pubid 2.0.0.pre.alpha.9 → 2.0.0.pre.alpha.12

data/lib/pubid.rb

@@ -15,6 +15,8 @@ require "pubid/lutaml/no_store_registration"
 require "pubid/prefixes_support"
 
 module Pubid
+  autoload :Schema, "pubid/schema"
+  autoload :Conformance, "pubid/conformance"
   # Upper bound on the length of an identifier string accepted by any +parse+
   # entry point. Real-world standards identifiers are well under 200 characters;
   # this limit exists purely to keep pathological, attacker-controlled inputs
@@ -30,12 +32,28 @@ module Pubid
   INPUT_TOO_LONG_MESSAGE =
     "identifier string exceeds maximum length of #{MAX_INPUT_LENGTH} characters"
 
+  # Raised (as an ArgumentError) when +parse+ gets something that is not a
+  # String. Without this check the input reaches +.length+ and the caller gets a
+  # NoMethodError ("undefined method 'length' for nil"), a Ruby internal error
+  # that says nothing about the identifier contract. Every public +parse+ entry
+  # point rejects a non-String before it measures the length.
+  INPUT_NOT_A_STRING_MESSAGE = "identifier must be a String"
+
   # Registry for tracking all loaded flavors
   class Registry
     @flavors = {}
 
     class << self
-      attr_reader :flavors
+      # The registered flavors, always key-sorted and frozen. Sorted so
+      # no behavior can observe registration (= module load) order -
+      # hosts autoload flavor modules in any order, and order-dependent
+      # iteration once routed the same identifier to different flavors
+      # per process. Frozen so the raw table is not public mutable
+      # state; #register is the only mutator.
+      # @return [Hash{String => Module}]
+      def flavors
+        @flavors.sort.to_h.freeze
+      end
 
       # Register a flavor with the registry
       # @param name [String, Symbol] Flavor name (e.g., :iso, :iec)
@@ -44,6 +62,15 @@ module Pubid
         @flavors[name.to_s.downcase] = flavor_module
       end
 
+      # Remove a flavor registration. Tests that register probe flavors
+      # MUST unregister them: registry-driven specs enumerate every
+      # registered flavor, so a leaked probe fails them for everyone
+      # else in the process.
+      # @param name [String, Symbol] Flavor name
+      def unregister(name)
+        @flavors.delete(name.to_s.downcase)
+      end
+
       # Get all registered flavor names
       # @return [Array<String>] Array of flavor names
       def flavor_names
@@ -57,30 +84,62 @@ module Pubid
         @flavors[name.to_s.downcase]
       end
 
+      # The canonical registered name of a module that may be registered
+      # under aliases (CenCenelec registers both cen_cenelec and cen).
+      # Canonical = the name the module's own prefix table is keyed by,
+      # so the answer cannot depend on registration or iteration order.
+      # @param flavor_module [Module]
+      # @return [String, nil]
+      def canonical_name(flavor_module)
+        if flavor_module.singleton_class.method_defined?(:prefix_flavor_key)
+          key = flavor_module.prefix_flavor_key.to_s
+          return key if @flavors[key] == flavor_module
+        end
+        @flavors.key(flavor_module)
+      end
+
       # Check if a flavor is registered
       # @param name [String, Symbol] Flavor name
       # @return [Boolean]
       def registered?(name)
         @flavors.key?(name.to_s.downcase)
       end
+
+      # Parse an identifier without naming its flavor.
+      #
+      # The pubid 1.x entry point, kept because consumers still call it —
+      # isodoc's +std_docid_semantic_parse+ among them. It delegates to
+      # {Pubid.parse}; it is not a second implementation.
+      #
+      # @param string [String] the identifier string
+      # @return [Pubid::Identifier]
+      # @raise [Parslet::ParseFailed] when no flavor can parse it
+      def parse(string, **opts)
+        Pubid.parse(string, **opts)
+      end
+
+      # Build an identifier from its hash without naming its flavor.
+      # It delegates to {Pubid.from_hash}, as {parse} delegates to
+      # {Pubid.parse}.
+      #
+      # @param data [Hash] an identifier hash
+      # @return [Pubid::Identifier]
+      # @raise [Pubid::Errors::InvalidInputError] see {Pubid.from_hash}
+      def from_hash(data)
+        Pubid.from_hash(data)
+      end
     end
   end
 
