pubid 2.0.0.pre.alpha.13 → 2.0.0.pre.alpha.15

data/lib/pubid/ieee/parser.rb

@@ -595,7 +595,10 @@ module Pubid
         # ISO/IEC/IEEE P26511/D8-2018 or ISO/IEEE P1003.1-2008 or IEC/IEEE P62582-1-2011
         # ALSO handle: IEC/IEEE P60780-323, CDV1 2014 (comma before stage code)
         # ALSO handle: IEEE/CSA P844.1/293.1/D2 (CSA dual numbering)
-        (str("ISO/IEC/IEEE") | str("ISO/IEEE") | str("IEC/IEEE") | str("IEEE/CSA")).as(:joint_publishers) >>
+        # ALSO handle: IEEE/ISO/IEC P42010/D=CD-2020 — the printed publisher
+        # order is the organization's perspective (pubid#469) and parses as
+        # printed; it is no longer rewritten to ISO-first.
+        (str("ISO/IEC/IEEE") | str("IEEE/ISO/IEC") | str("ISO/IEEE") | str("IEC/IEEE") | str("IEEE/CSA")).as(:joint_publishers) >>
           space >>
           # P = project (the document is a draft): identity-bearing, so it
           # is captured and preserved, never silently consumed.
@@ -605,14 +608,19 @@ module Pubid
           # CSA dual numbering: /293.1 (second number)
           (slash >> digits >> (dot >> digits).maybe >> (dash >> digits.as(:draft_version)).maybe).maybe >>
           (
-            # Variant 1b: the ordinal-less stage draft "D=CDV[:2020]" -
+            # Variant 1b: the ordinal-less stage draft "D=CDV-2020" -
             # D (draft) = CDV (the IEC stage it drafts). The year rides in
             # the draft clause (a distinct key, so the builder keeps the
-            # date inside the designator).
+            # date inside the designator). The year follows IEEE format:
+            # dash-joined ("D=CD-2020", the ruled canonical); the colon
+            # spelling ("D=CDV:2020") stays accepted as the alias it was
+            # frozen with. A dash-year is guarded to 19xx/20xx so a
+            # YYMM monthcode tail can never be read as a year.
             (slash >> str("D") >> str("=") >>
               (str("CDV") | str("FDIS") | str("PWI") | str("WD") |
                str("NP") | str("DIS") | str("CD")).as(:draft_iso_stage) >>
-              (str(":") >> year_digits.as(:draft_stage_year)).maybe) |
+              ((str(":") | str("-")) >>
+               ((str("19") | str("20")) >> digit >> digit).as(:draft_stage_year)).maybe) |
             # Variant 1: /D8 notation (original), with the compound
             # both-systems suffix "=DDIS.3" (docs/IEEE-DRAFT-STAGES.md §1.3)
             (slash >> str("D") >> digits.as(:draft_version) >>
@@ -1468,7 +1476,10 @@ module Pubid
         cleaned
       end
 
-      def self.parse(string)
+      # Pre-parse ingestion normalizations (R2): every parse path —
+      # parslet and PG artifact alike — feeds the grammar the same
+      # normalized string.
+      def self.normalize_input(string)
         # Strip .pdf extension if present (Pattern 3: File Extensions)
         cleaned = string.sub(/\.pdf$/i, "")
 
@@ -1822,11 +1833,10 @@ module Pubid
         # Fix 2H: "IEC XXXX First edition YYYY-MM; IEEE NNNN" -> normalize semicolon
         # Already handled by earlier semicolon normalization
 
-        # Fix 2I: "IEEE/ISO/IEC PXXX/DIS" -> normalize to "ISO/IEC/IEEE PXXX/DIS"
-        cleaned = cleaned.gsub(/^IEEE\/ISO\/IEC\s+(P[\w.-]+)/,
-                               'ISO/IEC/IEEE \1')
-        cleaned = cleaned.gsub(/^IEEE\/IEC\/ISO\s+(P[\w.-]+)/,
-                               'IEC/ISO/IEEE \1')
+        # (Fix 2I removed: "IEEE/ISO/IEC PXXX/…" is no longer rewritten to
+        # ISO-first — the printed publisher order is the organization's
+        # perspective (pubid#469) and the joint P-form rule parses it as
+        # printed.)
 
