Changelog¶

All notable changes to this project are documented in this file.

The format is based on Keep a Changelog, and this project adheres to Semantic Versioning.

Unreleased ¶

Added XML Schema validation of the ISO 23387 element grammar via Document::validate_schema, reporting structured SchemaViolation values with stable codes for unknown roots, undeclared elements and attributes, sequence order, minOccurs/maxOccurs, mandatory choice branches, enumeration and pattern facets, datatype lexical spaces, and character data in element-only content.
Added byte-identical round trips via Document::to_xml_string_exact, which reproduces the parsed bytes exactly — including attribute quoting, character reference spelling, intra-tag whitespace, CDATA versus escaped text, and empty-element syntax — and fails closed rather than normalizing when source provenance is unavailable.
Added Document::source and Document::without_source to inspect and drop the retained parse input.
Added validate-schema and rewrite-exact CLI commands.
Added scripts/generate-schema-tables.py, which regenerates the committed schema tables from a derived structural catalog without vendoring the ISO schema.

Relicensed repository-authored work from MIT to AGPL-3.0-or-later; historical releases remain under their published MIT terms, and third-party material retains its own terms.
Document, Element, and node lists now compare by semantic content; retained source spans are provenance and no longer affect equality.
Documented the two writers explicitly: to_xml_string is semantic and may normalize equivalent syntax, to_xml_string_exact is byte-identical.

Added validated GUID, XML Schema date-time/language/decimal/any-URI, multilingual text, reference, rational, data-type/value-list, unit-scale/base, and reusable ConceptType contracts.
Added a reusable arbitrary-precision XML Schema positiveInteger contract.
Modelled the four Annex E DataTypeType boundary variants and its repeating, implementation-defined regular-expression DataFormat values.
Added owned contracts for the ISO 23387 complex types imported by ISO 7817-3: subject, object type, property, group, quantity kind, reference document, dimension, unit, and data template.
Added a bounded, namespace-aware XML 1.0 parser and semantic writer retaining namespaces, ordering, comments, processing instructions, CDATA, and unknown elements and attributes.
Added typed views and owned wrappers for all five concrete global roots and the local Library child families represented by the Annex E XSD inventory.
Added structured built-in diagnostics with stable severity, category, and path fields while keeping parsing separate from validation.
Added inspect, validate, and rewrite CLI commands.
Added original synthetic XML fixtures with committed provenance and mutation probes for strict XML parsing, root classification, unknown-content retention, and safe CLI output replacement.

Replaced the unused openbim-core scaffold dependency with direct maintained roxmltree well-formedness checking and quick-xml event integration.
Updated documentation from namespace-scaffold status to the exact implemented 0.2 capability boundary without claiming XSD or clause-level conformance.

Reject malformed XML 1.0, duplicate expanded attributes, illegal namespace bindings, DTD declarations, undeclared entities and prefixes, invalid characters, and malformed qualified names; enforce configurable byte, depth, node, and per-element attribute limits.
Write CLI rewrites through an OS-random, exclusively created same-directory temporary file so pre-created symlinks cannot redirect the write.

Corrected global-root classification to the five concrete Annex E declarations while retaining typed local Library child views.
Made multilingual-text and reference diagnostics context-sensitive to the actual Annex E Symbol and *Ref declarations.
Made the owned Concept core require validated creation-date, name, and definition state and applied XML Schema whitespace collapsing to typed scalar values.
Reworked inherited namespace handling to use delta scopes and shared URI storage, preventing depth-amplified copies of ancestor namespace declarations.
Corrected DateTime extended-year and end-of-day fractional-second edges and aligned AnyUri with XML Schema 1.0's broad escaped lexical space.
Preserved tab, newline, and carriage-return character references in attribute values, plus carriage-return references in text, across repeated parse/write/ parse cycles.
Corrected ValueListType to require a validated language and one-or-more DataValue children whose optional order is bounded to XML Schema int, and restored missing-language diagnostics.
Connected the complete pre-standalone DT lineage across packages/openbim/openbim-dt, packages/openbim-dt, and packages/dt without rewriting the published v0.1.0 or v0.1.1 release commits.
Made the GitHub Pages upload include hidden files so the deployed artifact matches the tree validated by the documentation gate, including .nojekyll.

Established openbimrs/dt as the canonical standalone repository while preserving the data-template subtree history.
Corrected repository, homepage, documentation, and package metadata.
Added standalone CI, packaging gates, architecture documentation, roadmap, contributor guidance, and a verified GitHub Pages documentation pipeline.
Clarified that the released API is a namespace scaffold, not a complete ISO 23387 model, parser, writer, or validator.

Added fail-closed package and documentation checks preventing ISO/DIN/CEN references, schemas, PDFs, and annex examples from entering release artifacts.

Reserved openbim-dt with ISO 23387 edition 2 and draft-placeholder namespace constants.