-  # Canonical joint / co-publication leading tokens, injected symmetrically into
-  # every participating flavor's +prefixes+ (see {PrefixesSupport}). Single
-  # source of truth: editing an entry here updates both sides at once
-  # (e.g. +"ISO/IEC"+ appears in +Pubid::Iso.prefixes+ and +Pubid::Iec.prefixes+),
-  # so co-publication symmetry can never drift. Keyed by
+  # Canonical joint / co-publication leading tokens, sourced at boot from
+  # schema/core/joint_prefixes.yaml - the single source of truth. Injected
+  # symmetrically into every participating flavor's +prefixes+ (see
+  # {PrefixesSupport}) so co-publication symmetry can never drift. Keyed by
   # {PrefixesSupport#prefix_flavor_key}.
-  JOINT_PREFIXES = {
-    iso: ["ISO/IEC", "IEC/ISO", "ISO/IEC/IEEE"],
-    iec: ["ISO/IEC", "IEC/ISO", "ISO/IEC/IEEE"],
-    ieee: ["ISO/IEC/IEEE"],
-    ansi: ["ANSI/ASHRAE", "ANSI/AMCA"],
-    ashrae: ["ANSI/ASHRAE"],
-    amca: ["ANSI/AMCA"],
-  }.freeze
+  JOINT_PREFIXES = Schema::Loader.joint_prefixes_map
+    .transform_keys(&:to_sym).freeze
 
+  autoload :Errors, "pubid/errors"
   autoload :Parser, "pubid/parser"
   autoload :Components, "pubid/components"
   autoload :BundledIdentifier, "pubid/bundled_identifier"
@@ -119,6 +178,7 @@ module Pubid
   autoload :Csa, "pubid/csa"
   autoload :Ecma, "pubid/ecma"
   autoload :Etsi, "pubid/etsi"
+  autoload :Evs, "pubid/evs"
   autoload :Gost, "pubid/gost"
   autoload :Gb, "pubid/gb"
   autoload :Idf, "pubid/idf"
@@ -161,17 +221,47 @@ module Pubid
 
   # Unified parse entry point with auto-detection
   #
+  # A human-readable string is routed to its owning flavor by leading prefix
+  # token, using the {prefix_flavors} table. This restores the pubid 1.x
+  # +Pubid::Registry.parse+ capability, which isodoc still calls; before it,
+  # this method raised +No flavor specified+ for every human-readable string,
+  # so a reference whose flavor the caller did not already know could not be
+  # parsed at all.
+  #
   # @param string [String] The identifier string to parse
   # @param format [Symbol] :auto, :human, :mr_string, or :urn
   # @return [Identifier] The parsed identifier
+  # @raise [ArgumentError] for a non-String, or one over {MAX_INPUT_LENGTH}
+  # @raise [Parslet::ParseFailed] when no flavor can parse the string — the
+  #   class every flavor's own +parse+ raises (see the parse-failure contract)
   def self.parse(string, format: :auto)
-    raise ArgumentError, INPUT_TOO_LONG_MESSAGE if string.length > MAX_INPUT_LENGTH
+    unless string.is_a?(String)
+      raise Pubid::Errors::InvalidInputError,
+            Pubid::INPUT_NOT_A_STRING_MESSAGE
+    end
+    if string.length > MAX_INPUT_LENGTH
+      raise Pubid::Errors::InvalidInputError,
+            Pubid::INPUT_TOO_LONG_MESSAGE
+    end
 
     format = FormatDetector.detect(string) if format == :auto
 
     case format
     when :mr_string
-      Parsers::MrString.parse(string)
+      # The MR detector is a shape heuristic (`/\A[A-Z]{2,}[.-]/`), and some
+      # human-readable identifiers have that shape — "ITU-T G.711" is the
+      # standing example. Fall through to prefix routing when the MR parse
+      # fails, rather than reporting an MR error for a string that was never
+      # an MR string. A genuine MR failure still surfaces, from the second
+      # attempt, as Parslet::ParseFailed.
+      #
+      # Narrow, for the reason spelled out on {parse_by_prefix}: only a parse
+      # failure means "wrong shape". Anything else is a defect and propagates.
+      begin
+        Parsers::MrString.parse(string)
+      rescue Parslet::ParseFailed
+        parse_by_prefix(string)
+      end
     when :urn
       eager_load_flavors!
       flavor = detect_flavor_from_urn(string)
@@ -182,18 +272,239 @@ module Pubid
       end
 
