pubid 2.0.0.pre.alpha.8 → 2.0.0.pre.alpha.9

data/lib/pubid/ieee/parser.rb

@@ -57,6 +57,24 @@ module Pubid
         date_with_month_text | date_with_month_numeric | date_year_only
       end
 
+      # Trailing print/reaffirm date "<sep>Month YYYY" — captured under the
+      # distinct :trailing_month/:trailing_year keys so it never collides with a
+      # base -YYYY identity year (the builder promotes it only when no base year
+      # exists). Accepts BOTH a month name ("May 2014") and a month-first numeric
+      # month ("05 2014"). The two variants are unambiguous: month_name never
+      # starts with a digit, and month_numeric (01-12) never matches a bare
+      # 19xx/20xx year — so a bare trailing year still falls through to callers'
+      # own bare-year clause.
+      rule(:trailing_month_year) do
+        # The numeric branch accepts `-` as well as space before the year:
+        # preprocessing (the `(\d)\s+(\d{4})` gsub) rewrites a trailing
+        # "05 2014" → "05-2014", so a numeric month reaches the grammar
+        # dash-joined; the month-name branch escapes that gsub (the char before
+        # the space is a letter) and stays space-joined.
+        ((comma | space) >> month_name.as(:trailing_month) >> space >> year_digits.as(:trailing_year)) |
+          ((comma | space) >> month_numeric.as(:trailing_month) >> (space | dash) >> year_digits.as(:trailing_year))
+      end
+
       # Month patterns
       rule(:month_name) do
         # Period-suffixed abbreviations (longest first)
@@ -75,9 +93,17 @@ module Pubid
       # Organizations
       rule(:organization) do
         str("IEEE") | str("AIEE") | str("ANSI") | str("ASA") |
+          # ANS (American Nuclear Society) — a third co-publisher on nuclear
+          # standards ("ANSI/IEEE/ANS 7.4-3-2-1982"). Listed AFTER ANSI so it
+          # never shadows the longer token.
+          str("ANS") |
           str("IEC") | str("ISO") | str("ASTM") | str("CSA") | str("ASME") |
           str("NACE") | str("NSF") | str("ASHRAE") | str("NCTA") | str("AESC") |
-          str("EIA") # NEW Session 224: Add EIA support
+          str("EIA") | # NEW Session 224: Add EIA support
+          # Historical / foreign co-publishers seen in relaton-data-ieee
+          # (bucket 4): AMPP, USAS (both standalone), and the IEEE sub-board /
+          # partner co-publishers USEMCSC, EAB, MPAI.
+          str("USEMCSC") | str("AMPP") | str("USAS") | str("EAB") | str("MPAI")
       end
 
       # Complex organization prefixes (Category 5: ANSI Complex)
@@ -130,12 +156,42 @@ module Pubid
         (slash >> str("C") >> digits >> dot >> digits >> dot >> digits >> dash >> year_digits).as(:ieee_crossref)
       end
 
+      # IPCEA co-designation suffix (/IPCEA P-46-426-1962). Captured verbatim
+      # (leading slash included) so it round-trips through the `crossref`
+      # attribute. Used only by the S-designation rule below.
+      rule(:ipcea_copub) do
+        (slash >> str("IPCEA") >> space >>
+         match('[A-Za-z0-9.\-]').repeat(1)).as(:ipcea_copub)
+      end
+
+      # Historical IEEE/IPCEA co-published cable designation (e.g. S-135).
+      # The "S-<digits>" number has a dash between the letter series and the
+      # digits, which the shared `number` rule deliberately rejects (that
+      # tightening is what keeps a bare "IEEE S" from parsing). So this one-off
+      # family gets its own rule rather than loosening `number` and risking the
+      # 12k-row corpus. Requiring `str("S") >> dash >> digits` keeps "IEEE S"
+      # (no dash+digit) rejected. Handles: "IEEE Std S-135", "IEEE S-135",
+      # bare "S-135", the "/IPCEA …" slash co-designation, and the
+      # "(IPCEA …)" parenthetical variant.
+      rule(:s_designation) do
+        (publisher >> space).maybe >>
+          (type_word.as(:type) >> space?).maybe >>
+          (str("S") >> dash >> digits).as(:s_number) >>
+          ipcea_copub.maybe >>
+          parenthetical.maybe
+      end
+
       # Document number - support letters and digits, with optional prefix P
       # Complex multi-part numbers like P11073-10404-10419 should be fully captured
       # But simple cases like "623-1976" should not consume the dash before year
       rule(:number) do
         (str("P").maybe >>
-         (digits | upper).repeat(1) >> # The first component must be at least one digit
+         # The numeric core must contain at least one digit: an optional letter
+         # prefix (C, S, …) then a required `digits` run, then any digit/letter
+         # tail. The old `(digits | upper).repeat(1)` wrongly accepted an
+         # all-letter token like a bare "S" (a prefix with no number — e.g.
+         # "IEEE S"), which is not a valid IEEE identifier.
+         upper.repeat(0) >> digits >> (digits | upper).repeat(0) >>
          # Only consume dash+digits if followed by another dash+digits (multi-part pattern)
          # OR if the digits don't look like a year (not 4 digits starting with 19/20)
          # This prevents consuming "623-1976" as a number but allows "P11073-10404-10419"
@@ -170,7 +226,14 @@ module Pubid
 
