rfcxml 0.2.1 → 0.4.0

data/Rakefile

@@ -2,6 +2,10 @@
 
 require "bundler/gem_tasks"
 require "rspec/core/rake_task"
+require "open-uri"
+require "rubygems/package"
+require "zlib"
+require "fileutils"
 
 RSpec::Core::RakeTask.new(:spec)
 
@@ -10,3 +14,54 @@ require "rubocop/rake_task"
 RuboCop::RakeTask.new
 
 task default: %i[spec rubocop]
+
+RFC_XML_URL = "https://www.rfc-editor.org/in-notes/tar/xmlsource-all.tar.gz"
+RFC_XML_DIR = File.expand_path("spec/fixtures/xmlsource", __dir__)
+RFC_XML_TAR = File.expand_path("tmp/xmlsource-all.tar.gz", __dir__)
+
+namespace :rfc do
+  desc "Download and extract RFC XML files from rfc-editor.org"
+  task :download do
+    FileUtils.mkdir_p(RFC_XML_DIR)
+    FileUtils.mkdir_p(File.dirname(RFC_XML_TAR))
+
+    # Download tarball if not exists
+    unless File.exist?(RFC_XML_TAR)
+      puts "Downloading RFC XML tarball from #{RFC_XML_URL}..."
+      URI.open(RFC_XML_URL) do |remote|
+        File.binwrite(RFC_XML_TAR, remote.read)
+      end
+      puts "Downloaded to #{RFC_XML_TAR}"
+    end
+
+    # Extract tarball
+    puts "Extracting to #{RFC_XML_DIR}..."
+    tar_extract = Gem::Package::TarReader.new(Zlib::GzipReader.open(RFC_XML_TAR))
+    tar_extract.rewind
+    tar_extract.each do |entry|
+      next unless entry.file? && entry.full_name.end_with?(".xml")
+
+      # Extract only the filename, flatten directory structure
+      filename = File.basename(entry.full_name)
+      target_path = File.join(RFC_XML_DIR, filename)
+
+      File.binwrite(target_path, entry.read)
+    end
+    tar_extract.close
+
+    xml_count = Dir.glob(File.join(RFC_XML_DIR, "*.xml")).count
+    puts "Done! Extracted #{xml_count} XML files to #{RFC_XML_DIR}"
+  end
+
+  desc "Clean downloaded RFC XML files"
+  task :clean do
+    puts "Removing #{RFC_XML_DIR}..."
+    FileUtils.rm_rf(RFC_XML_DIR)
+    puts "Removing #{RFC_XML_TAR}..."
+    FileUtils.rm_f(RFC_XML_TAR)
+    puts "Done!"
+  end
+
+  desc "Re-download RFC XML files (clean + download)"
+  task redownload: %i[clean download]
+end

data/lib/rfcxml/v3/abstract.rb

@@ -2,11 +2,6 @@
 
 require "lutaml/model"
 
-require_relative "dl"
-require_relative "ol"
-require_relative "text"
-require_relative "ul"
-
 module Rfcxml
   module V3
     class Abstract < Lutaml::Model::Serializable
@@ -18,7 +13,8 @@ module Rfcxml
       attribute :ul, Ul, collection: true
 
       xml do
-        root "abstract"
+        element "abstract"
+        ordered
 
         map_attribute "anchor", to: :anchor
         map_attribute "pn", to: :pn

data/lib/rfcxml/v3/address.rb

@@ -2,12 +2,6 @@
 
 require "lutaml/model"
 
-require_relative "email"
-require_relative "facsimile"
-require_relative "phone"
-require_relative "postal"
-require_relative "uri"
-
 module Rfcxml
   module V3
     class Address < Lutaml::Model::Serializable
@@ -18,7 +12,8 @@ module Rfcxml
       attribute :uri, Uri
 
       xml do
-        root "address"
+        element "address"
+        ordered
 
         map_element "postal", to: :postal
         map_element "phone", to: :phone

data/lib/rfcxml/v3/annotation.rb

@@ -2,20 +2,6 @@
 
 require "lutaml/model"
 
