accomodate SHA-1 as the default checksum algorithm; closes #1555

app/services/build_batch_from_folder_ingest.rb

@@ -8,7 +8,7 @@ def initialize(user, filesystem, content_modeler, metadata_provider, checksum_pr
     @filesystem = filesystem
     @content_modeler = content_modeler
     @metadata_provider = metadata_provider
-    @checksum_provider = checksum_provider    
+    @checksum_provider = checksum_provider
     @batch_name = batch_name
     @batch_description = batch_description
   end
@@ -17,7 +17,7 @@ def call
     @batch = create_batch
     traverse_filesystem
     batch.update_attributes(status: Ddr::Batch::Batch::STATUS_READY)
-    batch    
+    batch
   end
 
   private
@@ -85,7 +85,7 @@ def add_content_datastream(batch_object, node)
       payload: full_filepath,
       payload_type: Ddr::Batch::BatchObjectDatastream::PAYLOAD_TYPE_FILENAME,
       checksum: checksum_provider.checksum(rel_filepath),
-      checksum_type: Ddr::Datastreams::CHECKSUM_TYPE_SHA256
+      checksum_type: Ddr::Datastreams::CHECKSUM_TYPE_SHA1
     )
     batch_object.batch_object_datastreams << ds
   end
@@ -99,4 +99,4 @@ def create_relationship(relationship_name, relationship_target_pid)
     )
   end
 
-end
+end

app/views/application/_select_content_file.html.erb

@@ -6,7 +6,7 @@
   <div class="col-md-3">
     <div class="form-group">
       <%= label_tag "content_checksum_type", "Checksum Type" %>
-      <%= select_tag "content[checksum_type]", options_for_select(Ddr::Datastreams::CHECKSUM_TYPES, "SHA-256"), class: "form-control" %>
+      <%= select_tag "content[checksum_type]", options_for_select(Ddr::Datastreams::CHECKSUM_TYPES, "SHA-1"), class: "form-control" %>
     </div>
   </div>
   <div class="col-md-9">

config/folder_ingest.yml.sample

@@ -4,7 +4,7 @@ config:
     target_folder: targets
     checksum_file:
         location: /dir/fixity/fedora_ingest/
-        type: SHA-256
+        type: SHA-1
 files:
     included_extensions:
         - .pdf

lib/dul_hydra/batch/scripts/process_simple_ingest.rb

@@ -4,7 +4,7 @@ class ProcessSimpleIngest
 
     attr_reader :batch_user, :configuration, :filepath
 
-    CHECKSUM_FILE = 'manifest-sha256.txt'
+    CHECKSUM_FILE = 'manifest-sha1.txt'
     METADATA_FILE = 'metadata.txt'
     DEFAULT_CONFIG_FILE = Rails.root.join('config', 'simple_ingest.yml')
     DEFAULT_ARGUMENTS = { config_file: DEFAULT_CONFIG_FILE }
@@ -81,4 +81,4 @@ def build_batch(filesystem)
     end
 
   end
-end
+end

lib/dul_hydra/version.rb

@@ -1,3 +1,3 @@
 module DulHydra
-  VERSION = "4.2.1"
+  VERSION = "4.2.2"
 end

spec/controllers/attachments_controller_spec.rb

@@ -75,7 +75,7 @@ def create_attachment opts={}
       end
       it "should validate the checksum when provided" do
         expect(controller).to receive(:validate_checksum)
-        create_attachment(checksum: "b3f5fc721b5b7ea0c1756a68ed4626463c610170aa199f798fb630ddbea87b18", checksum_type: "SHA-256")
+        create_attachment(checksum: "ff01aab0eada29d35bb423c5c73a9f67a22bc1fd", checksum_type: "SHA-1")
       end
     end
     describe "user cannot add attachments to object" do

spec/controllers/components_controller_spec.rb

@@ -87,7 +87,7 @@ def create_component opts={}
       end
       it "should validate the checksum when provided" do
         expect(controller).to receive(:validate_checksum)
-        create_component(checksum: "bda5fda452d0047c27e9e0048ed59428cb9e6d5d46fe9c27dff5c8e39b75a59e", checksum_type: "SHA-256")
+        create_component(checksum: "75e2e0cec6e807f6ae63610d46448f777591dd6b", checksum_type: "SHA-1")
       end
     end
     context "when the user cannot add children to the item" do

