Fix reading crate with missing hasPart (it is optional)

Also test handling of root data entity with non-default ID

Author: fbacall

lib/ro_crate/reader.rb

@@ -204,7 +204,7 @@ def self.initialize_crate(entity_hash, source, crate_class: ROCrate::Crate, cont
     # @param entity_hash [Hash] A Hash containing all the entities in the @graph, mapped by their @id.
     # @return [Array<ROCrate::File, ROCrate::Directory>] The extracted DataEntity objects.
     def self.extract_data_entities(crate, source, entity_hash)
-      crate.raw_properties['hasPart'].map do |ref|
+      (crate.raw_properties['hasPart'] || []).map do |ref|
         entity_props = entity_hash.delete(ref['@id'])
         next unless entity_props
         entity_class = ROCrate::DataEntity.specialize(entity_props)

test/fixtures/arcp/ro-crate-metadata.json

@@ -0,0 +1,21 @@
+{
+    "@context": "https://w3id.org/ro/crate/1.1/context",
+    "@graph": [
+        {
+            "@id": "arcp://name,somethingsomething",
+            "@type": "Dataset",
+            "datePublished": "2024-01-31T10:47:17+00:00"
+            
+        },
+        {
+            "@id": "ro-crate-metadata.json",
+            "@type": "CreativeWork",
+            "about": {
+                "@id": "arcp://name,somethingsomething"
+            },
+            "conformsTo": {
+                "@id": "https://w3id.org/ro/crate/1.1"
+            }
+        }
+    ]
+}

test/reader_test.rb

@@ -360,6 +360,13 @@ class ReaderTest < Test::Unit::TestCase
     assert_include e.message, 'not found in crate: file1.txt'
   end
 
+  test 'tolerates arcp identifier on root data entity (and missing hasPart)' do
+    crate = ROCrate::Reader.read(fixture_file('arcp').path)
+
+    assert_equal 'arcp://name,somethingsomething', crate.id
+    assert_empty crate.data_entities
+  end
+
   private
 
   def check_exception(exception_class)

test/writer_test.rb

@@ -230,36 +230,48 @@ class WriterTest < Test::Unit::TestCase
 
   test 'write crate with remote files and directories' do
     orig = ROCrate::Reader.read(fixture_file('uri_heavy_crate').path)
-      Tempfile.create do |file|
-        ROCrate::Writer.new(orig).write_zip(file)
+    Tempfile.create do |file|
+      ROCrate::Writer.new(orig).write_zip(file)
 
-        Zip::File.open(file) do |zipfile|
-          refute zipfile.find_entry('nih:sha-256;3a2c-8d14-a40b-3755-4abc-5af8-a56d-ba3a-e159-d688-c9b3-f169-6751-4b88-fbd2-6a9f;7')
-        end
+      Zip::File.open(file) do |zipfile|
+        refute zipfile.find_entry('nih:sha-256;3a2c-8d14-a40b-3755-4abc-5af8-a56d-ba3a-e159-d688-c9b3-f169-6751-4b88-fbd2-6a9f;7')
+      end
+
+      file.rewind
+
+      crate = ROCrate::Reader.read(file)
+
+      dir = crate.get('nih:sha-256;f70e-eb2e-89d0-b3dc-5c99-8541-fa4b-6e64-a194-cf9d-ebd8-ca58-24e7-c47a-553f-86fa;c/')
+      assert dir
+      assert dir.is_a?(ROCrate::Directory)
+      assert dir.remote?
+      assert_empty dir.payload
+
+      file = crate.get('nih:sha-256;3a2c-8d14-a40b-3755-4abc-5af8-a56d-ba3a-e159-d688-c9b3-f169-6751-4b88-fbd2-6a9f;7')
+      assert file
+      assert file.is_a?(ROCrate::File)
+      assert file.remote?
+      assert_empty file.payload
+
+      real_file = crate.get('main.nf')
+      assert real_file
+      assert real_file.is_a?(ROCrate::File)
+      refute real_file.remote?
+      assert_not_empty real_file.payload
+      refute real_file.payload.values.first.remote?
+      refute real_file.payload.values.first.directory?
+    end
+  end
+
+  test 'write crate with arcp root identifier' do
+    crate = ROCrate::Reader.read(fixture_file('arcp').path)
 
-        file.rewind
-
-        crate = ROCrate::Reader.read(file)
-
-        dir = crate.get('nih:sha-256;f70e-eb2e-89d0-b3dc-5c99-8541-fa4b-6e64-a194-cf9d-ebd8-ca58-24e7-c47a-553f-86fa;c/')
-        assert dir
-        assert dir.is_a?(ROCrate::Directory)
-        assert dir.remote?
-        assert_empty dir.payload
-
-        file = crate.get('nih:sha-256;3a2c-8d14-a40b-3755-4abc-5af8-a56d-ba3a-e159-d688-c9b3-f169-6751-4b88-fbd2-6a9f;7')
-        assert file
-        assert file.is_a?(ROCrate::File)
-        assert file.remote?
-        assert_empty file.payload
-
-        real_file = crate.get('main.nf')
-        assert real_file
-        assert real_file.is_a?(ROCrate::File)
-        refute real_file.remote?
-        assert_not_empty real_file.payload
-        refute real_file.payload.values.first.remote?
-        refute real_file.payload.values.first.directory?
+    assert_equal 'arcp://name,somethingsomething', crate.id
+    Dir.mktmpdir do |dir|
+      ROCrate::Writer.new(crate).write(dir)
+      Dir.chdir(dir) do
+        assert File.exist?('ro-crate-metadata.json')
       end
+    end
   end
 end