Parse small letters in the end of identifier number, transform draft …

…dates to the consistent format

lib/pubid/ieee.rb

@@ -8,4 +8,5 @@ module Ieee
 
 require_relative "ieee/errors"
 require_relative "ieee/parser"
+require_relative "ieee/transformer"
 require_relative "ieee/identifier"

lib/pubid/ieee/identifier.rb

@@ -9,8 +9,20 @@ class Identifier
                   :edition, :draft, :rev, :corr, :amd, :redline, :year, :month, :type, :alternative,
                   :draft_status
 
-    def initialize(**opts)
-      opts.each { |key, value| send("#{key}=", value.is_a?(Enumerable) && value || value.to_s) }
+    def initialize(organizations:, type_status:, number:, parameters:)
+      @number = number
+      [organizations, type_status, parameters].each do |data|
+        case data
+        when Hash
+          set_values(data)
+        when Array
+          set_values(Identifier.merge_parameters(data))
+        end
+      end
+    end
+
+    def set_values(hash)
+      hash.each { |key, value| send("#{key}=", value.is_a?(Enumerable) && value || value.to_s) }
     end
 
     def self.update_old_code(code)
@@ -37,7 +49,7 @@ def self.merge_parameters(params)
     end
 
     def self.parse(code)
-      new(**merge_parameters(Parser.new.parse(update_old_code(code))).to_h)
+      new(**merge_parameters(Transformer.new.apply(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}"

lib/pubid/ieee/parser.rb

@@ -13,7 +13,7 @@ class Parser < Parslet::Parser
     end
 
     rule(:number) do
-      (digits | match("[A-Z]")).repeat(1).as(:number)
+      ((digits | match("[A-Z]")).repeat(1) >> match("[a-z]").maybe).as(:number)
     end
 
     rule(:part) do
@@ -57,50 +57,55 @@ class Parser < Parslet::Parser
         ((str(", ") | str(" ")) >> match("[A-Za-z]").repeat(1).as(:month) >>
           (((str(" ") >> digits.as(:day)).maybe >>
           str(", ") >> match('\d').repeat(4, 4).as(:year)) |
-          ((str(" ") | str(", ")) >> match('\d').repeat(4, 4).as(:year)))
+          ((str(" ") | str(", ")) >> match('\d').repeat(2, 4).as(:year)))
         ).maybe
     end
 
+    rule(:part_subpart_year) do
+      # 802.15.22.3-2020
+      # 1073.1.1.1-2004
+      (part >> subpart.repeat(2, 2).as(:subpart) >> year) |
+        # C57.12.00-1993
+        (part >> subpart.as(:subpart) >> year) |
+        # N42.44-2008
+        # 1244-5.2000
+        # 11073-40102-2020
+        # C37.0781-1972
+        (part >> year) |
+        # C57.19.101
+        (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
+
+    end
+
+    rule(:dual_pubids) do
+      str(" ") >>
+        ((str("(") >> (identifier.as(:alternative) >> str(", ").maybe).repeat(1) >>
+          str(")")) | (str("and ") >> identifier.as(:alternative)) |
+          identifier.as(:alternative))
+    end
+
+    rule(:number_prefix) do
+      ((str("No") | str("no")) >> (str(".") | str(" "))).maybe >> str(" ").maybe
+    end
+
     rule(:identifier) do
-      organization.as(:publisher) >> ((str("/ ") | str("/")) >> organization.as(:copublisher)).repeat >>
-        draft_status.maybe >>
-        str(" ") >> (type.as(:type) >> str(" ")).maybe >> (
-        (str("No") | str("no")) >> (str(".") | str(" "))
-      ).maybe >> str(" ").maybe >>
-      number >>
-        # part/subpart/year patterns:
-        (
-          # 802.15.22.3-2020
-          # 1073.1.1.1-2004
-          (part >> subpart.repeat(2, 2).as(:subpart) >> year) |
-          # C57.12.00-1993
-          (part >> subpart.as(:subpart) >> year) |
-          # N42.44-2008
-          # 1244-5.2000
-          # 11073-40102-2020
-          # C37.0781-1972
-          (part >> year) |
-          # C57.19.101
-          (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 >>
+      (organization.as(:publisher) >> ((str("/ ") | str("/")) >> organization.as(:copublisher)).repeat)
+        .as(:organizations) >>
+        (draft_status.maybe >> (str(" ") >> type.as(:type) >> str(" ")).maybe).as(:type_status) >>
+        str(" ").maybe >> number_prefix >> number >> (part_subpart_year.maybe >> draft.as(:draft).maybe >>
         edition.as(:edition).maybe >>
         # dual-PubIDs
-        (str(" ") >>
-          ((str("(") >> (identifier.as(:alternative) >> str(", ").maybe).repeat(1) >>
-            str(")")) | (str("and ") >> identifier.as(:alternative)) |
-            identifier.as(:alternative))
-        ).maybe
+        dual_pubids.maybe).as(:parameters)
     end
 