         # Fix 2J: "IEEE/IEC PXXX D5" -> normalize space to slash before D
         cleaned = cleaned.gsub(/^(IEEE\/IEC P[\w.-]+)\s+D(\d)/, '\1/D\2')
@@ -1925,7 +1935,11 @@ module Pubid
         # Fix 2AF: "IEEE Std 1003.1/2003.l/lNT" -> fix typos
         # .l -> .1 and lNT -> INT handled by existing fixes
 
-        new.parse(cleaned)
+        cleaned
+      end
+
+      def self.parse(string)
+        new.parse(normalize_input(string))
       end
     end
   end

data/lib/pubid/ieee/renderer.rb

@@ -169,9 +169,11 @@ module Pubid
           # ("D08, September, 2018").
           if id.draft_obj
             printed = id.draft_obj.to_s
-            if id.publisher == "IEC" && id.copublisher == ["IEEE"]
-              printed = printed.split(", ").first
-            elsif id.publisher == "IEEE" && id.draft_status.to_s.empty?
+            # The draft designator carries its date on every lead
+            # (docs/IEEE-DRAFT-STAGES.md §3: "the draft designator with its
+            # date" — the IEC/IEEE-led date-drop contradicted the DCD
+            # convention and the raw records).
+            if id.publisher == "IEEE" && id.draft_status.to_s.empty?
               # The long comma form is for dated project drafts; an
               # unapproved-draft render keeps its pinned single-comma form
               # (pubid#318 idempotence).

data/lib/pubid/ietf/identifiers/base.rb

@@ -64,7 +64,7 @@ module Pubid
           raise Pubid::Errors::InvalidInputError, Pubid::INPUT_TOO_LONG_MESSAGE
         end
 
-        parsed = Parser.parse(identifier)
+        parsed = Pubid::Parg::Backend.parse(:ietf, identifier)
         Builder.build(parsed)
       end
     end

data/lib/pubid/isbn/identifier.rb

@@ -64,7 +64,7 @@ module Pubid
 
       # @raise [Pubid::Errors::ParseError] if the string is not a valid ISBN
       def self.build_identifier(identifier)
-        parsed = Parser.parse(identifier)
+        parsed = Pubid::Parg::Backend.parse(:isbn, identifier)
         Builder.build(parsed)
       rescue ArgumentError => e
         # The Builder validates length and check digit. Surface that as a parse

data/lib/pubid/iso/identifier.rb

@@ -304,7 +304,10 @@ module Pubid
         when :mr_string
           Pubid::Parsers::MrString.parse(string)
         else
-          parsed = Pubid::Iso::Parser.new.parse(string)
+          # R1 parser swap: the baked PG artifact is the identifier
+          # parser of record; the Builder consumes the same attribute
+          # hash it always has.
+          parsed = Pubid::Parg::Backend.parse(:iso, string)
           Pubid::Iso::Builder.new.build(parsed)
         end
       end

data/lib/pubid/iso/normalizer.rb

@@ -27,7 +27,10 @@ module Pubid
         private
 
         def parse_with_builder(string)
-          parsed = Pubid::Iso.parser.parse(string)
+          # R2 ingestion hook: normalized strings reach the model through
+          # the same parser of record as Identifier.parse (the baked PG
+          # artifact), so every Tier-3 normalization feeds one grammar.
+          parsed = Pubid::Parg::Backend.parse(:iso, string)
           Pubid::Iso.builder.build(parsed)
         end
 

