pubid 2.0.0.pre.alpha.22 → 2.0.0.pre.alpha.24

data/lib/pubid/ieee/parser.rb

@@ -1374,108 +1374,6 @@ module Pubid
         cleaned.sub(%r{(\d)[Rr][Ee][Vv][A-Za-z0-9]+\s*\z}, '\1')
       end
 
-      # Rewrite the mechanical spellings of the joint ISO-stage forms that no
-      # grammar branch reaches (pubid#216 residuals). Two safety rules keep
-      # this from stealing inputs other rules already parse:
-      #   1. only BROKEN separators fire — underscore/space glue between
-      #      number, draft and stage, a slash-SPACE before the stage, a
-      #      dash/underscore month after it. A plain "/FDIS" tail is
-      #      natively accepted by ieee_p_identifier (fdraft) and the
-      #      stage-LAST embedded rule, so it is never rewritten.
-      #   2. the number carries at most ONE part — joint_development_iso_format
-      #      reads a single optional part, while ieee_p_identifier takes
-      #      multi-part numbers ("P62271-37-013"); a rewrite that moved a
-      #      multi-part number onto the joint rule would break it.
-      def self.normalize_joint_stage_spellings(cleaned)
-        # The separator (slash-optional-space or a space) is INSIDE the
-        # capture so a rewrite re-emits it — an uncaptured separator was
-        # silently eaten, gluing the publisher to the stage.
-        pubs = %r{((?:ISO/IEC/IEEE|IEEE/ISO/IEC|IEEE/IEC/ISO|ISO/IEEE|IEC/IEEE|IEEE/IEC|ISO/IEC|IEEE)(?:/ ?| ))}
-        stage = /(FDIS|FCD|CDV|DIS\d?|CD\d?|WD|PWI|NP)/
-        num = /(P?\d+(?:[.-]\d+)?)/
-        # What the joint grammar can finish reading AFTER the stage: the end
-        # of the string, a (comma-)month-year date, or a dash-year[-month].
-        # Captured and re-emitted so a rewrite never drops the date.
-        tail = /(,? (?:Jan|Feb|Mar|Apr|May|Jun|Jul|Aug|Sep|Sept|Oct|Nov|Dec)[a-z]* \d{4}|\z|-\d{4}(?:-\d\d)?)/
-
-        # Draft glued to the number with an underscore: "P15289_D3" →
-        # "P15289/D3" (a digit must precede the underscore; "_FDIS" is the
-        # stage, not a draft, and is already slash-rewritten earlier).
-        cleaned = cleaned.gsub(%r{(\d)_D(\d)}, '\1/D\2')
-
-        # Crawl typo: a comma with no space before a month (",March 2021").
-        # In these identifiers a comma is always a date separator, so
-        # spacing it is safe.
-        cleaned = cleaned.sub(/,(?=[A-Za-z])/, ", ")
-
-        # Stage AFTER the draft, both glued with underscores (the early
-        # stage gsub already turned "_CD" into "/CD"):
-        # "…P24748-3/D3/FDIS, April 2020 (E)" → "…FDIS P24748-3/D3, …"
-        cleaned = cleaned.sub(
-          %r{\A#{pubs}#{num}/(D[\d.]+)/ ?#{stage}\b},
-          '\1\4 \2/\3',
-        )
-
-        # Slash-SPACE before the stage ("…P16085/ FDIS, August 2020" —
-        # ieee_p's fdraft has no inner space, so this spelling has no other
-        # home): → "…FDIS P16085, August 2020"
-        cleaned = cleaned.sub(
-          %r{\A#{pubs}#{num}/ #{stage}#{tail}},
-          '\1\3 \2\4',
-        )
-
-        # A space-year after a slash-stage ("…P26511.2/FDIS 2018" — no rule
-        # reads "/FDIS 2018"): → "…FDIS P26511.2-2018" (dash-year clause).
-        cleaned = cleaned.sub(
-          %r{\A#{pubs}#{num}/#{stage} (\d{4})\b},
-          '\1\3 \2-\4',
-        )
-
-        # Stage separated from the number by a SPACE ("IEC/IEEE P63113 CD4,
-        # April 2019"): → "IEC/IEEE CD4 P63113, April 2019". The tail guard
-        # keeps "…FDIS March 2019" (space-month-year, readable after the
-        # swap) while leaving prose suffixes alone.
-        cleaned = cleaned.sub(
-          %r{\A#{pubs}#{num} #{stage}#{tail}},
-          '\1\3 \2\4',
-        )
-
-        # The same space-stage with a space-DRAFT chained after it
-        # ("IEC/IEEE P60980-344 CDV D1, June 2019"): → stage-first with a
-        # slash draft ("…CDV P60980-344/D1, June 2019").
-        cleaned = cleaned.sub(
-          %r{\A#{pubs}#{num} #{stage} (D\d+)\b},
-          '\1\3 \2/\4',
-        )
-
-        # Slash-stage carrying a dash-date ("…P15288/CD2-2013-09 …" — the
-        # dash-year tail is what the joint grammar reads after the number, so
-        # this cannot steal the plain "/FDIS, June 2021" spellings
-        # ieee_p_identifier finishes): → "…CD2 P15288-2013-09 …". The date
-        # must be a plausible year — a 4-digit monthcode (YYMM, e.g.
-        # "…P24748-4/DIS-1404" = April 2014) is a DRAFT designator, not a
-        # dash-date, and rewriting it onto the number breaks the parse.
-        cleaned = cleaned.sub(
-          %r{\A#{pubs}#{num}/ ?#{stage}(-(?:19|20)\d\d(?:-\d\d)?)},
-          '\1\3 \2\4',
-        )
-
-        # Draft separated from a stage-first number by a SPACE:
-        # "…DIS P11073-10418 D13, January 2011" → "…DIS P11073-10418/D13, …"
-        cleaned = cleaned.sub(
-          %r{\A#{pubs}#{stage} #{num}/? ?(D\d+)\b},
-          '\1\2 \3/\4',
-        )
-
-        # NOT rewritten: a month glued to the stage with a dash/underscore
-        # ("…/FDIS_Dec 2012", "…/FDIS-Dec 2015"). Even routed onto the joint
-        # grammar, the joint render of a month+draft is itself not yet
-        # idempotent, and the bare-IEEE fdraft path drops the stage and the
-        # date from to_hash — rewriting would trade a visible parse failure
-        # for silent garbage. Deferred with the pubid#216 semantic residue.
-        cleaned
-      end
-
       # Pre-parse ingestion normalizations (R2): every parse path —
       # parslet and PG artifact alike — feeds the grammar the same
       # normalized string.
@@ -1488,23 +1386,10 @@ module Pubid
         # which standards are actually IEEE-related.
         # ISO-only standards are still filtered as they have separate handling.
 
