Handle remote file & directory data entities similarly. Fixes #28

lib/ro_crate/model/data_entity.rb

@@ -27,12 +27,25 @@ def self.specialize(props)
     # PLEASE NOTE, the new data entity will not be added to the crate. To do this, call Crate#add_data_entity.
     #
     # @param crate [Crate] The RO-Crate that owns this directory.
-    # @param source [nil] Ignored. For compatibility with the File and Directory constructor signatures.
-    # @param id [String, nil] An ID to identify this DataEntity, or nil to auto-generate an appropriate one,
-    #   (or determine via the properties param)
+    # @param source [String, Pathname, ::File, URI, nil, #read] The source on the disk (or on the internet if a URI)
+    # where the content of this DataEntity can be found.
+    # @param crate_path [String, nil] The relative path within the RO-Crate where this data entity should be written.
+    # Also used as the ID of the DataEntity. Will be taken from `properties` or generated if `crate_path` is nil.
     # @param properties [Hash{String => Object}] A hash of JSON-LD properties to associate with this DataEntity.
-    def initialize(crate, source = nil, id = nil, properties = {})
-      super(crate, id, properties)
+    def initialize(crate, source = nil, crate_path = nil, properties = {})
+      if crate_path.is_a?(Hash) && properties.empty?
+        properties = crate_path
+        crate_path = nil
+      end
+
+      @source = normalize_source(source)
+
+      if crate_path.nil?
+        crate_path = @source.basename.to_s if @source.respond_to?(:basename)
+        crate_path = @source.to_s if @source.is_a?(URI) && @source.absolute?
+      end
+
+      super(crate, crate_path, properties)
     end
 
     ##
@@ -51,5 +64,28 @@ def payload
     def filepath
       Addressable::URI.unescape(id.sub(/\A\//, '')).to_s # Remove initial / and decode %20 etc.
     end
+
+    private
+
+    ##
+    # Do some normalization of the given source. Coverts `::File` and `String` (if relative path) to `Pathname`
+    # and ensures they are expanded.
+    #
+    # @param source [String, Pathname, ::File, URI, nil, #read] The source on the disk (or on the internet if a URI).
+    # @return [Pathname, URI, nil, #read] An absolute Pathname or URI for the source.
+    def normalize_source(source)
+      if source.is_a?(String)
+        uri = URI(source) rescue nil
+        if uri&.absolute?
+          source = uri
+        else
+          source = Pathname.new(source)
+        end
+      elsif source.is_a?(::File)
+        source = Pathname.new(source)
+      end
+
+      source.is_a?(Pathname) ? source.expand_path : source
+    end
   end
 end

lib/ro_crate/model/directory.rb

@@ -11,20 +11,21 @@ def self.format_local_id(id)
     # Crate#add_data_entity, or just use Crate#add_directory.
     #
     # @param crate [Crate] The RO-Crate that owns this directory.
-    # @param source_directory [String, Pathname, ::File, nil] The source directory that will be included in the crate.
-    # @param crate_path [String] The relative path within the RO-Crate where this directory will be written.
+    # @param source [String, Pathname, ::File, nil] The source directory that will be included in the crate.
+    # @param crate_path [String, nil] The relative path within the RO-Crate where this directory will be written.
     # @param properties [Hash{String => Object}] A hash of JSON-LD properties to associate with this directory.
-    def initialize(crate, source_directory = nil, crate_path = nil, properties = {})
-      @directory_entries = {}
+    def initialize(crate, source = nil, crate_path = nil, properties = {})
+      super(crate, source, crate_path, properties)
 
-      if source_directory
-        source_directory = Pathname.new(::File.expand_path(source_directory))
-        @entry = Entry.new(source_directory)
-        populate_entries(source_directory)
-        crate_path = source_directory.basename.to_s if crate_path.nil?
+      @directory_entries = {}
+      if @source
+        if @source.is_a?(URI) && @source.absolute?
+          @entry = RemoteEntry.new(@source, directory: true)
+        else
+          @entry = Entry.new(@source)
+          populate_entries(@source)
+        end
       end
-
-      super(crate, nil, crate_path, properties)
     end
 
     ##
@@ -34,7 +35,7 @@ def initialize(crate, source_directory = nil, crate_path = nil, properties = {})
     # @return [Hash{String => Entry}>]
     def payload
       entries = {}
-      entries[filepath.chomp('/')] = @entry if @entry
+      entries[filepath.chomp('/')] = @entry if @entry && !remote?
 
       @directory_entries.each do |rel_path, entry|
         entries[full_entry_path(rel_path)] = entry
@@ -43,6 +44,10 @@ def payload
       entries
     end
 
+    def remote?
+      @entry.remote?
+    end
+
     private
 
     ##

lib/ro_crate/model/file.rb

@@ -15,33 +15,12 @@ def self.format_local_id(id)
     # @param crate_path [String] The relative path within the RO-Crate where this file will be written.
     # @param properties [Hash{String => Object}] A hash of JSON-LD properties to associate with this file.
     def initialize(crate, source, crate_path = nil, properties = {})
-      if crate_path.is_a?(Hash) && properties.empty?
-        properties = crate_path
-        crate_path = nil
-      end
-
-      if source.is_a?(String)
-        uri = URI(source) rescue nil
-        if uri.absolute?
-          source = uri
-        else
-          source = Pathname.new(source).expand_path
-        end
-      elsif source.is_a?(::File)
-        source = Pathname.new(source).expand_path
-      end
-
-      if crate_path.nil?
-        crate_path = source.basename.to_s if source.respond_to?(:basename)
-        crate_path = source.to_s if source.is_a?(URI) && source.absolute?
-      end
-
-      super(crate, nil, crate_path, properties)
+      super(crate, source, crate_path, properties)
 
-      if source.is_a?(URI) && source.absolute?
-        @entry = RemoteEntry.new(source)
+      if @source.is_a?(URI) && @source.absolute?
+        @entry = RemoteEntry.new(@source)
       else
-        @entry = Entry.new(source)
+        @entry = Entry.new(@source)
       end
     end
 

lib/ro_crate/model/remote_entry.rb

@@ -9,8 +9,9 @@ class RemoteEntry < Entry
     # Create a new RemoteEntry.
     #
     # @param uri [URI] An absolute URI.
-    def initialize(uri)
+    def initialize(uri, directory: false)
       @uri = uri
+      @directory = directory
     end
 
     ##
@@ -23,7 +24,7 @@ def source
     ##
     # Does this RemoteEntry point to a directory
     def directory?
-      false
+      @directory
     end
 
     ##

test/fixtures/uri_heavy_crate/main.nf

@@ -0,0 +1,2 @@
+# Redacted!
+# Thanks to Veit Schwämmle and Laura Rodriguez Navas for this example