bagit 0.3.5 → 0.4.4

data/lib/bagit/fetch.rb

@@ -1,50 +1,46 @@
-require 'open-uri'
+# frozen_string_literal: true
+
+require "open-uri"
 
 module BagIt
-  
   module Fetch
-    
     def fetch_txt_file
-      File.join @bag_dir, 'fetch.txt'
+      File.join @bag_dir, "fetch.txt"
     end
 
     def add_remote_file(url, path, size, sha1, md5)
-      open(fetch_txt_file, 'a') { |io| io.puts "#{url} #{size || '-'} #{path}" }
-      open(manifest_file('sha1'), 'a') { |io| io.puts "#{sha1} #{File.join 'data', path}" }
-      open(manifest_file('md5'), 'a') { |io| io.puts "#{md5} #{File.join 'data', path}" }
+      File.open(fetch_txt_file, "a") { |io| io.puts "#{url} #{size || "-"} #{path}" }
+      File.open(manifest_file("sha1"), "a") { |io| io.puts "#{sha1} #{File.join "data", path}" }
+      File.open(manifest_file("md5"), "a") { |io| io.puts "#{md5} #{File.join "data", path}" }
     end
 
     # feth all remote files
     def fetch!
-
-      open(fetch_txt_file) do |io|
-        
+      File.open(fetch_txt_file) do |io|
         io.readlines.each do |line|
-          
-          (url, length, path) = line.chomp.split(/\s+/, 3)
-          
-          add_file(path) do |io|
-            io.write open(url)
+          (url, _length, path) = line.chomp.split(/\s+/, 3)
+
+          add_file(path) do |file_io|
+            file_io.write URI.open(url)
           end
-          
         end
-        
       end
 
-      # rename the old fetch.txt
-      Dir["#{fetch_txt_file}.?*"].sort.reverse.each do |f|
-        
+      rename_old_fetch_txt(fetch_txt_file)
+      move_current_fetch_txt(fetch_txt_file)
+    end
+
+    def rename_old_fetch_txt(fetch_txt_file)
+      Dir["#{fetch_txt_file}.?*"].sort.reverse_each do |f|
         if f =~ /fetch.txt.(\d+)$/
-          new_f = File.join File.dirname(f), "fetch.txt.#{$1.to_i + 1}"
-          FileUtils::mv f, new_f
+          new_f = File.join File.dirname(f), "fetch.txt.#{Regexp.last_match(1).to_i + 1}"
+          FileUtils.mv f, new_f
         end
-        
       end
+    end
 
-      # move the current fetch_txt
-      FileUtils::mv fetch_txt_file, "#{fetch_txt_file}.0"
+    def move_current_fetch_txt(fetch_txt_file)
+      FileUtils.mv fetch_txt_file, "#{fetch_txt_file}.0"
     end
-    
   end
-  
 end

data/lib/bagit/file.rb

@@ -1,19 +1,16 @@
-class File
+# frozen_string_literal: true
 
+class File
   # Clean out all the empty dirs
-  def File.clean(file_name)
+  def self.clean(file_name)
+    return unless File.directory? file_name
+    # clean all subdirs
+    subdirs = Dir.entries(file_name).select { |p| File.directory?(File.join(file_name, p)) }
+    subdirs.reject! { |p| %w[. ..].include? p }
+    subdirs.each { |sd| File.clean File.join(file_name, sd) }
 
-    if File.directory? file_name
-      # clean all subdirs
-      subdirs = Dir.entries(file_name).select { |p| File.directory?(File.join(file_name, p)) }
-      subdirs.reject! { |p| %w{. ..}.include? p }
-      subdirs.each { |sd| File.clean File.join(file_name, sd) }
-
-      # if its empty then delete it
-      contents = Dir.entries(file_name).reject { |p| %w{. ..}.include? p }
-      Dir.delete file_name if contents.empty?
-    end
-    
+    # if its empty then delete it
+    contents = Dir.entries(file_name).reject { |p| %w[. ..].include? p }
+    Dir.delete file_name if contents.empty?
   end
-
 end

data/lib/bagit/info.rb

@@ -1,49 +1,47 @@
-require 'set'
+# frozen_string_literal: true
 
-module BagIt
+require "set"
 
