Add optional hidden file detection (#51)

lib/bagit/bag.rb

@@ -8,18 +8,40 @@
 require "bagit/valid"
 
 module BagIt
+  class FileFinder
+    def self.find(dir)
+      raise NotImplementedError
+    end
+  end
+
+  class StandardFileFinder < FileFinder
+    def self.find(dir)
+      Dir[File.join(dir, "**", "*")].select { |f| File.file? f }
+    end
+  end
+
+  class StandardWithHiddenFileFinder < FileFinder
+    def self.find(dir)
+      Dir.glob(File.join(dir, "**", "*"), File::FNM_DOTMATCH).select { |f| File.file? f }
+    end
+  end
+
   # Represents the state of a bag on a filesystem
   class Bag
     attr_reader :bag_dir
+    attr_reader :detect_hidden
 
     include Validity # Validity functionality
     include Info # bagit & bag info functionality
     include Manifest # manifest related functionality
     include Fetch # fetch related functionality
 
     # Make a new Bag based at path
-    def initialize(path, info = {}, _create = false)
+    def initialize(path, info = {}, _create = false, detect_hidden = false)
       @bag_dir = path
+      @detect_hidden = detect_hidden
+      @file_finder = @detect_hidden ? StandardWithHiddenFileFinder : StandardFileFinder
+
       # make the dir structure if it doesn't exist
       FileUtils.mkdir bag_dir unless File.directory? bag_dir
       FileUtils.mkdir data_dir unless File.directory? data_dir
@@ -37,7 +59,7 @@ def data_dir
 
     # Return the paths to each bag file relative to bag_dir
     def bag_files
-      Dir[File.join(data_dir, "**", "*")].select { |f| File.file? f }
+      @file_finder.find(data_dir)
     end
 
     # Return the paths to each tag file relative to bag_dir

lib/bagit/valid.rb

@@ -33,7 +33,11 @@ def complete?
 
       empty_manifests.each do |file|
         logger.error("#{file} is manifested but not present".red)
-        errors.add :completeness, "#{file} is manifested but not present"
+        error_message = "#{file} is manifested but not present"
+        if !detect_hidden && file.start_with?(File.join("data", "."))
+          error_message += "; consider turning on hidden file detection"
+        end
+        errors.add :completeness, error_message
       end
       tag_empty_manifests.each do |file|
         logger.error("#{file} is a manifested tag but not present".red)

spec/bagit_spec.rb

@@ -165,4 +165,31 @@
       end
     end
   end
+
+  describe "bag with hidden files" do
+    before do
+      @sandbox = Sandbox.new
+
+      # make the bag
+      @bag_path = File.join @sandbox.to_s, "the_bag"
+      @bag = described_class.new @bag_path, {}, false, true
+
+      # add some files
+      @bag.add_file(".keep") { |io| io.puts "" }
+      @bag.add_file("test.txt") { |io| io.puts "testing testing" }
+    end
+
+    after do
+      @sandbox.cleanup!
+    end
+
+    describe "#bag_files" do
+      it "returns an array including non-hidden and hidden files" do
+        files = @bag.bag_files.map { |f| f.sub(File.join(@bag_path, "data", ""), "") }
+        expect(files).to be_a_kind_of(Array)
+        expect(files).not_to be_empty
+        expect(files).to eq([".keep", "test.txt"])
+      end
+    end
+  end
 end

spec/validation_spec.rb

@@ -44,13 +44,17 @@
 
     it "is invalid if there are files that are in the manifest but not in the bag" do
       # add a file and then remove it through the back door
-      @bag.add_file("file-k") { |io| io.puts "time to go" }
+      file_name = "file-k"
+      @bag.add_file(file_name) { |io| io.puts "time to go" }
       @bag.manifest!
 
-      FileUtils.rm File.join(@bag.bag_dir, "data", "file-k")
+      FileUtils.rm File.join(@bag.bag_dir, "data", file_name)
 
       @bag.validate_only("true_for/completeness")
       expect(@bag.errors.on(:completeness)).not_to be_empty
+      expect(@bag.errors.on(:completeness)).to include(
+        "#{File.join("data", file_name)} is manifested but not present"
+      )
       expect(@bag).not_to be_valid
     end
 
@@ -139,4 +143,63 @@
       end
     end
   end
+
+  describe "a bag with unmanifested hidden files" do
+    before do
+      @sandbox = Sandbox.new
+
+      # make a bag with hidden files not manifested
+      @source_bag_path = File.join @sandbox.to_s, "the_bag"
+      @source_bag = described_class.new @source_bag_path, {}, false, false
+      @source_bag.add_file(".keep") { |io| io.puts "" }
+      @source_bag.add_file("test.txt") { |io| io.puts "testing testing" }
+      @source_bag.manifest!
+    end
+
+    after do
+      @sandbox.cleanup!
+    end
+
+    it "fails validation when hidden file detection is on" do
+      @aware_bag = described_class.new @source_bag_path, {}, false, true
+      expect(@aware_bag).to_not be_valid
+      expect(@aware_bag.errors.on(:completeness)).not_to be_empty
+    end
+
+    it "passes validation when hidden file detection is off" do
+      @unaware_bag = described_class.new @source_bag_path, {}, false, false
+      expect(@unaware_bag).to be_valid
+    end
+  end
+
+  describe "a bag with manifested hidden files" do
+    before do
+      @sandbox = Sandbox.new
+
+      # make a bag with hidden files manifested
+      @source_bag_path = File.join @sandbox.to_s, "the_bag"
+      @source_bag = described_class.new @source_bag_path, {}, false, true
+      @source_bag.add_file(".keep") { |io| io.puts "" }
+      @source_bag.add_file("test.txt") { |io| io.puts "testing testing" }
+      @source_bag.manifest!
+    end
+
+    after do
+      @sandbox.cleanup!
+    end
+
+    it "passes validation when hidden file detection is on" do
+      @aware_bag = described_class.new @source_bag_path, {}, false, true
+      expect(@aware_bag).to be_valid
+    end
+
+    it "fails validation when hidden file detection is off, with suggested fix offered" do
+      @unaware_bag = described_class.new @source_bag_path, {}, false, false
+      expect(@unaware_bag).to_not be_valid
+      expect(@unaware_bag.errors.on(:completeness)).not_to be_empty
+      expect(@unaware_bag.errors.on(:completeness)).to include(
+        "data/.keep is manifested but not present; consider turning on hidden file detection"
+      )
+    end
+  end
 end