       # Draft patterns
       rule(:draft_status) do
-        (str("Active Unapproved") | str("Unapproved") | str("Approved")) >> space
+        # "Active" (bucket 3) joins the generic draft-status path so
+        # "IEEE Active Std P… /D…" parses like the "Unapproved" forms, without
+        # touching ieee_approved_draft_identifier (which would break the
+        # issue-#209 unapproved-drops-Std rendering). "Active Approved" completes
+        # the two-word matrix alongside "Active Unapproved" (draft-grammar
+        # coverage: the "<status> Draft P…/D…" prefix forms). Longest token first.
+        (str("Active Unapproved") | str("Active Approved") | str("Unapproved") |
+         str("Approved") | str("Active")) >> space
       end
 
       rule(:draft_prefix) do
@@ -180,7 +243,11 @@ module Pubid
       rule(:draft_version) do
         # Enhanced to handle multiple draft notation patterns
         # D is optional to handle /08 style drafts (e.g., IEEE P1052/08)
-        (str("D") >> str("IS").absent?).maybe >> # Avoid matching "DIS" (ISO stage)
+        # A draft never begins with "R-" — that is the revision suffix
+        # (revision_suffix rule); guard so the D-less path doesn't swallow a
+        # bare "/R-<id>" (e.g. the no-draft "P1722/R-1") as a draft.
+        (str("R") >> dash).absent? >>
+          (str("D") >> str("IS").absent?).maybe >> # Avoid matching "DIS" (ISO stage)
           (
             # Pattern: D3.1 (decimal with 1-2 digits on each side) - MOST COMMON, put first
             # Also handles trailing letter: D7.3A, D2.0E
@@ -202,8 +269,17 @@ module Pubid
       rule(:draft_date) do
         # Enhanced to handle: ", Sept 2008" or " Sept 2008" or ", Month Year"
         ((comma | space) >> month_name.as(:month) >> space >> year_digits.as(:year)) |
+          # Numeric-month form ", 05 2007" / " 05 2007" — no text month name. The
+          # year separator may be a dash: preprocessing (parser.rb ~1278) rewrites
+          # a trailing " <digits> <year>" to "<digits>-<year>", so "05 2007"
+          # reaches the grammar as "05-2007".
+          ((comma | space) >> month_numeric.as(:month) >> (space | dash) >> year_digits.as(:year)) |
           (((space? >> comma >> space?) | space) >> month_name.as(:month) >>
           (
+            # "Month DD, Year" (day then comma) and "Month DD Year" (day, no
+            # comma) — the day/year separator may be a dash for the same
+            # preprocessing reason ("July 15 2012" → "July 15-2012").
+            ((space >> digits.as(:day)) >> ((comma | space | dash) >> year_digits.as(:year))) |
             ((space >> digits.as(:day)).maybe >> comma >> year_digits.as(:year)) |
             (comma >> space? >> year_digits.as(:year)) |
             (space >> year_digits.as(:year))
@@ -229,13 +305,32 @@ module Pubid
          draft_date.maybe).as(:draft)
       end
 
+      # Trailing revision suffix "/R-<id>". Normalization funnels every revision
+      # spelling here — IEEE's native inline "Rev<n>" (repositioned) and
+      # relaton's synthetic "/R-<x>" (kept in place) — so a single trailing rule
+      # captures them all. The id is alphanumeric ("2", "18", "i").
+      rule(:revision_suffix) do
+        slash >> str("R") >> dash >> match('[0-9A-Za-z]').repeat(1).as(:revision)
+      end
+
       # Edition - enhanced to support IEC formats like "Edition 1.0 2015-03"
+      #
+      # The edition's own year is captured as :edition_year (NOT :year). The
+      # generic Std branch carries a base "-YYYY" slot AND a separate trailing
+      # `edition.maybe`; a bare :year here would collide with the base :year
+      # when both fire, so Parslet warns "Duplicate subtrees … keys: [:year]"
+      # and drops the base identity year (the same failure #299 fixed for the
+      # trailing month/year clause). The builder promotes :edition_year to the
+      # identity year only when there is no base year.
       rule(:edition) do
-        (comma >> year_digits.as(:year) >> str(" Edition")) |
+        (comma >> year_digits.as(:edition_year) >> str(" Edition")) |
           ((space | dash) >> str("Edition ") >>
            (digits >> dot >> digits).as(:edition) >>
-           (space | str(" - ")) >>
-           year_digits.as(:year) >>
+           # Year separator: a space, " - ", or a bare dash (preprocessing
+           # rewrites "Edition 3.0 2016" -> "Edition 3.0-2016", the shape the
+           # normalized "/E-<n>-YYYY" suffix produces — nil-residue item 1).
+           (str(" - ") | space | dash) >>
+           year_digits.as(:edition_year) >>
            (dash >> digit.repeat(2, 2).as(:edition_month)).maybe) # Capture -MM as edition_month
       end
 
@@ -261,10 +356,16 @@ module Pubid
          ((dash | str(":") | space) >> year_digits.as(:cor_year)).maybe).as(:corrigendum)
       end
 
