iev 0.4.6 → 0.4.7

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: f5d3397238922342cdb3fb6b5adcf8fa969d2b3d6510742485d7d2a20b1d1a12
-  data.tar.gz: 6a82e64e1201a9a2f04cff66d024de1b23643030ce6e37f26ef09bb2ad334b10
+  metadata.gz: a965bb1653227f1b4ba8ebc27e9f3a67950d359cec2f809b46d4e0edebec18e9
+  data.tar.gz: 5414e80f45caf460970d75052dc7bd7ec931d697f267de0ea2df6abd6a2209ba
 SHA512:
-  metadata.gz: be3b984bd46c791580a73644e88dccca4b98b708ce0e28c50f72e822abb609573455201b633d0e4d791b9271cdecb20a555b11cc674dda46bde90fcf7eaccb74
-  data.tar.gz: e56796a853c58aae578dd5c4fe3fbaf3e18115e1a08e22feb7d5f876d9521dee4d27fb734fb05baa12e297b1dc9ac89b67197cad0b51695b2224c3fdb622f3a5
+  metadata.gz: 805143019d38cf0ed15e6526b8b727e681e8e3361b25d3c58ea71f6b2b4cf66eb3dda9882bcb8f038317bc4f1bd84554f6abdd14e4575774c2572a08b45c8865
+  data.tar.gz: 204a790f8cc565859a3abb9f64b137b235a559bbb411f52fee15989b467c5860ff2fe3073c65495f5c28e1e01a251375c533139cac07eaefb7341b5016669d1a

data/.rubocop_todo.yml

@@ -1,6 +1,6 @@
 # This configuration was generated by
 # `rubocop --auto-gen-config`
-# on 2026-06-13 04:18:23 UTC using RuboCop version 1.86.1.
+# on 2026-06-18 12:23:48 UTC using RuboCop version 1.86.1.
 # The point is for the user to remove these configuration records
 # one by one as the offenses are removed from the code base.
 # Note that changes in the inspected code, or installation of new
@@ -11,37 +11,7 @@ Gemspec/RequiredRubyVersion:
   Exclude:
     - 'iev.gemspec'
 
-# Offense count: 2
-# This cop supports safe autocorrection (--autocorrect).
-# Configuration parameters: EnforcedStyle, IndentationWidth.
-# SupportedStyles: with_first_argument, with_fixed_indentation
-Layout/ArgumentAlignment:
-  Exclude:
-    - 'spec/iev/exporter_spec.rb'
-
-# Offense count: 1
-# This cop supports safe autocorrection (--autocorrect).
-# Configuration parameters: EnforcedStyleAlignWith.
-# SupportedStylesAlignWith: either, start_of_block, start_of_line
-Layout/BlockAlignment:
-  Exclude:
-    - 'spec/iev/exporter_spec.rb'
-
-# Offense count: 1
-# This cop supports safe autocorrection (--autocorrect).
-Layout/BlockEndNewline:
-  Exclude:
-    - 'spec/iev/exporter_spec.rb'
-
-# Offense count: 2
-# This cop supports safe autocorrection (--autocorrect).
-# Configuration parameters: Width, EnforcedStyleAlignWith, AllowedPatterns.
-# SupportedStylesAlignWith: start_of_line, relative_to_receiver
-Layout/IndentationWidth:
-  Exclude:
-    - 'spec/iev/exporter_spec.rb'
-
-# Offense count: 65
+# Offense count: 64
 # This cop supports safe autocorrection (--autocorrect).
 # Configuration parameters: Max, AllowHeredoc, AllowURI, AllowQualifiedName, URISchemes, AllowRBSInlineAnnotation, AllowCopDirectives, AllowedPatterns, SplitStrings.
 # URISchemes: http, https
@@ -65,7 +35,7 @@ Lint/ShadowedException:
   Exclude:
     - 'lib/iev/source_parser.rb'
 
