tbx 0.1.1 → 0.1.2

data/lib/tbx/data_element.rb

@@ -0,0 +1,50 @@
+# frozen_string_literal: true
+
+module Tbx
+  # Shared attribute declarations for TBX data-category elements.
+  #
+  # Injects the common attributes defined by the `IDLangTgtDtyp` attribute
+  # class + `type` + text content pattern from TBXcoreStructV03.rng.
+  #
+  # @example Usage in a data-category element
+  #   class Admin < Lutaml::Model::Serializable
+  #     include Tbx::DataElement
+  #     include Tbx::DataElement::InlineContent
+  #
+  #     xml do
+  #       root "admin"
+  #       ...
+  #     end
+  #   end
+  module DataElement
+    def self.included(base)
+      base.class_eval do
+        attribute :id, :string
+        attribute :lang, Lutaml::Xml::W3c::XmlLangType
+        attribute :target, :string
+        attribute :datatype, :string
+        attribute :type, :string
+        attribute :content, :string, collection: true
+      end
+    end
+
+    # Inline child element attributes for data-category elements that
+    # support mixed content per the `entity.noteText` pattern in
+    # TBXcoreStructV03.rng.
+    #
+    # Included by: Admin, Descrip, TermNote (elements with noteText content).
+    # NOT included by: AdminNote, DescripNote, TransacNote, Transac, Ref
+    # (plain text only per RNG).
+    module InlineContent
+      def self.included(base)
+        base.class_eval do
+          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
+        end
+      end
+    end
+  end
+end

data/lib/tbx/date.rb

@@ -1,6 +1,12 @@
 # frozen_string_literal: true
 
 module Tbx
+  # Date element `<date>`.
+  #
+  # Records when a transaction occurred. Content must be a valid XML Schema
+  # `date` (YYYY-MM-DD) or `dateTime` (YYYY-MM-DDThh:mm:ss) value.
+  #
+  # Schema source: TBXcoreStructV03.rng `<define name="date">`
   class DateElement < Lutaml::Model::Serializable
     attribute :id, :string
     attribute :content, :string

data/lib/tbx/descrip.rb

@@ -1,18 +1,30 @@
 # frozen_string_literal: true
 
 module Tbx
+  # Descriptive data category element `<descrip>`.
+  #
+  # Carries conceptual and contextual information about a term, including
+  # definitions, subject field classifications, and usage contexts.
+  #
+  # Permitted `type` values (composed from all modules):
+  #
+  # * Min module: `subjectField` — field of special knowledge (DC-489),
+  #   level: conceptEntry
+  # * Basic module: `context` — text illustrating term usage (DC-149),
+  #   level: termSec; `definition` — narrative statement of concept (DC-168),
+  #   level: conceptEntry/langSec
+  #
+  # Content model: `entity.noteText` (mixed text + inline elements hi, ec,
+  # foreign, ph, sc).
+  #
+  # Schema source: TBXcoreStructV03.rng `<define name="descrip">`;
+  # data categories: Min.tbxmd, Basic.tbxmd.
   class Descrip < 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
+
+    TYPES = Modules::Min::DESCRIP_TYPES
+      .merge(Modules::Basic::DESCRIP_TYPES).freeze
 
     xml do
       root "descrip"

data/lib/tbx/descrip_grp.rb

@@ -1,6 +1,13 @@
 # frozen_string_literal: true
 
 module Tbx
+  # Groups a `<descrip>` element with its administrative data, notes, and
+  # source references.
+  #
+  # Commonly used to associate a `source` admin with a `definition` descrip,
+  # providing provenance for descriptive data.
+  #
+  # Schema source: TBXcoreStructV03.rng `<define name="descripGrp">`
   class DescripGrp < Lutaml::Model::Serializable
     attribute :id, :string
     attribute :descrip, ::Tbx::Descrip

