metanorma-document 0.2.0

data/lib/metanorma/iso_document/metadata/iso_bibliographic_item.rb

@@ -0,0 +1,175 @@
+# frozen_string_literal: true
+
+module Metanorma
+  module IsoDocument
+    module Metadata
+      # The bibliographical description of an ISO/IEC document.
+      # Language element with current attribute.
+      class LanguageElement < Lutaml::Model::Serializable
+        attribute :current, :string
+        attribute :value, :string
+
+        xml do
+          element "language"
+          map_attribute "current", to: :current
+          map_content to: :value
+        end
+      end
+
+      # Script element with current attribute.
+      class ScriptElement < Lutaml::Model::Serializable
+        attribute :current, :string
+        attribute :value, :string
+
+        xml do
+          element "script"
+          map_attribute "current", to: :current
+          map_content to: :value
+        end
+      end
+
+      class IsoBibliographicItem < Lutaml::Model::Serializable
+        # Identifier of the document.
+        attribute :doc_identifier, DocIdentifier, collection: true
+
+        # Collection of all title elements, grouped by language via consolidation mapping.
+        # Access: titles.items (raw), titles.per_language (grouped), titles.for_language("en")
+        attribute :titles, AbstractTitle, collection: TitleCollection
+
+        # Type of the document.
+        attribute :type, :string
+
+        # Fetched date
+        attribute :fetched, Metanorma::Document::Relaton::DateTime
+
+        # URIs associated with the document.
+        attribute :uri, Metanorma::Document::Relaton::TypedUri, collection: true
+
+        # Link/source
+        attribute :source, :string, collection: true
+
+        # Document number
+        attribute :docnumber, :string
+
+        # Contributors
+        attribute :contributor, Metanorma::Document::Relaton::ContributionInfo,
+                  collection: true
+
+        # Edition
+        attribute :edition, Metanorma::Document::Relaton::Edition,
+                  collection: true
+
+        # Version
+        attribute :version, Metanorma::Document::Relaton::VersionInfo
+
+        # Notes
+        attribute :note, Metanorma::Document::Relaton::TypedNote,
+                  collection: true
+
+        # Language codes
+        attribute :language, LanguageElement, collection: true
+
+        # Script codes
+        attribute :script, ScriptElement, collection: true
+
+        # Abstract of the document.
+        attribute :abstract, Metanorma::Document::Components::DataTypes::FormattedString,
+                  collection: true
+
+        # Publication status of the document.
+        attribute :status, IsoDocumentStatus
+
+        # Copyright information
+        attribute :copyright, Metanorma::Document::Relaton::CopyrightAssociation,
+                  collection: true
+
+        # Relations to other documents
+        attribute :relation, Metanorma::Document::Relaton::DocumentRelation,
+                  collection: true
+
+        # Formatted reference
+        attribute :formattedref, Metanorma::Document::Components::DataTypes::FormattedString
+
+        # Dates
+        attribute :date, Metanorma::Document::Relaton::BibliographicDate,
+                  collection: true
+
+        # Place of publication
+        attribute :place, :string, collection: true
+
+        # Extension point of the bibliographical description.
+        attribute :ext, IsoBibDataExtensionType
+        attribute :keyword, Metanorma::Document::Relaton::KeywordType,
+                  collection: true
+        attribute :series, Metanorma::Document::Relaton::SeriesType,
+                  collection: true
+        attribute :editorialgroup, Metanorma::StandardDocument::Metadata::EditorialGroupType
+        attribute :semx_id, :string
+
+        # Schema version (used by collection/entry bibdata, e.g. "v1.2.9").
+        attribute :schema_version, :string
+
+        xml do
+          element "bibdata"
+          ordered
+          map_attribute "type", to: :type
+          map_attribute "schema-version", to: :schema_version
+          map_element "fetched", to: :fetched
+          map_element "title", to: :titles
+          map_element "uri", to: :uri
+          map_element "link", to: :source
+          map_element "docidentifier", to: :doc_identifier
+          map_element "docnumber", to: :docnumber
+          map_element "contributor", to: :contributor
+          map_element "edition", to: :edition
+          map_element "version", to: :version
+          map_element "note", to: :note
+          map_element "language", to: :language
+          map_element "script", to: :script
+          map_element "abstract", to: :abstract
+          map_element "status", to: :status
+          map_element "copyright", to: :copyright
+          map_element "relation", to: :relation
+          map_element "formattedref", to: :formattedref
+          map_element "date", to: :date
+          map_element "place", to: :place
+          map_element "ext", to: :ext
+          map_element "keyword", to: :keyword
+          map_element "series", to: :series
+          map_element "editorial-group", to: :editorialgroup
+          map_attribute "semx-id", to: :semx_id
+        end
+
+        json do
+          map "doc_identifier", to: :doc_identifier
+          map "titles", to: :titles
+          map "type", to: :type
+          map "status", to: :status
+          map "source", to: :source
+          map "abstract", to: :abstract
+          map "ext", to: :ext
+        end
+
+        # Returns the title for a given language (defaults to English)
+        # Uses consolidation-mapped per_language grouping for efficiency
+        def title_for(language = "en")
+          return nil unless titles
+
+          if titles.is_a?(TitleCollection)
+            titles.for_language(language)
+          elsif titles.is_a?(Array)
+            titles.find { |t|
+              lang = safe_attr(t, :language) || safe_attr(t, :lang)
+              lang == language
+            }
+          end
+        end
+
+        # Primary title (English by default)
+        def title
+          @title ||= title_for("en")
+        end
+      end
+    end
+  end
+end

