tbx 0.1.0 → 0.1.2

data/lib/tbx/modules/min.rb

@@ -0,0 +1,46 @@
+# frozen_string_literal: true
+
+module Tbx
+  module Modules
+    # Data category definitions for the TBX-Min module.
+    #
+    # TBX-Min provides a minimal set of data categories for simple
+    # terminology exchange.
+    #
+    # Source: Min.tbxmd, Min.rng, Min.sch
+    module Min
+      ADMIN_TYPES = {
+        customer_subset: "customerSubset",
+      }.freeze
+
+      DESCRIP_TYPES = {
+        subject_field: "subjectField",
+      }.freeze
+
+      TERM_NOTE_TYPES = {
+        administrative_status: "administrativeStatus",
+        part_of_speech: "partOfSpeech",
+      }.freeze
+
+      TERM_NOTE_VALUES = {
+        administrative_status: {
+          admitted_term: "admittedTerm-admn-sts",
+          deprecated_term: "deprecatedTerm-admn-sts",
+          superseded_term: "supersededTerm-admn-sts",
+          preferred_term: "preferredTerm-admn-sts",
+        }.freeze,
+        part_of_speech: {
+          adjective: "adjective",
+          noun: "noun",
+          other: "other",
+          verb: "verb",
+          adverb: "adverb",
+        }.freeze,
+      }.freeze
+
+      XREF_TYPES = {
+        external_cross_reference: "externalCrossReference",
+      }.freeze
+    end
+  end
+end

data/lib/tbx/namespace.rb

@@ -0,0 +1,12 @@
+# frozen_string_literal: true
+
+module Tbx
+  # XML namespace for TBX elements per ISO 30042:2019.
+  #
+  # URI: `urn:iso:std:iso:30042:ed-2`
+  #
+  # All TBX structural elements use this namespace.
+  class Namespace < Lutaml::Xml::Namespace
+    uri "urn:iso:std:iso:30042:ed-2"
+  end
+end

data/lib/tbx/note.rb

@@ -1,6 +1,15 @@
 # frozen_string_literal: true
 
 module Tbx
+  # Generic note element `<note>`.
+  #
+  # Provides supplementary information at conceptEntry, langSec, termSec,
+  # and various group levels.
+  #
+  # Content model: `entity.noteText` (mixed text + inline elements hi, ec,
+  # foreign, ph, sc).
+  #
+  # Schema source: TBXcoreStructV03.rng `<define name="note">`
   class Note < Lutaml::Model::Serializable
     attribute :id, :string
     attribute :lang, Lutaml::Xml::W3c::XmlLangType
@@ -14,7 +23,7 @@ module Tbx
     xml do
       root "note"
       mixed_content
-      namespace ::Tbx::Namespaces::TbxNamespace
+      namespace ::Tbx::Namespace
 
       map_attribute "id", to: :id
       w3c_attributes :lang

data/lib/tbx/p.rb

@@ -1,6 +1,15 @@
 # frozen_string_literal: true
 
 module Tbx
+  # Paragraph element `<p>`.
+  #
+  # Used for prose text in header contexts (publicationStmt, sourceDesc,
+  # encodingDesc, change).
+  #
+  # Content model: `entity.noteText` (mixed text + inline elements hi, ec,
+  # foreign, ph, sc).
+  #
+  # Schema source: TBXcoreStructV03.rng `<define name="p">`
   class P < Lutaml::Model::Serializable
     attribute :id, :string
     attribute :lang, Lutaml::Xml::W3c::XmlLangType
@@ -15,7 +24,7 @@ module Tbx
     xml do
       root "p"
       mixed_content
-      namespace ::Tbx::Namespaces::TbxNamespace
+      namespace ::Tbx::Namespace
 
