PDF to EPUB

Pull the text out of a PDF, or paste text you copied, and make an e-book whose type size you can change. Nothing is uploaded; it is converted right here in your browser.

Drop a PDF here or click to choose Works best for single-column text: novels, essays, reports

The EPUB opens in Apple Books, Google Play Books, Kobo and most e-reader apps, and Kindle accepts it through Send to Kindle. Images, tables and equations are not included, and two-column layouts or scanned PDFs do not convert well.

About PDF to EPUB

What it is

A tool that pulls the text out of a PDF and makes an EPUB e-book. A PDF keeps the exact look of a printed page, so on a phone or e-reader the type is small and zooming in means scrolling sideways. As an EPUB the text flows to fit the screen, and you can change the type size and line spacing. Your file is never uploaded; it is converted only in this browser.

How to use it

Drop a PDF onto the page or click the box to choose one. Once every page has been read, a table of contents and a preview appear.

Choose Paste text to make a book from text copied from a web page, email or document, or from a TXT file. Headings on a web page become chapter titles; when tidying by hand, start a line with '# ' to make it a chapter title.

Title, author and language are filled in from the PDF's metadata and can be edited. For the cover, choose a picture of page 1, a cover made from the title, or none.

Chapters follow the PDF's bookmarks when it has them; otherwise the book is split at headings larger than the body text. You can also keep it as one chapter.

Click a chapter in the contents to check in the preview that paragraphs joined correctly, then press Make EPUB.

The EPUB opens in Apple Books, Google Play Books, Kobo and most e-reader apps, and Kindle accepts it through Send to Kindle.

Data and method

pdf.js (3.11.174), Mozilla's PDF engine, supplies the text fragments on each page with their coordinates. Fragments at the same height become lines, and lines become paragraphs based on line spacing, indentation, whether the previous line ended with sentence punctuation and whether it ran to the right margin. Text repeated in the same place on page after page (running headers and footers) and page numbers are removed. Whether a Korean word broken at a line end should be rejoined is decided by counting how often those characters appear joined or spaced elsewhere in the same document. The result is packaged as EPUB 3 (with a toc.ncx for older readers), zipped with JSZip (3.10.1).

Limits

Images, tables, equations and footnote placement are not carried over. Only text goes in.

Two-column layouts, magazines full of text boxes and vertical text can come out in the wrong order.

Scanned PDFs are pictures of pages with no text to extract, so they cannot be converted.

For Korean PDFs that break lines in the middle of words, spacing is inferred from statistics of the document. On a test document, 21 of 3,092 spaces were wrong; a PDF that breaks only between words had 1 error in the same test.

When pasted text is broken at every line, as text copied from a PDF viewer is, a paragraph is taken to end where the next line's first word would have fit on the line before. A paragraph whose last line is full can merge with the next one. On a test text, 72 of 80 paragraphs were recovered and none were split wrongly.

Converting a copyrighted book and sharing it can infringe copyright. Use it to make documents you own easier to read.