THotPDF.ValidatePDFUA2Structure Method

Runs bounded PDF/UA-2 structure validation over a producer or loaded object graph and returns stable structured diagnostics without mutating the document

Declaration

function ValidatePDFUA2Structure(
  const Options: THPDFUA2ValidationOptions;
  out Issues: THPDFUA2ValidationIssues): boolean; overload;

function ValidatePDFUA2Structure(
  const Options: THPDFUA2ValidationOptions;
  out Findings: THPDFComplianceFindings): boolean; overload;

Unit

HPDFDoc

Validation options

Use THPDFUA2ValidationOptions.Default for strict generation validation or adjust individual semantic policies for intake diagnostics

Field	Purpose
`HeadingPolicy`	Ignores, diagnoses or refuses skipped heading levels
`AllowDocumentFragmentHeadingStarts`	Resets heading context inside `DocumentFragment`
`CheckDocumentSemantics`	Checks Catalog language, metadata, displayed title, optional content and embedded files
`CheckHumanReadableStrings`	Refuses Unicode private-use characters in human-readable strings
`CheckAnnotationSemantics`	Checks page tab order, annotation ownership, roles, flags, descriptions and appearance streams
`CheckFormSemantics`	Checks AcroForm fields, widgets, labels, actions, options and rich values
`CheckNavigationSemantics`	Checks structure destinations, outlines and nested page-label number trees
`RequireListLabels`	Requires one `Lbl` for each numbered list item
`RequireExplicitTableHeaderAssociations`	Reports data cells that depend only on scope inference
`CheckListSemantics`	Checks numbering, continuation chains, captions, labels, bodies and direct real-content placement
`CheckTableSemantics`	Checks bounded grid geometry, row groups, spans, scopes and table-local header associations
`CheckRubyWarichuSequences`	Checks the exact direct-child sequences required for Ruby and Warichu
`CheckPronunciationHints`	Checks pronunciation entry types and indirect embedded PLS lexicons
`RequireStructureAssociatedFiles`	Requires every catalogue associated file to be reachable from structure
`RequireFormulaTextAlternative`	Requires `Alt`, `ActualText` or associated MathML for formulas
`MaxObjects`	Bounds indirect-object scanning and defaults to 1,000,000 objects
`MaxIssues`	Bounds retained diagnostics and defaults to 256 issues

Return value

Returns True only when no uisError diagnostic remains

Diagnostics

THPDFUA2IssueSeverity distinguishes warnings and errors, whilst each THPDFUA2ValidationIssue carries a stable namespaced code, applicable clause, object path, object number, zero-based page index when known and human-readable message in the THPDFUA2ValidationIssues array

Validation never creates a missing StructTreeRoot or namespace dictionary
Namespace-aware role maps are resolved transitively with cycle and depth detection
Document metadata, human-readable strings, optional content, structure elements, parent backlinks, attributes, annotations, forms, navigation, associated files, headings, lists, tables, Ruby, Warichu, pronunciation hints, formulas and MathML are checked in one bounded pass
Loaded annotation ownership is resolved from the physical ParentTree and supports scalar or array K entries without relying on producer state
Local navigation requires structure destinations, applies SD before a compatible D fallback and validates nested page-label number trees
XMP properties are resolved by namespace URI so valid alternate prefixes remain interoperable
Table diagnostics distinguish invalid spans, cell intersections, irregular rows and row groups, cross-table IDs, duplicate or cyclic header references and missing algorithmic associations
List diagnostics distinguish invalid numbering, unwrapped content, label and body structure, ordered same-level continuation references and continuation cycles
Expanded table geometry shares the MaxObjects bound and fails closed with ua2.table.limit.cells
ua2.limit.objects fails closed before scanning an oversized graph

Machine-readable findings

The THPDFComplianceFindings overload maps the same bounded validation pass into the shared compliance model without changing validation order or mutating the graph

Each finding identifies PDF/UA-2, severity, rule ID, standard clause, path, object number, page index, message and a crsNotAttempted repair state