       map_attribute "id", to: :id
       w3c_attributes :lang

data/lib/tbx/ph.rb

@@ -1,12 +1,18 @@
 # frozen_string_literal: true
 
 module Tbx
+  # XLIFF inline placeholder `<ph>`.
+  #
+  # Represents a code that does not have a distinct start and end.
+  # Empty element with optional `type` attribute.
+  #
+  # Schema source: TBXcoreStructV03.rng `<define name="ph">`
   class Ph < Lutaml::Model::Serializable
     attribute :type, :string
 
     xml do
       root "ph"
-      namespace ::Tbx::Namespaces::TbxNamespace
+      namespace ::Tbx::Namespace
 
       map_attribute "type", to: :type
     end

data/lib/tbx/publication_stmt.rb

@@ -1,6 +1,9 @@
 # frozen_string_literal: true
 
 module Tbx
+  # Publication statement `<publicationStmt>`.
+  #
+  # Schema source: TBXcoreStructV03.rng `<define name="publicationStmt">`
   class PublicationStmt < Lutaml::Model::Serializable
     attribute :id, :string
     attribute :p, ::Tbx::P, collection: true
@@ -8,7 +11,7 @@ module Tbx
     xml do
       root "publicationStmt"
       mixed_content
-      namespace ::Tbx::Namespaces::TbxNamespace
+      namespace ::Tbx::Namespace
 
       map_attribute "id", to: :id
       map_element "p", to: :p

data/lib/tbx/ref.rb

@@ -1,18 +1,29 @@
 # frozen_string_literal: true
 
 module Tbx
+  # Internal reference element `<ref>`.
+  #
+  # Used for cross-references within the same termbase. The `target`
+  # attribute holds an IDREF pointing to another element's `id`.
+  #
+  # Permitted `type` values:
+  #
+  # * Basic module: `crossReference` — reference to another concept entry
+  #   in the same termbase (DC-164), level: conceptEntry/termSec
+  #
+  # Content is plain text only (no inline elements per the core RNG).
+  #
+  # Schema source: TBXcoreStructV03.rng `<define name="ref">`;
+  # data category: Basic.tbxmd.
   class Ref < 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::REF_TYPES.freeze
 
     xml do
       root "ref"
       mixed_content
-      namespace ::Tbx::Namespaces::TbxNamespace
+      namespace ::Tbx::Namespace
 
       map_attribute "id", to: :id
       w3c_attributes :lang

data/lib/tbx/ref_object.rb

@@ -1,6 +1,12 @@
 # frozen_string_literal: true
 
 module Tbx
+  # A reference object `<refObject>` defining a picklist or value set.
+  #
+  # Contains `itemSet`, `itemGrp`, or `item` children that enumerate the
+  # permitted values for a data category's picklist.
+  #
+  # Schema source: TBXcoreStructV03.rng `<define name="refObject">`
   class RefObject < Lutaml::Model::Serializable
     attribute :id, :string
     attribute :item_set, ::Tbx::ItemSet, collection: true
@@ -10,7 +16,7 @@ module Tbx
     xml do
       root "refObject"
       mixed_content
-      namespace ::Tbx::Namespaces::TbxNamespace
+      namespace ::Tbx::Namespace
 
       map_attribute "id", to: :id
       map_element "itemSet", to: :item_set

data/lib/tbx/ref_object_sec.rb

@@ -1,6 +1,12 @@
 # frozen_string_literal: true
 
 module Tbx
+  # Reference object section `<refObjectSec>` in back matter.
+  #
+  # Contains reference objects that define picklists and value sets used
+  # by data category elements throughout the termbase.
+  #
+  # Schema source: TBXcoreStructV03.rng `<define name="refObjectSec">`
   class RefObjectSec < Lutaml::Model::Serializable
     attribute :id, :string
     attribute :type, :string
@@ -9,7 +15,7 @@ module Tbx
     xml do
       root "refObjectSec"
       mixed_content
-      namespace ::Tbx::Namespaces::TbxNamespace
+      namespace ::Tbx::Namespace
 
