Decide what shape you need
“Extract the text” can mean four different things, and choosing the right one saves a lot of cleaning. The EPUB Text Extractor offers all four on the same loaded book.
| You want | Mode | Output |
|---|---|---|
| Everything, to copy or search | Full text | One UTF-8 text file |
| One chapter at a time | One text file per chapter | ZIP with numbered files |
| Memorable lines | Quotations | List of passages inside quotation marks |
| Everything about a word or a name | Passages containing… | List with context |
Steps
- Choose the EPUB.
- Pick the mode; for passage search, type the word or phrase.
- Press Extract.
- Copy from the page or press the download button.
Everything runs in your browser. The book is not uploaded.
How each mode works
Full text walks the book’s reading order and writes headings, paragraphs, list items and quotations on separate lines. Chapters writes one file per reading-order item and names it with the table-of-contents entry when there is one. Quotations matches text between curly, straight or guillemet quotation marks that is 20 to 600 characters long; that excludes short exclamations and runaway matches. Passage search finds every occurrence ignoring capitalisation and shows around 160 characters on either side.
Working with the result
- For citations: keep the chapter file number or table-of-contents title with each excerpt. Page numbers do not exist in a reflowable EPUB, so cite by chapter.
- For translation or study: chapter files are easier to load into a translation memory or a flashcard tool than one huge file.
- For text-to-speech: the full text works. Remove the front-matter chapters first if you do not want the copyright page read aloud.
- For comparing editions: extract both and use a diff tool on the plain text; it ignores formatting differences.
Limits you should know
- Quotation detection depends on punctuation. A book that marks dialogue with dashes returns few or no results.
- Only content in the reading order is exported. Files that exist in the archive but are not in the spine are skipped.
- Footnotes are exported where the book places them; the clickable links are removed.
- DRM-protected books cannot be read.
- Scanned image-only books have no text to extract. They need OCR, which is a separate process.
If you want structure kept, such as headings and emphasis, convert to Markdown. For a single clean file with no choices, EPUB to TXT is quicker.
A research workflow
Suppose you are writing a paper that discusses one novel and need accurate quotations with locations. A workable routine with the extractor is:
- Extract one text file per chapter. Keep the ZIP with your notes; the file names carry the chapter order and title.
- Use passage search for each term or character name you are tracking. Copy the passages you want into your notes together with the chapter file name.
- Use quotations to find lines the author set in quotation marks, which is a fast way to list speech for dialogue analysis.
- Check each quotation against the book in the reader before it goes into a manuscript, because extraction can join or split text around footnotes.
Cleaning typical artefacts
| Artefact | Cause | Remedy |
|---|---|---|
| Words split with a hyphen at line ends | The book was converted from a print layout | Search for a hyphen followed by a line break and join the parts. |
| Footnote numbers glued to words | Superscript numbers become plain digits in text | Remove digits that directly follow a word with a regular expression. |
| Running headers repeated | Print page headers preserved in the file | Search for the repeated header line and delete it. |
| Missing paragraph breaks | The book uses line breaks instead of paragraphs | Add blank lines at sentence ends if you need paragraphs. |
What the passages output looks like
Each result shows the match with about 160 characters on either side, trimmed to whole spaces and marked with an ellipsis where the passage was cut. It is a reading aid: for a citation, go to the book and check the surrounding paragraph.
Respecting rights
Extracting text for your own study, accessibility or analysis is a different matter from republishing it. Check the copyright and licence of the book before sharing extracted text or large quotations.