sdr-client 0.4.0 → 0.5.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: b1d76c1422877b6b5f25f0a1cb46641e32771496771576f8f7c8fd2697c14b0e
-  data.tar.gz: eb116e73a114a3966254a0996c5d843fc960e1454146214ca68fae11eff89bda
+  metadata.gz: 565e7d34958ad818fcc048440816b0a360fa723ea6e62dc9437c6832ba8877c1
+  data.tar.gz: 14dfae8ecef2b6d05a256a8bdc7aebebfe801a278546c98bd3066169f7c98899
 SHA512:
-  metadata.gz: acf219c807f4f22f87437f779cf4d3ab2bb8091bc810a616146a302a9e2bab4b5be99685c91ea072056344a59b5b0f37e9ed5008105d836078d1b70032720c75
-  data.tar.gz: b6ec07336e2014a6fbf02470c08649c426384a51c40ee32a164f29c96ab75cd2057615ee5ee8081f7a70f406f794fed6c100fb93aba7848d629e8933fa91f589
+  metadata.gz: bc629b8837567ee95b01ce673a1f1b49596c3665810ab8d59573012b3b9b6a674db12dde7d15461a279f87a4ef88ed23d65d2457bbc7c0479330f45b3f2c1649
+  data.tar.gz: fe176665dadd92ae291b7dbf5b461bb40606600383ad81c9824a8a4bad83344db5e74b6568a10eced02f4fa14894188122605ac2b7ecd5176d7b25b241963da4

data/.rubocop.yml

@@ -13,3 +13,5 @@ Layout/LineLength:
 Metrics/BlockLength:
   Exclude:
     - 'spec/**/*'
+  ExcludedMethods:
+    - 'OptionParser.new'

data/exe/sdr

@@ -3,6 +3,7 @@
 
 $LOAD_PATH.unshift 'lib'
 require 'optparse'
+require 'sdr_client'
 
 options = {}
 global = OptionParser.new do |opts|
@@ -60,6 +61,20 @@ subcommands = {
       options[:source_id] = source_id
     end
 
+    opts.on('--strategy STRATEGY',
+            'The strategy to use for distributing files into filesets. Either "default" or "filename"') do |strategy|
+      strategy_class = case strategy
+                       when 'filename'
+                         SdrClient::Deposit::MatchingFileGroupingStrategy
+                       when 'default'
+                         SdrClient::Deposit::SingleFileGroupingStrategy
+                       else
+                         warn "Unknown strategy #{strategy}"
+                         exit(1)
+                       end
+      options[:grouping_strategy] = strategy_class
+    end
+
     opts.on('-h', '--help', 'Display this screen') do
       puts opts
       exit
@@ -75,6 +90,5 @@ end
 
 subcommands[command].order!
 
-require 'sdr_client'
 options[:files] = ARGV unless ARGV.empty?
 SdrClient::CLI.start(command, options)

data/lib/sdr_client/deposit.rb

@@ -3,13 +3,16 @@
 module SdrClient
   # The namespace for the "deposit" command
   module Deposit
+    # rubocop:disable Metrics/ParameterLists
     def self.run(label: nil,
                  type: 'http://cocina.sul.stanford.edu/models/book.jsonld',
                  apo:,
                  collection:,
                  catkey: nil,
                  source_id:,
-                 url:, files: [])
+                 url:,
+                 files: [],
+                 grouping_strategy: SingleFileGroupingStrategy)
       token = Credentials.read
 
       metadata = Request.new(label: label,
@@ -18,12 +21,14 @@ module SdrClient
                              collection: collection,
                              source_id: source_id,
                              catkey: catkey)
-      Process.new(metadata: metadata, url: url, token: token, files: files).run
+      Process.new(metadata: metadata, url: url, token: token, files: files, grouping_strategy: grouping_strategy).run
     end
+    # rubocop:enable Metrics/ParameterLists
   end
 end
 require 'json'
-require 'sdr_client/deposit/default_file_set_builder'
+require 'sdr_client/deposit/single_file_grouping_strategy'
+require 'sdr_client/deposit/matching_file_grouping_strategy'
 require 'sdr_client/deposit/files/direct_upload_request'
 require 'sdr_client/deposit/files/direct_upload_response'
 require 'sdr_client/deposit/file'

data/lib/sdr_client/deposit/file.rb

@@ -4,17 +4,21 @@ module SdrClient
   module Deposit
     # This represents the File metadata that we send to the server for doing a deposit
     class File
-      def initialize(external_identifier:, label:, filename:, access: 'dark', preserve: false, shelve: false)
+      def initialize(external_identifier:, label:, filename:, access: 'dark', preserve: false, shelve: false,
+                     md5: nil, sha1: nil)
         @external_identifier = external_identifier
         @label = label
         @filename = filename
         @access = access
         @preserve = preserve
         @shelve = shelve
+        @md5 = md5
+        @sha1 = sha1
       end
 
