Define the DVZ public interface and shrink it to what is used

Author: mjgiarlo

app/models/replication/druid_version_zip.rb

@@ -9,10 +9,6 @@ module Replication
   # See comment on part_paths method re: individual part suffixes.
   # Just a regular model, not an ActiveRecord-backed model
   class DruidVersionZip
-    attr_reader :druid, :version, :storage_location
-
-    delegate :base64digest, :hexdigest, to: :md5
-
     # the size used with "zip -s" to break up the zip into parts
     ZIP_SPLIT_SIZE = '10g'
 
@@ -26,34 +22,14 @@ def initialize(druid, version, storage_location = nil)
       @storage_location = storage_location
     end
 
-    # @return [String] Filename/key without extension, in common to all this object's zip parts
-    def base_key
-      @base_key ||= s3_key.sub(/.zip\z/, '')
-    end
-
-    # Checks to see whether a zip file already exists for this druid-version.  If it does, just touch the
-    # file to refresh atime and mtime, so the zip cache cleaning cron job doesn't see it as stale.  If it doesn't,
-    # create it.
-    # @raise [StandardError] if there's a zip file for this druid-version, but it looks too small to be complete.
-    def find_or_create_zip!
-      if exist?
-        raise "zip already exists, but size (#{total_part_size}) is smaller than the moab version size (#{moab_version_size})!" unless zip_size_ok?
-        FileUtils.touch(file_path)
-      else
-        create_zip!
-      end
-    end
-
-    def exist?
-      File.exist?(file_path)
-    end
-
     # @return [Boolean] true if there is a match between the zip part files and their md5 sidecar files
     def complete?
       # There is at least one part file
       return false if part_paths.empty?
+
       # The set of md5 sidecar files matches the set of part files
       return false unless part_keys.to_set == part_keys_from_md5_sidecars.to_set
+
       # Check each md5 sidecar file against the zip part file
       druid_version_zip_parts.all?(&:md5_match?)
     end
@@ -67,6 +43,7 @@ def create_zip!
       ensure_zip_directory!
       combined, status = Open3.capture2e(zip_command, chdir: work_dir.to_s)
       raise "zipmaker failure #{combined}" unless status.success?
+
       unless zip_size_ok?
         raise "zip size (#{total_part_size}) is smaller than the moab version size (#{moab_version_size})! zipmaker failure #{combined}"
       end
@@ -79,21 +56,6 @@ def create_zip!
       raise
     end
 
-    # Ensure the directory the zip will live in exists
-    # @return [Pathname] the existing or created directory
-    def ensure_zip_directory!
-      Pathname.new(file_path).tap { |pn| pn.dirname.mkpath }
-    end
-
-    # @param [Integer] count Number of parts
-    # @return [Array<String>] Ordered pathnames expected to correspond to a zip broken into a given number of parts
-    def expected_part_keys(count = 1)
-      raise ArgumentError, "count (#{count}) must be >= 1" if count < 1
-      [s3_key].concat(
-        (1..(count - 1)).map { |n| base_key + format('.z%02d', n) }
-      )
-    end
-
     # @param [String] suffix, e.g. '.zip', '.z01', '.z125', etc., including the dot
     # @return [String] s3_key for the zip part specified by suffix
     # @see [S3 key name performance implications] https://docs.aws.amazon.com/AmazonS3/latest/dev/request-rate-perf-considerations.html
@@ -102,60 +64,37 @@ def s3_key(suffix = '.zip')
       druid.tree.join('/') + ".#{v_version}#{suffix}"
     end
 