     rule(:root) { identifier }

lib/pubid/ieee/transformer.rb

@@ -0,0 +1,19 @@
+module Pubid::Ieee
+  class Transformer < Parslet::Transform
+    rule(draft: subtree(:draft)) do
+      result = draft
+      if draft[:month]
+
+        if draft[:year].length == 2
+          result[:year] = case draft[:year].to_i
+            when 0..25 then "20#{draft[:year]}"
+            when 26..99 then "19#{draft[:year]}"
+          end
+        end
+
+        result[:month] = Date.parse(draft[:month]).strftime("%B")
+      end
+      { draft: result }
+    end
+  end
+end

spec/fixtures/pubid-parsed.txt

@@ -5570,3 +5570,101 @@ IEEE Unapproved Draft Std PC57.12.90_D8, Sep 2008
 IEEE/IEC P62582-3_D1, April 2011
 IEEE/IEC P62659_D2
 IEEE/IEC P82079_D2, March 2017
+IEEE Approved Draft Std P1115a/D4, Feb 2007
+IEEE Approved Draft Std P1484.20.1/D7, May 07
+IEEE Approved Draft Std P1516.4/Dx.1, Jun 07
+IEEE Approved Draft Std P1542/D8, Aug 07
+IEEE Approved Draft Std P1613a/D3.2 Nov 2007
+IEEE Approved Draft Std P1617/D7.0, Jun 07
+IEEE Approved Draft Std P1636.1/D3, May 07
+IEEE Approved Draft Std P1661/D11, Jul 07
+IEEE Approved Draft Std P495/D13, Aug 07
+IEEE Approved Draft Std P532/D9.6, Apr 07
+IEEE Approved Draft Std P741/D2, Apr 07
+IEEE Approved Draft Std P99/D2.0, May 07
+IEEE Approved Draft Std PC37.20.7/D14, Dec 06
+IEEE Approved Draft Std PC37.230/D6.1, Jul 07
+IEEE Approved Draft Std PC37.235/D4, Mar 07
+IEEE Approved Draft Std PC57.12.36/D11, May 07
+IEEE Approved Draft Std PC57.129/D10, Jul 07
+IEEE P1048a/D002, February 2020
+IEEE P1048a/D003, December 2020
+IEEE P1538a/D3, February 2015
+IEEE P1538a/D4, February 2015
+IEEE P1538a/D5, July 2015
+IEEE P1547a.1/D4, August 2014
+IEEE P1547a/D1.3, December 2019
+IEEE P1547a/D1.4, January 2020
+IEEE P1547a/D2, June 2013
+IEEE P1547a/D3, December 2013
+IEEE P1588b/D1.0, August 2021
+IEEE P1588b/D1.1, December 2021
+IEEE P1657a/D4, January 2015
+IEEE P1657a/D5, January 2015
+IEEE P1708a/D2, February 2019
+IEEE P1708a/D3, April 2019
+IEEE P1801a/D01, May 2014
+IEEE P1801a/D03, May 2014
+IEEE P1857a/D1, December 2013
+IEEE P1857a/D2, January 2014
+IEEE P1901a/D3, December 2018