-# Offense count: 26
+# Offense count: 27
 # Configuration parameters: AllowedMethods, AllowedPatterns, CountRepeatedAttributes, Max.
 Metrics/AbcSize:
   Exclude:
@@ -128,15 +98,3 @@ Naming/MethodParameterName:
 Naming/VariableNumber:
   Exclude:
     - 'spec/iev/exporter_spec.rb'
-
-# Offense count: 5
-# This cop supports safe autocorrection (--autocorrect).
-# Configuration parameters: EnforcedStyle, ProceduralMethods, FunctionalMethods, AllowedMethods, AllowedPatterns, AllowBracesOnProceduralOneLiners, BracesRequiredMethods.
-# SupportedStyles: line_count_based, semantic, braces_for_chaining, always_braces
-# ProceduralMethods: benchmark, bm, bmbm, create, each_with_object, measure, new, realtime, tap, with_object
-# FunctionalMethods: let, let!, subject, watch
-# AllowedMethods: lambda, proc, it
-Style/BlockDelimiters:
-  Exclude:
-    - 'lib/iev/exporter.rb'
-    - 'spec/iev/exporter_spec.rb'

data/CLAUDE.md

@@ -33,7 +33,9 @@ This is a Ruby gem (`iev`) for working with the International Electrotechnical V
 - `SupersessionParser` — parses the REPLACES field for deprecated term relationships.
 - `SubjectAreas` — manages the IEV subject area/section hierarchy. Bundled `data/subject_areas.yaml` contains the area/section tree. URI scheme: `area-{code}` and `section-{code}`.
 - `SubjectAreaConcepts` — builds area and section hierarchy concepts. Uses `ConceptReference` with proper `ref_type` per `ConceptReferenceType`: `"domain"` for thematic area classification, `"section"` for structural section membership. Sets `ConceptData#domain` to area title text.
-- `Exporter` — full export pipeline (Excel/SQLite → Glossarist YAML). Assigns domain and section `ConceptReference` objects via `domain_references_for`. Uses `Glossarist::DatasetRegister` model for `register.yaml`. Sets `schema_version: "3"` on all exported concepts.
+- `Exporter` — full export pipeline (Excel/SQLite → Glossarist YAML). Assigns domain and section `ConceptReference` objects via `domain_references_for`. Uses `Glossarist::DatasetRegister` model for `register.yaml`. Sets `schema_version: "3"` on all exported concepts. Pipeline order: build → subject areas → section relations → figure extraction → reference enrichment → save concepts → save figures → save bibliography → save register.
+- `FigureBuilder` — destructive extraction pass that hoists AsciiDoc image macros (emitted from SIMG tags by `Utilities`) into dataset-shared `Glossarist::Figure` entities. Rewrites inline text to `{{fig:id, display}}` mentions and adds `FigureReference` entries to `ManagedConceptData#figures`. One Figure entity per unique image file; captions merge across languages.
+- `BibliographyBuilder` — collects unique `(source, id)` pairs from every concept's sources (localized and managed) into a `Glossarist::BibliographyData`. Entry ids are normalized with the same rules as `Glossarist::Validation::BibliographyIndex` so consumers can resolve anchors.
 - `Converter::MathmlToAsciimath` — converts MathML markup to AsciiMath using Plurimath.
 - `Utilities` — HTML processing: converts IEV cross-references (`<a href=IEV...>`) to `{{URN, term}}` format (ID first, display text last), handles figures, images, bold tags, and newline normalization.
 
@@ -45,6 +47,14 @@ Per the concept model's `ConceptReferenceType`:
 
 Each concept's `ManagedConceptData#domains` contains both refs. `ConceptData#domain` (a `LocalizedString`) holds the section/area title text. The `ManagedConcept#related` array holds `broader`/`narrower` relationships for the hierarchy tree.
 
