS
+        end
+      end
+
+      def headers
+        headers = {}
+
+        enc = encryption.to_s.upcase
+        headers["x-amz-server-side-encryption"] = enc unless enc.empty?
+
+        sc = storage_class.to_s.upcase
+        headers["x-amz-storage-class"] = sc unless sc.empty? || sc == "STANDARD"
+
+        headers
+      end
+
+      def adjusted_chunk_bytes(chunk_bytes, file_size)
+        return chunk_bytes if file_size / chunk_bytes.to_f <= 10_000
+
+        mb = orig_mb = chunk_bytes / 1024**2
+        mb += 1 until file_size / (1024**2 * mb).to_f <= 10_000
+        Logger.warn Error.new(<<-EOS)
+          Chunk Size Adjusted
+          Your original #chunk_size of #{orig_mb} MiB has been adjusted
+          to #{mb} MiB in order to satisfy the limit of 10,000 chunks.
+          To enforce your chosen #chunk_size, you should use the Splitter.
+          e.g. split_into_chunks_of #{mb * 10_000} (#chunk_size * 10_000)
+        EOS
+        1024**2 * mb
+      end
+
+      class Object
+        attr_reader :key, :etag, :storage_class
+
+        def initialize(cloud_io, data)
+          @cloud_io = cloud_io
+          @key  = data["Key"]
+          @etag = data["ETag"]
+          @storage_class = data["StorageClass"]
+        end
+
+        # currently 'AES256' or nil
+        def encryption
+          metadata["x-amz-server-side-encryption"]
+        end
+
+        private
+
+        def metadata
+          @metadata ||= @cloud_io.head_object(self).headers
+        end
+      end
+    end
+  end
+end

data/lib/backup/compressor/base.rb

@@ -0,0 +1,32 @@
+module Backup
+  module Compressor
+    class Base
+      include Utilities::Helpers
+      include Config::Helpers
+
+      ##
+      # Yields to the block the compressor command and filename extension.
+      def compress_with
+        log!
+        yield @cmd, @ext
+      end
+
+      private
+
+      ##
+      # Return the compressor name, with Backup namespace removed
+      def compressor_name
+        self.class.to_s.sub("Backup::", "")
+      end
+
+      ##
+      # Logs a message to the console and log file to inform
+      # the client that Backup is using the compressor
+      def log!
+        Logger.info "Using #{compressor_name} for compression.\n" \
+          "  Command: '#{@cmd}'\n" \
+          "  Ext: '#{@ext}'"
+      end
+    end
+  end
+end

data/lib/backup/compressor/bzip2.rb

@@ -0,0 +1,35 @@
+module Backup
+  module Compressor
+    class Bzip2 < Base
+      ##
+      # Specify the level of compression to use.
+      #
+      # Values should be a single digit from 1 to 9.
+      # Note that setting the level to either extreme may or may not
+      # give the desired result. Be sure to check the documentation
+      # for the compressor being used.
+      #
+      # The default `level` is 9.
+      attr_accessor :level
+
+      ##
+      # Creates a new instance of Backup::Compressor::Bzip2
+      def initialize(&block)
+        load_defaults!
+
+        @level ||= false
+
+        instance_eval(&block) if block_given?
+
+        @cmd = "#{utility(:bzip2)}#{options}"
+        @ext = ".bz2"
+      end
+
+      private
+
+      def options
+        " -#{@level}" if @level
+      end
+    end
+  end
+end

data/lib/backup/compressor/custom.rb

@@ -0,0 +1,49 @@
+module Backup
+  module Compressor
+    class Custom < Base
+      ##
+      # Specify the system command to invoke a compressor,
+      # including any command-line arguments.
+      # e.g. @compressor.command = 'pbzip2 -p2 -4'
+      #
+      # The data to be compressed will be piped to the command's STDIN,
+      # and it should write the compressed data to STDOUT.
+      # i.e. `cat file.tar | %command% > file.tar.%extension%`
+      attr_accessor :command
+
+      ##
+      # File extension to append to the compressed file's filename.
+      # e.g. @compressor.extension = '.bz2'
+      attr_accessor :extension
+
+      ##
+      # Initializes a new custom compressor.
+      def initialize(&block)
+        load_defaults!
+
+        instance_eval(&block) if block_given?
+
+        @cmd = set_cmd
+        @ext = set_ext
+      end
+
+      private
+
+      ##
+      # Return the command line using the full path.
+      # Ensures the command exists and is executable.
+      def set_cmd
+        parts = @command.to_s.split(" ")
+        parts[0] = utility(parts[0])
+        parts.join(" ")
+      end
+
+      ##
+      # Return the extension given without whitespace.
+      # If extension was not set, return an empty string
+      def set_ext
+        @extension.to_s.strip
+      end
+    end
+  end
+end