+IEEE P1901b/D1.1, August 2021
+IEEE P269a/D5.0, November 2011
+IEEE P269a/D5.2, January 2012
+IEEE P802.15.4q/D5.0, May, 20
+IEEE P802c/D2.2, April 2017
+IEEE P802d/D1.0 August, 2016
+IEEE P802d/D1.1 September 17, 2016
+IEEE P802d/D1.2, November 2016
+IEEE P835a/D2, October 2012
+IEEE Std 1012a-1998
+IEEE Std 1394a-2000
+IEEE Std 1613a-2008
+IEEE Std 1616a
+IEEE Std 1666 IEC61691-7 Edition 1.0 2009-12
+IEEE Std 4a-2001
+IEEE Std 524a-1993
+IEEE Std 802a-2003
+IEEE Std P1076c/D3.1
+IEEE Std P1115a/D3
+IEEE Std P1394c/D1.1
+IEEE Std P1394c/D1.2
+IEEE Std P1500/D11, Jan 06
+IEEE Std P1528a/D2.0
+IEEE Std P1528a/D2.1
+IEEE Std P269a_D5
+IEEE Std P269a_D8
+IEEE Std P802a/D4
+IEEE Std P802b/D3
+IEEE Unapproved Draft Std P1003.1_D3, Jun 07
+IEEE Unapproved Draft Std P1028/D5.1, Sep, 07
+IEEE Unapproved Draft Std P1189/D15, Jun 07
+IEEE Unapproved Draft Std P1407_D7, Mar 07
+IEEE Unapproved Draft Std P1561/D16, Jul 07
+IEEE Unapproved Draft Std P1610_D13, Aug 07
+IEEE Unapproved Draft Std P1613a/D3.2, Nov 2007
+IEEE Unapproved Draft Std P1616a/D3, Oct 2009
+IEEE Unapproved Draft Std P1616a/D4, Jan 2010
+IEEE Unapproved Draft Std P1619/D17, Jul 07
+IEEE Unapproved Draft Std P1631/D2, Sep 07
+IEEE Unapproved Draft Std P1636.1/D3, Jan 09
+IEEE Unapproved Draft Std P1686_D4, Jun 07
+IEEE Unapproved Draft Std P1900.2/D2.22, Jun 07
+IEEE Unapproved Draft Std P1900_D01, Jun 07
+IEEE Unapproved Draft Std P2600.1/D41b, Jan 09
+IEEE Unapproved Draft Std P2600.3/D41b, Jan 09
+IEEE Unapproved Draft Std P2600.4/D41b, Jan 09
+IEEE Unapproved Draft Std P532/D9.6, May 07
+IEEE Unapproved Draft Std P605a/D2, Sept 2009
+IEEE Unapproved Draft Std P605a/D3, Nov 2009
+IEEE Unapproved Draft Std P841/D5, Jan 09
+IEEE Unapproved Draft Std P841_D3, Jul 07
+IEEE Unapproved Draft Std PC37.1/D3, Jun 07
+IEEE Unapproved Draft Std PC37.1/D5, Sep 07
+IEEE Unapproved Draft Std PC37.12.1_D4.0, Sep 07
+IEEE Unapproved Draft Std PC37.16/D13, Apr 08
+IEEE Unapproved Draft Std PC37.43/D9, Oct 06
+IEEE Unapproved Draft Std PC57.12.38/D6.3, Aug 08
+IEEE Unapproved Draft Std PC57.13/D09, Aug 07
+IEEEPC37.74/D12, May 2014
+ISO/IEC/IEEE P42010_D8, June 2010