+### V3 Output Artifacts
+
+An export produces these files alongside the concepts/ directory:
+- `register.yaml` — `Glossarist::DatasetRegister` with section tree, languages, owner, URN.
+- `bibliography.yaml` — single `bibliography:` key wrapping an array of `BibliographyEntry` objects. Entry `id` is the normalized anchor that `Glossarist::Validation::BibliographyIndex` will resolve against.
+- `figures/{fig-id}.yaml` — one `Glossarist::Figure` per unique image. Each concept carries a `FigureReference` on `ManagedConceptData#figures` and an inline `{{fig:id, display}}` mention in the text where the figure appeared.
+- References on localized concepts are populated by `Glossarist::ConceptEnricher#inject_references`, which scans text for `{{urn:...}}`, `<<xref>>`, and `image::` patterns.
+
 ### Configuration
 
 `Iev.configure` yields a `Config` object with:

data/Gemfile

@@ -3,11 +3,12 @@
 source "https://rubygems.org"
 
 # Use local glossarist-ruby when available for development.
-# Otherwise falls back to released gem (requires >= 2.8.2 for tags support).
+# Otherwise falls back to released gem (requires >= 2.8.15 for
+# BibliographyData, Figure/NonVerbRep, and ConceptEnricher support).
 if File.directory?(File.expand_path("../glossarist-ruby", __dir__))
   gem "glossarist", path: "../glossarist-ruby"
 else
-  gem "glossarist", ">= 2.8.2"
+  gem "glossarist", ">= 2.8.15"
 end
 
 gem "benchmark"

data/iev.gemspec

@@ -23,7 +23,7 @@ Gem::Specification.new do |spec|
 
   spec.add_dependency "creek", "~> 2.6"
   spec.add_dependency "ferrum", "~> 0.15"
-  spec.add_dependency "glossarist", ">= 2.8.2"
+  spec.add_dependency "glossarist", ">= 2.8.15"
   spec.add_dependency "lutaml-model", "~> 0.8.0"
   spec.add_dependency "nokogiri", "~> 1.19"
   spec.add_dependency "plurimath"

data/lib/iev/bibliography_builder.rb

@@ -0,0 +1,87 @@
+# frozen_string_literal: true
+
+module Iev
+  # Builds a `Glossarist::BibliographyData` from the sources cited across a
+  # concept collection.
+  #
+  # Each unique `(source, id)` pair from a concept's `ConceptSource#origin`
+  # becomes one `BibliographyEntry`. The entry's `id` is the normalized
+  # anchor that `Glossarist::Validation::BibliographyIndex` uses for
+  # resolution — so the same normalization rules are applied here.
+  module BibliographyBuilder
+    module_function
+
+    # @param concepts [Enumerable<Glossarist::ManagedConcept>]
+    # @return [Glossarist::BibliographyData]
+    def build(concepts)
+      entries = collect_entries(concepts)
+      Glossarist::BibliographyData.new(entries: entries)
+    end
+
+    def collect_entries(concepts)
+      seen = {}
+      concepts.each do |concept|
+        concept.localizations.each do |l10n|
+          collect_from_l10n(l10n, seen)
+        end
+        Array(concept.sources).each { |src| add_source_entry(src, seen) }
+      end
+      seen.values.sort_by(&:id)
+    end
+    private_class_method :collect_entries
+
+    def collect_from_l10n(l10n, seen)
+      Array(l10n.all_sources).each { |src| add_source_entry(src, seen) }
+    end
+    private_class_method :collect_from_l10n
+
+    def add_source_entry(source, seen)
+      ref = source_origin_ref(source)
+      return unless ref
+
+      seen[entry_label(ref)] ||= build_entry(ref, source&.origin)
+    end
+    private_class_method :add_source_entry
+
+    def source_origin_ref(source)
+      ref = source&.origin&.ref
+      return unless ref&.source && !ref.source.strip.empty?
+
+      ref
+    end
+    private_class_method :source_origin_ref
+
+    def build_entry(ref, origin)
+      label = entry_label(ref)
+      Glossarist::BibliographyEntry.new(
+        id: normalize_anchor(label),
+        reference: label,
+        link: origin&.link,
+        type: type_for(ref.source),
+      )
+    end
+    private_class_method :build_entry
+
+    def entry_label(ref)
+      [ref.source, ref.id].compact.join(" ").strip
+    end
+    private_class_method :entry_label
+
+    # Mirrors `Glossarist::Validation::BibliographyIndex#normalize_anchor`
+    # so the id we write matches what the validator will look up.
+    def normalize_anchor(anchor)
+      anchor.to_s.gsub(/[ \/:]/, "_").gsub(/__+/, "_").downcase
+    end
+    private_class_method :normalize_anchor
+
+    def type_for(source)
+      case source.to_s
+      when /\A(IEV|VIM|JCGM)/ then "vocabulary"
+      when /\AITU/ then "recommendation"
+      when /\A(BIPM|BBIPM)/ then "brochure"
+      else "standard"
+      end
+    end
+    private_class_method :type_for
+  end
+end