-        # Pattern 3: Replace underscore before ISO stage codes with slash
-        # These are joint development drafts that use underscore instead of slash
-        cleaned = cleaned.gsub(/_(FDIS|CDV|CD|DIS|WD|PWI|NP)/, '/\1')
-
         # NEW: Normalize multiple spaces to single space
         # 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)
@@ -1520,17 +1405,9 @@ module Pubid
         cleaned = cleaned.gsub(/(\d)\s+-(\d{4})\b/, '\1-\2')  # "C37.101 -2006" → "C37.101-2006"
         cleaned = cleaned.gsub(/(\d)-\s+(\d{4})\b/, '\1-\2')  # "C62.35- 2010" → "C62.35-2010"
 
-        # NEW Session 171: HTML entity for en dash (&#x2013;)
-        # ONLY convert if not already followed by a dash (avoid creating --)
-        cleaned = cleaned.gsub(/&#x2013;(?!-)/, "-")  # En dash → regular hyphen (if not followed by dash)
-        cleaned = cleaned.gsub("&#x2013;-", "-")      # En-dash-dash → single dash
-
         # NEW Session 171: Remove wrong ! prefix
         cleaned = cleaned.gsub(/^!IEEE /, "IEEE ")
 
-        # NEW Session 171: Fix "IEEE/ ASTM" spacing (extra space after slash)
-        cleaned = cleaned.gsub("IEEE/ ASTM", "IEEE/ASTM")
-
         # NEW Phase 1: Handle HTML entities comprehensively
         cleaned = cleaned.gsub("&#x2122;", "™") # Trademark symbol
         cleaned = cleaned.gsub("&#x2019;", "'") # Smart apostrophe
@@ -1559,11 +1436,6 @@ module Pubid
           /\b(Jan|Feb|Mar|Apr|May|Jun|Jul|Aug|Sep|Sept|Oct|Nov|Dec)(\d{4})\b/, '\1 \2'
         )
 
-        # Rewrite the joint ISO-stage glued spellings (pubid#216) AFTER the
-        # month unglue above, so "Dec2015" is already spaced when the
-        # dash-month rewrite looks for it.
-        cleaned = normalize_joint_stage_spellings(cleaned)
-
         # 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

data/lib/pubid/version.rb

@@ -1,5 +1,5 @@
 # frozen_string_literal: true
 
 module Pubid
-  VERSION = "2.0.0.pre.alpha.22"
+  VERSION = "2.0.0.pre.alpha.24"
 end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: pubid
 version: !ruby/object:Gem::Version
-  version: 2.0.0.pre.alpha.22
+  version: 2.0.0.pre.alpha.24
 platform: ruby
 authors:
 - Ribose Inc.