-      # Amendment
+      # Amendment — mirrors the corrigendum rule's separator flexibility so
+      # IEEE-format "/Amd 2-2004" and "/Amd2-2004" parse the same way as
+      # their Cor counterparts (issue #210).
       rule(:amendment) do
-        (slash >> str("Amd") >> digits.as(:amd_number) >>
-         (dash >> year_digits.as(:amd_year)).maybe).as(:amendment)
+        ((str("_") | slash | dash | space) >>
+         (str("Amendment") | str("Amd")) >>
+         (dash | dot | space).maybe >>
+         space? >>
+         digits.as(:amd_number).maybe >>
+         ((dash | str(":") | space) >> year_digits.as(:amd_year)).maybe).as(:amendment)
       end
 
       # Interpretation notation (/INT)
@@ -283,9 +384,14 @@ module Pubid
         ).as(:reaffirmed)
       end
 
-      # Redline
+      # Redline suffix at the very end. Accepts relaton's canonical " Redline"
+      # (space, no dash) and pubid's older " - Redline" (space-dash-space),
+      # case-insensitive. Captured (presence only) so the builder sets a
+      # `redline: true` flag the renderer restores — a redline is a distinct
+      # document and must not collapse to its base standard.
       rule(:redline) do
-        str(" - Redline").as(:redline)
+        (space >> (dash >> space).maybe >>
+         (str("Redline") | str("REDLINE") | str("redline"))).as(:redline)
       end
 
       # Book nickname (e.g., "[The Orange Book]", "[IEEE Gold Book]")
@@ -346,8 +452,11 @@ module Pubid
           str(" and ").absent? >>
           str(", ").absent? >>
           str(" as amended by ").absent? >>
-          str(" / ").absent? >>
-          str("; ").absent? >>
+          # Stop at any "/", ";", or "-" that introduces another relationship
+          # (look-ahead: separator + relationship_type keyword). This lets the
+          # slash inside "IEEE Std 525-2007/Cor 1-2015" stay part of the
+          # identifier, while the slash before "Incorporates ..." splits.
+          relationship_break.absent? >>
           str(")").absent? >>
           match(".")
         ).repeat(1)
@@ -372,15 +481,38 @@ module Pubid
           str(" and its approved amendments").as(:approved_amendments)
       end
 
+      # A character sequence that may separate two relationships inside the
+      # parenthetical: a "/", ";", or "-" with optional surrounding spaces.
+      # Standalone it is permissive — the actual decision to split is gated by
+      # `relationship_break`, which requires another relationship_type to
+      # follow. This way the slash in "IEEE Std 525-2007/Cor 1-2015" stays
+      # part of the related identifier.
+      rule(:relationship_separator) do
+        (space.maybe >> str("/") >> space.maybe) |
+          (space.maybe >> str(";") >> space.maybe) |
+          (space.maybe >> str("-") >> space.maybe)
+      end
+
+      # Look-ahead: a separator followed by another relationship_type keyword.
+      # Used as an absent? guard in identifier_string so the parser stops
+      # consuming characters right before a new relationship begins.
+      rule(:relationship_break) do
+        relationship_separator >> space.maybe >> relationship_type
+      end
+
       # Relationship clause (handles all relationship types)
       rule(:relationship_clause) do
         space.maybe >> str("(") >>
           relationship_type.as(:relationship_type) >>
           identifier_list.as(:related_ids) >>
           as_amended_by_clause.maybe >>
-          # Handle multiple relationships separated by " / " OR "; "
+          # Additional relationships separated by "/", ";", or "-" (optionally
+          # surrounded by spaces). The separator is only honored when followed
+          # by another relationship_type — identifier-internal slashes like
+          # "/Cor 1-2015" are not mistaken for relationship breaks because
+          # identifier_string stops at relationship_break ahead.
           (
-            (str(" / ") | str("; ")) >> # Support both separators
+            relationship_separator >>
             relationship_type.as(:relationship_type) >>
             identifier_list.as(:related_ids) >>
             as_amended_by_clause.maybe
@@ -435,6 +567,11 @@ module Pubid
         str("IEC/IEEE") >>
           space >>
           str("P").absent? >> # NOT a P prefix (would be joint development)
+          # The copublished number must contain at least one digit. This rejects
+          # an all-letter placeholder like "IEC/IEEE TR" (no real document
+          # number) — the same "require a digit" tightening the Standard number
+          # rule got, applied to the copublished number grammar.
+          (match("[^0-9\n]").repeat >> digit).present? >>
           match("[^\n]").repeat(1).as(:content)
       end
 
@@ -456,20 +593,100 @@ module Pubid
             # Variant 2: , CDV1 notation (comma before stage code)
             (comma >> (str("CDV") | str("FDIS") | str("CD") | str("DIS")).as(:iec_stage) >> digits.maybe.as(:stage_iteration))
           ).maybe >>
+          # Optional edition, from relaton's "/E-<n>" suffix normalized to
+          # "Edition <n>.0[ YYYY]" (nil-residue hand-off item 1). The edition
+          # rule carries its own year, so the year clause below simply doesn't
+          # fire when an edition is present.
+          edition.maybe >>
           ((dash >> year_digits.as(:year)) | # Either -YEAR
-           (comma.maybe >> space >> month_name.as(:month) >> space.maybe >> year_digits.as(:year))).maybe # Or Month YEAR (with optional comma)
+           (comma.maybe >> space >> month_name.as(:month) >> space.maybe >> year_digits.as(:year))).maybe >> # Or Month YEAR (with optional comma)
+          revision_suffix.maybe
       end
 