data/lib/iev/exporter.rb

@@ -56,13 +56,18 @@ module Iev
       collection = build_collection(dataset)
       add_subject_area_concepts(collection) if @include_areas
       build_section_narrower_relations(collection) if @include_areas
+      figures = FigureBuilder.extract!(collection)
+      enrich_references(collection)
       save_collection(collection)
+      save_figures(figures)
+      save_bibliography(BibliographyBuilder.build(collection))
       save_register
       elapsed = Process.clock_gettime(Process::CLOCK_MONOTONIC) - start_time
 
       @stats = {
         concept_count: collection.count,
         localized_count: localized_count(collection),
+        figure_count: figures.length,
         elapsed_seconds: elapsed,
       }
       collection
@@ -175,6 +180,34 @@ module Iev
       collection.save_grouped_concepts_to_files(concepts_dir.to_s)
     end
 
+    def save_figures(figures)
+      return if figures.empty?
+
+      figures_dir = output_dir.expand_path.join("figures")
+      FileUtils.mkdir_p(figures_dir)
+      figures.each do |figure|
+        path = figures_dir.join("#{figure.id}.yaml")
+        File.write(path, figure.to_yaml, encoding: "utf-8")
+      end
+      puts "Written #{figures.length} figures to figures/" if $stdout.tty?
+    end
+
+    def save_bibliography(bibliography)
+      return if bibliography.entries.empty?
+
+      path = output_dir.expand_path.join("bibliography.yaml")
+      FileUtils.mkdir_p(path.dirname)
+      File.write(path, bibliography.to_yaml, encoding: "utf-8")
+      count = bibliography.entries.length
+      puts "Written bibliography.yaml with #{count} entries" if $stdout.tty?
+    end
+
+    def enrich_references(collection)
+      return if collection.none?
+
+      Glossarist::ConceptEnricher.new.inject_references(collection.to_a)
+    end
+
     def save_register
       areas = SubjectAreas.all
       sections = build_section_tree(areas)