+      # rubocop:disable Metrics/MethodLength
       def as_json
-        {
+        json = {
           "type": 'http://cocina.sul.stanford.edu/models/file.jsonld',
           label: @label,
           filename: @filename,
@@ -27,6 +31,25 @@ module SdrClient
             shelve: @shelve
           }
         }
+        json['hasMessageDigests'] = message_digests unless message_digests.empty?
+        json
+      end
+      # rubocop:enable Metrics/MethodLength
+
+      private
+
+      def message_digests
+        @message_digests ||= [].tap do |message_digests|
+          message_digests << create_message_digest('md5', @md5) unless @md5.nil?
+          message_digests << create_message_digest('sha1', @sha1) unless @sha1.nil?
+        end
+      end
+
+      def create_message_digest(algorithm, digest)
+        {
+          "type": algorithm,
+          digest: digest
+        }
       end
     end
   end

data/lib/sdr_client/deposit/file_set.rb

@@ -4,11 +4,11 @@ module SdrClient
   module Deposit
     # This represents the FileSet metadata that we send to the server for doing a deposit
     class FileSet
-      def initialize(uploads: [], files: [], label:)
+      def initialize(uploads: [], uploads_metadata: {}, files: [], label:)
         @label = label
         @files = if !uploads.empty?
                    uploads.map do |upload|
-                     File.new(external_identifier: upload.signed_id, label: upload.filename, filename: upload.filename)
+                     File.new(file_args(upload, uploads_metadata.fetch(upload.filename, {})))
                    end
                  else
                    files
@@ -28,6 +28,15 @@ module SdrClient
       private
 
       attr_reader :files, :label
+
+      def file_args(upload, upload_metadata)
+        args = {
+          external_identifier: upload.signed_id,
+          label: upload.filename,
+          filename: upload.filename
+        }
+        args.merge(upload_metadata)
+      end
     end
   end
 end

data/lib/sdr_client/deposit/matching_file_grouping_strategy.rb

@@ -0,0 +1,14 @@
+# frozen_string_literal: true
+
+module SdrClient
+  module Deposit
+    # This strategy is for building one file set per set of similarly prefixed uploaded files
+    class MatchingFileGroupingStrategy
+      # @param [Array<SdrClient::Deposit::Files::DirectUploadResponse>] uploads the uploaded files to attach.
+      # @return [Array<Array<SdrClient::Deposit::Files::DirectUploadResponse>>] uploads the grouped uploaded files.
+      def self.run(uploads: [])
+        uploads.group_by { |ul| ::File.basename(ul.filename, '.*') }
+      end
+    end
+  end
+end

data/lib/sdr_client/deposit/process.rb

@@ -9,19 +9,22 @@ module SdrClient
       BLOB_PATH = '/v1/direct_uploads'
       DRO_PATH = '/v1/resources'
       # @param [Request] metadata information about the object
-      # @param [#build] file_set_builder a strategy for constructing FileSets
+      # @param [Class] grouping_strategy class whose run method groups an array of uploads
       # @param [String] url the server to send to
       # @param [String] token the bearer auth token for the server
       # @param [Array<String>] files a list of file names to upload
+      # @param [Hash<String, Hash<String, String>>] files_metadata file name, hash of additional file metadata
+      # Additional metadata includes access, preserve, shelve, md5, sha1
       # @param [Logger] logger the logger to use
-      def initialize(metadata:, file_set_builder: DefaultFileSetBuilder, url:,
-                     token:, files: [], logger: Logger.new(STDOUT))
+      def initialize(metadata:, grouping_strategy: SingleFileGroupingStrategy, url:,
+                     token:, files: [], files_metadata: {}, logger: Logger.new(STDOUT))
         @files = files
         @url = url
         @token = token
         @metadata = metadata
         @logger = logger
-        @file_set_builder = file_set_builder
+        @grouping_strategy = grouping_strategy
+        @files_metadata = files_metadata
       end
 