       urn_parser = flavor_module.const_get(:UrnParser)
-      urn_parser.parse(string) else
-      # Default to MR string parser for MR format, human-readable otherwise
-      # The MR string parser converts to human-readable and delegates to flavor.parse
-      raise ArgumentError,
-            "No flavor specified. Use Pubid::Iso.parse() or another flavor-specific parse method."
+      urn_parser.parse(string)
+    else
+      parse_by_prefix(string)
+    end
+  end
+
+  # Build an identifier from its serialized hash, without naming its flavor —
+  # the hash counterpart of {parse}.
+  #
+  # The hash is what {Identifier#to_hash} emits, and what a relaton-data
+  # +index-v2.yaml+ row stores under +:id+. Its +_type+ key
+  # ("pubid:<flavor>:<type>") names the concrete class, so no prefix guessing
+  # is necessary. The class is resolved by {TypeResolver}, and the class's own
+  # +from_hash+ does the rest, nested identifiers of other flavors included.
+  #
+  # @example
+  #   Pubid.from_hash("_type" => "pubid:iso:international-standard",
+  #                   "number" => "9001", "year" => "2015").to_s
+  #   # => "ISO 9001:2015"
+  #
+  # @param data [Hash] an identifier hash. String and Symbol keys are both
+  #   accepted.
+  # @return [Identifier] an instance of the class that +_type+ names
+  # @raise [Pubid::Errors::InvalidInputError] when +data+ is not a Hash, has no
+  #   +_type+, or has a +_type+ that no registered flavor defines. These are
+  #   the only failures it translates: once +_type+ names a class, an error
+  #   from that class's own +from_hash+ (a malformed field, or a bad +_type+
+  #   in a NESTED hash) propagates unchanged, so a flavor defect is not
+  #   disguised as bad input.
+  def self.from_hash(data)
+    unless data.is_a?(Hash)
+      raise Pubid::Errors::InvalidInputError,
+            "identifier data must be a Hash, got #{data.class}"
+    end
+
+    identifier_class_for(data).from_hash(data)
+  end
+
+  # The concrete class that the +_type+ of +data+ names.
+  #
+  # @param data [Hash] an identifier hash
+  # @return [Class<Identifier>]
+  # @raise [Pubid::Errors::InvalidInputError] no +_type+, or an unknown one
+  # @api private
+  def self.identifier_class_for(data)
+    type = data["_type"] || data[:_type]
+    unless type
+      raise Pubid::Errors::InvalidInputError,
+            "identifier hash has no _type key: #{data.inspect[0, 200]}"
+    end
+
+    TypeResolver.resolve(type) ||
+      raise(Pubid::Errors::InvalidInputError,
+            "unknown identifier _type: #{type.inspect}")
+  end
+  private_class_method :identifier_class_for
+
+  # Route a human-readable identifier to its flavor by leading prefix token.
+  #
+  # Three passes, widening each time — the pubid 1.x +Registry.parse+ shape:
+  #
+  #   1. flavors owning the longest matching prefix ("ISO/IEC" before "ISO")
+  #   2. every other flavor, in registry order
+  #
+  # A longest-match-first order matters because prefixes nest: "ISO/IEC 2131"
+  # must not be handed to the flavor that merely owns "ISO".
+  #
+  # @param string [String] a human-readable identifier
+  # @return [Identifier]
+  # @raise [Parslet::ParseFailed] when no flavor accepts the string
+  # @api private
+  def self.parse_by_prefix(string)
+    eager_load_flavors!
+
+    first_error = nil
+    fallback = nil
+
+    prefix_owners = candidates_by_prefix(string)
+    (prefix_owners + other_candidates(prefix_owners)).each do |mod|
+      begin
+        parsed = mod.parse(string)
+      rescue Parslet::ParseFailed => e
+        # ONLY a parse failure means "not this flavor's identifier". Every
+        # other exception is a defect in that flavor and must propagate.
+        #
+        # A blanket `rescue StandardError` here would swallow exactly the
+        # `ArgumentError: unknown attribute…` that the constructor contract on
+        # `Identifier` now raises — the error that turned three silent
+        # data-loss bugs (CIE `s_prefix`, IEEE `adopted_identifier`,
+        # IEEE `csa_identifier`) into visible ones. Catching it one layer up
+        # would hand it straight back its disguise: a real builder bug in any
+        # flavor would read as "this string is not an identifier".
+        first_error ||= e
+        next
+      end
+
+      # An exact round trip is the strongest evidence the string belongs to
+      # this flavor. It matters because some grammars accept anything by
+      # design — adobe and iana take any slug — so without it the first such
+      # flavor in registry order claims every unclaimed string and invents
+      # structure that was not in the input. Same test isodoc applies before
+      # it will annotate a parsed id.