       rule(:joint_development_iso_format) do
-        # ISO/IEC/IEEE FDIS 26511:2018 (ISO-led format)
-        (str("ISO/IEC/IEEE") | str("ISO/IEEE") | str("IEC/IEEE")).as(:joint_publishers) >>
+        # ISO-led stage designations. Two spellings:
+        #   colon form  : "ISO/IEC/IEEE FDIS 26511:2018"           (already used)
+        #   corpus form : "ISO/IEC/IEEE FDIS P26515-2018-05"       (historical)
+        # The corpus form adds a leading "P" on the number, a trailing
+        # "-YYYY[-MM]" date (instead of ":YYYY"), multi-digit committee-draft
+        # stage codes (CD1..CD4) plus CDV, and a wider set of joint publishers.
+        # (roadmap items 2/3, phase 1). longest publisher token first.
+        (str("ISO/IEC/IEEE") | str("IEEE/ISO/IEC") | str("IEEE/IEC/ISO") |
+         str("ISO/IEEE") | str("IEC/IEEE") | str("IEEE/IEC") | str("ISO/IEC") |
+         str("IEEE")).as(:joint_publishers) >>
           space >>
-          # ISO stage codes
-          (str("FDIS") | str("DIS") | str("CD") | str("WD") | str("PWI") | str("NP")).as(:iso_stage) >>
+          # ISO stage codes: FDIS, FCD, CDV; DIS/CD with an optional round digit
+          # (DIS2, CD1..CD4); WD/PWI/NP. (FCD before FDIS is fine — distinct.)
+          (str("FDIS") | str("FCD") | str("CDV") |
+           (str("DIS") >> digit.maybe) |
+           (str("CD") >> digit.maybe) |
+           str("WD") | str("PWI") | str("NP")).as(:iso_stage) >>
+          # optional " Std" noise word after the stage (e.g. "FDIS Std P15288")
+          (space >> str("Std")).maybe >>
           space >>
+          str("P").maybe >> # optional project marker on the number
           digits.as(:number) >>
-          ((dot | dash) >> digits.as(:part)).maybe >> # Optional part
-          (str(":") >> year_digits.as(:year)).maybe
+          # part must not swallow the trailing year (year_digits.absent?)
+          ((dot | dash) >> year_digits.absent? >> digits.as(:part)).maybe >>
+          (
+            (str(":") >> year_digits.as(:year)) |
+            (dash >> year_digits.as(:year) >>
+             (dash >> month_numeric.as(:month)).maybe) |
+            # Trailing text date ", April 2015" / " April 2015" (build_joint_development
+            # already reads :month/:year). No :year collision — the iso rule has no
+            # other :year capture (edition uses :edition_year).
+            ((comma | space) >> month_name.as(:month) >> space >> year_digits.as(:year))
+          ).maybe >>
+          # Optional /D<draft> tail. normalize_relaton_suffixes repositions the
+          # historical "…/D-3-2017" onto the number as "…-2017/D3", so by the
+          # time this rule runs the draft usually trails the date (bucket 5);
+          # a date-less "/D-4" keeps its hyphen (bucket 7), hence dash.maybe.
+          (slash >> str("D") >> dash.maybe >>
+           match('[0-9.]').repeat(1).as(:draft_version)).maybe >>
+          # Optional edition, from relaton's "/E-<n>" suffix normalized to
+          # "Edition <n>.0[ YYYY]" (nil-residue hand-off item 1).
+          edition.maybe >>
+          revision_suffix.maybe
+      end
+
+      # Embedded (stage-LAST) ISO-led designations: the corpus writes the ISO
+      # stage AFTER the (dotted or dashed) part —
+      #   "ISO/IEC/IEEE 29119.4.FDIS, April 2015"   (dot part .4, stage .FDIS)
+      #   "ISO/IEC/IEEE 24748-5.CD3, February 2015" (dash part -5, stage .CD3)
+      #   "IEEE P24748.5.CD3, July 2015"            (bare IEEE + P)
+      # — rather than before the number (joint_development_iso_format). Same
+      # meaning as the stage-first form: `29119.4`/`29119-4` = "29119 part 4" and
+      # the trailing `.FDIS` is the ISO stage, NOT a second part. Routes through
+      # the same build_joint_development (via :joint_publishers/:iso_stage), so the
+      # stage is modeled correctly and the numeric part stays separate.
+      rule(:joint_development_embedded_stage) do
+        # publisher set mirrors joint_development_iso_format (incl. bare IEEE);
+        # longest token first.
+        (str("ISO/IEC/IEEE") | str("IEEE/ISO/IEC") | str("IEEE/IEC/ISO") |
+         str("ISO/IEEE") | str("IEC/IEEE") | str("IEEE/IEC") | str("ISO/IEC") |
+         str("IEEE")).as(:joint_publishers) >>
+          space >>
+          str("P").maybe >> # optional project marker on the number
+          digits.as(:number) >>
+          # optional numeric part (dot or dash); must not swallow a year
+          ((dot | dash) >> year_digits.absent? >> digits.as(:part)).maybe >>
+          # the embedded stage, dot-separated, closed vocab (same as iso_stage).
+          # The (digit|letter) look-ahead keeps it at a token boundary so "CD"
+          # can't match inside a longer token and a glued no-space date
+          # (".DISMay2013") is left to fall through.
+          dot >>
+          (str("FDIS") | str("FCD") | str("CDV") |
+           (str("DIS") >> digit.maybe) |
+           (str("CD") >> digit.maybe) |
+           str("WD") | str("PWI") | str("NP")).as(:iso_stage) >>
+          (digit | match("[A-Za-z]")).absent? >>
+          # optional trailing date: :YYYY, -YYYY[-MM], or text "Month YYYY"
+          (
+            (str(":") >> year_digits.as(:year)) |
+            (dash >> year_digits.as(:year) >>
+             (dash >> month_numeric.as(:month)).maybe) |
+            ((comma | space) >> month_name.as(:month) >> space >> year_digits.as(:year))
+          ).maybe
       end
 