       def run
@@ -29,13 +32,14 @@ module SdrClient
         file_metadata = collect_file_metadata
         upload_responses = upload_file_metadata(file_metadata)
         upload_files(upload_responses)
-        request = file_set_builder.run(request: metadata, uploads: upload_responses.values)
+        file_sets = build_filesets(uploads: upload_responses.values, files_metadata: files_metadata)
+        request = metadata.with_file_sets(file_sets)
         upload_metadata(request.as_json)
       end
 
       private
 
-      attr_reader :metadata, :files, :url, :token, :logger, :file_set_builder
+      attr_reader :metadata, :files, :url, :token, :logger, :grouping_strategy, :files_metadata
 
       def check_files_exist
         logger.info('checking to see if files exist')
@@ -114,6 +118,18 @@ module SdrClient
           conn.adapter :net_http
         end
       end
+
+      # @param [Array<SdrClient::Deposit::Files::DirectUploadResponse>] uploads the uploaded files to attach.
+      # @param [Hash<String,Hash<String, String>>] files_metadata filename, hash of additional file metadata.
+      # @return [Array<SdrClient::Deposit::FileSet>] the uploads transformed to filesets
+      def build_filesets(uploads:, files_metadata:)
+        grouped_uploads = grouping_strategy.run(uploads: uploads)
+        grouped_uploads.each_with_index.map do |upload_group, i|
+          metadata_group = {}
+          upload_group.each { |upload| metadata_group[upload.filename] = files_metadata.fetch(upload.filename, {}) }
+          FileSet.new(uploads: upload_group, uploads_metadata: metadata_group, label: "Object #{i + 1}")
+        end
+      end
     end
   end
 end

data/lib/sdr_client/deposit/single_file_grouping_strategy.rb

@@ -0,0 +1,14 @@
+# frozen_string_literal: true
+
+module SdrClient
+  module Deposit
+    # This strategy is for building one file set per uploaded file
+    class SingleFileGroupingStrategy
+      # @param [Array<SdrClient::Deposit::Files::DirectUploadResponse>] uploads the uploaded files to attach.
+      # @return [Array<Array<SdrClient::Deposit::Files::DirectUploadResponse>>] uploads the grouped uploaded files.
+      def self.run(uploads: [])
+        uploads.map { |upload| [upload] }
+      end
+    end
+  end
+end

data/lib/sdr_client/version.rb

@@ -1,5 +1,5 @@
 # frozen_string_literal: true
 
 module SdrClient
-  VERSION = '0.4.0'
+  VERSION = '0.5.0'
 end

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: sdr-client
 version: !ruby/object:Gem::Version
-  version: 0.4.0
+  version: 0.5.0
 platform: ruby
 authors:
 - Justin Coyne
 autorequire: 
 bindir: exe
 cert_chain: []
-date: 2020-01-24 00:00:00.000000000 Z
+date: 2020-01-27 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: dry-monads
@@ -162,13 +162,14 @@ files:
 - lib/sdr_client/cli.rb
 - lib/sdr_client/credentials.rb
 - lib/sdr_client/deposit.rb
-- lib/sdr_client/deposit/default_file_set_builder.rb
 - lib/sdr_client/deposit/file.rb
 - lib/sdr_client/deposit/file_set.rb
 - lib/sdr_client/deposit/files/direct_upload_request.rb
 - lib/sdr_client/deposit/files/direct_upload_response.rb
+- lib/sdr_client/deposit/matching_file_grouping_strategy.rb
 - lib/sdr_client/deposit/process.rb
 - lib/sdr_client/deposit/request.rb
+- lib/sdr_client/deposit/single_file_grouping_strategy.rb
 - lib/sdr_client/login.rb
 - lib/sdr_client/login_prompt.rb
 - lib/sdr_client/version.rb

data/lib/sdr_client/deposit/default_file_set_builder.rb

@@ -1,18 +0,0 @@
-# frozen_string_literal: true
-
-module SdrClient
-  module Deposit
-    # This strategy is for building one file set per uploaded file
-    class DefaultFileSetBuilder
-      # @return [Request] request The initial request
-      # @param [Array<SdrClient::Deposit::Files::DirectUploadResponse>] uploads the uploaded files to attach.
-      # @return [Request] a clone of this request with the uploads added
-      def self.run(request:, uploads: [])
-        file_sets = uploads.each_with_index.map do |upload, i|
-          FileSet.new(uploads: [upload], label: "Object #{i + 1}")
-        end
-        request.with_file_sets(file_sets)
-      end
-    end
-  end
-end