data/lib/tbx/descrip_note.rb

@@ -1,13 +1,13 @@
 # frozen_string_literal: true
 
 module Tbx
+  # Note element associated with a `<descrip>` description.
+  #
+  # Content is plain text only (no inline elements per the core RNG).
+  #
+  # Schema source: TBXcoreStructV03.rng `<define name="descripNote">`
   class DescripNote < 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
 
     xml do
       root "descripNote"

data/lib/tbx/document.rb

@@ -1,6 +1,18 @@
 # frozen_string_literal: true
 
 module Tbx
+  # Root element of a TBX (Termbase Exchange) document per ISO 30042:2019.
+  #
+  # The `type` attribute declares the dialect (e.g. "TBX-Basic", "TBX-Min",
+  # "TBX-Core", "TBX-Linguist"), and `style` declares the serialization
+  # style ("dca" for Data Category Archive or "dct" for Data Category Tagging).
+  #
+  # @example Parsing a TBX file
+  #   doc = Tbx::Document.from_xml(File.read("glossary.tbx"))
+  #   doc.type   # => "TBX-Basic"
+  #   doc.style  # => "dca"
+  #
+  # Schema source: TBXcoreStructV03.rng `<define name="tbx">`
   class Document < Lutaml::Model::Serializable
     attribute :lang, Lutaml::Xml::W3c::XmlLangType
     attribute :type, :string

data/lib/tbx/ec.rb

@@ -1,6 +1,22 @@
 # frozen_string_literal: true
 
 module Tbx
+  # XLIFF inline end-code marker `<ec>`.
+  #
+  # Used for representing inline formatting codes from source formats.
+  # Paired with `<sc>` via the `startRef` attribute.
+  #
+  # Attributes:
+  # * `startRef` (IDREF) — references the id of the corresponding `<sc>`
+  # * `isolated` (yes/no) — whether this code is isolated
+  # * `disp` — display text
+  # * `equiv` — equivalent text
+  # * `id` — unique identifier
+  # * `type` (fmt, ui, quote, link, image, other) — code type
+  # * `subtype` — sub-classification
+  # * `target` — target reference
+  #
+  # Schema source: TBXcoreStructV03.rng `<define name="ec">`
   class Ec < Lutaml::Model::Serializable
     attribute :start_ref, :string
     attribute :isolated, :string

data/lib/tbx/encoding_desc.rb

@@ -1,6 +1,9 @@
 # frozen_string_literal: true
 
 module Tbx
+  # Encoding description `<encodingDesc>`.
+  #
+  # Schema source: TBXcoreStructV03.rng `<define name="encodingDesc">`
   class EncodingDesc < Lutaml::Model::Serializable
     attribute :id, :string
     attribute :p, ::Tbx::P, collection: true

data/lib/tbx/file_desc.rb

@@ -1,6 +1,12 @@
 # frozen_string_literal: true
 
 module Tbx
+  # File description `<fileDesc>`.
+  #
+  # Contains file-level metadata including publication statement, title,
+  # and source description.
+  #
+  # Schema source: TBXcoreStructV03.rng `<define name="fileDesc">`
   class FileDesc < Lutaml::Model::Serializable
     attribute :id, :string
     attribute :publication_stmt, ::Tbx::PublicationStmt

data/lib/tbx/foreign.rb

@@ -1,6 +1,12 @@
 # frozen_string_literal: true
 
 module Tbx
+  # Foreign-language text element `<foreign>`.
+  #
+  # Marks text in a foreign language. The `xml:lang` attribute indicates
+  # the language. Content model: `entity.noteText` (mixed text + inline
+  # elements).
+  # Schema source: TBXcoreStructV03.rng `<define name="foreign">`
   class Foreign < Lutaml::Model::Serializable
     attribute :id, :string
     attribute :lang, Lutaml::Xml::W3c::XmlLangType