       # Number-first pattern: "1873-2015 IEEE Standard..."
@@ -493,12 +710,25 @@ module Pubid
           (slash >> digits.as(:draft_version)).as(:digit_draft).maybe >>
           # FDIS and other ISO stage codes without D prefix (Pattern 3)
           fdraft.maybe >>
-          # Enhanced: Accept both comma and space before month/year
-          ((comma | space) >> month_name.as(:month) >> space >> year_digits.as(:year)).maybe >>
+          # Trailing "Month YYYY" print/reaffirm date. Captured under distinct
+          # keys so it never collides with the base -YYYY identity year (a
+          # collision made Parslet drop the base year and warn "Duplicate
+          # subtrees … keys: [:year]"). The builder promotes it to the identity
+          # only when there is no base year.
+          trailing_month_year.maybe >>
           corrigendum.maybe >>
           draft.maybe >>
+          # Revision trails the draft (before any date), matching normalization's
+          # ".../D<n>/R-<x>" repositioning of "P802.16Rev2/D3 Feb 2008".
+          revision_suffix.maybe >>
           # ALSO accept month/year after draft (some patterns like /DX, Month YEAR)
-          ((comma | space) >> month_name.as(:month) >> space >> year_digits.as(:year)).maybe >>
+          trailing_month_year.maybe >>
+          # Trailing corrigendum AFTER the draft+date ("…/D1, Jan 2007/Cor. 1").
+          # The draft's own draft_date has already consumed the date, leaving
+          # "/Cor. 1" here; the resulting flat corrigendum+draft tree routes to
+          # build_flat_corrigendum (disjoint from the pre-draft corrigendum slot,
+          # which only fires when no draft precedes it).
+          corrigendum.maybe >>
           parenthetical.maybe
       end
 
@@ -508,27 +738,50 @@ module Pubid
           str("P") >> space.maybe >> # Make space after P optional
           number >>
           (part_subpart_year | edition).maybe >>
-          # Enhanced: Accept both comma and space before month/year
-          ((comma | space) >> month_name.as(:month) >> space >> year_digits.as(:year)).maybe >>
+          # Trailing "Month YYYY"/bare-year date under distinct keys so they
+          # never collide with the base -YYYY identity year (see ieee_p_identifier).
+          trailing_month_year.maybe >>
           corrigendum.maybe >>
           draft.maybe >>
+          revision_suffix.maybe >>
           # ALSO accept month/year after draft
-          ((comma | space) >> month_name.as(:month) >> space >> year_digits.as(:year)).maybe >>
+          trailing_month_year.maybe >>
           # Accept bare year after draft: ", 2015"
-          ((comma | space) >> year_digits.as(:year)).maybe >>
+          ((comma | space) >> year_digits.as(:trailing_year)).maybe >>
+          # Trailing corrigendum after the draft+date (parity with ieee_p_identifier).
+          corrigendum.maybe >>
           parenthetical.maybe
       end
 
       # IEEE Draft P pattern: "IEEE Draft P802.11..." OR "Draft P802.11..." (IEEE prefix optional)
+      # An optional status phrase may precede "Draft" — "IEEE Unapproved Draft P…",
+      # "IEEE Active Approved Draft P…" — the largest index-v2 draft bucket. The
+      # status is captured (draft_status ends in a space, so "Approved Draft"
+      # splits cleanly) and round-trips as the draft_status attribute; the literal
+      # "Draft" stays a bare marker (dropped on render, like the plain form).
       rule(:ieee_draft_p_identifier) do
         (str("IEEE").as(:publisher) >> space).maybe >> # Make IEEE prefix optional
+          draft_status.as(:draft_status).maybe >>
           str("Draft") >> space >>
-          str("P") >>
+          # `P` is optional — a status-word draft may carry a bare number
+          # ("IEEE Unapproved Draft 802.1ah/D4.2"), mirroring
+          # ieee_approved_draft_identifier's str("P").maybe. The `number` rule
+          # already accepts the bare forms (802.1ah, C57.15, 11073-10471).
+          str("P").maybe >>
           number >>
           (part_subpart_year | edition).maybe >>
-          # Enhanced: Accept month/year after draft number
-          (space >> month_name.as(:month) >> space >> year_digits.as(:year)).maybe >>
+          # Trailing "Month YYYY" date under distinct keys so it never collides
+          # with the base -YYYY identity year (see ieee_p_identifier).
+          ((space >> month_name.as(:trailing_month) >> space >> year_digits.as(:trailing_year)) |
+           (space >> month_numeric.as(:trailing_month) >> (space | dash) >> year_digits.as(:trailing_year))).maybe >>
           draft.maybe >>
+          revision_suffix.maybe >>
+          # Trailing corrigendum after the draft ("…/D2.0/Cor. 1", or
+          # "…/D1.0, Dec 2007/Cor. 1" where the draft's own draft_date consumes
+          # the date, leaving "/Cor. N"). The flat corrigendum+draft tree (no
+          # :base) routes to build_flat_corrigendum, which rebuilds the base
+          # standard (carrying the draft + draft_status) and wraps it.
+          corrigendum.maybe >>
           parenthetical.maybe
       end
 
