EPUB to PDF

An EPUB is a ZIP of XHTML documents, and the order they appear in the archive means nothing. The book's reading order lives in the <spine> element of the package document, which lists the chapters by reference.

A converter that walks the ZIP alphabetically produces a book with chapter 10 after chapter 1, and an appendix in the middle. This one reads META-INF/container.xml to find the package document, parses the manifest and the spine, and follows the spine exactly. Chapter breaks start new pages.

This conversion throws away the single best property of an EPUB: text that reflows to fit any screen and any font size. A PDF has fixed pages, so on a phone you are pinching and panning where the EPUB simply reflowed.

That trade is the point of converting, and it cannot be undone afterwards — a PDF made from an EPUB does not become reflowable again. Keep the EPUB if you might want to read it on a small screen. Convert when you need to print it, or hand it to something that only accepts PDF.

Set the page size and font size to suit the destination. A5 at 11 pt reads like a paperback; A4 at 12 pt is what a printer expects.

Kept: the text, the reading order, headings at three levels, lists, block quotes, preformatted blocks, and the title and author from the book's metadata.

Not kept: the publisher's CSS — fonts, colours, drop caps, page furniture. Images are omitted, because placing them back into reflowed text needs to know where they sat in the original design and guessing produces something worse than leaving them out. Footnote links become plain text. The table of contents is not turned into PDF bookmarks.

What you get is a clean, readable, typeset document rather than a facsimile of the publisher's design.

PDF has fourteen built-in fonts that every reader already has, and they cover WinAnsi — Latin letters, accented European characters, curly quotes, the euro and pound signs. Nothing else.

Writing Chinese, Japanese, Korean, Arabic, Hebrew, Thai, Devanagari or Cyrillic means embedding a font that contains those glyphs, and a font with usable CJK coverage is several megabytes. Downloading that on the chance a visitor needs it is not a reasonable trade for a page that mostly converts English.

So instead of producing a document full of blank boxes, unencodable characters become a question mark and the conversion tells you exactly how many were affected and which script they belonged to. If your text is in one of those scripts, this is the wrong tool and you should use a desktop converter — that is a more useful answer than a broken file.

A book bought from Kindle, Kobo, Google Play Books or most other stores is DRM-protected, and its text is encrypted. It cannot be converted here or anywhere else without stripping the protection, which this site does not do. Such a file is detected and refused with an explanation rather than producing an empty PDF.

Books from Project Gutenberg, Standard Ebooks, most technical publishers and anything you made yourself are DRM-free and convert normally.