data/lib/pubid/itu/CLAUDE.md

@@ -67,3 +67,49 @@ on any call. The base now derives the key from the class name
 `underscore` is not a dependency. metanorma-itu constructs through this
 lookup; its flavor-local `pubid_contribution.rb` render override can be
 deleted once it migrates.
+
+## relaton's query forms — RR, OB sector, publication ids
+
+Four forms relaton's `Relaton::Itu::Pubid` parsed and `Pubid::Itu` did not
+(hand-off itu-relaton-query-forms). None of them occurs in the published
+`relaton-data-itu` index — no `series: RR`, no OB row, no six-digit part — and
+a replay of all 24,382 rows and every ITU pass fixture showed 0 changes.
+
+- **Radio Regulations** — `Identifiers::RadioRegulations`
+  (`pubid:itu:radio-regulations`): `ITU-R RR`, `ITU-R RR (2020)`, and the URL
+  spelling `ITU-R RR-2020`, which used to build a *wrong* Recommendation
+  (series `RR`, number `2020`) with no error. "RR" is the series; there is no
+  code, so `#number` returns the series and `root.number` is `"RR"`. The rule
+  ends in `any.absent?`: it sits before `with_series`, and PEG never re-enters
+  the alternation, so a partial match on `ITU-R RR.1` must fail inside it.
+- **Operational Bulletins keep their sector.** This reverses the old
+  "cross-bureau, sector must not be set" rule: `validate_ob_no_sector!` is
+  gone, `ITU-T OB.1096 (2016)` renders back as it is, and the sector-less
+  `ITU OB No. 1096` is unchanged. The long forms with a sector (`ITU-T OB No.
+  1096`) now normalise to `ITU-T OB.1096`. `No.` stays in the default
+  render — it is how ITU's bulletin site and pubid v1 write it — and
+  metanorma-itu's `ITU OB 1000` / `Annex to ITU OB 1000` (its i18n template
+  omits `No.`) is accepted on parse (`ob_bare_body`) and rendered with it.
+  The sector is a spelling, not
+  identity: `SpecialPublication#==` skips it, the URN keeps `urn:itu:itu:…`
+  and `mr_type` stays nil, so both spellings are one bulletin on every
+  surface except `to_s`/`to_hash`. The printed date `- 15.III.2016` sets
+  `date.day`, which only this form does, so `render_ob_date` uses the day as
+  the spelling marker; `day_to_kv` emits only then.
+- **`-YYYYMM` is a date, never a part.** `part` refuses a six-digit run with a
+  19xx/20xx year and a 01–12 month (`yyyymm_shape`), and `id_date` reads it as
+  year+month. `-200313`, `-180001` stay parts. `ITU-T REC T.4` drops the
+  uncaptured `rec_word`; `T-REC-T.4-200307-I` is `publication_id`, last in
+  `identifier` (nothing else starts with a bare sector letter), and requires
+  the date. The trailing status letter (`I` in force, `S` superseded) is
+  parsed and dropped: it names the state of an edition, not the edition.
+  **`S` is also the Spanish language suffix**, so it is a status only inside
+  the full `T-REC-…` id (`id_status`); after an `ITU-T …-YYYYMM` print form
+  only `-I` is (`print_id_status`), and `ITU-T Z.100-199911-S` keeps language
+  `S`. The `-YYYYMM` date and `REC` word are also in `base_with_series`/
+  `base_without_series`: the part guard applies there too, so without them
+  `ITU-T G.989-200307 Amd 1` — a (wrong) Amendment on `main` — stopped parsing.
+  The day of an OB date reaches the URN (`…:15/03/2016`), since it is in `==`.
+  **Not done:** an RR supplement (`ITU-R RR (2020) Amd 1` fails; the
+  base-less `ITU-R RR Amd 1` still builds an Amendment on series `RR`), and
+  `ITU-R RR-E` is still a Recommendation numbered `E` — both as on `main`.

data/lib/pubid/itu/builder.rb

@@ -50,6 +50,16 @@ module Pubid
           return sp
         end
 
+        # Radio Regulations — "ITU-R RR (2020)"
+        if data[:radio_regulations]
+          return Identifiers::RadioRegulations.new(
+            sector: Components::Sector.new(sector: data[:sector].to_s),
+            series: Components::Series.new(series: "RR"),
+            date: data[:year] ? build_date(data) : nil,
+            language: data[:language]&.to_s,
+          )
+        end
+
         # Check if this is a supplement identifier
         if data[:supplement_type]
           supp = build_supplement(data)
@@ -153,11 +163,12 @@ module Pubid
         nil
       end
 
-      # Build Special Publication (OB). Sector is silently dropped — OB is a
-      # cross-bureau publication and `Identifier` rejects sector+OB
-      # in its constructor.
+      # Build Special Publication (OB). The sector of the TSB spelling
+      # ("ITU-T OB.1096") is kept so the bulletin renders back as it was cited;
+      # SpecialPublication#== ignores it, since OB is cross-bureau.
       def build_special_publication(data)
         Identifiers::SpecialPublication.new(
+          sector: (Components::Sector.new(sector: data[:sector].to_s) if data[:sector]),
           series: Components::Series.new(series: "OB"),
           code: data[:number] ? build_code(data) : nil,
           date: data[:year] ? build_date(data) : nil,
@@ -344,10 +355,19 @@ module Pubid
         )
       end
 