-    # @return [String] Path to the local temporary transfer root (.zip) part
-    def file_path
-      File.join(zip_storage, s3_key)
-    end
-
-    # WITHOUT (re)digesting the file, convert a hexdigest MD5 value to base64-endcoded equivalent.
-    # Motivation: we store hexdigest and S3 requires base64 in headers.
-    # @param [String] hex
-    # @return [String] base64 equivalent
-    def hex_to_base64(hex)
-      [[hex].pack('H*')].pack('m0')
-    end
-
-    # @raise [StandardError] if storage_location is not available (should have been provided in constructor)
-    def moab_version_path
-      raise "cannot determine moab_version_path for #{druid.id} v#{version}, storage_location not provided" unless storage_location
-      @moab_version_path ||= "#{druid.path}/#{v_version}"
+    def moab_version_size
+      moab_version_files.sum { |f| File.size(f) }
     end
 
-    # @return [Array<String>] relative paths, i.e. s3_part_keys for existing parts
-    def part_keys
-      part_paths.map { |part| part.relative_path_from(zip_storage).to_s }
+    # Deletes all zip part files and their md5 sidecar files from local zip storage
+    def cleanup_zip_parts!
+      FileUtils.rm_f(parts_and_checksums_paths)
     end
 
-    # NOTE: if there is only ONE part, it will end .zip;  if there are multiple parts,
-    #  the last one will end .zip, so two parts is:  .z01, zip. (this agrees with zip utility)
-    # @return [Array<Pathname>] Existing pathnames for zip parts based on glob (.zip, .z01, .z02, etc.)
-    def part_paths
-      Pathname.glob(file_path.sub(/.zip\z/, '.z*')).reject do |path|
-        path.to_s =~ /.md5\z/
+    # @return [Array<DruidVersionZipPart>] all parts for this DruidVersionZip
+    def druid_version_zip_parts
+      part_keys.map do |part_key|
+        Replication::DruidVersionZipPart.new(self, part_key)
       end
     end
 
-    # @return [Array<Pathname>] all extant zip part and checksum files for this dvz (e.g. bc123df4567.zip, bc123df4567.z01, bc123df4567.zip.md5,
-    #  bc123df4567.z01.md5, etc)
-    def parts_and_checksums_paths
-      Pathname.glob(File.join(zip_storage, s3_key('.*')))
-    end
+    private
 
-    # @return [String] "v" with zero-padded 4-digit version, e.g., v0001
-    def v_version
-      format('v%04d', version)
-    end
+    attr_reader :druid, :version, :storage_location
 
-    # @return [Pathname] The proper directory in which to execute zip_command
-    # @raise [StandardError] if storage_location is not available (should have been provided in constructor)
-    def work_dir
-      Pathname.new(moab_version_path).parent.parent
+    # This assumes that the zip file will be at least as large as the Moab version being zipped. Why? Because
+    # we don't enable compression (see zip_command). Why no compression? We thought it might make extraction
+    # from zips more reliable in the distant future. For further explanation, see https://github.com/sul-dlss/preservation_catalog/wiki/Zip-Creation
+    def zip_size_ok?
+      total_part_size > moab_version_size
     end
 
-    # @return [String] shell command to unzip
-    # def unzip_command
-    #   "unzip #{file_path} -d #{place_to_unzip}"
-    # end
+    # @return [Pathname]
+    def zip_storage
+      @zip_storage ||= Pathname.new(Settings.zip_storage)
+    end
 
     # Presumes execution just "above" the druid dir in the druid tree, i.e. if the Moab is:
     #   /storage_trunk_01/bj/102/hs/9687/bj102hs9687/v0003/...
@@ -167,39 +106,54 @@ def zip_command
       "zip -r0X -s #{ZIP_SPLIT_SIZE} #{file_path} #{druid.id}/#{v_version}"
     end
 
-    def zip_version
-      @zip_version ||= fetch_zip_version
+    # @return [Pathname] The proper directory in which to execute zip_command
+    # @raise [StandardError] if storage_location is not available (should have been provided in constructor)
+    def work_dir
+      Pathname.new(moab_version_path).parent.parent
     end
 
-    # @return [Pathname]
-    def zip_storage
-      @zip_storage ||= Pathname.new(Settings.zip_storage)
+    # @return [String] "v" with zero-padded 4-digit version, e.g., v0001
+    def v_version
+      format('v%04d', version)
     end
 