data/lib/backup/compressor/gzip.rb

@@ -0,0 +1,73 @@
+module Backup
+  module Compressor
+    class Gzip < Base
+      class Error < Backup::Error; end
+      extend Utilities::Helpers
+
+      ##
+      # Specify the level of compression to use.
+      #
+      # Values should be a single digit from 1 to 9.
+      # Note that setting the level to either extreme may or may not
+      # give the desired result. Be sure to check the documentation
+      # for the compressor being used.
+      #
+      # The default `level` is 6.
+      attr_accessor :level
+
+      ##
+      # Use the `--rsyncable` option with `gzip`.
+      #
+      # This option directs `gzip` to compress data using an algorithm that
+      # allows `rsync` to efficiently detect changes. This is especially useful
+      # when used to compress `Archive` or `Database` backups that will be
+      # stored using Backup's `RSync` Storage option.
+      #
+      # The `--rsyncable` option is only available on patched versions of `gzip`.
+      # While most distributions apply this patch, this option may not be
+      # available on your system. If it's not available, Backup will log a
+      # warning and continue to use the compressor without this option.
+      attr_accessor :rsyncable
+
+      ##
+      # Determine if +--rsyncable+ is supported and cache the result.
+      def self.has_rsyncable?
+        return @has_rsyncable unless @has_rsyncable.nil?
+        cmd = "#{utility(:gzip)} --rsyncable --version >/dev/null 2>&1; echo $?"
+        @has_rsyncable = `#{cmd}`.chomp == "0"
+      end
+
+      ##
+      # Creates a new instance of Backup::Compressor::Gzip
+      def initialize(&block)
+        load_defaults!
+
+        @level ||= false
+        @rsyncable ||= false
+
+        instance_eval(&block) if block_given?
+
+        @cmd = "#{utility(:gzip)}#{options}"
+        @ext = ".gz"
+      end
+
+      private
+
+      def options
+        opts = ""
+        opts << " -#{@level}" if @level
+        if @rsyncable
+          if self.class.has_rsyncable?
+            opts << " --rsyncable"
+          else
+            Logger.warn Error.new(<<-EOS)
+              'rsyncable' option ignored.
+              Your system's 'gzip' does not support the `--rsyncable` option.
+            EOS
+          end
+        end
+        opts
+      end
+    end
+  end
+end

data/lib/backup/config.rb

@@ -0,0 +1,128 @@
+require "backup/config/dsl"
+require "backup/config/helpers"
+
+module Backup
+  module Config
+    class Error < Backup::Error; end
+
+    DEFAULTS = {
+      config_file: "config.rb",
+      data_path: ".data",
+      tmp_path: ".tmp"
+    }
+
+    GEM_NAME = /\w+/
+    CONFIG_FILE = /[\w\/\.]+/
+
+    class << self
+      include Utilities::Helpers
+
+      attr_reader :user, :root_path, :data_path, :tmp_path
+      attr_accessor :config_file
+
+      # Loads the user's +config.rb+ and all model files.
+      def load(options = {})
+        update(options) # from the command line
+
+        unless File.exist?(config_file)
+          raise Error, "Could not find configuration file: '#{config_file}'."
+        end
+
+        config = File.read(config_file)
+        version = Backup::VERSION.split(".").first
+        models = File.join(File.dirname(config_file), "models", "*.rb")
+        if config =~ /^# Backup v#{ version }\.x Configuration\[(#{GEM_NAME})\]\[(#{CONFIG_FILE})\]$/
+          gem_name, self.config_file = $1, $2
+          spec = Bundler.load.specs.find{ |s| s.name == gem_name }
+          self.config_file = File.join(spec.full_gem_path, config_file)
+          config = File.read(config_file)
+        elsif !(config =~ /^# Backup v#{ version }\.x Configuration$/)
+          raise Error, <<-EOS
+            Invalid Configuration File
+            The configuration file at '#{config_file}'
+            does not appear to be a Backup v#{version}.x configuration file.
+            If you have upgraded to v#{version}.x from a previous version,
+            you need to upgrade your configuration file.
+            Please see the instructions for upgrading in the Backup documentation.
+          EOS
+        end
+
+        dsl = DSL.new
+        dsl.instance_eval(config, config_file)
+
+        update(dsl._config_options)  # from config.rb
+        update(options)              # command line takes precedence
+
+        Dir[models].each do |model|
+          dsl.instance_eval(File.read(model), model)
+        end
+      end
+
+      def hostname
+        @hostname ||= run(utility(:hostname))
+      end
+
+      private
+
+      # If :root_path is set in the options, all paths will be updated.
+      # Otherwise, only the paths given will be updated.
+      def update(options = {})
+        root_path = options[:root_path].to_s.strip
+        new_root = root_path.empty? ? false : set_root_path(root_path)
+
+        DEFAULTS.each do |name, ending|
+          set_path_variable(name, options[name], ending, new_root)
+        end
+      end
+
+      # Sets the @root_path to the given +path+ and returns it.
+      # Raises an error if the given +path+ does not exist.
+      def set_root_path(path)
+        # allows #reset! to set the default @root_path,
+        # then use #update to set all other paths,
+        # without requiring that @root_path exist.
+        return @root_path if path == @root_path
+
+        path = File.expand_path(path)
+        unless File.directory?(path)
+          raise Error, <<-EOS
+            Root Path Not Found
+            When specifying a --root-path, the path must exist.
+            Path was: #{path}
+          EOS
+        end
+        @root_path = path
+      end
+
+      def set_path_variable(name, path, ending, root_path)
+        # strip any trailing '/' in case the user supplied this as part of
+        # an absolute path, so we can match it against File.expand_path()
+        path = path.to_s.sub(/\/\s*$/, "").lstrip
+        new_path = false
+        # If no path is given, the variable will not be set/updated
+        # unless a root_path was given. In which case the value will
+        # be updated with our default ending.
+        if path.empty?
+          new_path = File.join(root_path, ending) if root_path
+        else
+          # When a path is given, the variable will be set/updated.
+          # If the path is relative, it will be joined with root_path (if given),
+          # or expanded relative to PWD.
+          new_path = File.expand_path(path)
+          unless path == new_path
+            new_path = File.join(root_path, path) if root_path
+          end
+        end
+        instance_variable_set(:"@#{name}", new_path) if new_path
+      end
+
+      def reset!
+        @user      = ENV["USER"] || Etc.getpwuid.name
+        @root_path = File.join(File.expand_path(ENV["HOME"] || ""), "Backup")
+        update(root_path: @root_path)
+      end
+    end
+
+    reset! # set defaults on load
+  end
+end