+      # The Roman month of a bulletin date ("15.III.2016") is stored as the
+      # two-digit month every other ITU date uses; the day marks the spelling.
       def build_date(data)
+        month = if data[:roman_month]
+                  (Identifiers::SpecialPublication::ROMAN_MONTHS.index(data[:roman_month].to_s) + 1)
+                    .to_s.rjust(2, "0")
+                else
+                  data[:month]&.to_s
+                end
         Pubid::Components::Date.new(
           year: data[:year].to_s,
-          month: data[:month]&.to_s,
+          month: month,
+          day: data[:day]&.to_s,
         )
       end
 

data/lib/pubid/itu/identifiers/base.rb

@@ -15,7 +15,7 @@ module Pubid
           raise Pubid::Errors::InvalidInputError, Pubid::INPUT_TOO_LONG_MESSAGE
         end
 
-        parsed = Parser.parse(normalize_whitespace(identifier))
+        parsed = Pubid::Parg::Backend.parse(:itu, normalize_whitespace(identifier))
         Builder.build(parsed)
       end
 
@@ -101,8 +101,6 @@ module Pubid
         end
 
         super
-
-        validate_ob_no_sector!
       end
 
       # The document number lives on the `code` component for ITU; surface it at
@@ -462,6 +460,16 @@ module Pubid
         model.code ||= Components::Code.new
       end
 
+      # The day is set only by the printed bulletin date ("15.III.2016"), so
+      # it emits only there and no existing index row gains a key.
+      def day_to_kv(model, doc)
+        emit_kv(doc, "day", model.date&.day)
+      end
+
+      def day_from_kv(model, value)
+        date_for(model).day = value.to_s
+      end
+
       def date_for(model)
         model.date ||= Pubid::Components::Date.new
       end
@@ -472,21 +480,6 @@ module Pubid
         str = value.to_s
         LANGUAGES[str] || str
       end
-
-      # OB (Operational Bulletin) is a cross-bureau ITU publication and
-      # must not have a sector. Direct construction with both raises;
-      # the parser silently drops sector for legacy strings like
-      # "ITU-T OB.X" (handled in Builder).
-      def validate_ob_no_sector!
-        return unless series&.series == "OB"
-        return if sector.nil?
-        return if sector.is_a?(Components::Sector) && (sector.sector.nil? || sector.sector.to_s.empty?)
-
-        raise ArgumentError,
-              "OB (Operational Bulletin) is a cross-bureau ITU publication; " \
-              "sector must not be set"
-      end
     end
-
   end
 end

data/lib/pubid/itu/identifiers/radio_regulations.rb

@@ -0,0 +1,27 @@
+# frozen_string_literal: true
+
+module Pubid
+  module Itu
+    module Identifiers
+      # The ITU Radio Regulations — the treaty text revised by each World
+      # Radiocommunication Conference.
+      # Format: ITU-R RR [(YYYY)]
+      # Example: ITU-R RR (2020)
+      #
+      # It has no document number: "RR" is the whole designation, stored as
+      # the series. `#number` returns it, so `root.number` (the relaton-index
+      # key) is not empty.
+      class RadioRegulations < Identifier
+        include StandardSerialization
+
+        def number
+          series&.series
+        end
+
+        def render_base(**_opts)
+          "#{publisher}-#{sector} #{series}#{render_date_suffix}"
+        end
+      end
+    end
+  end
+end

data/lib/pubid/itu/identifiers/special_publication.rb

@@ -4,26 +4,60 @@ module Pubid
   module Itu
     module Identifiers
       # ITU Special Publication — currently models the Operational Bulletin (OB).
-      # OB is a cross-bureau publication (no sector) rendered as
-      # "ITU OB No. {number}" with optional date.
-      #
-      # Pattern: "ITU OB No. 1283 (01/2024)"
+      # OB is a cross-bureau publication. ITU prints it without a sector
+      # ("ITU OB No. 1283 (01/2024)"); the TSB spelling carries one
+      # ("ITU-T OB.1096 (2016)"), which is kept and rendered back. The sector
+      # is not part of the identity: `==`, the URN and the MR slug ignore it,
+      # so both spellings name one bulletin.
       class SpecialPublication < Identifier
         include StandardSerialization
 
