reader.rb in ro-crate-0.5.2

- old
+ new

@@ -7,11 +7,10 @@
     #
     # @param source [String, ::File, Pathname, #read] The location of the zip or directory, or an IO-like object containing a zip.
     # @param target_dir [String, ::File, Pathname] The target directory where the crate should be unzipped (if its a Zip file).
     # @return [Crate] The RO-Crate.
     def self.read(source, target_dir: Dir.mktmpdir)
-      raise "Not a directory!" unless ::File.directory?(target_dir)
       begin
         is_dir = ::File.directory?(source)
       rescue TypeError
         is_dir = false
       end
@@ -80,10 +79,12 @@
     #
     # @param source [String, ::File, Pathname, #read] The location of the zip file, or an IO-like object.
     # @param target_dir [String, ::File, Pathname] The target directory where the crate should be unzipped.
     # @return [Crate] The RO-Crate.
     def self.read_zip(source, target_dir: Dir.mktmpdir)
+      raise ROCrate::ReadException, "Target is not a directory!" unless ::File.directory?(target_dir)
+
       unzip_to(source, target_dir)
 
       # Traverse the unzipped directory to try and find the crate's root
       root_dir = detect_root_directory(target_dir)
 
@@ -94,25 +95,30 @@
     # Reads an RO-Crate from a directory.
     #
     # @param source [String, ::File, Pathname] The location of the directory.
     # @return [Crate] The RO-Crate.
     def self.read_directory(source)
-      raise "Not a directory!" unless ::File.directory?(source)
+      raise ROCrate::ReadException, "Source is not a directory!" unless ::File.directory?(source)
 
       source = ::File.expand_path(source)
       metadata_file = Dir.entries(source).detect { |entry| entry == ROCrate::Metadata::IDENTIFIER ||
           entry == ROCrate::Metadata::IDENTIFIER_1_0 }
 
       if metadata_file
         metadata_json = ::File.read(::File.join(source, metadata_file))
-        metadata = JSON.parse(metadata_json)
+        begin
+          metadata = JSON.parse(metadata_json)
+        rescue JSON::ParserError => e
+          raise ROCrate::ReadException.new("Error parsing metadata", e)
+        end
+
         entities = entities_from_metadata(metadata)
         context = metadata['@context']
 
         build_crate(entities, source, context: context)
       else
-        raise 'No metadata found!'
+        raise ROCrate::ReadException, "No metadata found!"
       end
     end
 
     ##
     # Extracts all the entities from the @graph of the RO-Crate Metadata.
@@ -129,18 +135,18 @@
           entities[entity['@id']] = entity
         end
 
         # Do some normalization...
         entities[ROCrate::Metadata::IDENTIFIER] = extract_metadata_entity(entities)
-        raise "No metadata entity found in @graph!" unless entities[ROCrate::Metadata::IDENTIFIER]
+        raise ROCrate::ReadException, "No metadata entity found in @graph!" unless entities[ROCrate::Metadata::IDENTIFIER]
         entities[ROCrate::Preview::IDENTIFIER] = extract_preview_entity(entities)
         entities[ROCrate::Crate::IDENTIFIER] = extract_root_entity(entities)
-        raise "No root entity (with @id: #{entities[ROCrate::Metadata::IDENTIFIER].dig('about', '@id')}) found in @graph!" unless entities[ROCrate::Crate::IDENTIFIER]
+        raise ROCrate::ReadException, "No root entity (with @id: #{entities[ROCrate::Metadata::IDENTIFIER].dig('about', '@id')}) found in @graph!" unless entities[ROCrate::Crate::IDENTIFIER]
 
         entities
       else
-        raise "No @graph found in metadata!"
+        raise ROCrate::ReadException, "No @graph found in metadata!"
       end
     end
 
     ##
     # Create and populate crate from the given set of entities.
@@ -196,11 +202,11 @@
     # @param crate [Crate] The RO-Crate being read.
     # @param source [String, ::File, Pathname] The location of the RO-Crate being read.
     # @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)
         entity = create_data_entity(crate, entity_class, source, entity_props)
         next if entity.nil?
@@ -232,25 +238,25 @@
     # @param entity_props [Hash] A Hash containing the entity's properties, including its @id.
     # @return [ROCrate::File, ROCrate::Directory, nil] The DataEntity object,
     #          or nil if it referenced a local file that wasn't found.
     def self.create_data_entity(crate, entity_class, source, entity_props)
       id = entity_props.delete('@id')
+      raise ROCrate::ReadException, "Data Entity missing '@id': #{entity_props.inspect}" unless id
       decoded_id = URI.decode_www_form_component(id)
       path = nil
       uri = URI(id) rescue nil
       if uri&.absolute?
         path = uri
         decoded_id = nil
-      else
+      elsif !id.start_with?('#')
         [id, decoded_id].each do |i|
           fullpath = ::File.join(source, i)
           path = Pathname.new(fullpath) if ::File.exist?(fullpath)
         end
-        # unless path
-        #   warn "Missing file/directory: #{id}, skipping..."
-        #   return nil
-        # end
+        if path.nil?
+          raise ROCrate::ReadException, "Local Data Entity not found in crate: #{id}"
+        end
       end
 
       entity_class.new(crate, path, decoded_id, entity_props)
     end
 
@@ -288,10 +294,10 @@
     # https://www.researchobject.org/ro-crate/1.1/root-data-entity.html#finding-the-root-data-entity
     # @return [Hash{String => Hash}] A Hash containing (hopefully) one value, the root entity's properties,
     # mapped by its @id.
     def self.extract_root_entity(entities)
       root_id = entities[ROCrate::Metadata::IDENTIFIER].dig('about', '@id')
-      raise "Metadata entity does not reference any root entity" unless root_id
+      raise ROCrate::ReadException, "Metadata entity does not reference any root entity" unless root_id
       entities.delete(root_id)
     end
 
     ##
     # Finds an RO-Crate's root directory (where `ro-crate-metdata.json` is located) within a given directory.