hybag 0.0.7 → 0.0.8

checksums.yaml

@@ -1,15 +1,15 @@
 ---
 !binary "U0hBMQ==":
   metadata.gz: !binary |-
-    YWRkN2Q0ZGNmODNmOTQ5NmVjOGY4ZWQ3MzhhZTc1MTU1NDFlZGYzNA==
+    MjEwZDk1ODkxNDQ4OThiMzhhM2U0ZDI3MDdkM2M3YTAwZjM0MzI1Yw==
   data.tar.gz: !binary |-
-    Y2ZlMDIwYWE1OTExOGY4NzY2MTI2MTgxODliODhhOTdhZDM1Y2NkNg==
+    OTlmYmU4Y2UzMjBjY2Y1ODg0NmJhN2VjMzFkZTkyYjEyYzM3OGYzYQ==
 !binary "U0hBNTEy":
   metadata.gz: !binary |-
-    OTYwNDA0OWRlZGE3MWJkNjk5ZDM3YWMxMjA3NjRkNjM2ZmNiZGI4MmYwNTk3
-    ZGQzMzk3NjI3ZDkxMjg3Y2UxYWM5ZWM4NThmMjA4NjM3YWY0NmJhYzBiZWRj
-    ZWNjNzJiYTdmZDI4ZTM4NDZkNTc3YjNiZmRjYjhlNzI5OTE0ZjQ=
+    NjhhZWY3NDczZmE2MzM4MDE2NjkyZGJjOWYxYzA0ZjgwYjhlMDAxM2Y0NWM0
+    ZmVhNjYyYmVkNzM1NzMyZDAzYTNhNWZkYjY3ZjEzNzRhMTM0YzA2YzVlMjcx
+    MTgzYjk1MzlmYzQxMTZmYTY0NmJhZWQxMjE4YmMxMWEyNjZhMWU=
   data.tar.gz: !binary |-
-    NWUwMTFjMDM0NTc5ZTdlOWE4YjYyNjgwNDI4M2JlMGZkY2IzMjZkNzUzOTE2
-    ZDNlYjZiMzNjODk5ODRjM2IwNzQ0YTM4OGFkM2JkNmZjZTI4M2Q3MzUxMjIx
-    MzNlYjYyZTU3Y2RhZDYzY2Q5MzcxMjc4MGJiYmM0ZGNkNDI1YmE=
+    ZTNmMzdkNjJjNmE0MGQwYzA4NTU0NjM5Yzk1OTYwZGU2YmFmOWJkNTY0YzI5
+    YzQ5MzRmNzM0MTg3NzhlZjhiYzI1NDJmOGRkMmRkNjc1MjhhZmYyNjhhNjY5
+    OWNhMjIzMjIyMGU0MzE0ZjI4OGVjZTJkMjBlZmUyZjRkOThmYmM=

data/lib/hybag/ingester.rb

@@ -11,7 +11,7 @@ module Hybag
       raise "Unable to determine model from bag" if model_name.blank?
       new_object = ActiveFedora.class_from_string(model_name.to_s).new
       # Assign a pid
-      new_object.inner_object.pid = ActiveFedora::Base.assign_pid(new_object)
+      new_object.inner_object.pid = new_object.inner_object.assign_pid
       set_metadata_streams(new_object)
       set_file_streams(new_object)
       return new_object
@@ -19,58 +19,64 @@ module Hybag
 
     private
 
-    # TODO: What to do if the bag has files that don't have model definitions?
     # TODO: Add some sort of configuration to map bag filenames -> dsids.
     def set_metadata_streams(object)
-      object.metadata_streams.each do |ds|
-        if bag_has_metastream?(ds.dsid)
-          ds.content = bag_metastream(ds.dsid).read.strip
-          # Assume the first subject in the metadata is about this object.
-          # TODO: Move this to configuration?
-          first_subject = ds.graph.first_subject
-          new_repository = RDF::Repository.new
-          ds.graph.each_statement do |statement|
-            subject = statement.subject
-            subject = ds.rdf_subject if subject == first_subject
-            new_repository << [subject, statement.predicate, statement.object]
-          end
-          ds.instance_variable_set(:@graph,new_repository)
-        end
+      bag_tag_files.each do |tag_file|
+        add_bag_file_to_object(object, tag_file, false)
       end
     end
 
-    def set_file_streams(object)
-      file_streams = object.datastreams.select{|k, ds| !ds.metadata?}.values
-      file_streams.each do |ds|
-        if bag_has_datastream?(ds.dsid)
-          ds.content = bag_datastream(ds.dsid).read
-        end
-      end
+
+    # Returns all registered tag files except those generated for the bag
+    # These includes the bag_info.txt, bagit.txt, and manifest files.
+    def bag_tag_files
+      bag.tag_files - [bag.bag_info_txt_file] - bag.manifest_files - [bag.bagit_txt_file]
     end
 
