feature: Import preservation_copy in mountpoint-aware format. (#870)

This is a new feature and should increment the minor version number.

Gemfile.lock

@@ -937,4 +937,4 @@ RUBY VERSION
    ruby 2.5.7p206
 
 BUNDLED WITH
-   1.17.3
+   1.17.3

app/assets/markdown/importer_guide.md

@@ -87,18 +87,23 @@
 
 ### File Name (required)
 
-A _full file path_ to the file in the "Masters" NetApp volume. Currently this must be single-valued. If a `Work` has multiple files associated with it, then each file should be given its own line with the object type `Page` and a `Parent ARK` value that refers to the parent `Work`.
+A _full file path_ to the file, beginning with the NetApp volume in the form `[volume].in.library.ucla.edu`. Currently this must be single-valued. If a `Work` has multiple files associated with it, then each file should be given its own line with the object type `Page` and a `Parent ARK` value that refers to the parent `Work`.
 
-If the File Name starts with "Masters/", it will be used as is. Otherwise, it will be prepended with `Masters/dlmasters/`, in order to match the content of DLCS exports.
+If the File Name does not start with `[volume].in.library.ucla.edu`, it is assumed to refer to `masters.in.library.ucla.edu`. If the first directory is explicitly `Masters/`, then it is interpreted as the volume root. Otherwise, it will be prepended with `masters.in.library.ucla.edu/dlmasters/`, in order to match the content of DLCS exports.
+
+For all formats, any number of leading `/` characters will be ignored.
 
 This field is a string. **This field is required**.
 
 Examples:
 
+- `masters.in.library.ucla.edu/dlmasters/postcards/masters/21198-zz00090nrm-1-master.tif`
 - `Masters/dlmasters/postcards/masters/21198-zz00090ntn-1-master.tif`
+  <br> (Imported as `masters.in.library.ucla.edu/dlmasters/postcards/masters/21198-zz00090ntn-1-master.tif`)
 - `postcards/masters/21198-zz00090nn2-1-master.tif`
-  <br> (Imported as `Masters/dlmasters/postcards/masters/21198-zz00090nn2-1-master.tif`)
-- `Masters/DLTempSecure/ABC/xyz/file_123.tif`
+  <br> (Imported as `masters.in.library.ucla.edu/dlmasters/postcards/masters/21198-zz00090nn2-1-master.tif`)
+- `//othermount.in.library.ucla.edu/ABC/xyz/file_123.tif`
+  <br> (Imported as `othermount.in.library.ucla.edu/ABC/xyz/file_123.tif`)
 
 ### Title (required)
 

app/importers/californica_mapper.rb

@@ -173,7 +173,7 @@ def visibility_mapping
   end
 
   def access_copy
-    map_field(:access_copy).first || preservation_copy
+    map_field(:access_copy).to_a.first
   end
 
   def ark
@@ -294,12 +294,18 @@ def ladnn?
   end
 
   def preservation_copy
-    path = map_field(:preservation_copy).first.to_s.strip.sub(/^\//, '')
-    return nil if path.empty?
-    if path.start_with?('Masters/')
+    path = map_field(:preservation_copy).first.to_s.strip.sub(/^\/+/, '')
+    if path.empty?
+      nil
+    elsif path.start_with?(/[^\/]+.in.library.ucla.edu\//)
+      # Standard format: must specify netapp volume
       path
+    elsif path.start_with?('Masters/')
+      # Legacy standard format: everything starts with "Masters/"
+      'masters.in.library.ucla.edu/' + path.sub(/^\/?Masters\//, '')
     else
-      'Masters/dlmasters/' + path
+      # paths coming from DLCSExport
+      'masters.in.library.ucla.edu/dlmasters/' + path
     end
   end
 

app/uploaders/csv_manifest_validator.rb

@@ -187,8 +187,12 @@ def validate_records
 
       # Row has a File Name that doesn't exist
       if @mapper.preservation_copy
-        full_path = File.join(file_uri_base_path, @mapper.preservation_copy)
-        this_row_warnings << "Rows contain a File Name that does not exist. Incorrect values may be imported." unless File.exist?(full_path)
+        if @mapper.preservation_copy.start_with?('masters.in.library.ucla.edu/')
+          full_path = File.join(file_uri_base_path, 'Masters', @mapper.preservation_copy.sub(/^masters.in.library.ucla.edu\//, ''))
+          this_row_warnings << "Rows contain a File Name that does not exist. Incorrect values may be imported." unless File.exist?(full_path)
+        else
+          this_row_warnings << "Unable to check that file exists. Only files in masters.in.library.ucla.edu/ can be checked at this time." unless File.exist?(full_path)
+        end
       end
 
       # Row has improperly formatted date values

spec/importers/californica_mapper_spec.rb

@@ -283,8 +283,8 @@
           'access_copy' => '' }
       end
 
-      it 'uses preservation_copy' do
-        expect(mapper.access_copy).to eq 'Masters/dlmasters/abc/xyz.tif'
+      it 'is nil' do
+        expect(mapper.access_copy).to be_nil
       end
     end
   end
@@ -295,32 +295,6 @@
     end
   end
 