@@ -542,6 +795,7 @@ module Pubid
           number >>
           (part_subpart_year | edition).maybe >>
           draft.maybe >>
+          revision_suffix.maybe >>
           parenthetical.maybe
       end
 
@@ -656,7 +910,7 @@ module Pubid
           (type_word.as(:type) >> space?).maybe >>
           number >>
           part_subpart_year.maybe # This captures the full identifier before /Cor
-        ).as(:base_identifier) >>
+        ).as(:base) >>
           # Now match the corrigendum portion
           (slash | dash | space) >>
           str("Cor") >>
@@ -677,7 +931,7 @@ module Pubid
           (type_word.as(:type) >> space?).maybe >>
           number >>
           part_subpart_year.maybe
-        ).as(:base_identifier) >>
+        ).as(:base) >>
           # Now match the interpretation portion
           (slash | dash | space) >>
           str("INT") >>
@@ -695,7 +949,7 @@ module Pubid
           (type_word.as(:type) >> space?).maybe >>
           number >>
           part_subpart_year.maybe
-        ).as(:base_identifier) >>
+        ).as(:base) >>
           # Now match the conformance portion
           (slash | dash | space) >>
           str("Conformance") >>
@@ -757,11 +1011,13 @@ module Pubid
           ieee_astm_si_psi | # NEW Session 171: Add IEEE/ASTM SI/PSI support
           multi_numbered_identifier | # NEW: Try multi-numbered identifiers before generic patterns
           csa_dual_published | # NEW: Try CSA dual published before generic patterns
+          s_designation | # Historical IEEE/IPCEA cable designation (S-135)
           corrigendum_identifier | # NEW: Try corrigendum before generic patterns
           interpretation_identifier | # NEW: Try interpretation identifier before generic patterns
           conformance_identifier | # NEW: Try conformance identifier before generic patterns
           joint_development_ieee_format |
           joint_development_iso_format |
+          joint_development_embedded_stage | # stage-LAST embedded form (before generic)
           iec_ieee_copublished |
           number_first_identifier |
           ieee_approved_draft_identifier |
@@ -781,8 +1037,16 @@ module Pubid
           ashrae_copub.maybe >> # NEW: Add /ASHRAE Guideline support
           ieee_crossref.maybe >> # NEW: Add /C62.22.1-1996 cross-reference support
           draft.maybe >>
-          # Enhanced: Accept both comma and space before month/year
-          ((comma | space) >> month_name.as(:month) >> space >> year_digits.as(:year)).maybe >>
+          revision_suffix.maybe >>
+          # Trailing "Month YYYY" print/reaffirm date under distinct keys so it
+          # never collides with the base -YYYY identity year (see
+          # ieee_p_identifier). The builder promotes it only when no base year.
+          trailing_month_year.maybe >>
+          # Trailing corrigendum after the draft+date ("IEEE Approved P1015/D1,
+          # Jan 2007/Cor. 1"): the generic bucket is the only path a status-word
+          # form reaches. Routes via build_flat_corrigendum (disjoint from the
+          # pre-draft corrigendum slot at line 954).
+          corrigendum.maybe >>
           edition.maybe >>
           parenthetical.maybe >>  # REVERT: Back to single parenthetical
           book_nickname.maybe >>  # NEW: Add book nickname support
@@ -794,6 +1058,133 @@ module Pubid
 
       root(:identifier)
 