       map_attribute "id", to: :id
       map_attribute "type", to: :type

data/lib/tbx/revision_desc.rb

@@ -1,6 +1,9 @@
 # frozen_string_literal: true
 
 module Tbx
+  # Revision description `<revisionDesc>`.
+  #
+  # Schema source: TBXcoreStructV03.rng `<define name="revisionDesc">`
   class RevisionDesc < Lutaml::Model::Serializable
     attribute :id, :string
     attribute :lang, Lutaml::Xml::W3c::XmlLangType
@@ -9,7 +12,7 @@ module Tbx
     xml do
       root "revisionDesc"
       mixed_content
-      namespace ::Tbx::Namespaces::TbxNamespace
+      namespace ::Tbx::Namespace
 
       map_attribute "id", to: :id
       w3c_attributes :lang

data/lib/tbx/sc.rb

@@ -1,22 +1,25 @@
 # 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"
-      namespace ::Tbx::Namespaces::TbxNamespace
+      namespace ::Tbx::Namespace
 
       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
@@ -9,7 +12,7 @@ module Tbx
     xml do
       root "sourceDesc"
       mixed_content
-      namespace ::Tbx::Namespaces::TbxNamespace
+      namespace ::Tbx::Namespace
 
       map_attribute "id", to: :id
       w3c_attributes :lang

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
@@ -10,7 +17,7 @@ module Tbx
     xml do
       root "tbxHeader"
       mixed_content
-      namespace ::Tbx::Namespaces::TbxNamespace
+      namespace ::Tbx::Namespace
 
       map_attribute "id", to: :id
       map_element "fileDesc", to: :file_desc

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
@@ -9,7 +15,7 @@ module Tbx
     xml do
       root "term"
       mixed_content
-      namespace ::Tbx::Namespaces::TbxNamespace
+      namespace ::Tbx::Namespace
 
       map_attribute "id", to: :id
       map_content to: :content

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

@@ -1,23 +1,56 @@
 # 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"
       mixed_content
-      namespace ::Tbx::Namespaces::TbxNamespace
+      namespace ::Tbx::Namespace
 
       map_attribute "id", to: :id
       w3c_attributes :lang

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
@@ -14,7 +21,7 @@ module Tbx
     xml do
       root "termNoteGrp"
       mixed_content
-      namespace ::Tbx::Namespaces::TbxNamespace
+      namespace ::Tbx::Namespace
 
       map_attribute "id", to: :id
       map_element "termNote", to: :term_note

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
@@ -18,7 +26,7 @@ module Tbx
     xml do
       root "termSec"
       mixed_content
-      namespace ::Tbx::Namespaces::TbxNamespace
+      namespace ::Tbx::Namespace
 
       map_attribute "id", to: :id
       map_element "term", to: :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
@@ -9,7 +14,7 @@ module Tbx
     xml do
       root "text"
       mixed_content
-      namespace ::Tbx::Namespaces::TbxNamespace
+      namespace ::Tbx::Namespace
 
       map_attribute "id", to: :id
       map_element "body", to: :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
@@ -9,7 +15,7 @@ module Tbx
     xml do
       root "title"
       mixed_content
-      namespace ::Tbx::Namespaces::TbxNamespace
+      namespace ::Tbx::Namespace
 
       map_attribute "id", to: :id
       w3c_attributes :lang

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
@@ -10,7 +13,7 @@ module Tbx
     xml do
       root "titleStmt"
       mixed_content
-      namespace ::Tbx::Namespaces::TbxNamespace
+      namespace ::Tbx::Namespace
 
       map_attribute "id", to: :id
       w3c_attributes :lang

data/lib/tbx/transac.rb

@@ -1,18 +1,29 @@
 # 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"
       mixed_content
-      namespace ::Tbx::Namespaces::TbxNamespace
+      namespace ::Tbx::Namespace
 
       map_attribute "id", to: :id
       w3c_attributes :lang