+        ROMAN_MONTHS = %w[I II III IV V VI VII VIII IX X XI XII].freeze
+
         def render_base(**_opts)
           number = code&.number
-          result = "#{publisher} #{series} No. #{number}"
-
-          if date
-            result += if date.month
-                        " (#{date.month.to_s.rjust(2, '0')}/#{date.year})"
-                      else
-                        " (#{date.year})"
-                      end
-          end
+          result = if sector
+                     "#{publisher}-#{sector} #{series}.#{number}"
+                   else
+                     "#{publisher} #{series} No. #{number}"
+                   end
+
+          result + render_ob_date
+        end
+
+        # Cross-bureau: the sector is how one bureau cites the bulletin, not
+        # which bulletin it is.
+        def ==(other)
+          return false unless other.instance_of?(self.class)
+
+          series == other.series &&
+            code == other.code &&
+            date == other.date &&
+            language == other.language &&
+            common_text_twin == other.common_text_twin
+        end
+
+        # Keep the MR slug sector-free, like `==` (it was always so, because
+        # the sector used to be dropped).
+        def mr_type
+          nil
+        end
 
-          result
+        private
+
+        # " (MM/YYYY)" or " (YYYY)"; the day-bearing date is the printed
+        # bulletin form " - 15.III.2016", the only spelling that sets a day.
+        def render_ob_date
+          return "" unless date
+
+          if date.day && date.month
+            roman = ROMAN_MONTHS[date.month.to_i - 1]
+            " - #{date.day.to_s.rjust(2, '0')}.#{roman}.#{date.year}"
+          elsif date.month
+            " (#{date.month.to_s.rjust(2, '0')}/#{date.year})"
+          else
+            " (#{date.year})"
+          end
         end
       end
     end

data/lib/pubid/itu/identifiers/standard_serialization.rb

@@ -46,6 +46,8 @@ module Pubid
                 with: { to: :year_to_kv, from: :year_from_kv }
             map "month",
                 with: { to: :month_to_kv, from: :month_from_kv }
+            map "day",
+                with: { to: :day_to_kv, from: :day_from_kv }
             map "language", to: :language
             map "common_text_twin",
                 with: { to: :common_text_twin_to_kv,

data/lib/pubid/itu/identifiers.rb

@@ -16,6 +16,7 @@ module Pubid
       autoload :Errata, "#{__dir__}/identifiers/errata"
       autoload :Handbook, "#{__dir__}/identifiers/handbook"
       autoload :Question, "#{__dir__}/identifiers/question"
+      autoload :RadioRegulations, "#{__dir__}/identifiers/radio_regulations"
       autoload :Recommendation, "#{__dir__}/identifiers/recommendation"
       autoload :Report, "#{__dir__}/identifiers/report"
       autoload :SpecialPublication, "#{__dir__}/identifiers/special_publication"

data/lib/pubid/itu/parser.rb

@@ -125,9 +125,48 @@ module Pubid
         dash >> (letter.repeat(1, 3) >> dot >> digits).as(:range_end)
       end
 
-      # Parts
+      # A "-YYYYMM" approval date — the "200307" of "T-REC-T.4-200307-I" and
+      # "ITU-T T.4-200307". ITU's own edition suffix is short ("-5"), so six
+      # digits that read as a plausible year (19xx/20xx) and month (01-12) are
+      # the date, never a part. A six-digit run that fails either test
+      # ("-200313", "-180001") is still a part, as it was before.
+      rule(:yyyymm_year) { (str("19") | str("20")) >> digit >> digit }
+      rule(:yyyymm_month) do
+        (str("0") >> match["1-9"]) | (str("1") >> match["0-2"])
+      end
+      rule(:yyyymm_shape) { yyyymm_year >> yyyymm_month >> digit.absent? }
+
+      # The status letter that trails the date in a publication id — "I" (in
+      # force) or "S" (superseded). It names the state of the edition, not the
+      # edition, so it is parsed and dropped. "S" is also the Spanish language
+      # suffix, so it is a status ONLY in the full "T-REC-…" id, where ITU
+      # always writes one; after an "ITU-T …-YYYYMM" print form only "I" is,
+      # and "-S" stays the language ("ITU-T Z.100-199911-S").
+      rule(:id_status) do
+        dash >> match["IS"] >> match["A-Za-z0-9"].absent?
+      end
+
+      rule(:print_id_status) do
+        dash >> str("I") >> match["A-Za-z0-9"].absent?
+      end
+
+      rule(:yyyymm_date) do
+        dash >> yyyymm_year.as(:year) >> yyyymm_month.as(:month) >>
+          digit.absent?
+      end
+
+      rule(:id_date) { yyyymm_date >> print_id_status.maybe }
+
+      # Either date spelling of a Recommendation.
+      rule(:document_date) { date_part | id_date }
+
+      # "ITU-T REC T.4", "ITU-T REC-T.4" — the redundant type word of ITU's
+      # own URLs. Not captured: a Recommendation is the default type.
+      rule(:rec_word) { str("REC") >> (space | dash) }
+
+      # Parts. The yyyymm guard keeps the approval date out of the part list.
       rule(:part) do
-        dash >> digits.as(:part)
+        dash >> yyyymm_shape.absent? >> digits.as(:part)
       end
 