+module BagIt
   module Info
-
     @@bag_info_headers = {
-      :agent => 'Bag-Software-Agent',
-      :org => 'Source-Organization',
-      :org_addr => 'Organization-Address',
-      :contact_name => 'Contact-Name',
-      :contact_phone => 'Contact-Phone',
-      :contact_email => 'Contact-Email',
-      :ext_desc => 'External-Description',
-      :ext_id => 'External-Identifier',
-      :size => 'Bag-Size',
-      :group_id => 'Bag-Group-Identifier',
-      :group_count => 'Bag-Count',
-      :sender_id => 'Internal-Sender-Identifier',
-      :int_desc => 'Internal-Sender-Description',
-      :date => 'Bagging-Date',
-      :oxum => 'Payload-Oxum'
+      agent: "Bag-Software-Agent",
+      org: "Source-Organization",
+      org_addr: "Organization-Address",
+      contact_name: "Contact-Name",
+      contact_phone: "Contact-Phone",
+      contact_email: "Contact-Email",
+      ext_desc: "External-Description",
+      ext_id: "External-Identifier",
+      size: "Bag-Size",
+      group_id: "Bag-Group-Identifier",
+      group_count: "Bag-Count",
+      sender_id: "Internal-Sender-Identifier",
+      int_desc: "Internal-Sender-Description",
+      date: "Bagging-Date",
+      oxum: "Payload-Oxum"
     }
 
     def bag_info_txt_file
-      File.join bag_dir, 'bag-info.txt'
+      File.join bag_dir, "bag-info.txt"
     end
 
     def bag_info
-      begin
-        read_info_file bag_info_txt_file
-      rescue
-        {}
-      end
+      read_info_file bag_info_txt_file
+    rescue
+      {}
     end
 
-    def write_bag_info(hash={})
+    def write_bag_info(hash = {})
       hash = bag_info.merge(hash)
-      hash[@@bag_info_headers[:agent]] = "BagIt Ruby Gem (http://bagit.rubyforge.org)" if hash[@@bag_info_headers[:agent]].nil?
-      hash[@@bag_info_headers[:date]] = Date.today.strftime('%Y-%m-%d') if hash[@@bag_info_headers[:date]].nil?
+      hash[@@bag_info_headers[:agent]] = "BagIt Ruby Gem (https://github.com/tipr/bagit)" if hash[@@bag_info_headers[:agent]].nil?
+      hash[@@bag_info_headers[:date]] = Date.today.strftime("%Y-%m-%d") if hash[@@bag_info_headers[:date]].nil?
       hash[@@bag_info_headers[:oxum]] = payload_oxum
       write_info_file bag_info_txt_file, hash
     end
 
     def bagit_txt_file
-      File.join bag_dir, 'bagit.txt'
+      File.join bag_dir, "bagit.txt"
     end
 
     def bagit
@@ -55,54 +53,44 @@ module BagIt
     end
 
     def update_bag_date
-      hash["Bagging-Date"] = Date.today.strftime('%Y-%m-%d')
+      hash["Bagging-Date"] = Date.today.strftime("%Y-%m-%d")
       write_bag_info(hash)
     end
 
     protected
 
     def read_info_file(file)
-
       File.open(file) do |io|
-
-        entries = io.read.split /\n(?=[^\s])/
+        entries = io.read.split(/\n(?=[^\s])/)
 
         entries.inject({}) do |hash, line|
-          name, value = line.chomp.split /\s*:\s*/, 2
-          hash.merge({name => value})
+          name, value = line.chomp.split(/\s*:\s*/, 2)
+          hash.merge(name => value)
         end
-
       end
-
     end
 
     def write_info_file(file, hash)
-
-      dups = hash.keys.inject(Set.new) do |acc, key|
+      dups = hash.keys.inject(Set.new) { |acc, key|
         a = hash.keys.grep(/#{key}/i)
         acc + (a.size > 1 ? a : [])
-      end
+      }
 
-      raise "Multiple labels (#{dups.to_a.join ', '}) in #{file}" unless dups.empty?
-
-      File.open(file, 'w') do |io|
+      raise "Multiple labels (#{dups.to_a.join ", "}) in #{file}" unless dups.empty?
 