data/lib/metanorma/iso_document/metadata/iso_document_id.rb

@@ -0,0 +1,46 @@
+# frozen_string_literal: true
+
+module Metanorma
+  module IsoDocument
+    module Metadata
+      # Representation of the identifier for an ISO/IEC document, giving its individual semantic components.
+      class IsoDocumentId < Lutaml::Model::Serializable
+        # N-numbers assigned to the document by technical committees.
+        attribute :tc_document_number, :integer, collection: true
+
+        # Numerical component uniquely identifying the document.
+        attribute :project_number, :integer
+
+        # Representation in the identifier of the document part, if this is a document part.
+        attribute :part_number, :integer
+
+        # Representation in the identifier of the document subpart, if this is a document subpart.
+        # Only applicable to IEC documents.
+        attribute :subpart_number, :integer
+
+        # Representation in the identifier of the type of document amendment, if this is a document amendment.
+        attribute :amendment_number, :integer
+
+        # Representation in the identifier of the type of document technical corrigendum, if this is a
+        # document technical corrigendum.
+        attribute :corrigendum_number, :integer
+
+        # Year of publication of the document in its current version or edition.
+        attribute :originalyear, :integer
+        attribute :semx_id, :string
+
+        xml do
+          element "docidentifier"
+          map_element "tc-document-number", to: :tc_document_number
+          map_element "project-number", to: :project_number
+          map_element "part-number", to: :part_number
+          map_element "subpart-number", to: :subpart_number
+          map_element "amendment-number", to: :amendment_number
+          map_element "corrigendum-number", to: :corrigendum_number
+          map_element "originalyear", to: :originalyear
+          map_attribute "semx-id", to: :semx_id
+        end
+      end
+    end
+  end
+end

data/lib/metanorma/iso_document/metadata/iso_document_stage_codes.rb

@@ -0,0 +1,26 @@
+# frozen_string_literal: true
+
+module Metanorma
+  module IsoDocument
+    module Metadata
+      # Stage-level code in the International Harmonized Stage Codes
+      # used by ISO/IEC for their documents.
+      #
+      # NOTE: See https://www.iso.org/stage-codes.html
+      class IsoDocumentStageCodes < Lutaml::Model::Serializable
+        attribute :value, :string
+        attribute :semx_id, :string
+
+        def self.values
+          %w[00 10 20 30 40 50 60 90 95]
+        end
+
+        xml do
+          element "iso-document-stage-codes"
+          map_content to: :value
+          map_attribute "semx-id", to: :semx_id
+        end
+      end
+    end
+  end
+end