       rule(:parts) { part.repeat(0).as(:parts) }
@@ -273,6 +312,7 @@ module Pubid
         itu_prefix >>
           sector >>
           space >>
+          rec_word.maybe >>
           series >> dot >>
           code >>
           range_end.maybe >>
@@ -281,7 +321,7 @@ module Pubid
           series_word.maybe >>
           attachment.maybe >>
           version_part.maybe >>
-          date_part.maybe
+          document_date.maybe
       end
 
       rule(:base_without_series) do
@@ -292,7 +332,7 @@ module Pubid
           code_suffixes >>
           attachment.maybe >>
           version_part.maybe >>
-          date_part.maybe
+          document_date.maybe
       end
 
       # A series-code document — "EMC-5", "MES-2", "QOS-2", "IMPL-8",
@@ -311,12 +351,10 @@ module Pubid
       # The number stays in `code.number`, so `root.number` — the field
       # relaton-index bsearches on — is "5" for EMC-5 and "QKD" for SEC-QKD
       # rather than nil.
-      # The OB guard keeps "ITU-T OB-1" a clean parse failure. Without it the
-      # string reaches Builder#build's Recommendation fallback, whose
-      # validate_ob_no_sector! raises an ArgumentError that escapes
-      # Identifier.parse's Parslet::ParseFailed rescue — turning a rejected
-      # input into a crash for callers. It guards "OB" + dash specifically, so
-      # a genuine two-letter mnemonic starting "OB" would still parse.
+      # The OB guard keeps "ITU-T OB-1" a clean parse failure rather than a
+      # Recommendation of a series "OB" — the Operational Bulletin's series
+      # name. It guards "OB" + dash specifically, so a genuine two-letter
+      # mnemonic starting "OB" would still parse.
       rule(:series_code_body) do
         (str("OB") >> dash).absent? >>
           letter.repeat(2).as(:series) >> dash.as(:series_dash) >>
@@ -365,10 +403,8 @@ module Pubid
       # Builder#build's `combined` branch, which builds a CombinedIdentifier and
       # would silently drop the marker — a clean parse failure is better than a
       # Report that comes back as a Recommendation. The OB guard mirrors
-      # series_code_body's: an Operational Bulletin is cross-bureau, and
-      # "Report ITU-T OB.1" would otherwise route to SpecialPublication (marker
-      # dropped) or hit validate_ob_no_sector!, whose ArgumentError escapes
-      # Identifier.parse's Parslet::ParseFailed rescue.
+      # series_code_body's: "Report ITU-T OB.1" would otherwise route to
+      # SpecialPublication with the marker dropped.
       rule(:report_body) do
         (str("OB") >> dot).absent? >>
           (series >> dot).maybe >>
@@ -535,6 +571,7 @@ module Pubid
         itu_prefix >>
           sector >>
           space >>
+          rec_word.maybe >>
           series >> dot >>
           code >>
           range_end.maybe >>
@@ -543,7 +580,7 @@ module Pubid
           series_word.maybe >>
           attachment.maybe >>
           version_part.maybe >>
-          date_part.maybe >>
+          document_date.maybe >>
           language.maybe
       end
 
