tbx 0.1.1 → 0.1.2

data/lib/tbx/sc.rb

@@ -1,12 +1,18 @@
 # frozen_string_literal: true
 
 module Tbx
+  # XLIFF inline start-code marker `<sc>`.
+  #
+  # Used for representing inline formatting codes from source formats.
+  # The `id` is required and is referenced by a corresponding `<ec>`
+  # element's `startRef`.
+  #
+  # Attributes per core RNG: `id` (required, type ID), `isolated` (optional,
+  # yes/no).
+  # Schema source: TBXcoreStructV03.rng `<define name="sc">`
   class Sc < Lutaml::Model::Serializable
     attribute :id, :string
     attribute :isolated, :string
-    attribute :type, :string
-    attribute :subtype, :string
-    attribute :target, :string
 
     xml do
       root "sc"
@@ -14,9 +20,6 @@ module Tbx
 
       map_attribute "id", to: :id
       map_attribute "isolated", to: :isolated
-      map_attribute "type", to: :type
-      map_attribute "subtype", to: :subtype
-      map_attribute "target", to: :target
     end
   end
 end

data/lib/tbx/source_desc.rb

@@ -1,6 +1,9 @@
 # frozen_string_literal: true
 
 module Tbx
+  # Source description `<sourceDesc>`.
+  #
+  # Schema source: TBXcoreStructV03.rng `<define name="sourceDesc">`
   class SourceDesc < Lutaml::Model::Serializable
     attribute :id, :string
     attribute :lang, Lutaml::Xml::W3c::XmlLangType

data/lib/tbx/tbx_header.rb

@@ -1,6 +1,13 @@
 # frozen_string_literal: true
 
 module Tbx
+  # Document header `<tbxHeader>`.
+  #
+  # Contains file description, encoding information, and revision history.
+  # Children: `fileDesc` (required), `encodingDesc` (optional),
+  # `revisionDesc` (optional).
+  #
+  # Schema source: TBXcoreStructV03.rng `<define name="tbxHeader">`
   class TbxHeader < Lutaml::Model::Serializable
     attribute :id, :string
     attribute :file_desc, ::Tbx::FileDesc

data/lib/tbx/term.rb

@@ -1,6 +1,12 @@
 # frozen_string_literal: true
 
 module Tbx
+  # The term designation element `<term>`.
+  #
+  # Contains the actual term text. Content is limited to plain text and
+  # `<hi>` inline highlighting elements per the `entity.basicText` pattern.
+  #
+  # Schema source: TBXcoreStructV03.rng `<define name="term">`
   class Term < Lutaml::Model::Serializable
     attribute :id, :string
     attribute :content, :string, collection: true

data/lib/tbx/term_comp.rb

@@ -0,0 +1,26 @@
+# frozen_string_literal: true
+
+module Tbx
+  # A single component of a decomposed term `<termComp>`.
+  #
+  # Used for representing hyphenation, syllabification, morphological
+  # elements, lemmata, or other logically significant portions of a term.
+  # Content is plain text only.
+  #
+  # Schema source: TermComp-namespace.rng `<define name="termComp">`
+  class TermComp < Lutaml::Model::Serializable
+    attribute :id, :string
+    attribute :lang, Lutaml::Xml::W3c::XmlLangType
+    attribute :content, :string, collection: true
+
+    xml do
+      root "termComp"
+      mixed_content
+      namespace ::Tbx::Namespace
+
+      map_attribute "id", to: :id
+      w3c_attributes :lang
+      map_content to: :content
+    end
+  end
+end

data/lib/tbx/term_comp_grp.rb

@@ -0,0 +1,24 @@
+# frozen_string_literal: true
+
+module Tbx
+  # Term component group `<termCompGrp>`.
+  #
+  # Groups a `<termComp>` with associated `<note>` elements.
+  #
+  # Schema source: TermComp-namespace.rng `<define name="termCompGrp">`
+  class TermCompGrp < Lutaml::Model::Serializable
+    attribute :id, :string
+    attribute :term_comp, ::Tbx::TermComp
+    attribute :note, ::Tbx::Note, collection: true
+
+    xml do
+      root "termCompGrp"
+      mixed_content
+      namespace ::Tbx::Namespace
+
+      map_attribute "id", to: :id
+      map_element "termComp", to: :term_comp
+      map_element "note", to: :note
+    end
+  end
+end

