Article to EPUB
Paste an article — plain text or cleaned HTML — and download a valid EPUB 3 ebook with proper metadata, chapters split at your headings and a working table of contents. Built in your browser; nothing is uploaded.
About this tool
This converter packages an article you already have — the text you drafted, or the cleaned HTML you saved from a reader view — into a valid EPUB 3 ebook you can side-load onto a Kindle, Kobo, Apple Books, Calibre or any other reader.
Everything happens in your browser. The article is never uploaded, there is no account, and the file you download is built locally by a small WebAssembly module.
What you get
Every book contains a complete OCF container: the mimetype entry stored first
and uncompressed as the spec requires, META-INF/container.xml, a content.opf
package document carrying your metadata, an EPUB 3 navigation document and
an EPUB 2 toc.ncx so older readers still show the table of contents, a
readable stylesheet, and one XHTML file per chapter.
Worked example
Paste this as plain text, set the title to Field Notes and leave the chapter
split on “Heading 1”:
# Tide pools
At low tide the rocks turn into a museum.
- Ochre sea stars
- Green anemones
# Getting there
Park early; the lot fills by nine.
You get field-notes.epub: two chapters (“Tide pools” and “Getting there”), a
two-entry table of contents, the bullets as a real <ul> list, and 24 words
counted in the summary line above the download button.
Reading messy HTML
Pasted article HTML is rarely tidy. Unclosed <p> tags, a stray </div>, a
leftover <script> block or a raw will all make a strict reader refuse
to open a book. This converter re-emits your markup as well-formed XHTML:
tags are balanced with a proper stack, scripts, styles, forms and embeds are
dropped along with their contents, unsafe links are unwrapped (their text is
kept), and named entities such as , — and é become real
characters.
The tags that survive are the ones books are made of: headings, paragraphs,
lists, blockquotes, preformatted blocks, tables, and inline emphasis, code,
sub/superscript and http(s)/mailto links.
Plain text and Markdown-ish input
In text mode, blank lines separate paragraphs, wrapped lines are joined back
into flowing prose, and a few Markdown conventions are recognised: # to ######
headings, -, *, + or • bullets, 1. numbered items, and --- for a
horizontal rule. Inline Markdown (**bold**, [links](…)) is not converted —
it stays as literal text.
Limits and behaviour worth knowing
- Up to 2,000,000 characters of input and 2,000 chapters per book.
- Images are not packaged. A pasted
<img>points at a remote file this offline converter can't embed, so images are dropped and theiralttext is kept inline as[Image: …]. The count is reported with the download. - Output is deterministic: the same article and metadata always produce a byte-identical file.
- Cover images, embedded fonts and MOBI/AZW3 output are out of scope here.
FAQ
Is the EPUB actually valid, or just a renamed ZIP?
It is a real OCF container. The mimetype entry is written first and stored
uncompressed, META-INF/container.xml points at an EPUB 3 content.opf, and the
book ships both an EPUB 3 nav.xhtml and an EPUB 2 toc.ncx. Every chapter is
well-formed XHTML with a declared language, which is what strict readers check
before they will open a file.
How do chapters and the table of contents get decided?
By the Start a new chapter at setting. With Heading 1 (the default) every
<h1> — or # line in text mode — begins a new chapter, and the heading text
becomes that chapter's entry in the table of contents. Choose Heading 2,
Heading 1 and 2 for a finer split, or No split to keep the whole article in a
single chapter. Content that appears before the first heading becomes the opening
chapter, titled after the book.
What happens to images in my article?
They are removed, and any alt text is kept in the flow as [Image: description].
An <img> in a pasted article points at a URL on someone else's server; a
browser-local converter can't download and embed those bytes, and an EPUB that
links to remote images shows blank frames on most e-readers. The summary line
tells you how many images were dropped.
Do I have to fill in the title, author and language?
No. Leave the title empty and it is taken from the HTML <title> element, then
the first heading, then the first line of text — and it is also used for the
download filename. Author and publisher are simply omitted when blank. Language
defaults to en; set any BCP-47 tag such as de, fr or pt-BR so readers
hyphenate and read the book aloud correctly.
Should I set a base font size?
Usually not. Leaving it at 0 lets the reading app apply the reader's own font
preference, which is what e-reader users expect. Set a value in points (say
12) only when you are producing a book for a fixed setup — it is written into
the book's stylesheet, and most apps still let the reader override it.
Can it fetch an article straight from a URL?
Not here — this tool is deliberately offline and never makes network requests. Copy the article text, or use a reader-view/extraction tool first and paste the cleaned HTML. In chat you can chain a fetch or HTML-extraction skill into this one.
Developer & Automation Access
Run it from the terminal
Same engine as this page, headless — via the gizza CLI:
gizza tool article-to-epub "# The Tide Pools of Muir Beach
At low tide the rocks north of the parking lot turn into a museum.
## What to look for
- Ochre sea stars
- Green anemones"New to the CLI? Get gizza →
Open it by URL
Pre-fill and auto-run this tool with query parameters — the names match the API/CLI:
https://gizza.ai/tools/article-to-epub/?content=%23%20The%20Tide%20Pools%20of%20Muir%20Beach%0A%0AAt%20low%20tide%20the%20rocks%20north%20of%20the%20parking%20lot%20turn%20into%20a%20museum.%0A%0A%23%23%20What%20to%20look%20for%0A%0A-%20Ochre%20sea%20stars%0A-%20Green%20anemones&input_format=auto&title=The%20Tide%20Pools%20of%20Muir%20Beach&author=Jane%20Doe&language=en&publisher=Coastal%20Review&split_level=h1&include_title_page=true&base_font_size=0Machine-readable descriptor: tool.json — title + parameters JSON Schema for agents.