+      return parsed if parsed.to_s == string
+
+      # A non-exact parse is still usable — several flavors legitimately
+      # normalise on render, so the round trip is a preference, not a
+      # requirement (OGC prints `06-121r9` for `OGC 06-121r9`, and 32 ASHRAE /
+      # ASME / CSA / IEEE identifiers route only this way). What it must NOT
+      # come from is a flavor that accepts anything: that is how `iec.60050`
+      # came back as `IANA iec.60050`. Reporting that no flavor could be
+      # determined beats inventing one.
+      fallback ||= parsed unless accepts_anything?(mod)
+    end
+
+    return fallback if fallback
+    raise first_error if first_error
+
+    raise Parslet::ParseFailed,
+          "no registered flavor could parse #{string.inspect}"
+  end
+  private_class_method :parse_by_prefix
+
+  # Flavor modules that own a prefix +string+ starts with, longest prefix
+  # first. Deduplicated by module identity, so an alias registration is not
+  # tried twice.
+  #
+  # These are the flavors with an actual claim on the string; a parse from one
+  # of them is trusted even when the flavor normalises on render.
+  #
+  # @param string [String] a human-readable identifier
+  # @return [Array<Module>]
+  # @api private
+  def self.candidates_by_prefix(string)
+    ordered = []
+    routing_table.each do |prefix, modules|
+      ordered.concat(modules) if prefix_match?(string, prefix)
+    end
+    ordered.compact.uniq
+  end
+  private_class_method :candidates_by_prefix
+
+  # Every other registered flavor, for the exhaustive second pass.
+  #
+  # @param exclude [Array<Module>] modules already tried
+  # @return [Array<Module>]
+  # @api private
+  def self.other_candidates(exclude)
+    ordered = []
+    each_prefix_flavor_module { |mod| ordered << mod }
+    ordered.compact.uniq - exclude
+  end
+  private_class_method :other_candidates
+
+  # A meaningless string no real identifier scheme should claim. Used to find
+  # the flavors whose grammar accepts anything.
+  ROUTING_PROBE = "zzqx-nonsense-9714"
+  private_constant :ROUTING_PROBE
+
+  # True when +mod+'s grammar accepts an arbitrary slug.
+  #
+  # `adobe` and `iana` do, by design — their identifiers genuinely are free
+  # text. That makes them unsafe as a fallback: whichever came first in
+  # registry order would claim every string no other flavor matched, and
+  # answer with structure the input never had.
+  #
+  # Detected by probing rather than by a hardcoded list, so a flavor that
+  # acquires (or loses) an anything-goes grammar is classified correctly
+  # without anyone remembering to edit a constant. Computed once per module.
+  #
+  # @param mod [Module] a flavor module
+  # @return [Boolean]
+  # @api private
+  def self.accepts_anything?(mod)
+    @accepts_anything ||= {}
+    return @accepts_anything[mod] if @accepts_anything.key?(mod)
+
+    @accepts_anything[mod] =
+      begin
+        mod.parse(ROUTING_PROBE)
+        true
+      rescue StandardError, Parslet::ParseFailed
+        false
+      end
+  end
+  private_class_method :accepts_anything?
+
+  # Prefix => flavor modules, longest prefix first, built once.
+  #
+  # {prefix_flavors} rebuilds its index on every call, iterating every
+  # registered flavor; routing consulted it twice per parse. The registry is
+  # static once +eager_load_flavors!+ has run, so the table is memoised here
+  # rather than making the public method's cost implicit.
+  #
+  # The memo is never invalidated, which is safe for every flavor shipped in
+  # this gem — they are all +Pubid+-namespace constants that
+  # +eager_load_flavors!+ loads before the first parse. A flavor registered
+  # from OUTSIDE that namespace after the first {parse} call would miss its
+  # prefix priority (it is still tried in the exhaustive second pass, so it is
+  # reachable, just not preferred). Bust the memo in +Registry.register+ if
+  # out-of-namespace flavors ever become a supported extension point.
+  #
+  # @return [Hash{String => Array<Module>}]
+  # @api private
+  def self.routing_table
+    @routing_table ||= begin
+      index = prefix_flavors
+      index.keys
+           .sort_by { |prefix| -prefix.length }
+           .to_h { |prefix| [prefix, index[prefix].map { |key| Registry.get(key) }] }
     end
   end
+  private_class_method :routing_table
+
+  # True when +string+ starts with +prefix+ at a token boundary, so "ISO" does
+  # not claim "ISOFIX" and "BS" does not claim "BSI".
+  # @api private
+  def self.prefix_match?(string, prefix)
+    return false unless string.start_with?(prefix)
+
+    rest = string[prefix.length]
+    rest.nil? || !/[A-Za-z0-9]/.match?(rest)
+  end
+  private_class_method :prefix_match?
 
