Parse draft documents

lib/pubid/ieee/identifier.rb

@@ -1,4 +1,7 @@
 require 'date'
+require "yaml"
+
+UPDATE_CODES = YAML.load_file(File.join(File.dirname(__FILE__), "../../../update_codes.yaml"))
 
 module Pubid::Ieee
   class Identifier
@@ -9,6 +12,13 @@ def initialize(**opts)
       opts.each { |key, value| send("#{key}=", value.is_a?(Enumerable) && value || value.to_s) }
     end
 
+    def self.update_old_code(code)
+      UPDATE_CODES.each do |from, to|
+        code = code.gsub(from.match?(/^\/.*\/$/) ? Regexp.new(from[1..-2]) : from, to)
+      end
+      code
+    end
+
     def self.merge_parameters(params)
       return params unless params.is_a?(Array)
 
@@ -26,14 +36,14 @@ def self.merge_parameters(params)
     end
 
     def self.parse(code)
-      new(**merge_parameters(Parser.new.parse(code)).to_h)
+      new(**merge_parameters(Parser.new.parse(update_old_code(code))).to_h)
 
     rescue Parslet::ParseFailed => failure
       raise Pubid::Ieee::Errors::ParseError, "#{failure.message}\ncause: #{failure.parse_failure_cause.ascii_tree}"
     end
 
     def to_s
-      "#{publisher}#{copublisher} #{type}#{number}#{part}#{subpart}#{year}#{edition}#{alternative}"
+      "#{publisher}#{copublisher} #{type}#{number}#{part}#{subpart}#{year}#{draft}#{edition}#{alternative}"
     end
 
     def copublisher
@@ -89,5 +99,14 @@ def edition
       result += "-#{@edition[:day]}" if @edition[:day]
       result
     end
+
+    def draft
+      return "" unless @draft
+
+      result = "/D#{@draft[:version]}"
+      result += ".#{@draft[:revision]}" if @draft[:revision]
+      result += ", #{@draft[:month]} #{@draft[:year]}" if @draft[:month]
+      result
+    end
   end
 end

lib/pubid/ieee/parser.rb

@@ -21,7 +21,7 @@ class Parser < Parslet::Parser
     end
 
     rule(:subpart) do
-      (str(".") | str("-")) >> match('\d').repeat(1)
+      (str(".") | str("-")) >> match('[\da-z]').repeat(1)
     end
 
     rule(:type) do
@@ -45,6 +45,15 @@ class Parser < Parslet::Parser
 
     end
 
+    rule(:draft) do
+      # /D14, April 2020
+      # /D7 November, 2019
+      str(" ").maybe >> str("/D") >> match('[\dA-Za-z]').repeat(1).as(:version) >>
+        ((str(".") | str("r")) >> digits.as(:revision)).maybe >>
+        ((str(", ") | str(" ")) >> match("[A-Za-z]").repeat(1).as(:month) >>
+          (str(" ") | str(", ")) >> match('\d').repeat(4, 4).as(:year)).maybe
+    end
+
     rule(:identifier) do
       organization.as(:publisher) >> ((str("/ ") | str("/")) >> organization.as(:copublisher)).repeat >>
         str(" ") >> (type.as(:type) >> str(" ")).maybe >> (
@@ -64,14 +73,19 @@ class Parser < Parslet::Parser
           # C37.0781-1972
           (part >> year) |
           # C57.19.101
-          (part >> subpart) |
+          (part >> subpart.as(:subpart)) |
+          # IEEE P11073-10101
+          # IEEE P11073-10420/D4D5
+          # trick to avoid being partially parsed by year
+          (str("-") >> match('[\dA-Z]').repeat(5).as(:part)) |
           # 581.1978
           year |
           # IEC 62525-Edition 1.0 - 2007
           edition.as(:edition) |
           # 61691-6
           part
         ).maybe >>
+        draft.as(:draft).maybe >>
         edition.as(:edition).maybe >>
         # dual-PubIDs
         (str(" ") >>