data/lib/tbx/hi.rb

@@ -1,7 +1,25 @@
 # frozen_string_literal: true
 
 module Tbx
+  # Inline highlighting element `<hi>`.
+  #
+  # Used within term, note, and other text content to mark up special text
+  # ranges. The `entailedTerm` type with `target` attribute creates inline
+  # cross-references to other concepts.
+  #
+  # Permitted `type` values (from Core RNG):
+  #
+  # * `entailedTerm` — cross-reference to another concept within term text
+  # * `hotkey` — keyboard shortcut designation
+  # * `italics` — italic text emphasis
+  # * `bold` — bold text emphasis
+  # * `superscript` — superscript text
+  # * `subscript` — subscript text
+  # * `math` — mathematical notation
+  #
+  # Schema source: TBXcoreStructV03.rng `<define name="hi">`
   class Hi < Lutaml::Model::Serializable
+    TYPES = Modules::CoreTypes::HI_TYPES.freeze
     attribute :target, :string
     attribute :type, :string
     attribute :content, :string, collection: true

data/lib/tbx/item.rb

@@ -1,6 +1,12 @@
 # frozen_string_literal: true
 
 module Tbx
+  # A single picklist value item `<item>`.
+  #
+  # Content model: `entity.noteText` (mixed text + inline elements:
+  # hi, ec, foreign, ph, sc).
+  #
+  # Schema source: TBXcoreStructV03.rng `<define name="item">`
   class Item < Lutaml::Model::Serializable
     attribute :id, :string
     attribute :type, :string

data/lib/tbx/item_grp.rb

@@ -1,6 +1,12 @@
 # frozen_string_literal: true
 
 module Tbx
+  # Item group `<itemGrp>` associating a picklist item with administrative data.
+  #
+  # Groups an `item` with its associated administrative data, notes,
+  # references, and transaction records.
+  #
+  # Schema source: TBXcoreStructV03.rng `<define name="itemGrp">`
   class ItemGrp < Lutaml::Model::Serializable
     attribute :id, :string
     attribute :item, ::Tbx::Item

data/lib/tbx/item_set.rb

@@ -1,6 +1,12 @@
 # frozen_string_literal: true
 
 module Tbx
+  # A set of items forming a picklist `<itemSet>`.
+  #
+  # Contains `itemGrp` or `item` children that enumerate individual
+  # picklist values.
+  #
+  # Schema source: TBXcoreStructV03.rng `<define name="itemSet">`
   class ItemSet < Lutaml::Model::Serializable
     attribute :id, :string
     attribute :type, :string

data/lib/tbx/lang_sec.rb

@@ -1,6 +1,12 @@
 # frozen_string_literal: true
 
 module Tbx
+  # Language section `<langSec>`.
+  #
+  # Groups all term-level data for a single language within a concept entry.
+  # The `xml:lang` attribute is required and specifies the language.
+  #
+  # Schema source: TBXcoreStructV03.rng `<define name="langSec">`
   class LangSec < Lutaml::Model::Serializable
     attribute :lang, Lutaml::Xml::W3c::XmlLangType
     attribute :admin, ::Tbx::Admin, collection: true