   def self.detect_flavor_from_urn(urn)
     # urn:iso:std:... → "iso"
     # urn:iec:std:... → "iec"
+    # urn:mrn:iala:pub:... → "iala": MRN URNs carry an assigning authority
+    # where ISO-style URNs carry their namespace, so routing follows the
+    # authority (works for any future MRN assignee, not only IALA).
     parts = urn.downcase.split(":")
+    return parts[2] if parts[1] == "mrn"
+
     parts[1] # The namespace part after "urn"
   end
 
@@ -267,4 +578,71 @@ module Pubid
       yield mod
     end
   end
+
+  # Typed-stage lookup by abbreviation, across flavors.
+  #
+  # Each flavor module memoises +all_typed_stages+ over its own +Identifiers+
+  # namespace, so an abbreviation only one flavor uses is invisible from any
+  # other: +Pubid::Iso.locate_stage("ADTS")+ is nil while
+  # +Pubid::Iec.locate_stage("ADTS")+ finds it. A consumer that holds an
+  # abbreviation but not the owning flavor had nowhere to ask.
+  #
+  # @param abbr [String, Symbol] the stage abbreviation (case-insensitive)
+  # @param flavor [Symbol, String, nil] search this flavor only; nil searches
+  #   every registered flavor and returns the first match
+  # @return [Object, nil] the flavor's typed-stage object, or nil. The class is
+  #   flavor-specific: most return {Pubid::Components::TypedStage}, but IEEE
+  #   has its own +Ieee::Components::TypedStage+, which is not a subclass.
+  # @raise [ArgumentError] if +flavor+ names a flavor that is not registered
+  def self.locate_stage(abbr, flavor: nil)
+    return locate_stage_in(Registry.get(flavor), abbr, flavor) if flavor
+
+    each_stage_flavor_module do |mod|
+      stage = mod.locate_stage(abbr)
+      return stage if stage
+    end
+    nil
+  end
+
+  # Every flavor that knows +abbr+ — the diagnostic companion to
+  # {locate_stage}, for telling a consumer which module owns a stage.
+  #
+  # @param abbr [String, Symbol] the stage abbreviation (case-insensitive)
+  # @return [Array<Symbol>] registered flavor keys, sorted
+  def self.locate_stage_flavors(abbr)
+    found = []
+    each_stage_flavor_module do |mod, flavor_name|
+      found << flavor_name.to_sym if mod.locate_stage(abbr)
+    end
+    found.sort
+  end
+
+  # Yields each unique flavor module that implements +locate_stage+, exactly
+  # once. Three registered flavors (adobe, easc, gost) define no typed stages
+  # at all, so a cross-flavor sweep must skip them rather than raise on the
+  # first one it reaches; alias registrations are collapsed by module identity,
+  # as in {each_prefix_flavor_module}.
+  #
+  # @yieldparam mod [Module] a flavor module
+  # @yieldparam flavor_name [String] the registry key it was reached under
+  def self.each_stage_flavor_module
+    eager_load_flavors!
+    seen = {}
+    Registry.flavor_names.each do |flavor_name|
+      mod = Registry.get(flavor_name)
+      next if seen.key?(mod) || !mod.respond_to?(:locate_stage)
+
+      seen[mod] = true
+      yield mod, flavor_name
+    end
+  end
+
+  # @api private
+  def self.locate_stage_in(mod, abbr, flavor)
+    raise ArgumentError, "unknown flavor: #{flavor.inspect}" unless mod
+    return nil unless mod.respond_to?(:locate_stage)
+
+    mod.locate_stage(abbr)
+  end
+  private_class_method :locate_stage_in
 end