-require_relative "bcp14"
-require_relative "cref"
-require_relative "em"
-require_relative "eref"
-require_relative "iref"
-require_relative "relref"
-require_relative "spanx"
-require_relative "strong"
-require_relative "sub"
-require_relative "sup"
-require_relative "tt"
-require_relative "u"
-require_relative "xref"
-
 module Rfcxml
   module V3
     class Annotation < Lutaml::Model::Serializable
@@ -35,7 +21,8 @@ module Rfcxml
       attribute :xref, Xref, collection: true
 
       xml do
-        root "annotation", mixed: true
+        element "annotation"
+        mixed_content
 
         map_content to: :content
         map_element "bcp14", to: :bcp14

data/lib/rfcxml/v3/area.rb

@@ -8,7 +8,7 @@ module Rfcxml
       attribute :content, :string
 
       xml do
-        root "area"
+        element "area"
 
         map_content to: :content
       end

data/lib/rfcxml/v3/artset.rb

@@ -2,8 +2,6 @@
 
 require "lutaml/model"
 
-require_relative "artwork"
-
 module Rfcxml
   module V3
     class Artset < Lutaml::Model::Serializable
@@ -12,7 +10,8 @@ module Rfcxml
       attribute :artwork, Artwork, collection: true
 
       xml do
-        root "artset"
+        element "artset"
+        ordered
 
         map_attribute "anchor", to: :anchor
         map_attribute "pn", to: :pn

data/lib/rfcxml/v3/artwork.rb

@@ -8,19 +8,21 @@ module Rfcxml
       attribute :content, :string
       attribute :anchor, :string
       attribute :pn, :string
-      attribute :name, :string, default: -> { "" }
-      attribute :type, :string, default: -> { "" }
+      attribute :name, :string
+      attribute :type, :string
       attribute :src, :string
-      attribute :align, :string, default: -> { "left" }
-      attribute :alt, :string, default: -> { "" }
-      attribute :width, :string, default: -> { "" }
-      attribute :height, :string, default: -> { "" }
+      attribute :align, :string, values: %w[left center right]
+      attribute :alt, :string
+      attribute :width, :string
+      attribute :height, :string
       attribute :original_src, :string
 
       xml do
-        root "artwork"
+        element "artwork"
 
-        map_content to: :content
+        # Use map_all to capture both text content AND child elements like <svg>
+        # map_content only captures text nodes, not child elements
+        map_all to: :content
         map_attribute "anchor", to: :anchor
         map_attribute "pn", to: :pn
         map_attribute "name", to: :name, value_map: { to: { empty: :empty } }
@@ -28,8 +30,9 @@ module Rfcxml
         map_attribute "src", to: :src
         map_attribute "align", to: :align
         map_attribute "alt", to: :alt, value_map: { to: { empty: :empty } }
-        map_attribute "width", to: :width
-        map_attribute "height", to: :height
+        map_attribute "width", to: :width, value_map: { to: { empty: :empty } }
+        map_attribute "height", to: :height,
+                                value_map: { to: { empty: :empty } }
         map_attribute "originalSrc", to: :original_src
       end
     end

data/lib/rfcxml/v3/aside.rb

@@ -2,17 +2,6 @@
 
 require "lutaml/model"
 
-require_relative "artset"
-require_relative "artwork"
-require_relative "blockquote"
-require_relative "dl"
-require_relative "figure"
-require_relative "iref"
-require_relative "ol"
-require_relative "text"
-require_relative "table"
-require_relative "ul"
-
 module Rfcxml
   module V3
     class Aside < Lutaml::Model::Serializable
@@ -30,7 +19,8 @@ module Rfcxml
       attribute :ul, Ul, collection: true
 
       xml do
-        root "aside"
+        element "aside"
+        ordered
 
         map_attribute "anchor", to: :anchor
         map_attribute "pn", to: :pn

data/lib/rfcxml/v3/author.rb

@@ -2,9 +2,6 @@
 
 require "lutaml/model"
 