data/lib/tbx/term_comp_sec.rb

@@ -0,0 +1,45 @@
+# frozen_string_literal: true
+
+module Tbx
+  # Term component section `<termCompSec>`.
+  #
+  # Section containing decomposed term components. The `type` attribute
+  # indicates the decomposition method.
+  #
+  # Permitted `type` values (from TermComp-namespace.rng):
+  #
+  # * `hyphenation` — parts of a term as it would be hyphenated
+  # * `lemma` — base form (dictionary entry form)
+  # * `morphologicalElement` — individual morphemes comprising the term
+  # * `syllabification` — individual syllables comprising the term
+  # * `termElement` — any other logically significant portion
+  #
+  # Schema source: TermComp-namespace.rng `<define name="termCompSec">`
+  class TermCompSec < Lutaml::Model::Serializable
+    TYPES = {
+      hyphenation: "hyphenation",
+      lemma: "lemma",
+      morphological_element: "morphologicalElement",
+      syllabification: "syllabification",
+      term_element: "termElement",
+    }.freeze
+
+    attribute :id, :string
+    attribute :type, :string
+    attribute :note, ::Tbx::Note, collection: true
+    attribute :term_comp, ::Tbx::TermComp, collection: true
+    attribute :term_comp_grp, ::Tbx::TermCompGrp, collection: true
+
+    xml do
+      root "termCompSec"
+      mixed_content
+      namespace ::Tbx::Namespace
+
+      map_attribute "id", to: :id
+      map_attribute "type", to: :type
+      map_element "note", to: :note
+      map_element "termComp", to: :term_comp
+      map_element "termCompGrp", to: :term_comp_grp
+    end
+  end
+end

data/lib/tbx/term_note.rb

@@ -1,18 +1,51 @@
 # frozen_string_literal: true
 
 module Tbx
+  # Term-level data category element `<termNote>`.
+  #
+  # Carries grammatical, administrative, and usage information about a term.
+  # Only allowed as a child of `<termSec>` or `<termNoteGrp>`.
+  #
+  # Permitted `type` values (composed from all modules):
+  #
+  # * Min module: `administrativeStatus` (picklist: admittedTerm-admn-sts,
+  #   deprecatedTerm-admn-sts, supersededTerm-admn-sts, preferredTerm-admn-sts),
+  #   `partOfSpeech` (picklist: adjective, noun, other, verb, adverb)
+  # * Basic module: `geographicalUsage` (string), `grammaticalGender`
+  #   (picklist: masculine, feminine, neuter, other), `termLocation`
+  #   (picklist: 18 UI element types), `termType` (picklist: fullForm,
+  #   acronym, abbreviation, shortForm, variant, phrase)
+  # * Linguist module: `grammaticalNumber` (picklist: singular, plural, dual,
+  #   mass, otherNumber), `register` (picklist: colloquialRegister,
+  #   neutralRegister, technicalRegister, in-houseRegister,
+  #   bench-levelRegister, slangRegister, vulgarRegister),
+  #   `transferComment` (string)
+  #
+  # Content model: `entity.noteText` (mixed text + inline elements hi, ec,
+  # foreign, ph, sc).
+  #
+  # @see TYPES
+  #
+  # Schema source: TBXcoreStructV03.rng `<define name="termNote">`;
+  # data categories: Min.tbxmd, Basic.tbxmd, Linguist.tbxmd;
+  # permitted types validated by DCA Schematron (e.g. TBX-Basic_DCA.sch).
   class TermNote < Lutaml::Model::Serializable
-    attribute :id, :string
-    attribute :lang, Lutaml::Xml::W3c::XmlLangType
-    attribute :target, :string
-    attribute :datatype, :string
-    attribute :type, :string
-    attribute :content, :string, collection: true
-    attribute :hi, ::Tbx::Hi, collection: true
-    attribute :ec, ::Tbx::Ec, collection: true
-    attribute :foreign, ::Tbx::Foreign, collection: true
-    attribute :ph, ::Tbx::Ph, collection: true
-    attribute :sc, ::Tbx::Sc, collection: true
+    include Tbx::DataElement
+    include Tbx::DataElement::InlineContent
+
+    # Permitted termNote/@type values per TBX-Basic DCA Schematron
+    # (ISO 30042:2019, TBX-Basic dialect, DCA style).
+    #
+    # Source: reference-docs/schemas/TBX-Basic_DCA.sch line 54-56
+    #   "Permitted type value(s): administrativeStatus, partOfSpeech,
+    #    geographicalUsage, grammaticalGender, termLocation, termType"
+    TYPES = Modules::Min::TERM_NOTE_TYPES
+      .merge(Modules::Basic::TERM_NOTE_TYPES)
+      .merge(Modules::Linguist::TERM_NOTE_TYPES).freeze
+
+    VALUES = Modules::Min::TERM_NOTE_VALUES
+      .merge(Modules::Basic::TERM_NOTE_VALUES)
+      .merge(Modules::Linguist::TERM_NOTE_VALUES).freeze
 