data/lib/tasks/conformance.rake

@@ -0,0 +1,34 @@
+# frozen_string_literal: true
+
+require "yaml"
+require "fileutils"
+
+namespace :conformance do
+  desc "Generate the neutral corpus from ground-truth fixtures"
+  task :generate, [:flavor] do |_t, args|
+    require "pubid"
+    require "pubid/conformance"
+    flavor = args[:flavor] || "iso"
+    tests_repo = ENV.fetch("PUBID_TESTSUITE_PATH",
+                           File.expand_path("../../../pubid-testsuite", 
+                                            __dir__))
+    output_dir = File.join(tests_repo, "tests", flavor)
+    FileUtils.mkdir_p(output_dir)
+    results = Pubid::Conformance::Generator.new(flavor)
+      .generate(output_dir: output_dir)
+    stats = results
+    puts "#{flavor}: #{stats[:files]} files, #{stats[:cases]} cases, " \
+         "#{stats[:aliases]} aliases, #{stats[:duplicates]} duplicates, " \
+         "#{stats[:reclassify]} reclassify, #{stats[:negative]} negative, " \
+         "#{stats[:debt]} debt, #{stats[:roundtrip_failures]} rt-failures"
+  end
+
+  desc "Run the neutral corpus against this implementation"
+  task :run do
+    require "pubid"
+    require "pubid/conformance"
+    failures = Pubid::Conformance::Runner.new.run.flatten
+    puts "TOTAL failures: #{failures.size}"
+    exit(failures.empty? ? 0 : 1)
+  end
+end

data/lib/tasks/docs.rake

@@ -1,24 +1,26 @@
-# frozen_string_literal
-
-require_relative "../../lib/pubid/core/pattern_doc_generator"
+# frozen_string_literal: true
 
 namespace :docs do
   desc "Generate identifier pattern reference docs"
   task :patterns do
+    require "pubid"
+
     output_dir = File.join(__dir__, "..", "..", "docs", "identifier-patterns")
     FileUtils.mkdir_p(output_dir)
 
-    flavors = Dir[File.join(__dir__, "..", "..", "lib", "pubid",
-                            "*")].select do |d|
-      File.directory?(d)
-    end
-      .map do |d|
-      File.basename(d)
-    end
-      .reject do |f|
-      %w[core components rendering
-         parser serializable utils].include?(f)
-    end
+    lib_root = File.join(__dir__, "..", "..", "lib", "pubid")
+    # The registry is the authoritative flavor list. This replaces a
+    # filesystem glob over lib/pubid/* directories, whose reject list kept
+    # going stale and swept internal modules (parsers, renderers, lutaml,
+    # builder, utils, ...) in as "flavors". The module-derived directory
+    # (CenCenelec -> cen_cenelec, W3c -> w3c) must exist on disk.
+    Pubid.eager_load_flavors!
+    flavors = Pubid::Registry.flavor_names.filter_map do |name|
+      mod = Pubid::Registry.get(name)
+      dir = mod.name.split("::").last
+               .gsub(/([a-z\d])([A-Z])/, '\1_\2').downcase
+      dir if File.directory?(File.join(lib_root, dir))
+    end.uniq.sort
 
     flavors.each do |flavor|
       puts "Generating docs for #{flavor}..."