+      File.open(file, "w") do |io|
         hash.each do |name, value|
-          simple_entry = "#{name}: #{value.gsub /\s+/, ' '}"
+          simple_entry = "#{name}: #{value.gsub(/\s+/, " ")}"
 
           entry = if simple_entry.length > 79
-                    simple_entry.wrap(77).indent(2)
-                  else
-                    simple_entry
-                  end
+            simple_entry.wrap(77).indent(2)
+          else
+            simple_entry
+          end
 
           io.puts entry
         end
-
       end
-
     end
-
   end
-
 end

data/lib/bagit/manifest.rb

@@ -1,22 +1,22 @@
-require 'pathname'
-require 'digest/sha1'
-require 'digest/md5'
+# frozen_string_literal: true
 
-module BagIt
+require "pathname"
+require "digest/sha1"
+require "digest/md5"
 
+module BagIt
   # Requires response to bag_dir, tag_files, bag_files
   module Manifest
     def encode_filename(s)
-     s = s.gsub(/\r/, '%0D')
-     s = s.gsub(/\n/,'%0A')
-     return s
+      s = s.gsub(/\r/, "%0D")
+      s = s.gsub(/\n/, "%0A")
+      s
     end
 
-  
     # All tag files that are bag manifest files (manifest-[algorithm].txt)
     def manifest_files
-      files = Dir[File.join(@bag_dir, '*')].select { |f|
-        File.file? f and File.basename(f) =~ /^manifest-.*.txt/
+      files = Dir[File.join(@bag_dir, "*")].select { |f|
+        File.file?(f) && File.basename(f) =~ /^manifest-.*.txt$/
       }
       files
     end
@@ -27,30 +27,59 @@ module BagIt
     end
 
     # Generate manifest files for all the bag files
-    def manifest!
-
+    def manifest!(algo: "default")
       # nuke all the existing manifest files
-      manifest_files.each { |f| FileUtils::rm f }
+      manifest_files.each { |f| FileUtils.rm f }
 
       # manifest each tag file for each algorithm
       bag_files.each do |f|
         rel_path = encode_filename(Pathname.new(f).relative_path_from(Pathname.new(bag_dir)).to_s)
-        
-        # sha1
-        sha1 = Digest::SHA1.file f
-        File.open(manifest_file(:sha1), 'a') { |io| io.puts "#{sha1} #{rel_path}" }
-
-        # md5
-        md5 = Digest::MD5.file f
-        File.open(manifest_file(:md5), 'a') { |io| io.puts "#{md5} #{rel_path}" }
+
+        write_checksum(checksum_algo: algo, relative_path: rel_path, file: f)
       end
       tagmanifest!
     end
 
+    def write_checksum(checksum_algo:, relative_path:, file:)
+      case checksum_algo
+      when "sha1"
+        write_sha1(file, relative_path)
+      when "md5"
+        write_md5(file, relative_path)
+      when "sha256"
+        write_sha256(file, relative_path)
+      when "sha512"
+        write_sha256(file, relative_path)
+      when "default"
+        write_sha1(file, relative_path)
+        write_md5(file, relative_path)
+      end
+    end
+
+    def write_sha1(f, rel_path)
+      sha1 = Digest::SHA1.file f
+      File.open(manifest_file(:sha1), "a") { |io| io.puts "#{sha1} #{rel_path}" }
+    end
+
+    def write_md5(f, rel_path)
+      md5 = Digest::MD5.file f
+      File.open(manifest_file(:md5), "a") { |io| io.puts "#{md5} #{rel_path}" }
+    end
+
+    def write_sha256(f, rel_path)
+      sha256 = Digest::SHA256.file f
+      File.open(manifest_file(:sha256), "a") { |io| io.puts "#{sha256} #{rel_path}" }
+    end
+
+    def write_sha512(f, rel_path)
+      sha512 = Digest::SHA512.file f
+      File.open(manifest_file(:sha512), "a") { |io| io.puts "#{sha512} #{rel_path}" }
+    end
+
     # All tag files that are bag manifest files (tagmanifest-[algorithm].txt)
     def tagmanifest_files
-      files = Dir[File.join(@bag_dir, '*')].select { |f|
-        File.file? f and File.basename(f) =~ /^tagmanifest-.*.txt/
+      files = Dir[File.join(@bag_dir, "*")].select { |f|
+        File.file?(f) && File.basename(f) =~ /^tagmanifest-.*.txt$/
       }
       files
     end
@@ -62,12 +91,11 @@ module BagIt
 