-require_relative "address"
-require_relative "organization"
-
 module Rfcxml
   module V3
     class Author < Lutaml::Model::Serializable
@@ -14,22 +11,29 @@ module Rfcxml
       attribute :surname, :string
       attribute :ascii_surname, :string
       attribute :fullname, :string
-      attribute :role, :string
+      attribute :role, :string, values: %w[editor]
       attribute :ascii_fullname, :string
       attribute :organization, Organization
       attribute :address, Address
 
       xml do
-        root "author"
+        element "author"
+        ordered
 
         map_attribute "anchor", to: :anchor
-        map_attribute "initials", to: :initials
-        map_attribute "asciiInitials", to: :ascii_initials
-        map_attribute "surname", to: :surname
-        map_attribute "asciiSurname", to: :ascii_surname
-        map_attribute "fullname", to: :fullname
-        map_attribute "role", to: :role
-        map_attribute "asciiFullname", to: :ascii_fullname
+        map_attribute "initials", to: :initials,
+                                  value_map: { to: { empty: :empty } }
+        map_attribute "asciiInitials", to: :ascii_initials,
+                                       value_map: { to: { empty: :empty } }
+        map_attribute "surname", to: :surname,
+                                 value_map: { to: { empty: :empty } }
+        map_attribute "asciiSurname", to: :ascii_surname,
+                                      value_map: { to: { empty: :empty } }
+        map_attribute "fullname", to: :fullname,
+                                  value_map: { to: { empty: :empty } }
+        map_attribute "role", to: :role, value_map: { to: { empty: :empty } }
+        map_attribute "asciiFullname", to: :ascii_fullname,
+                                       value_map: { to: { empty: :empty } }
         map_element "organization", to: :organization
         map_element "address", to: :address
       end

data/lib/rfcxml/v3/back.rb

@@ -2,10 +2,6 @@
 
 require "lutaml/model"
 
-require_relative "displayreference"
-require_relative "references"
-require_relative "section"
-
 module Rfcxml
   module V3
     class Back < Lutaml::Model::Serializable
@@ -14,7 +10,8 @@ module Rfcxml
       attribute :section, Section, collection: true
 
       xml do
-        root "back"
+        element "back"
+        ordered
 
         map_element "displayreference", to: :displayreference
         map_element "references", to: :references

data/lib/rfcxml/v3/bcp14.rb

@@ -8,7 +8,7 @@ module Rfcxml
       attribute :content, :string
 
       xml do
-        root "bcp14"
+        element "bcp14"
 
         map_content to: :content
       end

data/lib/rfcxml/v3/blockquote.rb

@@ -2,33 +2,8 @@
 
 require "lutaml/model"
 
-require_relative "artset"
-require_relative "artwork"
-require_relative "bcp14"
-require_relative "br"
-require_relative "cref"
-require_relative "dl"
-require_relative "em"
-require_relative "eref"
-require_relative "figure"
-require_relative "iref"
-# require_relative "ol"
-require_relative "relref"
-require_relative "sourcecode"
-require_relative "strong"
-require_relative "sub"
-require_relative "sup"
-require_relative "text"
-require_relative "tt"
-require_relative "u"
-require_relative "ul"
-require_relative "xref"
-
 module Rfcxml
   module V3
-    class Ol < Lutaml::Model::Serializable; end
-    class Dl < Lutaml::Model::Serializable; end
-
     class Blockquote < Lutaml::Model::Serializable
       attribute :content, :string
       attribute :anchor, :string
@@ -58,7 +33,8 @@ module Rfcxml
       attribute :xref, Xref, collection: true
 
       xml do
-        root "blockquote"
+        element "blockquote"
+        ordered
 
         map_content to: :content
         map_attribute "anchor", to: :anchor

data/lib/rfcxml/v3/boilerplate.rb

@@ -2,15 +2,14 @@
 
 require "lutaml/model"
 