-    # This assumes that the zip file will be at least as large as the Moab version being zipped. Why? Because
-    # we don't enable compression (see zip_command). Why no compression? We thought it might make extraction
-    # from zips more reliable in the distant future. For further explanation, see https://github.com/sul-dlss/preservation_catalog/wiki/Zip-Creation
-    def zip_size_ok?
-      total_part_size > moab_version_size
+    # @return [Array<Pathname>] all extant zip part and checksum files for this dvz (e.g. bc123df4567.zip, bc123df4567.z01, bc123df4567.zip.md5,
+    #  bc123df4567.z01.md5, etc)
+    def parts_and_checksums_paths
+      Pathname.glob(File.join(zip_storage, s3_key('.*')))
     end
 
-    def moab_version_size
-      moab_version_files.sum { |f| File.size(f) }
+    # NOTE: if there is only ONE part, it will end .zip;  if there are multiple parts,
+    #  the last one will end .zip, so two parts is:  .z01, zip. (this agrees with zip utility)
+    # @return [Array<Pathname>] Existing pathnames for zip parts based on glob (.zip, .z01, .z02, etc.)
+    def part_paths
+      Pathname.glob(file_path.sub(/.zip\z/, '.z*')).reject do |path|
+        path.to_s =~ /.md5\z/
+      end
     end
 
-    # Deletes all zip part files and their md5 sidecar files from local zip storage
-    def cleanup_zip_parts!
-      FileUtils.rm_f(parts_and_checksums_paths)
+    # @return [Array<String>] relative paths, i.e. s3_part_keys for existing parts
+    def part_keys
+      part_paths.map { |part| part.relative_path_from(zip_storage).to_s }
     end
 
-    # @return [Array<DruidVersionZipPart>] all parts for this DruidVersionZip
-    def druid_version_zip_parts
-      part_keys.map do |part_key|
-        Replication::DruidVersionZipPart.new(self, part_key)
-      end
+    # @raise [StandardError] if storage_location is not available (should have been provided in constructor)
+    def moab_version_path
+      raise "cannot determine moab_version_path for #{druid.id} v#{version}, storage_location not provided" unless storage_location
+
+      @moab_version_path ||= "#{druid.path}/#{v_version}"
     end
 
-    private
+    # @return [String] Path to the local temporary transfer root (.zip) part
+    def file_path
+      File.join(zip_storage, s3_key)
+    end
+
+    # Ensure the directory the zip will live in exists
+    # @return [Pathname] the existing or created directory
+    def ensure_zip_directory!
+      Pathname.new(file_path).tap { |pn| pn.dirname.mkpath }
+    end
 
     # Throws an error if any of the files in the moab are not yet readable.  For example due to
     # Ceph MDS instance for a pres cat worker VM thinking that a file is a stray as a result of our
@@ -217,24 +171,8 @@ def total_part_size
 
     def moab_version_files
       raise "Moab version does not exist: #{moab_version_path}" unless File.exist?(moab_version_path)
-      Dir
-        .glob("#{moab_version_path}/**/*")
-        .select { |path| File.file?(path) }
-    end
-
-    # @return [String] e.g. 'Zip 3.0 (July 5th 2008)' or 'Zip 3.0.1'
-    def fetch_zip_version
-      match = nil
-      IO.popen('zip -v') do |io|
-        re = zip_version_regexp
-        io.find { |line| match = line.match(re) }
-      end
-      return match[1] if match && match[1].present?
-      raise 'No version info matched from `zip -v` ouptut'
-    end
 
-    def zip_version_regexp
-      /This is (Zip \d+(\.\d)+\s*(\(.*\d{4}\))?)/
+      Dir.glob("#{moab_version_path}/**/*").select { |path| File.file?(path) }
     end
 
     # @return [Array<String>] relative paths, i.e. s3_part_keys for existing parts based on the md5 sidecar files