     xml do
       root "termNote"

data/lib/tbx/term_note_grp.rb

@@ -1,6 +1,13 @@
 # frozen_string_literal: true
 
 module Tbx
+  # Groups a `<termNote>` with its associated administrative data and
+  # references.
+  #
+  # Children: `termNote` (required), then entity.noteLinkInfo (admin,
+  # adminGrp, note, ref, transacGrp, xref).
+  #
+  # Schema source: TBXcoreStructV03.rng `<define name="termNoteGrp">`
   class TermNoteGrp < Lutaml::Model::Serializable
     attribute :id, :string
     attribute :term_note, ::Tbx::TermNote

data/lib/tbx/term_sec.rb

@@ -1,6 +1,14 @@
 # frozen_string_literal: true
 
 module Tbx
+  # Term section `<termSec>`.
+  #
+  # Contains a single term designation and its associated term-level notes
+  # and administrative data. Children: `term` (required), then
+  # termNote/termNoteGrp, then auxInfo elements (admin, adminGrp, descrip,
+  # descripGrp, note, ref, transacGrp, xref).
+  #
+  # Schema source: TBXcoreStructV03.rng `<define name="termSec">`
   class TermSec < Lutaml::Model::Serializable
     attribute :id, :string
     attribute :term, ::Tbx::Term

data/lib/tbx/text_element.rb

@@ -1,6 +1,11 @@
 # frozen_string_literal: true
 
 module Tbx
+  # Wrapper element `<text>` for the termbase body and back matter.
+  #
+  # Children: `body` (optional), `back` (optional).
+  #
+  # Schema source: TBXcoreStructV03.rng `<define name="text">`
   class TextElement < Lutaml::Model::Serializable
     attribute :id, :string
     attribute :body, ::Tbx::Body

data/lib/tbx/title.rb

@@ -1,6 +1,12 @@
 # frozen_string_literal: true
 
 module Tbx
+  # Title element `<title>`.
+  #
+  # Used within `<titleStmt>` to specify the title of the TBX file.
+  # Content is plain text only.
+  #
+  # Schema source: TBXcoreStructV03.rng `<define name="title">`
   class Title < Lutaml::Model::Serializable
     attribute :id, :string
     attribute :lang, Lutaml::Xml::W3c::XmlLangType

data/lib/tbx/title_stmt.rb

@@ -1,6 +1,9 @@
 # frozen_string_literal: true
 
 module Tbx
+  # Title statement `<titleStmt>`.
+  #
+  # Schema source: TBXcoreStructV03.rng `<define name="titleStmt">`
   class TitleStmt < Lutaml::Model::Serializable
     attribute :id, :string
     attribute :lang, Lutaml::Xml::W3c::XmlLangType

data/lib/tbx/transac.rb

@@ -1,13 +1,24 @@
 # frozen_string_literal: true
 
 module Tbx
+  # Transaction element `<transac>`.
+  #
+  # Records an action performed on a terminological entry. The content
+  # indicates the type of transaction (origination or modification).
+  #
+  # Permitted `type` values:
+  #
+  # * Basic module: `transactionType` (picklist: `origination`, `modification`)
+  #
+  # Content is plain text only (no inline elements per the core RNG).
+  #
+  # Schema source: TBXcoreStructV03.rng `<define name="transac">`;
+  # data category: Basic.tbxmd.
   class Transac < Lutaml::Model::Serializable
-    attribute :id, :string
-    attribute :lang, Lutaml::Xml::W3c::XmlLangType
-    attribute :target, :string
-    attribute :datatype, :string
-    attribute :type, :string
-    attribute :content, :string, collection: true
+    include Tbx::DataElement
+
+    TYPES = Modules::Basic::TRANSAC_TYPES.freeze
+    VALUES = Modules::Basic::TRANSAC_VALUES.freeze
 
     xml do
       root "transac"