spec/controllers/ingest_folders_controller_spec.rb

@@ -43,8 +43,8 @@
       expect(assigns[:ingest_folder].model).to eql(IngestFolder.default_file_model)
       expect(assigns[:ingest_folder].base_path).to eql("base/path/")
       expect(assigns[:ingest_folder].sub_path).to eql('/subpath/subsubpath/')
-      expect(assigns[:ingest_folder].checksum_file).to eql(File.join(IngestFolder.default_checksum_file_location, "subpath-base-sha256.txt"))
-      expect(assigns[:ingest_folder].checksum_type).to eql(Ddr::Datastreams::CHECKSUM_TYPE_SHA256)
+      expect(assigns[:ingest_folder].checksum_file).to eql(File.join(IngestFolder.default_checksum_file_location, "subpath-base-sha1.txt"))
+      expect(assigns[:ingest_folder].checksum_type).to eql(Ddr::Datastreams::CHECKSUM_TYPE_SHA1)
     end
 
   end

spec/controllers/items_controller_spec.rb

@@ -91,7 +91,7 @@ def create_item_and_component opts={}
         end
         it "should validate the checksum if provided" do
           expect_any_instance_of(Component).to receive(:validate_checksum!)
-          create_item_and_component(checksum: "bda5fda452d0047c27e9e0048ed59428cb9e6d5d46fe9c27dff5c8e39b75a59e", checksum_type: "SHA-256")
+          create_item_and_component(checksum: "75e2e0cec6e807f6ae63610d46448f777591dd6b", checksum_type: "SHA-1")
         end
       end
     end

spec/factories/batch_object_datastream_factories.rb

@@ -20,8 +20,8 @@
         name Ddr::Datastreams::CONTENT
         payload File.join(Rails.root, 'spec', 'fixtures', 'batch_ingest', 'miscellaneous', 'id001.tif')
         payload_type Ddr::Batch::BatchObjectDatastream::PAYLOAD_TYPE_FILENAME
-        checksum "120ad0814f207c45d968b05f7435034ecfee8ac1a0958cd984a070dad31f66f3"
-        checksum_type Ddr::Datastreams::CHECKSUM_TYPE_SHA256
+        checksum "257fa1025245d2d2a60ae81ac7922ca9581ca314"
+        checksum_type Ddr::Datastreams::CHECKSUM_TYPE_SHA1
       end
 
     end

spec/factories/ingest_folder_factories.rb

@@ -3,7 +3,7 @@
     model "TestChild"
     base_path "base/path/"
     sub_path "subpath"
-    checksum_type Ddr::Datastreams::CHECKSUM_TYPE_SHA256
+    checksum_type Ddr::Datastreams::CHECKSUM_TYPE_SHA1
     collection_pid "test:456"
     add_parents true
     parent_id_length 1

spec/fixtures/batch_ingest/miscellaneous/checksums.xml

@@ -2,8 +2,8 @@
 <checksums>
   <checksum>
     <id>id001</id>
-    <value>120ad0814f207c45d968b05f7435034ecfee8ac1a0958cd984a070dad31f66f3</value>
-    <type>SHA-256</type>
+    <value>257fa1025245d2d2a60ae81ac7922ca9581ca314</value>
+    <type>SHA-1</type>
     <filename>id001.tif</filename>
   </checksum>
 </checksums>

spec/fixtures/batch_ingest/miscellaneous/checksums_bad_elements.xml

@@ -2,8 +2,8 @@
 <checksums>
   <checksum>
     <id>id001</id>
-    <misnamedvalue>120ad0814f207c45d968b05f7435034ecfee8ac1a0958cd984a070dad31f66f3</misnamedvalue>
-    <misnamedtype>SHA-256</misnamedtype>
+    <misnamedvalue>257fa1025245d2d2a60ae81ac7922ca9581ca314</misnamedvalue>
+    <misnamedtype>SHA-1</misnamedtype>
     <filename>id001.tif</filename>
   </checksum>
 </checksums>

spec/fixtures/batch_ingest/miscellaneous/incorrect_checksums.xml

@@ -2,8 +2,8 @@
 <checksums>
   <checksum>
     <componentid>id001</componentid>
-    <value>badad0814f207c45d968b05f7435034ecfee8ac1a0958cd984a070dad31f66f3</value>
-    <type>SHA-256</type>
+    <value>badfa1025245d2d2a60ae81ac7922ca9581ca314</value>
+    <type>SHA-1</type>
     <filename>id001.tif</filename>
   </checksum>
 </checksums>

spec/fixtures/batch_ingest/simple_ingest/example/manifest-sha1.txt

@@ -0,0 +1 @@
+548bd2678027f3acb4d4c5ccedf6f92ca07f74bd data/itemA/image1.tiff

spec/fixtures/batch_ingest/simple_ingest/manifest-sha1.txt

@@ -0,0 +1,5 @@
+7cc5abd7ed8c1c907d86bba5e6e18ed6c6ec995c data/file01001.tif
+75e2e0cec6e807f6ae63610d46448f777591dd6b data/file01002.tif
+a6ae0d815c1a2aef551b45fe34a35ceea1828a4d data/pdf/file01.pdf
+2cf23f0035c12b6242093e93d0f7eeba0b1e08e8 data/targets/T001.tiff
+ff01aab0eada29d35bb423c5c73a9f67a22bc1fd data/targets/T002.tiff