data/lib/iev/figure_builder.rb

@@ -0,0 +1,186 @@
+# frozen_string_literal: true
+
+module Iev
+  # Hoists IEV figure references into dataset-shared Figure entities.
+  #
+  # IEV source data carries figures as inline SIMG tags, which Utilities
+  # rewrites to AsciiDoc image macros (+image::/assets/images/parts/{area}/
+  # FILE[Figure N - caption]+). This builder walks every concept's
+  # localizations, finds those image macros, promotes each to a
+  # dataset-shared Glossarist::Figure entity, and rewrites the inline text
+  # to a V3 figure mention (+{{fig:id, display}}+).
+  #
+  # The Figure entity is shared across concepts and languages — captions
+  # from different localizations merge into the same {lang => text} hash.
+  # The structural link from concept to figure is a FigureReference entry
+  # on ManagedConceptData#figures.
+  #
+  # Extraction is destructive: it mutates DetailedDefinition#content and
+  # appends FigureReference entries. Returns the unique Figure entities so
+  # the exporter can persist them to figures/{id}.yaml.
+  module FigureBuilder
+    # URL path prefix emitted by Utilities when converting SIMG tags.
+    # Kept in sync with Utilities::IMAGE_PATH_PREFIX (without the macro).
+    PATH_PREFIX = "/assets/images/parts"
+    private_constant :PATH_PREFIX
+
+    # Matches AsciiDoc image macros emitted by Utilities#process_simg_figures.
+    IMAGE_MACRO_REGEX = /
+      image::#{Regexp.escape(PATH_PREFIX)}
+      \/(?<area>\d+)\/(?<file>[\w.-]+)\[(?<caption>[^\]]*)\]
+    /x
+
+    # Captures "Figure N" label and the trailing caption text.
+    CAPTION_REGEX = /\A(?<label>Figure\s+\d+)\s*[–-]\s*(?<text>.+)\z/m
+
+    module_function
+
+    # @param collection [Glossarist::ManagedConceptCollection]
+    # @return [Array<Glossarist::Figure>] unique figures, sorted by id
+    def extract!(collection)
+      figures_by_id = {}
+
+      collection.each do |concept|
+        concept.localizations.each do |l10n|
+          process_localization(l10n, concept, figures_by_id)
+        end
+      end
+
+      figures_by_id.values.sort_by(&:id)
+    end
+
+    def process_localization(l10n, concept, figures_by_id)
+      lang = l10n.data&.language_code
+      return unless lang && lang.length == 3
+
+      Glossarist::ConceptData.detailed_definition_fields.each do |field|
+        process_field(l10n, field, lang, concept, figures_by_id)
+      end
+    end
+    private_class_method :process_localization
+
+    def process_field(l10n, field, lang, concept, figures_by_id)
+      l10n.data.public_send(field).each do |dd|
+        next unless dd.content&.include?("image::")
+
+        rewritten, hits = extract_from_text(dd.content, lang)
+        next if hits.empty?
+
+        dd.content = rewritten
+        hits.each { |hit| register_figure(hit, concept, figures_by_id) }
+      end
+    end
+    private_class_method :process_field
+
+    # @return [Array<(String, Array<Hash>)>] rewritten text and per-match
+    #   figure descriptors ({ id:, identifier:, caption:, lang:, image: })
+    def extract_from_text(text, lang)
+      hits = []
+      rewritten = text.gsub(IMAGE_MACRO_REGEX) do
+        hit = build_hit(Regexp.last_match, lang)
+        hits << hit
+        mention_for(hit)
+      end
+      [rewritten, hits]
+    end
+    private_class_method :extract_from_text
+
+    def build_hit(match, lang)
+      identifier, caption = parse_caption(match[:caption])
+      {
+        id: figure_id_for(match[:file]),
+        identifier: identifier,
+        caption: caption,
+        lang: lang,
+        image: build_image(match[:area], match[:file]),
+      }
+    end
+    private_class_method :build_hit
+
+    def build_image(area, file)
+      Glossarist::FigureImage.new(
+        src: "#{PATH_PREFIX}/#{area}/#{file}",
+        format: format_for(file),
+      )
+    end
+    private_class_method :build_image
+
+    def parse_caption(bracket)
+      stripped = bracket.to_s.strip
+      return [nil, nil] if stripped.empty?
+
+      if (m = stripped.match(CAPTION_REGEX))
+        label = m[:label].gsub(/\s+/, " ")
+        [label, m[:text].strip]
+      elsif stripped.match?(/\AFigure\s+\d+\z/)
+        [stripped, nil]
+      else
+        [nil, stripped]
+      end
+    end
+    private_class_method :parse_caption
+
+    def figure_id_for(file)
+      "fig-#{file.sub(/\.[^.]+\z/, '')}"
+    end
+    private_class_method :figure_id_for
+
+    def format_for(file)
+      File.extname(file).delete_prefix(".").downcase
+    end
+    private_class_method :format_for
+
+    def mention_for(hit)
+      parts = [hit[:identifier], hit[:caption]].compact
+      return "{{fig:#{hit[:id]}}}" if parts.empty?
+
+      "{{fig:#{hit[:id]}, #{parts.join(' - ')}}}"
+    end
+    private_class_method :mention_for
+
+    # Add or merge a figure descriptor into the shared index, and ensure the
+    # concept carries a FigureReference to it.
+    def register_figure(hit, concept, figures_by_id)
+      figure = figures_by_id[hit[:id]] ||= build_figure(hit)
+      merge_caption!(figure, hit)
+      add_image_if_missing(figure, hit[:image])
+      add_figure_reference(concept, hit[:id], hit[:identifier])
+    end
+    private_class_method :register_figure
+
+    def build_figure(hit)
+      Glossarist::Figure.new(
+        id: hit[:id],
+        identifier: hit[:identifier],
+        images: [],
+        caption: {},
+      )
+    end
+    private_class_method :build_figure
+
+    def merge_caption!(figure, hit)
+      return unless hit[:caption]
+
+      figure.caption ||= {}
+      figure.caption[hit[:lang]] ||= hit[:caption]
+    end
+    private_class_method :merge_caption!
+
+    def add_image_if_missing(figure, image)
+      return if figure.images.any? { |i| i.src == image.src }
+
+      figure.images << image
+    end
+    private_class_method :add_image_if_missing
+
+    def add_figure_reference(concept, figure_id, display)
+      refs = Array(concept.data.figures)
+      return if refs.any? { |r| r.entity_id == figure_id }
+
+      concept.data.figures = refs + [
+        Glossarist::FigureReference.new(entity_id: figure_id, display: display),
+      ]
+    end
+    private_class_method :add_figure_reference
+  end
+end