data/lib/tbx/transac_grp.rb

@@ -1,6 +1,12 @@
 # frozen_string_literal: true
 
 module Tbx
+  # Groups a `<transac>` transaction element with its date and
+  # responsibility information.
+  #
+  # Children: `transac` (required), then date, note, ref, transacNote, xref.
+  #
+  # Schema source: TBXcoreStructV03.rng `<define name="transacGrp">`
   class TransacGrp < Lutaml::Model::Serializable
     attribute :id, :string
     attribute :transac, ::Tbx::Transac

data/lib/tbx/transac_note.rb

@@ -1,13 +1,24 @@
 # frozen_string_literal: true
 
 module Tbx
+  # Transaction note element `<transacNote>`.
+  #
+  # Provides additional information about a transaction, typically identifying
+  # the person or organization responsible.
+  #
+  # Permitted `type` values:
+  #
+  # * Basic module: `responsibility` — person/org responsible for the
+  #   transaction (DC-451)
+  #
+  # Content is plain text only (no inline elements per the core RNG).
+  #
+  # Schema source: TBXcoreStructV03.rng `<define name="transacNote">`;
+  # data category: Basic.tbxmd.
   class TransacNote < Lutaml::Model::Serializable
-    attribute :id, :string
-    attribute :lang, Lutaml::Xml::W3c::XmlLangType
-    attribute :target, :string
-    attribute :datatype, :string
-    attribute :type, :string
-    attribute :content, :string, collection: true
+    include Tbx::DataElement
+
+    TYPES = Modules::Basic::TRANSAC_NOTE_TYPES.freeze
 
     xml do
       root "transacNote"

data/lib/tbx/version.rb

@@ -1,5 +1,5 @@
 # frozen_string_literal: true
 
 module Tbx
-  VERSION = "0.1.1"
+  VERSION = "0.1.2"
 end

data/lib/tbx/xref.rb

@@ -1,7 +1,23 @@
 # frozen_string_literal: true
 
 module Tbx
+  # External cross-reference element `<xref>`.
+  #
+  # The `target` attribute holds a URI (http/https) pointing to an external
+  # resource. Unlike `Ref`, which uses IDREF for internal references, Xref
+  # uses anyURI for external resources.
+  #
+  # Permitted `type` values:
+  #
+  # * Min module: `externalCrossReference` — reference to an external resource
+  # * Basic module: `externalCrossReference` — external reference (DC-226),
+  #   `xGraphic` — graphic image associated with the concept (DC-2920)
+  #
+  # Schema source: TBXcoreStructV03.rng `<define name="xref">`;
+  # data categories: Basic.tbxmd.
   class Xref < Lutaml::Model::Serializable
+    TYPES = Modules::Min::XREF_TYPES
+      .merge(Modules::Basic::XREF_TYPES).freeze
     attribute :id, :string
     attribute :target, :string
     attribute :type, :string

data/lib/tbx.rb

@@ -4,6 +4,13 @@ require "lutaml/model"
 require_relative "tbx/version"
 
 module Tbx
+  module Modules
+    autoload :Min, "#{__dir__}/tbx/modules/min"
+    autoload :Basic, "#{__dir__}/tbx/modules/basic"
+    autoload :Linguist, "#{__dir__}/tbx/modules/linguist"
+    autoload :CoreTypes, "#{__dir__}/tbx/modules/core_types"
+  end
+
   autoload :Namespace, "#{__dir__}/tbx/namespace"
 
   # Inline elements
@@ -18,6 +25,9 @@ module Tbx
   autoload :P, "#{__dir__}/tbx/p"
   autoload :Title, "#{__dir__}/tbx/title"
 
+  # Shared concerns
+  autoload :DataElement, "#{__dir__}/tbx/data_element"
+
   # Data elements
   autoload :Admin, "#{__dir__}/tbx/admin"
   autoload :AdminGrp, "#{__dir__}/tbx/admin_grp"
@@ -45,6 +55,9 @@ module Tbx
 