+      # Rewrite relaton's historical IEEE serialization into canonical pubid
+      # spellings. relaton's own formatter (Relaton::Ieee::PubId::Id#to_s) emits
+      # suffix tokens that differ from pubid's grammar:
+      #
+      #   /D-N-YYYY[-MM]  draft + trailing numeric date  (the dominant form)
+      #   /E-N[-YYYY[-MM]] edition
+      #   /R-N[-YYYY]      revision (pubid has no revision suffix)
+      #   " Redline"       redline suffix without the " - " pubid expects
+      #
+      # The draft/edition trailing date is repositioned onto the document number
+      # as a base year/month (a form pubid already parses), which also keeps the
+      # draft component clean so it round-trips through to_hash/from_hash.
+      def self.normalize_relaton_suffixes(cleaned)
+        # NOTE: the trailing " Redline"/" - Redline" suffix is NO LONGER stripped
+        # here — the grammar's `redline` rule captures it into a redline flag so
+        # a redline id stays distinct from its base standard.
+
+        # Combined draft + corrigendum: relaton emits "…/D-N/CorM-YYYY" (draft
+        # then corrigendum), but pubid's grammar accepts the corrigendum first.
+        # Swap them so the corrigendum keeps its own year and the draft trails.
+        # The hyphen after "D" is mandatory here: relaton's formatter always
+        # emits "/D-<draft>", whereas pubid's own canonical joint-development
+        # form is "/D<draft>-<year>" (no hyphen, year kept on the draft) — which
+        # already parses and must not be repositioned. A trailing corrigendum
+        # month (the "-MM" in "/CorM-YYYY-MM") is intentionally dropped: pubid's
+        # corrigendum model carries only a year.
+        cleaned = cleaned.sub(
+          %r{\A(.*)/D-([0-9A-Za-z][0-9A-Za-z.+]*?)/Cor\.?[ ]?(\d+)(?:-((?:19|20)\d\d))?(?:-\d\d)?\z},
+        ) do
+          base, draft, cor, year = Regexp.last_match.captures
+          "#{base}/Cor #{cor}#{year ? "-#{year}" : ''}/D#{draft}"
+        end
+
+        # Combined draft + revision, and the empty-draft revision-only form:
+        #   "…/D-<d>/R-<x>-YYYY[-MM]"  and  "…/D-/R-<x>-YYYY"   (nil-residue #2).
+        # Reposition the base publication date onto the number (pubid's
+        # "-YYYY[-MM]" shape), keep the draft as "/D<d>" (dropped when the draft
+        # is empty), and leave a trailing "/R-<x>" the grammar captures as the
+        # revision. Runs before the plain "/D-…" reposition, which the embedded
+        # "/R-" would otherwise defeat.
+        cleaned = cleaned.sub(
+          %r{\A(.*?)/D-([0-9A-Za-z.+]*)/R-([0-9A-Za-z]+)(?:-((?:19|20)\d\d)(?:-(0[1-9]|1[0-2]))?)?\z},
+        ) do
+          base, draft, rev, year, month = Regexp.last_match.captures
+          date = year ? "-#{year}#{month ? "-#{month}" : ''}" : ""
+          draft_part = draft.to_s.empty? ? "" : "/D#{draft}"
+          "#{base}#{date}#{draft_part}/R-#{rev}"
+        end
+
+        # /D-N drafts with a trailing numeric date, when the draft is the last
+        # suffix: reposition the -YYYY[-MM] date onto the number. A following
+        # /Cor, /Amd, /R or /E suffix carries its own year, so the `\z` anchor
+        # keeps this from firing on those combined forms.
+        cleaned = cleaned.sub(
+          %r{\A(.*)/D-([0-9A-Za-z][0-9A-Za-z.+]*?)-((?:19|20)\d\d)(?:-(0[1-9]|1[0-2]))?\z},
+        ) do
+          base, draft, year, month = Regexp.last_match.captures
+          "#{base}-#{year}#{month ? "-#{month}" : ''}/D#{draft}"
+        end
+
+        # /E-N editions: relaton's "/E-2-2023-02" → pubid's "Edition 2.0 2023-02".
+        cleaned = cleaned.sub(
+          %r{\A(.*?)/E-(\d+)(?:-((?:19|20)\d\d)(?:-(0[1-9]|1[0-2]))?)?\z},
+        ) do
+          base, edition, year, month = Regexp.last_match.captures
+          date = year ? " #{year}#{month ? "-#{month}" : ''}" : ""
+          "#{base} Edition #{edition}.0#{date}"
+        end
+
+        # /R-N revisions: PRESERVE them (the grammar's revision_suffix rule now
+        # captures a trailing "/R-<x>" into the `revision` attribute). Just
+        # reposition any trailing publication year onto the number, keeping the
+        # "/R-<x>" in place for the grammar.
+        cleaned.sub(
+          %r{\A(.*?)/R-([0-9A-Za-z]+)(?:-((?:19|20)\d\d))?\z},
+        ) do
+          base, rev, year = Regexp.last_match.captures
+          "#{year ? "#{base}-#{year}" : base}/R-#{rev}"
+        end
+      end
+
+      # Strip the IEEE rawbib revision-notation dialects. `REV`/`Rev`
+      # (case-insensitive) + a trailing revision id `[A-Za-z0-9]+`, glued to the
+      # number or separated by `-`, `/`, `_`, `.`, or a space, and preceding the
+      # draft. pubid's canonical "<num>/D<n>/R-<x>" form already drops the
+      # revision on render (normalize_relaton_suffixes strips a trailing /R-x),
+      # so the revision-less result is *the same identifier* — and stripping
+      # (rather than reordering) leaves any trailing date/parenthetical intact,
+      # which is why forms that already parse (`Draft P…-REVmb/D3.0, Mar 2010`)
+      # are NOT disturbed. Examples:
+      #   "P802.16.2-REVa/D8" -> "P802.16.2/D8"
+      #   "P802.16/REVd/D5"   -> "P802.16/D5"
+      #   "P802.15.1REVa/D5"  -> "P802.15.1/D5"
+      #   "P802.11REVmb"      -> "P802.11"   (no draft)
+      def self.normalize_revision_notation(cleaned)
+        # NUMBERED revisions ("Rev<digits>") are PRESERVED — repositioned to a
+        # trailing "/R-<n>" suffix the grammar captures as the `revision`
+        # attribute (IEEE's native inline spelling; numbered-revision hand-off).
+        # A "\d+" right after "Rev" both selects the numbered subset and keeps
+        # these off the English word "Revision". Three source positions:
+        #   after a draft : "PC37.30.2/D043 Rev 18" -> ".../D043/R-18"
+        cleaned = cleaned.sub(
+          %r{(/D[0-9A-Za-z.]*)\s+[Rr][Ee][Vv]\s*(\d+)}, '\1/R-\2'
+        )
+        #   before a draft: "P802.16Rev2/D3" -> "P802.16/D3/R-2"
+        cleaned = cleaned.sub(
+          %r{[-/_.]?\s?[Rr][Ee][Vv][-\s]?(\d+)(/D[0-9A-Za-z.]*)}, '\2/R-\1'
+        )
+        #   no draft, trailing: "P1722-rev1" -> "P1722/R-1"
+        cleaned = cleaned.sub(
+          %r{(\d)[-._]?\s?[Rr][Ee][Vv]\s*(\d+)\s*\z}, '\1/R-\2'
+        )
+
+        # LETTERED inline revisions ("REVa", "REVmb") have no pubid model and are
+        # still STRIPPED (unchanged behaviour). The numbered forms above already
+        # became "/R-<n>", so these regexes only see the lettered residue.
+        # Revision token that PRECEDES a draft: drop it (keep the /D…).
+        cleaned = cleaned.sub(
+          %r{[-/_.]?\s?[Rr][Ee][Vv][-\s]?[A-Za-z0-9]+(?=/D[0-9])},
+          "",
+        )
+        # Trailing revision glued to the number with no draft ("P802.11REVmb");
+        # a digit must immediately precede REV so a trailing English word like
+        # "…Revision" can't match.
+        cleaned.sub(%r{(\d)[Rr][Ee][Vv][A-Za-z0-9]+\s*\z}, '\1')
+      end
+
       def self.parse(string)
         # Strip .pdf extension if present (Pattern 3: File Extensions)
         cleaned = string.sub(/\.pdf$/i, "")
@@ -811,6 +1202,24 @@ module Pubid
         # No valid IEEE identifier pattern needs more than 1 space
         cleaned = cleaned.gsub(/\s+/, " ")
 