spec/models/ingest_folder_spec.rb

@@ -69,7 +69,7 @@
   let(:mount_point_path) { "/mount/" }
   let(:base_path) { File.join(mount_point_name, "base/path/") }
   let(:checksum_directory) { "/fixity/fedora_ingest/" }
-  let(:checksum_type) { "sha256" }
+  let(:checksum_type) { "sha1" }
   let(:user) { FactoryGirl.create(:user) }
   before do
     allow(File).to receive(:readable?).and_return(true)

spec/models/simple_ingest_checksum_spec.rb

@@ -2,15 +2,15 @@
 
 RSpec.describe SimpleIngestChecksum, type: :model, batch: true, simple_ingest: true do
 
-  let(:checksum_filepath) { Rails.root.join('spec', 'fixtures', 'batch_ingest', 'simple_ingest', 'manifest-sha256.txt') }
+  let(:checksum_filepath) { Rails.root.join('spec', 'fixtures', 'batch_ingest', 'simple_ingest', 'manifest-sha1.txt') }
   let(:sic) { SimpleIngestChecksum.new(checksum_filepath) }
 
   describe "checksum" do
     it "should provide the recorded checksum" do
-      expect(sic.checksum('file01001.tif')).to eq('120ad0814f207c45d968b05f7435034ecfee8ac1a0958cd984a070dad31f66f3')
-      expect(sic.checksum('targets/T001.tiff')).to eq('6890139bcc7b9cc0341fee77e8149aad4c7368eb5d0511ae480609dabd48b0c5')
+      expect(sic.checksum('file01001.tif')).to eq('7cc5abd7ed8c1c907d86bba5e6e18ed6c6ec995c')
+      expect(sic.checksum('targets/T001.tiff')).to eq('2cf23f0035c12b6242093e93d0f7eeba0b1e08e8')
       expect(sic.checksum('image001.tif')).to be nil
     end
   end
 
-end
+end

spec/scripts/batch_fixity_check_spec.rb

@@ -49,8 +49,8 @@ module Scripts
             createDate = Time.parse(row["dsCreateDate"])
             expect(createDate).to be_a Time
             expect(createDate).to_not be_utc
-            expect(row["dsChecksumType"]).to eq("SHA-256")
-            expect(row["dsChecksum"]).to match(/^\h{64}$/)
+            expect(row["dsChecksumType"]).to eq("SHA-1")
+            expect(row["dsChecksum"]).to match(/^\h{40}$/)
             expect(row["dsChecksumValid"]).to match(/^(true|false)$/)
           end
         end

spec/scripts/process_simple_ingest_spec.rb

@@ -48,10 +48,10 @@ module DulHydra::Batch::Scripts
         expect(batch_objects[2].batch_object_attributes.where(name: 'title').first.value).to eq('Component 1 Title')
         content_ds = batch_objects[2].batch_object_datastreams.where(name: Ddr::Datastreams::CONTENT).first
         expect(content_ds.payload).to eq(File.join(folder_path, 'data', 'itemA', 'image1.tiff'))
-        expect(content_ds.checksum).to eq('bda5fda452d0047c27e9e0048ed59428cb9e6d5d46fe9c27dff5c8e39b75a59e')
+        expect(content_ds.checksum).to eq('548bd2678027f3acb4d4c5ccedf6f92ca07f74bd')
       end
     end
 
   end
 
-end
+end

spec/services/build_batch_from_folder_ingest_spec.rb

@@ -33,7 +33,7 @@
 
     it "should build an appropriate batch" do
       batch = batch_builder.call
-    
+
       # Batch expectations
       expect(batch.id).to be_present
       expect(batch.name).to eq(batch_name)
@@ -92,7 +92,7 @@
         content_datastreams = obj.batch_object_datastreams.where(
                                   name: Ddr::Datastreams::CONTENT)
         expect(content_datastreams.size).to eq(1)
-        expect(content_datastreams.first.checksum_type).to eq(Ddr::Datastreams::CHECKSUM_TYPE_SHA256)
+        expect(content_datastreams.first.checksum_type).to eq(Ddr::Datastreams::CHECKSUM_TYPE_SHA1)
         component_filepaths << content_datastreams.first.payload
         component_checksums << content_datastreams.first.checksum
       end
@@ -112,4 +112,4 @@
 
   end
 
-end
+end