Table of Contents

• 1. Core Principle
• 2. Schema System
• 3. Model Generation
• 4. Generated Classes
• 5. Benefits

Uniword v2.0 uses a schema-driven architecture where document classes are generated from complete OOXML specification coverage. This eliminates hardcoded XML and provides perfect round-trip fidelity.

1. Core Principle

NO RAW XML STORAGE — EVER. Every OOXML element is a proper lutaml-model class generated from YAML schemas.

2. Schema System

OOXML elements are defined in YAML files under config/ooxml/schemas/. Each schema file describes a namespace, its elements, attributes, and relationships.

Schema definition example (config/ooxml/schemas/wordprocessingml.yml)

# config/ooxml/schemas/wordprocessingml.yml
namespace:
  uri: 'http://schemas.openxmlformats.org/wordprocessingml/2006/main'
  prefix: 'w'

elements:
  p:
    class_name: Paragraph
    description: 'Paragraph - block-level text element'
    attributes:
      - name: properties
        type: ParagraphProperties
        xml_name: pPr
      - name: runs
        type: Run
        collection: true
        xml_name: r

The YAML schema defines:

• Namespace — URI and prefix for XML serialization

• Elements — Each element’s class name, description, and attributes

• Attribute types — References to other model classes or primitive types

• Collections — Whether an attribute can appear multiple times

3. Model Generation

Classes are automatically generated from schemas using the model generator:

require 'uniword/schema/model_generator'

generator = Uniword::Schema::ModelGenerator.new('wordprocessingml')
generator.generate_all
# => Generates 200+ lutaml-model classes from YAML schema

The generator produces Ruby classes that:

• Inherit from Lutaml::Model::Serializable

• Declare attributes using lutaml-model’s attribute DSL

• Define XML mappings in an xml do block

• Automatically enforce Pattern 0 (attributes before xml blocks)

4. Generated Classes

Generated classes provide full lutaml-model integration:

require 'uniword'

# Main document classes
doc = Uniword::Document.new
para = doc.add_paragraph("Hello World", bold: true)

# All classes support lutaml-model serialization automatically
xml = doc.to_xml                          # Automatic XML generation
doc2 = Uniword::Document.from_xml(xml)    # Automatic deserialization

5. Benefits

The schema-driven approach provides significant advantages:

100% ISO 29500 coverage

All 760 OOXML elements across 22 namespaces are modeled.

Zero hardcoding

All structure is defined in YAML, not in Ruby code. XML generation is handled entirely by lutaml-model.

Perfect round-trip

Complete modeling guarantees that documents can be loaded, modified, and saved without data loss.

Easy extensibility

New OOXML elements are added by editing YAML schemas, not by writing Ruby code.

Community contributions

Schema editing with YAML is simpler than writing Ruby model classes, lowering the barrier for contributors.

Type safety

All attributes have declared types with strong typing enforced by lutaml-model.