+        # A joint ISO-led publisher list is sometimes crawled with a stray slash
+        # (or slash+space) before the ISO stage code — "ISO/IEC/IEEE/ FDIS …" or
+        # "ISO/IEC/IEEE/FDIS …". Restore the space separator so the stage parses
+        # (bucket 7).
+        cleaned = cleaned.gsub(
+          %r{\b(ISO/IEC/IEEE|IEEE/ISO/IEC|IEEE/IEC/ISO|ISO/IEEE|IEC/IEEE|IEEE/IEC|ISO/IEC)/ ?(FDIS|FCD|CDV|DIS\d?|CD\d?|WD|PWI|NP)\b},
+          '\1 \2',
+        )
+
+        # Rewrite the rawbib revision-notation dialects (REVa/REVd/glued) into
+        # the canonical /R-<x> form before the suffix normalization below.
+        cleaned = normalize_revision_notation(cleaned)
+
+        # Normalize relaton's bespoke historical serialization (the spellings
+        # emitted by Relaton::Ieee::PubId::Id#to_s) into canonical pubid forms
+        # so `relaton-data-ieee` parses. See #normalize_relaton_suffixes.
+        cleaned = normalize_relaton_suffixes(cleaned)
+
         # NEW Session 171: CONSERVATIVE data quality fixes for TODO.IEEE-MUST-DO.txt
         # Only fix clear typos: space before dash + 4-digit year, OR dash + space + 4-digit year
         # Do NOT touch " - " (space-dash-space) which is valid formatting
@@ -859,10 +1268,21 @@ module Pubid
         # NEW: Convert IEC/IEEE space-separated to semicolon format
         # Pattern: "IEC 61523-3 First edition 2004-09; IEEE 1497" → already semicolon
         # Pattern: "IEC 62539 First Edition 2007-07 IEEE 930" → needs semicolon
-        # Match: IEC identifier (with edition) + space + IEEE identifier
-        # Be conservative: only convert if IEC has "First edition" or similar and followed by IEEE
+        # Pattern: "IEC 60076-21:2011 Edition 1.0 2011-12 IEEE Std C57.15" → needs semicolon (issue #202)
+        # Match: IEC identifier (with optional colon-year, optional "First"/numeric
+        # Edition + YYYY-MM) + space + IEEE identifier.
+        cleaned = cleaned.gsub(
+          /(IEC\s+\d+(?:-\d+)?(?::\d{4})?(?:\s+(?:First\s+)?[Ee]dition\s+\d+(?:\.\d+)?\s+\d{4}-\d{2})?)\s+(IEEE\s+Std\s+\S+|IEEE\s+\S+)/,
+          '\1; \2'
+        )
+
+        # Strip ":YYYY" from IEC numbers when an Edition clause follows — the
+        # IEEE parser's number rule doesn't accept the colon-year form, but
+        # the year is preserved in the "Edition N.M YYYY-MM" suffix.
+        # (issue #202)
         cleaned = cleaned.gsub(
-          /(IEC\s+\d+(?:-\d+)?(?:\s+First?\s+Edition\s+\d{4}-\d{2})?)\s+(IEEE\s+\S+)/, '\1; \2'
+          /^(IEC\s+\d+(?:-\d+)?):\d{4}(\s+(?:First\s+)?[Ee]dition\s+\d+(?:\.\d+)?\s+\d{4}-\d{2})/,
+          '\1\2'
         )
 
         # NEW Phase 1 (Session 141): Remove literal trademark symbol
@@ -1078,8 +1498,6 @@ module Pubid
         # Remove period after "Std": "IEEE Std." -> "IEEE Std"
         cleaned = cleaned.gsub(/\bStd\.\s+/, "Std ")
 
-        # Redline Suffix Removal: " - Redline" at end
-        cleaned = cleaned.gsub(/\s+-\s+Redline\b.*$/, "")
 
         # Title portion removal after year: "YYYY - IEEE Standard for..."
         cleaned = cleaned.gsub(