data/lib/tasks/schema.rake

@@ -0,0 +1,88 @@
+# frozen_string_literal: true
+
+require "json"
+require "yaml"
+require "fileutils"
+
+# Extracts scheme declarations from the live model into schema/{flavor}.yaml.
+# Seed tooling for TODO.restructure/01-02; TODO.restructure/07 inverts the
+# direction (the identifier classes read these files instead of carrying the
+# data in constants and class methods).
+class SchemeExtractor
+  def initialize(flavor)
+    require "pubid"
+    require "pubid/export"
+
+    @flavor = flavor.to_s
+    Pubid.eager_load_flavors!
+    @flavor_module = Pubid::Registry.get(@flavor)
+    raise ArgumentError, "Unknown flavor: #{@flavor}" unless @flavor_module
+
+    exporter = Pubid::Export::FlavorExporter.new(@flavor.to_sym)
+    @export = JSON.parse(JSON.generate(exporter.export.to_h))
+  end
+
+  def payload
+    types = @export["identifier_types"].map { |type| type_entry(type) }
+    {
+      "flavor" => @flavor,
+      "schema_version" => "1.0.0",
+      "prefixes" => own_prefixes,
+      "identifier_types" => types,
+    }
+  end
+
+  def write(output_dir)
+    FileUtils.mkdir_p(output_dir)
+    path = File.join(output_dir, "#{@flavor}.yaml")
+    File.write(path, header + YAML.dump(payload))
+    puts "Extracted #{payload['identifier_types'].size} types to #{path}"
+  end
+
+  private
+
+  def own_prefixes
+    if @flavor_module.const_defined?(:PREFIXES, false)
+      @flavor_module.const_get(:PREFIXES).to_a
+    else
+      @flavor_module.prefixes
+    end
+  end
+
+  def type_entry(type)
+    stages = (type["typed_stages"] || []).map { |stage| stage_entry(stage) }
+    {
+      "key" => type["key"],
+      "title" => type["title"],
+      "short" => type["short"],
+      "abbr" => type["abbr"],
+      "typed_stages" => stages,
+    }.compact
+  end
+
+  def stage_entry(stage)
+    {
+      "stage_code" => stage["stage_code"],
+      "type_code" => stage["type_code"],
+      "abbr" => stage["abbr"],
+      "name" => stage["name"],
+      "harmonized_stages" => stage["harmonized_stages"],
+    }.compact
+  end
+
+  def header
+    [
+      "# Generated by rake schema:extract[#{@flavor}] - SSOT seed.",
+      "# Joint prefixes live in schema/core/joint_prefixes.yaml.\n",
+    ].join("\n")
+  end
+end
+
+namespace :schema do
+  desc "Extract scheme declarations into schema/{flavor}.yaml"
+  task :extract, [:flavor] do |_t, args|
+    flavor = args[:flavor] || "iso"
+    output_dir = File.expand_path("../../schema", __dir__)
+    SchemeExtractor.new(flavor).write(output_dir)
+  end
+end

data/schema/core/joint_prefixes.yaml

@@ -0,0 +1,24 @@
+# Canonical joint / co-publication leading tokens, mirroring
+# Pubid::JOINT_PREFIXES (lib/pubid.rb). Keyed by flavor; merged into each
+# participating flavor's prefixes at load time so co-publication symmetry can
+# never drift (an ISO/IEC token appears in both the iso and iec lists).
+# TODO.restructure/05 makes this file the only home for joint tokens.
+schema_version: 1.0.0
+joint_prefixes:
+  iso:
+    - ISO/IEC
+    - IEC/ISO
+    - ISO/IEC/IEEE
+  iec:
+    - ISO/IEC
+    - IEC/ISO
+    - ISO/IEC/IEEE
+  ieee:
+    - ISO/IEC/IEEE
+  ansi:
+    - ANSI/ASHRAE
+    - ANSI/AMCA
+  ashrae:
+    - ANSI/ASHRAE
+  amca:
+    - ANSI/AMCA