Extract Text from EPUB

This tool pulls the readable prose out of an EPUB and hands it back as plain text or Markdown, split by chapter. It suits a handful of concrete jobs rather than everything. Writers and self-publishers use it to recover the running text of a manuscript when the original document is lost and only the EPUB remains, then paste that text back into a word processor for revision. Students and researchers use it to lift passages or whole chapters for quoting, note-taking, or feeding into a search across their own files.

On the data-prep side, the Markdown output is convenient when you want chapter text in a format that drops cleanly into a documentation site, a static-site generator, or a notes app that understands headings. Anyone reviewing a book's actual word content - editors checking what shipped, localisers pulling source strings, or someone building a plain-text archive of titles they own - can get a clean, chapter-by-chapter copy without opening a reader app.

It is a reading-and-extraction tool, not a conversion or layout tool. If you need the styled book itself, a PDF, or a re-flowed e-reader file, this is the wrong tool; it deliberately discards formatting and keeps only the words.

You drop an EPUB onto the page or pick one with the file dialog; the file is read in your browser using a ZIP reader and an HTML parser, so nothing is uploaded. The tool reads the book's spine in reading order, opens each linear HTML/XHTML chapter, removes navigation, script, and style elements, and keeps the visible text. Each chapter gets a title - taken from the table of contents where available, otherwise the first heading, otherwise the file name - plus a word count.

You then see chapters listed with per-chapter word counts and tick boxes, with All and None shortcuts, and a running total of selected chapters and words. Choose Plain text (.txt), which writes each title with an underline rule, or Markdown (.md), which writes each title as a level-one heading. From there you download a file named after the original book, or copy the assembled text to the clipboard.

It accepts only .epub (EPUB 2 and EPUB 3). It does not handle MOBI, AZW, KFX, or PDF, and it will stop with a clear message if the file is DRM-protected, since that text cannot legally or technically be read. It extracts text only - images, tables, footnote layout, and styling are not preserved.

The biggest surprise is that structure is approximate. Chapters follow the EPUB's spine and table of contents, so a poorly built book may merge two chapters into one file, split one chapter across several, or surface front matter, copyright pages, and the TOC itself as separate entries. Titles fall back to a heading or file name when the TOC is thin, so they will not always match what you see in a reader.

Because only text is kept, anything that depended on layout is lost or flattened: tables become run-together lines, image captions may appear without their image, footnotes can land mid-sentence, and drop caps or small-caps styling vanish. Poetry and code blocks lose their line breaks, since whitespace is collapsed. Empty or near-empty sections (ten characters or fewer) are dropped entirely, which is usually right but can silently remove a short interstitial page.

Common mistakes: expecting MOBI or PDF to work, expecting a DRM-protected library or store book to extract, and forgetting that deselecting chapters changes both the download and the clipboard copy. If you get a 'no readable text' message, the book is likely image-only scans rather than real text.

Open the .txt or .md and skim from the top: confirm the chapter order matches the book, that titles line up, and that no front matter or the TOC slipped in as an unwanted 'chapter'. Spot-check a few chapter boundaries to make sure nothing was merged or split, and glance at the per-chapter word counts in the tool against the size you expect. Pay extra attention to tables, lists, footnotes, and any verse or code, since those are where flattening shows up.

If you only needed certain chapters, re-check that the selection you ticked is what actually came through, and that the word total looks sensible. Search the output for a distinctive phrase you know is in the book to confirm the text really extracted rather than producing empty or garbled sections.

For anything official, financial, or legal - contracts, regulatory filings, anything you will quote or act on - treat the extracted text as a draft and verify it line by line against the source EPUB. Automatic extraction can drop or reorder content, so a human read-through against the original is essential before the text is used in a record or relied upon.