data/lib/metanorma/iso_document/metadata/iso_document_status.rb

@@ -0,0 +1,56 @@
+# frozen_string_literal: true
+
+module Metanorma
+  module IsoDocument
+    module Metadata
+      # Publication status of an ISO/IEC document.
+      # Stage element in presentation XML: <stage abbreviation="..." language="..." type="...">value</stage>
+      class StageElement < Lutaml::Model::Serializable
+        attribute :abbreviation, :string
+        attribute :language, :string
+        attribute :type, :string
+        attribute :value, :string, collection: true
+        attribute :br, Metanorma::Document::Components::Inline::BrElement,
+                  collection: true
+
+        xml do
+          element "stage"
+          mixed_content
+          map_attribute "abbreviation", to: :abbreviation
+          map_attribute "language", to: :language, render_empty: true
+          map_attribute "type", to: :type
+          map_content to: :value
+          map_element "br", to: :br
+        end
+      end
+
+      class IsoDocumentStatus < Lutaml::Model::Serializable
+        # Publication stage of the ISO/IEC document, expressed as an International Harmonized Stage Code.
+        attribute :stage, StageElement, collection: true
+
+        # Standard abbreviation of the publication stage.
+        attribute :stage_abbreviation, :string
+
+        # Substage code for the ISO/IEC document.
+        attribute :substage, IsoDocumentSubstageCodes
+
+        # Standard abbreviation of the publication substage.
+        attribute :substage_abbreviation, :string
+
+        # Number of the iteration of the publication stage that the document is in.
+        attribute :iteration, :integer
+        attribute :semx_id, :string
+
+        xml do
+          element "status"
+          map_element "stage", to: :stage
+          map_element "stage-abbreviation", to: :stage_abbreviation
+          map_element "substage", to: :substage
+          map_element "substage-abbreviation", to: :substage_abbreviation
+          map_element "iteration", to: :iteration
+          map_attribute "semx-id", to: :semx_id
+        end
+      end
+    end
+  end
+end

data/lib/metanorma/iso_document/metadata/iso_document_substage_codes.rb

@@ -0,0 +1,26 @@
+# frozen_string_literal: true
+
+module Metanorma
+  module IsoDocument
+    module Metadata
+      # Substage-level code in the International Harmonized Stage Codes.
+      # These codes are used by ISO and IEC to indicate development stage.
+      #
+      # NOTE: See https://www.iso.org/stage-codes.html for the full list.
+      class IsoDocumentSubstageCodes < Lutaml::Model::Serializable
+        attribute :value, :string
+        attribute :semx_id, :string
+
+        def self.values
+          %w[00 20 60 90 92 93 98 99]
+        end
+
+        xml do
+          element "iso-document-substage-codes"
+          map_content to: :value
+          map_attribute "semx-id", to: :semx_id
+        end
+      end
+    end
+  end
+end

data/lib/metanorma/iso_document/metadata/iso_document_type.rb

@@ -0,0 +1,25 @@
+# frozen_string_literal: true
+
+module Metanorma
+  module IsoDocument
+    module Metadata
+      # Type of ISO/IEC document.
+      class IsoDocumentType < Lutaml::Model::Serializable
+        attribute :value, :string
+        attribute :semx_id, :string
+
+        def self.values
+          %w[internationalStandard technicalSpecification technicalReport
+             publiclyAvailableSpecification internationalWorkshopAgreement guide
+             interpretationSheet amendment technical-corrigendum]
+        end
+
+        xml do
+          element "iso-document-type"
+          map_content to: :value
+          map_attribute "semx-id", to: :semx_id
+        end
+      end
+    end
+  end
+end

data/lib/metanorma/iso_document/metadata/iso_localized_title.rb

@@ -0,0 +1,57 @@
+# frozen_string_literal: true
+
+module Metanorma
+  module IsoDocument
+    module Metadata
+      # Title of ISO/IEC document for a single language.
+      # Groups all title parts (title_intro, title_main, title_part, etc.) for one language.
+      # This is the GroupClass used by TitleCollection's consolidation mapping.
+      class IsoLocalizedTitle < Lutaml::Model::Serializable
+        # Language code (en, fr, ru, etc.)
+        attribute :language, :string
+
+        # Full title (type="main")
+        attribute :title_full, AbstractTitle
+
+        # Introductory component of title.
+        attribute :title_intro, AbstractTitle
+
+        # Main component of title.
+        attribute :title_main, AbstractTitle
+
+        # Title of document part, if applicable.
+        attribute :title_part, AbstractTitle
+
+        # Title part prefix (e.g., "Part 1")
+        attribute :title_part_prefix, AbstractTitle
+
+        # Amendment description text
+        attribute :title_amd, AbstractTitle
+
+        # Amendment prefix (e.g., "AMENDMENT 1")
+        attribute :title_amendment_prefix, AbstractTitle
+        attribute :semx_id, :string
+
+        # Compose all parts into a single title string
+        def to_s
+          parts = []
+          parts << title_intro&.value if title_intro
+          parts << title_main&.value if title_main
+          parts << title_full&.value if title_full && !title_main
+          if title_part
+            prefix = title_part_prefix&.value.to_s.strip
+            part_val = title_part.value
+            if !prefix.empty?
+              sep = prefix.end_with?(":") ? " " : ": "
+              parts << "#{prefix}#{sep}#{part_val}"
+            else
+              parts << part_val
+            end
+          end
+          return parts.join(" — ") unless parts.empty?
+          ""
+        end
+      end
+    end
+  end
+end