data/lib/tbx/modules/basic.rb

@@ -0,0 +1,92 @@
+# frozen_string_literal: true
+
+module Tbx
+  module Modules
+    # Data category definitions for the TBX-Basic module.
+    #
+    # TBX-Basic extends Min with additional data categories for standard
+    # terminology exchange including definitions, context, term types,
+    # and transaction tracking.
+    #
+    # Source: Basic.tbxmd, Basic.rng, Basic.sch
+    module Basic
+      ADMIN_TYPES = {
+        project_subset: "projectSubset",
+        source: "source",
+      }.freeze
+
+      DESCRIP_TYPES = {
+        context: "context",
+        definition: "definition",
+      }.freeze
+
+      TERM_NOTE_TYPES = {
+        geographical_usage: "geographicalUsage",
+        grammatical_gender: "grammaticalGender",
+        term_location: "termLocation",
+        term_type: "termType",
+      }.freeze
+
+      TERM_NOTE_VALUES = {
+        grammatical_gender: {
+          masculine: "masculine",
+          feminine: "feminine",
+          neuter: "neuter",
+          other: "other",
+        }.freeze,
+        term_location: {
+          check_box: "checkBox",
+          combo_box: "comboBox",
+          combo_box_element: "comboBoxElement",
+          dialog_box: "dialogBox",
+          group_box: "groupBox",
+          informative_message: "informativeMessage",
+          interactive_message: "interactiveMessage",
+          menu_item: "menuItem",
+          progress_bar: "progressBar",
+          push_button: "pushButton",
+          radio_button: "radioButton",
+          slider: "slider",
+          spin_box: "spinBox",
+          tab: "tab",
+          table_text: "tableText",
+          text_box: "textBox",
+          tool_tip: "toolTip",
+          user_defined_type: "user-definedType",
+        }.freeze,
+        term_type: {
+          full_form: "fullForm",
+          acronym: "acronym",
+          abbreviation: "abbreviation",
+          short_form: "shortForm",
+          variant: "variant",
+          phrase: "phrase",
+        }.freeze,
+      }.freeze
+
+      REF_TYPES = {
+        cross_reference: "crossReference",
+      }.freeze
+
+      XREF_TYPES = {
+        external_cross_reference: "externalCrossReference",
+        x_graphic: "xGraphic",
+      }.freeze
+
+      TRANSAC_TYPES = {
+        transaction_type: "transactionType",
+      }.freeze
+
+      TRANSAC_VALUES = {
+        transaction_type: {
+          origination: "origination",
+          modification: "modification",
+        }.freeze,
+      }.freeze
+
+      TRANSAC_NOTE_TYPES = {
+        responsibility: "responsibility",
+      }.freeze
+    end
+  end
+end

data/lib/tbx/modules/core_types.rb

@@ -0,0 +1,23 @@
+# frozen_string_literal: true
+
+module Tbx
+  module Modules
+    # Data category definitions from the TBX core RNG schema.
+    #
+    # These types are not module-specific; they are defined directly in
+    # TBXcoreStructV03.rng and apply to all dialects.
+    module CoreTypes
+      # Permitted `<hi>` element type values per TBXcoreStructV03.rng
+      # `<define name="hi">`.
+      HI_TYPES = {
+        entailed_term: "entailedTerm",
+        hotkey: "hotkey",
+        italics: "italics",
+        bold: "bold",
+        superscript: "superscript",
+        subscript: "subscript",
+        math: "math",
+      }.freeze
+    end
+  end
+end

data/lib/tbx/modules/linguist.rb

@@ -0,0 +1,46 @@
+# frozen_string_literal: true
+
+module Tbx
+  module Modules
+    # Data category definitions for the TBX-Linguist module.
+    #
+    # TBX-Linguist extends Basic with additional linguistic data categories
+    # including grammatical number, register, and transfer comments.
+    #
+    # Source: Linguist.tbxmd, Linguist.rng, Linguist.sch
+    module Linguist
+      ADMIN_TYPES = {
+        reading: "reading",
+      }.freeze
+
+      ADMIN_NOTE_TYPES = {
+        reading_note: "readingNote",
+      }.freeze
+
+      TERM_NOTE_TYPES = {
+        grammatical_number: "grammaticalNumber",
+        register: "register",
+        transfer_comment: "transferComment",
+      }.freeze
+
+      TERM_NOTE_VALUES = {
+        grammatical_number: {
+          singular: "singular",
+          plural: "plural",
+          dual: "dual",
+          mass: "mass",
+          other_number: "otherNumber",
+        }.freeze,
+        register: {
+          colloquial: "colloquialRegister",
+          neutral: "neutralRegister",
+          technical: "technicalRegister",
+          in_house: "in-houseRegister",
+          bench_level: "bench-levelRegister",
+          slang: "slangRegister",
+          vulgar: "vulgarRegister",
+        }.freeze,
+      }.freeze
+    end
+  end
+end

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

@@ -1,6 +1,11 @@
 # 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

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

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

data/lib/tbx/ph.rb

@@ -1,6 +1,12 @@
 # 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
 

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

data/lib/tbx/ref.rb

@@ -1,13 +1,24 @@
 # 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"

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

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

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