@@ -556,24 +593,71 @@ module Pubid
           code_suffixes >>
           attachment.maybe >>
           version_part.maybe >>
-          date_part.maybe >>
+          document_date.maybe >>
+          language.maybe
+      end
+
+      # ITU's publication id — "T-REC-T.4-200307-I",
+      # "R-REC-BO.1130-5-202602-I": <sector>-REC-<number>[-<edition>]-<YYYYMM>
+      # [-<status>], the name ITU gives each edition in its URLs and PDF
+      # files. It builds the plain Recommendation it names and renders in the
+      # print form ("ITU-T T.4 (07/2003)"). The date is required: without it
+      # the string names no edition. No other rule starts with a bare sector
+      # letter, so the slot is free.
+      rule(:publication_id) do
+        sector >> dash >> str("REC") >> dash >>
+          series >> dot >> code >> yyyymm_date >> id_status.maybe >>
           language.maybe
       end
 
+      # The Radio Regulations — "ITU-R RR", "ITU-R RR (2020)", and the URL
+      # spelling "ITU-R RR-2020". Always ITU-R. The trailing any.absent? is
+      # load-bearing: PEG ordered choice never re-enters the alternation once
+      # an alternative succeeds, so a partial match on "ITU-R RR.1" must fail
+      # here and fall through to with_series.
+      rule(:radio_regulations) do
+        itu_prefix >> str("R").as(:sector) >> space >>
+          str("RR").as(:radio_regulations) >>
+          (date_part | (dash >> digit.repeat(4, 4).as(:year))).maybe >>
+          language.maybe >> any.absent?
+      end
+
       # OB (Operational Bulletin) — Special Publication.
-      # OB is a cross-bureau ITU publication; sector, when present in legacy
-      # strings like "ITU-T OB.1096", is silently dropped by the builder.
+      # OB is a cross-bureau ITU publication. The TSB spelling carries a
+      # sector ("ITU-T OB.1096 (2016)"); the builder keeps it, and it renders
+      # back, but it is not part of the bulletin's identity.
       rule(:ob_series) { str("OB").as(:series) }
 
       rule(:ob_dot_body) { dot >> number }
       rule(:ob_no_body) { space >> str("No.") >> space >> number }
+      # "ITU OB 1000" — metanorma-itu's docidentifier ("Annex to ITU OB %").
+      # Accepted as an input spelling only; it renders "ITU OB No. 1000", the
+      # form ITU's own bulletin site uses.
+      rule(:ob_bare_body) { space >> number }
+
+      # The date as a bulletin prints it — "ITU-T OB.1096 - 15.III.2016": day,
+      # Roman month, year. The months are tried longest first, because PEG
+      # takes the first alternative that matches and "I" would otherwise win
+      # on "III"; "XIII" matches "XII", then fails on the required dot.
+      rule(:roman_month) do
+        %w[XII XI X IX VIII VII VI V IV III II I]
+          .map { |m| str(m) }.reduce(:|)
+      end
+
+      rule(:ob_roman_date) do
+        str(" - ") >> digit.repeat(2, 2).as(:day) >> dot >>
+          roman_month.as(:roman_month) >> dot >>
+          digit.repeat(4, 4).as(:year)
+      end
+
+      rule(:ob_date) { date_part | ob_roman_date }
 
       rule(:ob_with_sector) do
         itu_prefix >>
           (sector >> space).maybe >>
           ob_series >>
-          (ob_dot_body | ob_no_body) >>
-          date_part.maybe >>
+          (ob_dot_body | ob_no_body | ob_bare_body) >>
+          ob_date.maybe >>
           language.maybe
       end
 
@@ -585,7 +669,7 @@ module Pubid
           str("Operational Bulletin").as(:_op_bull) >>
           space >> str("No.") >> space >>
           number >>
-          date_part.maybe >>
+          ob_date.maybe >>
           language.maybe
       end
 
@@ -671,13 +755,15 @@ module Pubid
           handbook |
           numeric_question |
           letter_question |
+          radio_regulations |
           with_series |
           contribution |
           # Unreachable earlier: special_publication needs the literal "OB",
           # handbook/numeric_question need leading digits, letter_question
           # needs series >> dot, and contribution needs "-C" after the series.
           series_code_identifier |
-          without_series
+          without_series |
+          publication_id
       end
 
       # Common-text form: an ITU identifier followed by "| ISO/IEC ...".