OfficeIMO.Reader.Xps 3.5.1

OfficeIMO.Reader.Xps

OfficeIMO.Reader.Xps registers .xps and .oxps ingestion through the native OfficeIMO.Xps engine. It reads literal Unicode, native story order, page identity and tables without rendering or requiring a desktop XPS viewer.

using OfficeIMO.Reader;
using OfficeIMO.Reader.Xps;

OfficeDocumentReader reader = new OfficeDocumentReaderBuilder()
    .AddXpsHandler()
    .Build();

OfficeDocumentReadResult result = reader.ReadDocument("document.oxps");
foreach (OfficeDocumentBlock block in result.EnumerateBlocks())
    Console.WriteLine($"Page {block.Location.Page}: {block.Text}");

An already loaded XpsDocument supports ToOfficeDocumentReadResult() without serialization or reopening. XpsDocument.ToOfficeDocumentModel() owns the reusable projection, including recursive native structure, list markers and cell spans. Reader retains bounded text chunks, physical page locations, native links, rectangular tables, diagnostics and optional SVG preview assets. Spanned table positions are blank in the rectangular projection; the native structure retains the span values. Grids exceeding 1000 columns or containing row spans beyond the available rows are omitted with a diagnostic; their text and recursive native structure remain available. Page Markdown emits each represented cell once, including nested tables. Rows outside a truncated grid remain as text. Logical tables that span pages remain at document scope; each physical page retains its own text instead of receiving another page's cells.

Authored Path and Canvas accessibility descriptions are payload-free graphic-description assets. Title retains AutomationProperties.Name; AltText contains the name followed by AutomationProperties.HelpText, separated by a newline when both are present. The assets retain page locations and survive JSON transport. Descriptions do not create text blocks, chunks or native story structure. Missing descriptions are not inferred.

ReaderXpsOptions.ReadOptions controls package expansion, part, page and XML limits. Registration snapshots these limits, and ReaderOptions.MaxInputBytes can tighten them. Stream input follows Reader's whole-stream snapshot contract, restores a seekable stream's position and leaves the caller's stream open. Source hashes cover the captured package bytes. A mutable loaded document has a stable logical source ID but no inferred package hash.

var readerWithPreviews = new OfficeDocumentReaderBuilder()
    .AddXpsHandler(new ReaderXpsOptions { IncludeSvgPreviewAssets = true })
    .Build();

Previews use strict native SVG conversion; unsupported paint fails instead of producing an incomplete preview. Preview projection is bounded to 512 pages and 128 MiB of preview payload. Description-only assets do not consume the preview page limit. The native package's rendering support matrix applies.

Native story order can cross physical pages. ReaderLocation.LogicalOrder retains that order through canonical block/content traversal and JSON transport. Missing structure falls back to page/markup order with a diagnostic. Unreferenced Unicode follows declared stories. Headers and footers retain their own block kinds. Glyph IDs are not reverse-mapped to Unicode, and text inside brush visuals is excluded. Overlapping text ownership in native structure is rejected.

Content-based detection recognizes bounded atomic OPC root relationships. Packages with interleaved root relationships use their .xps or .oxps extension with the default detection mode; their contents are assembled by the native package loader.

The adapter references only OfficeIMO.Reader.Core and OfficeIMO.Xps. See the native support matrix and PDF export API.