data/lib/metanorma/iso_document/metadata/iso_project_group.rb

@@ -0,0 +1,32 @@
+# frozen_string_literal: true
+
+module Metanorma
+  module IsoDocument
+    module Metadata
+      # Editorial groups associated with the production of an ISO/IEC document.
+      class IsoProjectGroup < Lutaml::Model::Serializable
+        # Technical committees.
+        attribute :technical_committee, IsoSubGroup, collection: true
+
+        # Subcommittees.
+        attribute :subcommittee, IsoSubGroup, collection: true
+
+        # Workgroups.
+        attribute :workgroup, IsoSubGroup, collection: true
+
+        # Secretariat.
+        attribute :secretariat, :string
+        attribute :semx_id, :string
+
+        xml do
+          element "editorial-group"
+          map_element "technical-committee", to: :technical_committee
+          map_element "subcommittee", to: :subcommittee
+          map_element "workgroup", to: :workgroup
+          map_attribute "secretariat", to: :secretariat
+          map_attribute "semx-id", to: :semx_id
+        end
+      end
+    end
+  end
+end

data/lib/metanorma/iso_document/metadata/iso_sub_group.rb

@@ -0,0 +1,38 @@
+# frozen_string_literal: true
+
+module Metanorma
+  module IsoDocument
+    module Metadata
+      # Editorial group set up in ISO or IEC to work on producing documents.
+      class IsoSubGroup < Lutaml::Model::Serializable
+        # Type of editorial group.
+        attribute :type, :string
+
+        # Number used to identify editorial group.
+        attribute :number, :integer
+
+        # Prefix used to differentiate editorial group for others of its class.
+        attribute :prefix, :string
+
+        # Full identifier used to identify editorial group.
+        attribute :identifier, :string
+
+        # Name of editorial group.
+        attribute :name, :string
+        attribute :semx_id, :string
+        attribute :text, :string
+
+        xml do
+          element "subgroup"
+          map_attribute "type", to: :type
+          map_attribute "number", to: :number
+          map_attribute "prefix", to: :prefix
+          map_attribute "identifier", to: :identifier
+          map_attribute "name", to: :name
+          map_attribute "semx-id", to: :semx_id
+          map_content to: :text
+        end
+      end
+    end
+  end
+end

data/lib/metanorma/iso_document/metadata/metanorma_extension.rb

@@ -0,0 +1,20 @@
+# frozen_string_literal: true
+
+module Metanorma
+  module IsoDocument
+    module Metadata
+      # Container for metanorma-specific extensions (semantic-metadata,
+      # presentation-metadata, UnitsML). Uses map_all_content to preserve
+      # all child XML as raw string since these elements have complex
+      # namespace-qualified structures.
+      class MetanormaExtension < Lutaml::Model::Serializable
+        attribute :content, :string
+
+        xml do
+          element "metanorma-extension"
+          map_all_content to: :content
+        end
+      end
+    end
+  end
+end

data/lib/metanorma/iso_document/metadata/structured_identifier.rb

@@ -0,0 +1,33 @@
+# frozen_string_literal: true
+
+module Metanorma
+  module IsoDocument
+    module Metadata
+      # Project number with optional part attribute.
+      class ProjectNumber < Lutaml::Model::Serializable
+        attribute :part, :string
+        attribute :amendment, :string
+        attribute :origyr, :string
+        attribute :value, :string
+
+        xml do
+          element "project-number"
+          map_attribute "part", to: :part
+          map_attribute "amendment", to: :amendment
+          map_attribute "origyr", to: :origyr
+          map_content to: :value
+        end
+      end
+
+      # Structured identifier for an ISO document (project number, part, etc).
+      class StructuredIdentifier < Lutaml::Model::Serializable
+        attribute :project_number, ProjectNumber
+
+        xml do
+          element "structuredidentifier"
+          map_element "project-number", to: :project_number
+        end
+      end
+    end
+  end
+end