-require_relative "section"
-
 module Rfcxml
   module V3
     class Boilerplate < Lutaml::Model::Serializable
       attribute :section, Section, collection: true
 
       xml do
-        root "boilerplate"
+        element "boilerplate"
+        ordered
 
         map_element "section", to: :section
       end

data/lib/rfcxml/v3/br.rb

@@ -6,7 +6,7 @@ module Rfcxml
   module V3
     class Br < Lutaml::Model::Serializable
       xml do
-        root "br"
+        element "br"
       end
     end
   end

data/lib/rfcxml/v3/c.rb

@@ -2,12 +2,6 @@
 
 require "lutaml/model"
 
-require_relative "cref"
-require_relative "eref"
-require_relative "iref"
-require_relative "spanx"
-require_relative "xref"
-
 module Rfcxml
   module V3
     class C < Lutaml::Model::Serializable
@@ -19,7 +13,8 @@ module Rfcxml
       attribute :xref, Xref, collection: true
 
       xml do
-        root "c"
+        element "c"
+        ordered
 
         map_content to: :content
         map_element "cref", to: :cref

data/lib/rfcxml/v3/city.rb

@@ -9,7 +9,7 @@ module Rfcxml
       attribute :ascii, :string
 
       xml do
-        root "city"
+        element "city"
 
         map_content to: :content
         map_attribute "ascii", to: :ascii

data/lib/rfcxml/v3/cityarea.rb

@@ -9,7 +9,7 @@ module Rfcxml
       attribute :ascii, :string
 
       xml do
-        root "cityarea"
+        element "cityarea"
 
         map_content to: :content
         map_attribute "ascii", to: :ascii

data/lib/rfcxml/v3/code.rb

@@ -9,7 +9,7 @@ module Rfcxml
       attribute :ascii, :string
 
       xml do
-        root "code"
+        element "code"
 
         map_content to: :content
         map_attribute "ascii", to: :ascii

data/lib/rfcxml/v3/contact.rb

@@ -2,9 +2,6 @@
 
 require "lutaml/model"
 
-require_relative "address"
-require_relative "organization"
-
 module Rfcxml
   module V3
     class Contact < Lutaml::Model::Serializable
@@ -19,7 +16,8 @@ module Rfcxml
       attribute :address, Address
 
       xml do
-        root "contact"
+        element "contact"
+        ordered
 
         map_attribute "anchor", to: :anchor
         map_attribute "initials", to: :initials

data/lib/rfcxml/v3/country.rb

@@ -9,7 +9,7 @@ module Rfcxml
       attribute :ascii, :string
 
       xml do
-        root "country"
+        element "country"
 
         map_content to: :content
         map_attribute "ascii", to: :ascii

data/lib/rfcxml/v3/cref.rb

@@ -2,16 +2,6 @@
 
 require "lutaml/model"
 
-require_relative "br"
-require_relative "em"
-require_relative "eref"
-require_relative "relref"
-require_relative "strong"
-require_relative "sub"
-require_relative "sup"
-require_relative "tt"
-require_relative "xref"
-
 module Rfcxml
   module V3
     class Cref < Lutaml::Model::Serializable
@@ -30,7 +20,8 @@ module Rfcxml
       attribute :xref, Xref, collection: true
 
       xml do
-        root "cref"
+        element "cref"
+        ordered
 
         map_content to: :content
         map_attribute "anchor", to: :anchor

data/lib/rfcxml/v3/date.rb

@@ -11,12 +11,12 @@ module Rfcxml
       attribute :year, :string
 
       xml do
-        root "date"
+        element "date"
 
         map_content to: :content
-        map_attribute "day", to: :day
-        map_attribute "month", to: :month
-        map_attribute "year", to: :year
+        map_attribute "day", to: :day, value_map: { to: { empty: :empty } }
+        map_attribute "month", to: :month, value_map: { to: { empty: :empty } }
+        map_attribute "year", to: :year, value_map: { to: { empty: :empty } }
       end
     end
   end

data/lib/rfcxml/v3/dd.rb

@@ -2,30 +2,6 @@
 
 require "lutaml/model"
 