-  describe '#preservation_copy' do
-    context 'when the path starts with a \'/\'' do
-      let(:metadata) { { 'File Name' => '/Masters/dlmasters/abc/xyz.tif' } }
-
-      it 'gets removed' do
-        expect(mapper.preservation_copy).to eq 'Masters/dlmasters/abc/xyz.tif'
-      end
-    end
-
-    context 'when the path starts with \'Masters/\'' do
-      let(:metadata) { { 'File Name' => 'Masters/dlmasters/abc/xyz.tif' } }
-
-      it 'imports as is' do
-        expect(mapper.preservation_copy).to eq 'Masters/dlmasters/abc/xyz.tif'
-      end
-    end
-
-    context 'when the path doesn\'t start with \'Masters\'' do
-      let(:metadata) { { 'File Name' => 'abc/xyz.tif' } }
-
-      it 'prepends \'Masters/dlmasters\'' do
-        expect(mapper.preservation_copy).to eq 'Masters/dlmasters/abc/xyz.tif'
-      end
-    end
-  end
-
   describe '#ark' do
     it "maps the required ark field" do
       expect(mapper.ark).to eq('ark:/21198/zz0002nq4w')
@@ -376,27 +350,67 @@
   end
 
   describe '#preservation_copy' do
-    context 'when the path starts with a \'/\'' do
-      let(:metadata) { { 'File Name' => '/Masters/dlmasters/abc/xyz.tif' } }
+    context 'when the path begins with [volume].in.library.ucla.edu/' do
+      let(:metadata) { { 'File Name' => 'masters.in.library.ucla.edu/dlmasters/abc/xyz.tif' } }
 
-      it 'gets removed' do
-        expect(mapper.preservation_copy).to eq 'Masters/dlmasters/abc/xyz.tif'
+      it 'imports it unchanged' do
+        expect(mapper.preservation_copy).to eq 'masters.in.library.ucla.edu/dlmasters/abc/xyz.tif'
+      end
+
+      context 'when the path starts with any number of initial `/` characters' do
+        let(:metadata) { { 'File Name' => '//masters.in.library.ucla.edu/dlmasters/abc/xyz.tif' } }
+
+        it 'ignores them' do
+          expect(mapper.preservation_copy).to eq 'masters.in.library.ucla.edu/dlmasters/abc/xyz.tif'
+        end
+      end
+    end
+
+    context 'when the path is not supplied' do
+      let(:metadata) { {} }
+
+      it 'returns nil' do
+        expect(mapper.preservation_copy).to be_nil
+      end
+    end
+
+    context 'when the path is blank' do
+      let(:metadata) { { 'File Name' => '' } }
+
+      it 'returns nil' do
+        expect(mapper.preservation_copy).to be_nil
       end
     end
 
     context 'when the path starts with \'Masters/\'' do
       let(:metadata) { { 'File Name' => 'Masters/dlmasters/abc/xyz.tif' } }
 
-      it 'imports as is' do
-        expect(mapper.preservation_copy).to eq 'Masters/dlmasters/abc/xyz.tif'
+      it 'is interpreted relative to masters.in.library.ucla.edu/' do
+        expect(mapper.preservation_copy).to eq 'masters.in.library.ucla.edu/dlmasters/abc/xyz.tif'
+      end
+
+      context 'when the path starts with any number of initial `/` characters' do
+        let(:metadata) { { 'File Name' => '/Masters/dlmasters/abc/xyz.tif' } }
+
+        it 'ignores them' do
+          expect(mapper.preservation_copy).to eq 'masters.in.library.ucla.edu/dlmasters/abc/xyz.tif'
+        end
       end
     end
 
-    context 'when the path doesn\'t start with \'Masters\'' do
+    context 'when the path doesn\'t start with \'[volume].in.library.ucla.edu/\' or \'Masters\'' do
       let(:metadata) { { 'File Name' => 'abc/xyz.tif' } }
 
-      it 'prepends \'Masters/dlmasters\'' do
-        expect(mapper.preservation_copy).to eq 'Masters/dlmasters/abc/xyz.tif'
+      it 'is interpreted relative to masters.in.library.ucla.edu/dlmasters/ (DLCSExport CSVs)' do
+        expect(mapper.preservation_copy).to eq 'masters.in.library.ucla.edu/dlmasters/abc/xyz.tif'
+      end
+
+      context 'when the path starts with \'/\'' do
+        let(:metadata) { { 'File Name' => '/abc/xyz.tif' } }
+
+        it 'is interpreted the same way' do
+          expect(mapper.preservation_copy).to eq 'masters.in.library.ucla.edu/dlmasters/abc/xyz.tif'
+        end
       end
     end
   end

spec/spec_helper.rb

@@ -102,16 +102,16 @@
   # as the one that triggered the failure.
   Kernel.srand config.seed
 
-  # Silence normal program output (cf. https://stackoverflow.com/questions/15430551/suppress-console-output-during-rspec-tests#15432948)
-  original_stderr = $stderr
-  original_stdout = $stdout
-  config.before(:all) do
-    # Redirect stderr and stdout
-    $stderr = File.open(File::NULL, "w")
-    $stdout = File.open(File::NULL, "w")
-  end
-  config.after(:all) do
-    $stderr = original_stderr
-    $stdout = original_stdout
-  end
+  # # Silence normal program output (cf. https://stackoverflow.com/questions/15430551/suppress-console-output-during-rspec-tests#15432948)
+  # original_stderr = $stderr
+  # original_stdout = $stdout
+  # config.before(:all) do
+  #   # Redirect stderr and stdout
+  #   $stderr = File.open(File::NULL, "w")
+  #   $stdout = File.open(File::NULL, "w")
+  # end
+  # config.after(:all) do
+  #   $stderr = original_stderr
+  #   $stdout = original_stdout
+  # end
 end