Publishing quality · 12 min read
EPUB Structure, Metadata, Navigation and Accessibility Guide
Understand the EPUB container, package metadata, manifest, spine, navigation document, language tags, validation, and accessibility review.
Published and technically reviewed by the PDFStack Team ·
An EPUB is a publication package, not a renamed document
An EPUB file is a ZIP-based container with defined resources and relationships. A reading system locates the package document through META-INF/container.xml. The package identifies metadata, declared resources in the manifest, and the primary reading sequence in the spine.
PDFStack creates the required mimetype entry, container file, EPUB 3 package document, navigation document, and one XHTML content document per PDF page. It also generates a unique identifier and modification timestamp for each conversion.
Metadata helps reading systems identify the publication
The package includes a title, language, identifier, and optional creator. A meaningful title is better than a raw filename, and creator metadata should reflect verified authorship rather than a guessed PDF property.
Language should be a valid BCP 47 tag such as en, en-US, ar, or fr-CA. Correct language metadata helps reading systems, search tools, pronunciation engines, and assistive technology. The current converter applies one language value to the full publication; multilingual books may need chapter-level editorial refinement.
- ✓Use a human-readable publication title.
- ✓Add author only when it is known.
- ✓Choose a valid language tag for the majority content.
- ✓Treat metadata as bibliographic information, not decoration.
The manifest and spine have different jobs
The manifest lists resources that belong to the publication. PDFStack declares the navigation document and every page chapter. The spine determines the linear reading order and references those chapters in PDF page order.
Changing a filename inside an EPUB without updating its manifest and links can break the book. When editing the output, use an EPUB-aware tool or update package references consistently.
Navigation must be useful as well as valid
EPUB 3 requires a navigation document with a table-of-contents nav element. PDFStack creates that document and links every page chapter as Page 1, Page 2, and so on. This provides deterministic movement through the converted source.
A publication-ready ebook should replace generic page labels with meaningful chapter and section names where the source supports them. A page-by-page contents list is accurate to the conversion but less helpful than semantic navigation for a long book.
Reflow improves adaptability but can remove visual meaning
Reflowable content adapts to different screens and reader settings. That can improve comfort and access, but complex layouts may depend on spatial relationships that plain paragraphs cannot express. Tables, callouts, equations, charts, footnotes, and side-by-side comparisons may lose meaning when reduced to extracted text.
Review whether the source is suitable for reflow. A novel or simple report is a stronger candidate than a graphic textbook, form, brochure, or engineering drawing. Fixed-layout EPUB or manually authored semantic HTML may be more appropriate for design-dependent content.
Accessibility requires semantic and editorial work
A technically readable EPUB is not automatically accessible. Meaningful headings, logical reading order, descriptive links, image alternatives, correct language, sufficient metadata, and predictable navigation all contribute to access.
The current converter creates page headings and paragraphs but does not infer a full heading hierarchy, landmarks, image descriptions, accessibility metadata, or table semantics. Use the output as a structured draft, then assess it against EPUB Accessibility and WCAG requirements appropriate to the publication.
Validate structure and test behavior
Run EPUBCheck to find packaging, XML, manifest, navigation, and specification problems. Then open the book in multiple reading systems, resize text, switch themes, use the table of contents, search content, and test keyboard or screen-reader navigation when required.
Conformance, accessibility, and editorial accuracy are related but separate. A book can pass EPUBCheck and still contain OCR mistakes or weak navigation; it can also read well in one application while exposing compatibility problems in another.
Frequently asked questions
What is the EPUB spine?
The spine lists publication resources in their primary reading order. PDFStack adds page chapters to the spine in source page order.
Does EPUBCheck guarantee an accessible ebook?
No. EPUBCheck evaluates technical conformance; accessibility and editorial quality require additional testing and review.