Original Reflow Ebook Builder

How to use Original Reflow Ebook Builder

Turn a small set of original text chapters into one portable reading draft. Each file supplies a title line and literal body text. Set the publication metadata and chapter sequence, then download a self-contained EPUB or its complete build evidence without uploading your writing.

1. Choose your own UTF-8 text chapters, or load the two original sample files. Put a nonblank chapter title on the first line of each file and body text below it.
2. Enter a book title, a language tag, a stable urn:uuid identifier and a real UTC modification date. Add the creator only when you intend to publish that name.
3. List exact filenames in the intended order, or leave the order empty to keep the provided selection sequence. No chapter is dropped, sorted or guessed.
4. Run the local build. Review chapter titles, file hashes, normalization and resource order in the evidence. Download the EPUB and inspect its contents in an EPUB 3 reader before distributing it.

Good to know

• This is an original text-only EPUB 3 draft builder, not an ebook reader, existing-book editor, Office/PDF converter, full conformance checker, bookstore submission service or accessibility certification. Reader fonts, navigation widgets, line breaking and display vary. A completed download does not establish acceptance by a particular reading system or store.
• Choose 1–20 distinct local chapter basenames, each with 1–65,536 UTF-8 bytes; aggregate source is at most 524,288 bytes. Each decoded source has at most 50,000 UTF-16 units and the batch at most 200,000. Combined local reads have one 5-second deadline. A read already requested from the browser may finish later, but it cannot restore a failed result.
• The batch permits 10,000 logical lines and 5,000 nonblank paragraphs. Chapter/book titles and optional creator have at most 160 UTF-16 units; filenames have 200 and no path separators. Invalid UTF-8, unpaired surrogate input, C0/C1 controls other than source tab/line endings, Unicode noncharacters and invalid or ambiguous order are refused atomically.
• One initial BOM is an encoding marker. CRLF and bare CR are normalized to LF. The first line becomes the heading; spaces/tabs-only blank lines separate paragraphs and repeated blank separators collapse. Nonblank lines inside a paragraph retain an explicit line break. This is a declared text mapping rather than source-byte preservation or a rich-text format. Original files remain unchanged.
• Only fixed generated XHTML, CSS, metadata, table-of-contents links and archive paths are created. User text is escaped; literal scripts, URLs and tags remain text. No remote resource, image, embedded font, JavaScript, encryption or DRM is added, mounted, fetched or evaluated. It does not import existing EPUB or HTML files.
• Required metadata uses one Intl-canonicalizable language-tag subset and an explicit urn:uuid identifier. The UTC date must be a real Gregorian YYYY-MM-DDThh:mm:ssZ with a year from 0001 through 9999; leap-second spelling is unsupported. Choosing a stable identifier and meaningful date remains your authoring decision.
• Both complete artifacts are checked before either mode returns: EPUB at most 3 MiB; build evidence at most 250,000 UTF-16 units and UTF-8 bytes. JSON records original basenames, authored metadata, source/resource hashes, text normalization and reading order. It is not a backup of the original raw files. ZIP timestamps use fixed DOS date fields and the authored UTC metadata is separate.
• All processing is local with no login, persistence, analytics of the manuscript or publication action. The EPUB intentionally contains the authored book content and metadata; review both it and the separately downloaded evidence before sharing.