-require_relative "artset"
-require_relative "artwork"
-require_relative "aside"
-require_relative "bcp14"
-require_relative "br"
-require_relative "cref"
-require_relative "dl"
-require_relative "em"
-require_relative "eref"
-require_relative "figure"
-require_relative "iref"
-require_relative "ol"
-require_relative "relref"
-require_relative "sourcecode"
-require_relative "strong"
-require_relative "sub"
-require_relative "sup"
-require_relative "text"
-require_relative "table"
-require_relative "tt"
-require_relative "u"
-require_relative "ul"
-require_relative "xref"
-
 module Rfcxml
   module V3
     class Dd < Lutaml::Model::Serializable
@@ -58,7 +34,8 @@ module Rfcxml
       attribute :aside, Aside, collection: true
 
       xml do
-        root "dd", mixed: true
+        element "dd"
+        mixed_content
 
         map_content to: :content
         map_attribute "anchor", to: :anchor

data/lib/rfcxml/v3/displayreference.rb

@@ -9,7 +9,7 @@ module Rfcxml
       attribute :to, :string
 
       xml do
-        root "displayreference"
+        element "displayreference"
 
         map_attribute "target", to: :target
         map_attribute "to", to: :to

data/lib/rfcxml/v3/dl.rb

@@ -2,14 +2,8 @@
 
 require "lutaml/model"
 
-require_relative "dd"
-require_relative "dt"
-
 module Rfcxml
   module V3
-    # class Dd < Lutaml::Model::Serializable; end
-    # class Dt < Lutaml::Model::Serializable; end
-
     class Dl < Lutaml::Model::Serializable
       attribute :anchor, :string
       attribute :spacing, :string, default: -> { "normal" }
@@ -20,7 +14,8 @@ module Rfcxml
       attribute :dd, Dd, collection: true
 
       xml do
-        root "dl", mixed: true
+        element "dl"
+        mixed_content
 
         map_attribute "anchor", to: :anchor
         map_attribute "spacing", to: :spacing

data/lib/rfcxml/v3/dt.rb

@@ -2,19 +2,6 @@
 
 require "lutaml/model"
 
-require_relative "bcp14"
-require_relative "br"
-require_relative "cref"
-require_relative "em"
-require_relative "eref"
-require_relative "iref"
-require_relative "relref"
-require_relative "strong"
-require_relative "sub"
-require_relative "sup"
-require_relative "tt"
-require_relative "xref"
-
 module Rfcxml
   module V3
     class Dt < Lutaml::Model::Serializable
@@ -35,7 +22,8 @@ module Rfcxml
       attribute :xref, Xref, collection: true
 
       xml do
-        root "dt"
+        element "dt"
+        ordered
 
         map_content to: :content
         map_attribute "anchor", to: :anchor

data/lib/rfcxml/v3/email.rb

@@ -9,7 +9,7 @@ module Rfcxml
       attribute :ascii, :string
 
       xml do
-        root "email"
+        element "email"
 
         map_content to: :content
         map_attribute "ascii", to: :ascii

data/lib/rfcxml/v3/eref.rb

@@ -10,7 +10,7 @@ module Rfcxml
       attribute :target, :string
 
       xml do
-        root "eref"
+        element "eref"
 
         map_content to: :content
         map_attribute "brackets", to: :brackets

data/lib/rfcxml/v3/extaddr.rb

@@ -9,7 +9,7 @@ module Rfcxml
       attribute :ascii, :string
 
       xml do
-        root "extaddr"
+        element "extaddr"
 
         map_content to: :content
         map_attribute "ascii", to: :ascii

data/lib/rfcxml/v3/facsimile.rb

@@ -8,7 +8,7 @@ module Rfcxml
       attribute :content, :string
 
       xml do
-        root "facsimile"
+        element "facsimile"
 
         map_content to: :content
       end