data/lib/iev/version.rb

@@ -1,5 +1,5 @@
 # frozen_string_literal: true
 
 module Iev
-  VERSION = "0.4.6"
+  VERSION = "0.4.7"
 end

data/lib/iev.rb

@@ -21,6 +21,7 @@ module Iev
   # IEV dataset URN — single source of truth for all concept references.
   IEV_SOURCE = "urn:iec:std:iec:60050"
 
+  autoload :BibliographyBuilder, "iev/bibliography_builder"
   autoload :Cli, "iev/cli"
   autoload :Config, "iev/config"
   autoload :Converter, "iev/converter"
@@ -28,6 +29,7 @@ module Iev
   autoload :DataSource, "iev/data_source"
   autoload :DbWriter, "iev/db_writer"
   autoload :Exporter, "iev/exporter"
+  autoload :FigureBuilder, "iev/figure_builder"
   autoload :IevCode, "iev/iev_code"
   autoload :Iso639Code, "iev/iso_639_code"
   autoload :Profiler, "iev/profiler"

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: iev
 version: !ruby/object:Gem::Version
-  version: 0.4.6
+  version: 0.4.7
 platform: ruby
 authors:
 - Ribose Inc.
 autorequire:
 bindir: exe
 cert_chain: []
-date: 2026-06-13 00:00:00.000000000 Z
+date: 2026-06-18 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: creek
@@ -44,14 +44,14 @@ dependencies:
     requirements:
     - - ">="
       - !ruby/object:Gem::Version
-        version: 2.8.2
+        version: 2.8.15
   type: :runtime
   prerelease: false
   version_requirements: !ruby/object:Gem::Requirement
     requirements:
     - - ">="
       - !ruby/object:Gem::Version
-        version: 2.8.2
+        version: 2.8.15
 - !ruby/object:Gem::Dependency
   name: lutaml-model
   requirement: !ruby/object:Gem::Requirement
@@ -195,6 +195,7 @@ files:
 - exe/iev
 - iev.gemspec
 - lib/iev.rb
+- lib/iev/bibliography_builder.rb
 - lib/iev/cli.rb
 - lib/iev/cli/command.rb
 - lib/iev/cli/command_helper.rb
@@ -206,6 +207,7 @@ files:
 - lib/iev/data_source.rb
 - lib/iev/db_writer.rb
 - lib/iev/exporter.rb
+- lib/iev/figure_builder.rb
 - lib/iev/iev_code.rb
 - lib/iev/iso_639_2.yaml
 - lib/iev/iso_639_code.rb