-    # TODO: Might consider decoration at some point.
-    def bag_filename_to_label(bag_filename)
-      Pathname.new(bag_filename).basename.sub_ext('').to_s
+    def add_bag_file_to_object(object, bag_file, binary=true)
+      parsed_name = bag_filename_to_label(bag_file)
+      found_datastream = object.datastreams.values.find{|x| x.dsid.downcase == bag_filename_to_label(bag_file).downcase}
+      content = File.open(bag_file).read
+      content = transform_content(content) unless binary
+      if found_datastream
+        found_datastream = replace_subject(content, found_datastream)
+      else
+        object.add_file_datastream(content, :dsid => parsed_name)
+      end
     end
 
-    def bag_has_datastream?(label)
-      bag.bag_files.any?{|x| bag_filename_to_label(x) == label}
+    def transform_content(content)
+      content = content.strip
     end
 
-    def bag_datastream(label)
-      bag_file = bag.bag_files.select{|x| bag_filename_to_label(x) == label}.first
-      result = File.open(bag_file) unless bag_file.blank?
-      return result
+    # Replaces the subject in RDF files with the datastream's rdf_subject.
+    # TODO: Deal with what happens when there's no defined datastream.
+    def replace_subject(content, ds)
+      ds.content = content
+      if ds.respond_to?(:rdf_subject)
+        # Assume the first subject in the metadata is about this object.
+        # TODO: Move this to configuration?
+        first_subject = ds.graph.first_subject
+        new_repository = RDF::Repository.new
+        ds.graph.each_statement do |statement|
+          subject = statement.subject
+          subject = ds.rdf_subject if subject == first_subject
+          new_repository << [subject, statement.predicate, statement.object]
+        end
+        ds.instance_variable_set(:@graph,new_repository)
+      end
+      return ds
     end
 
-    def bag_has_metastream?(label)
-      bag.tag_files.any?{|x| bag_filename_to_label(x) == label}
+    def set_file_streams(object)
+      bag.bag_files.each do |bag_file|
+        add_bag_file_to_object(object, bag_file)
+      end
     end
 
-    def bag_metastream(label)
-      tag_file = bag.tag_files.select{|x| bag_filename_to_label(x) == label}.first
-      result = File.open(tag_file) unless tag_file.blank?
-      return result
+    # TODO: Might consider decoration at some point.
+    def bag_filename_to_label(bag_filename)
+      Pathname.new(bag_filename).basename.sub_ext('').to_s
     end
 
     def model_name

data/lib/hybag/version.rb

@@ -1,3 +1,3 @@
 module Hybag
-  VERSION = "0.0.7"
+  VERSION = "0.0.8"
 end

data/spec/lib/hybag/ingester_spec.rb

@@ -112,6 +112,49 @@ describe Hybag::Ingester do
         it "should not be persisted" do
           expect(built_model).not_to be_persisted
         end
+        context "when there is a file datastream and no matching datastream defined" do
+          before(:each) do
+            # Add the hydra.png from fixture
+            FakeFS.deactivate!
+            hybag_content = File.read(File.join(FIXTURE_PATH,"hydra.png"))
+            FakeFS.activate!
+            File.open(File.join(bag.data_dir,"new_content.png"),'wb') {|f| f.puts hybag_content}
+          end
+          it "should add that datastream" do
+            expect(built_model.datastreams.keys).to include("new_content")
+          end
+        end
+        context "when there is a metadata stream and no matching datastream defined" do
+          before(:each) do
+            # Add the example_datastream.nt from fixture
+            FakeFS.deactivate!
+            @hybag_content = File.read(File.join(FIXTURE_PATH,"example_datastream.nt"))
+            FakeFS.activate!
+            bag.add_tag_file("example_datastream.nt") do |f|
+              f.write @hybag_content
+            end
+          end
+          it "should add that file as a datastream" do
+            expect(built_model.datastreams.keys).to include("example_datastream")
+            expect(built_model.datastreams.values.find{|x| x.dsid == "example_datastream"}.content).to eq @hybag_content.strip
+          end
+          context "and it's an RDF datastream" do
+            it "should replace the subject"
+          end
+        end
+        context "when there is an unregistered tag file and no matching datastream defined" do
+          before(:each) do
+            FakeFS.deactivate!
+            @hybag_content = File.read(File.join(FIXTURE_PATH,"example_datastream.nt"))
+            FakeFS.activate!
+            File.open(File.join(bag.bag_dir, "new_tag_file.nt"), 'wb') {|f| f.puts @hybag_content}
+          end
+          # TODO: Write this when bagit supports returning unmarked tag files.
+          xit "should add that datastream" do
+            expect(built_model.datastreams.keys).to include("new_tag_file")
+            expect(built_model.datastreams.values.find{|x| x.dsid == "new_tag_file"}.content).to eq @hybag_content.strip
+          end
+        end
       end
     end
   end

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: hybag
 version: !ruby/object:Gem::Version
-  version: 0.0.7
+  version: 0.0.8
 platform: ruby
 authors:
 - Trey Terrell
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2013-08-08 00:00:00.000000000 Z
+date: 2013-09-16 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: bundler