data/lib/metanorma/iso_document/metadata/title_amd.rb

@@ -0,0 +1,12 @@
+# frozen_string_literal: true
+
+module Metanorma
+  module IsoDocument
+    module Metadata
+      # Title with type="title-amd" - amendment description text
+      class TitleAmd < AbstractTitle
+        attribute :semx_id, :string
+      end
+    end
+  end
+end

data/lib/metanorma/iso_document/metadata/title_amendment_prefix.rb

@@ -0,0 +1,12 @@
+# frozen_string_literal: true
+
+module Metanorma
+  module IsoDocument
+    module Metadata
+      # Title with type="title-amendment-prefix" - amendment prefix (e.g. "AMENDMENT 1")
+      class TitleAmendmentPrefix < AbstractTitle
+        attribute :semx_id, :string
+      end
+    end
+  end
+end

data/lib/metanorma/iso_document/metadata/title_collection.rb

@@ -0,0 +1,46 @@
+# frozen_string_literal: true
+
+module Metanorma
+  module IsoDocument
+    module Metadata
+      # Collection for title elements with consolidation mapping.
+      # Groups titles by language and routes by type to typed attributes.
+      #
+      # Usage:
+      #   bibdata.titles.items        # => All raw AbstractTitle instances
+      #   bibdata.titles.per_language # => Array of IsoLocalizedTitle grouped by language
+      #
+      class TitleCollection < Lutaml::Model::Collection
+        attribute :semx_id, :string
+        instances :items, AbstractTitle
+        organizes :per_language, IsoLocalizedTitle
+
+        xml do
+          root "titles"
+          map_instances to: :items
+
+          consolidate_map by: :language, to: :per_language do
+            gather :language, to: :language
+            dispatch_by :_type do
+              route "main" => :title_full
+              route "title-intro" => :title_intro
+              route "intro" => :title_intro
+              route "title-main" => :title_main
+              route "title-part" => :title_part
+              route "title-part-prefix" => :title_part_prefix
+              route "title-amd" => :title_amd
+              route "title-amendment-prefix" => :title_amendment_prefix
+              route "title-abbrev" => :title_full
+            end
+          end
+          map_attribute "semx-id", to: :semx_id
+        end
+
+        # Find the localized title for a given language
+        def for_language(language = "en")
+          per_language.find { |g| g.language == language }
+        end
+      end
+    end
+  end
+end

data/lib/metanorma/iso_document/metadata/title_full.rb

@@ -0,0 +1,12 @@
+# frozen_string_literal: true
+
+module Metanorma
+  module IsoDocument
+    module Metadata
+      # Title with type="main" - full/complete title
+      class TitleFull < AbstractTitle
+        attribute :semx_id, :string
+      end
+    end
+  end
+end

data/lib/metanorma/iso_document/metadata/title_intro.rb

@@ -0,0 +1,12 @@
+# frozen_string_literal: true
+
+module Metanorma
+  module IsoDocument
+    module Metadata
+      # Title with type="title-intro" - introductory component
+      class TitleIntro < AbstractTitle
+        attribute :semx_id, :string
+      end
+    end
+  end
+end

data/lib/metanorma/iso_document/metadata/title_main.rb

@@ -0,0 +1,12 @@
+# frozen_string_literal: true
+
+module Metanorma
+  module IsoDocument
+    module Metadata
+      # Title with type="title-main" - main component
+      class TitleMain < AbstractTitle
+        attribute :semx_id, :string
+      end
+    end
+  end
+end

data/lib/metanorma/iso_document/metadata/title_part.rb

@@ -0,0 +1,12 @@
+# frozen_string_literal: true
+
+module Metanorma
+  module IsoDocument
+    module Metadata
+      # Title with type="title-part" - part component
+      class TitlePart < AbstractTitle
+        attribute :semx_id, :string
+      end
+    end
+  end
+end

data/lib/metanorma/iso_document/metadata/title_part_prefix.rb

@@ -0,0 +1,12 @@
+# frozen_string_literal: true
+
+module Metanorma
+  module IsoDocument
+    module Metadata
+      # Title with type="title-part-prefix" - part prefix (e.g., "Part 1")
+      class TitlePartPrefix < AbstractTitle
+        attribute :semx_id, :string
+      end
+    end
+  end
+end