     # Generate manifest files for all the tag files (except the tag
     # manifest files)
-    def tagmanifest!(tags=nil)
-
-      tags = tag_files if tags == nil
+    def tagmanifest!(tags = nil)
+      tags = tag_files if tags.nil?
 
       # nuke all the existing tagmanifest files
-      tagmanifest_files.each { |f| FileUtils::rm f }
+      tagmanifest_files.each { |f| FileUtils.rm f }
 
       # ensure presence of manfiest files
       manifest_files.each do |manifest|
@@ -85,38 +113,37 @@ module BagIt
       tag_files
     end
 
-    def add_tag_file(path, src_path=nil)
-
+    def add_tag_file(path, src_path = nil)
       f = File.join(@bag_dir, path)
       raise "Tag file already in manifest: #{path}" if tag_files.include?(f)
 
-      if not File.exist? f
-        FileUtils::mkdir_p File.dirname(f)
+      if !File.exist? f
+        FileUtils.mkdir_p File.dirname(f)
 
         # write file
         if src_path.nil?
-          File.open(f, 'w') { |io| yield io }
+          File.open(f, "w") { |io| yield io }
         else
-          FileUtils::cp src_path, f
+          FileUtils.cp src_path, f
         end
         # this adds the manifest and bag info files on initial creation
         # it must only run when the manifest doesn't already exist or it will
         # infinitely recall add_tag_file. Better way of doing this?
         tagmanifest!
-      elsif not src_path.nil?
+      elsif !src_path.nil?
         raise "Tag file already exists, will not overwrite: #{path}\n Use add_tag_file(path) to add an existing tag file."
       end
 
-      data = File.open(f) { |io| io.read }
+      data = File.open(f, &:read)
       rel_path = Pathname.new(f).relative_path_from(Pathname.new(bag_dir)).to_s
 
       # sha1
       sha1 = Digest::SHA1.hexdigest data
-      File.open(tagmanifest_file(:sha1), 'a') { |io| io.puts "#{sha1} #{rel_path}" }
+      File.open(tagmanifest_file(:sha1), "a") { |io| io.puts "#{sha1} #{rel_path}" }
 
       # md5
       md5 = Digest::MD5.hexdigest data
-      File.open(tagmanifest_file(:md5), 'a') { |io| io.puts "#{md5} #{rel_path}" }
+      File.open(tagmanifest_file(:md5), "a") { |io| io.puts "#{md5} #{rel_path}" }
       tag_files
     end
 
@@ -131,7 +158,7 @@ module BagIt
       filepath = File.join(@bag_dir, path)
       raise "Tag file does not exist: #{path}" unless File.exist? filepath
       remove_tag_file(path) if tag_files.include?(path)
-      FileUtils::rm filepath
+      FileUtils.rm filepath
     end
 
     # Returns true if all present manifested files' message digests
@@ -141,33 +168,29 @@ module BagIt
         # extract the algorithm
         mf =~ /manifest-(.+).txt$/
 
-        algo = case $1
+        algo = case Regexp.last_match(1)
                when /sha1/i
                  Digest::SHA1
                when /md5/i
                  Digest::MD5
                else
                  :unknown
-               end
+        end
 
         # check it, an unknown algorithm is always true
-        unless algo == :unknown
-          lines = File.open(mf) { |io| io.readlines }
+        if algo == :unknown
+          true
+        else
+          lines = File.open(mf, &:readlines)
 
           lines.all? do |line|
-            manifested_digest, path = line.chomp.split /\s+/, 2
+            manifested_digest, path = line.chomp.split(/\s+/, 2)
             actual_digest = File.open(File.join(@bag_dir, path)) { |io| algo.hexdigest io.read }
             actual_digest == manifested_digest
           end
 
-        else
-          true
         end
-
       end
-
     end
-
   end
-
 end