   # Structural elements
   autoload :Term, "#{__dir__}/tbx/term"
+  autoload :TermComp, "#{__dir__}/tbx/term_comp"
+  autoload :TermCompGrp, "#{__dir__}/tbx/term_comp_grp"
+  autoload :TermCompSec, "#{__dir__}/tbx/term_comp_sec"
   autoload :TermSec, "#{__dir__}/tbx/term_sec"
   autoload :LangSec, "#{__dir__}/tbx/lang_sec"
   autoload :ConceptEntry, "#{__dir__}/tbx/concept_entry"

data/reference-docs/README.adoc

@@ -0,0 +1,109 @@
+= Reference Documentation
+
+This directory contains official TBX schema, dialect, and module reference
+files from https://ltac-global.org[LTAC Global] (LTAC-Global on GitHub).
+
+== Source repositories
+
+All files below are sourced from the
+https://github.com/LTAC-Global[LTAC-Global GitHub organization] and are
+used under their respective licenses for reference purposes.
+
+=== Dialects
+
+Dialects define a specific profile of TBX with Schematron constraints,
+RELAX NG schemas, and example files.
+
+|===
+| Directory | Source repository | Description
+
+| `schemas/TBX-Basic_dialect_v1/`
+| https://github.com/LTAC-Global/TBX-Basic_dialect[TBX-Basic_dialect]
+| TBX-Basic dialect (DCA + DCT schemas, example files, bundled modules)
+
+| `schemas/TBX-Min_dialect/`
+| https://github.com/LTAC-Global/TBX-Min_dialect[TBX-Min_dialect]
+| TBX-Min dialect (DCA + DCT Schematron, NVDL)
+
+| `schemas/TBX-Core_dialect/`
+| https://github.com/LTAC-Global/TBX-Core_dialect[TBX-Core_dialect]
+| TBX-Core dialect (integrated RNG, Schematron, example)
+
+| `schemas/TBX-Linguist_dialect/`
+| https://github.com/LTAC-Global/TBX-Linguist_dialect[TBX-Linguist_dialect]
+| TBX-Linguist dialect (DCA + DCT Schematron, integrated RNG, examples)
+|===
+
+=== Modules
+
+Modules define data categories and structural rules for a specific level
+of terminological detail.
+
+|===
+| Directory | Source repository | Key files
+
+| `schemas/TBX-Basic_dialect_v1/Modules/TBX_min_module/`
+| https://github.com/LTAC-Global/TBX_min_module[TBX_min_module]
+| `Min.rng`, `Min.sch`, `Min.tbxmd`
+
+| `schemas/TBX-Basic_dialect_v1/Modules/TBX_basic_module/`
+| https://github.com/LTAC-Global/TBX_basic_module[TBX_basic_module]
+| `Basic.rng`, `Basic.sch`, `Basic.tbxmd`
+
+| `schemas/TBX_linguist_module/`
+| https://github.com/LTAC-Global/TBX_linguist_module[TBX_linguist_module]
+| `Linguist.rng`, `Linguist.sch`, `Linguist.tbxmd`
+
+| `schemas/TBX_termComp_module/`
+| https://github.com/LTAC-Global/TBX_termComp_module[TBX_termComp_module]
+| `TermComp-namespace.rng`
+|===
+
+=== Core schema
+
+|===
+| File | Source repository | Description
+
+| `schemas/TBXcoreStructV03.rng`
+| https://github.com/LTAC-Global/TBX_core_module[TBX_core_module]
+| TBX core structural RELAX NG schema
+
+| `schemas/TBX_core.xsd`
+| https://github.com/LTAC-Global/TBX_core_module[TBX_core_module]
+| TBX core XML Schema (XSD)
+|===
+
+=== Module description format
+
+|===
+| Directory | Source repository | Key files
+
+| `schemas/TBX_module_description_xml/`
+| https://github.com/LTAC-Global/TBX_module_description_xml[TBX_module_description_xml]
+| `tbxmd.rng` -- RELAX NG schema for `.tbxmd` module definition files
+|===
+
+=== Implementation guide
+
+|===
+| Directory | Source repository | Key files
+
+| `schemas/TBX-Basic_ImplementationGuide/`
+| https://github.com/LTAC-Global/TBX-Basic_ImplementationGuide[TBX-Basic_ImplementationGuide]
+| `TBX-Basic_ImplementationGuide.pdf`
+|===
+
+== TBX module hierarchy
+
+The TBX standard (ISO 30042:2019) defines a modular architecture:
+
+....
+TBX-Core          structural foundation (element definitions)
+  +-- TBX-Min     minimal data categories for basic terminology
+  +-- TBX-Basic   standard data categories for general terminology
+  +-- TBX-Linguist  extended data categories for linguistic detail
+  +-- TBX-TermComp  term composition and decomposition
+....
+
+Each dialect (TBX-Min, TBX-Basic, etc.) combines the core structure with
+one or more modules and adds dialect-specific Schematron constraints.