Turn text into a PDF document

Runs on your device — nothing is uploaded

A PDF made by printing a web page looks like a printed web page. This one is measured instead: real margins, a readable line length, headings that step down in size, bullets that line up, and page numbers at the foot. Type it in, or bring text in from a file — or straight from the image tool, if what you have is a photograph.

How it works

1. Write or paste the textType it into the box, or drop a .txt or .md file on the panel. Pasted text keeps its line breaks. Sinhala and Tamil are laid out and printed properly, not as boxes or question marks.
2. Choose how it should lookPage size, margins, typeface and text size. Headings and bullets are found for you, and you can overrule them with two characters.
3. Download the documentA real PDF, ready to print or send. Nothing was uploaded and nothing was installed.

The two characters you need to know

Headings, bullets and emphasis are all understood from the way people already type. A line starting with # becomes a heading, ## a subheading. A line starting with - becomes a bullet, and 1. becomes a numbered item. Two asterisks around a word make it **bold**, one makes it *italic*, and two underscores make it __underlined__.

None of that is required. Plain text is laid out as paragraphs with sensible spacing, and if the text looks like it has headings in it — short lines, no full stop, sitting on their own — they are picked out and set in bold at the right size without being asked. The checkbox turns that guess off if it gets something wrong, and the # overrules it either way.

What it does not pretend to do

This is not a word processor. There are no tables, no images, no columns, no footnotes and no page-break controls, and there is no live preview — press the button and look at the PDF. A document that needs any of those needs a program built for it, and saying so is cheaper than half-supporting them.

The typefaces are the fourteen that every PDF reader already contains, which is why an English document adds nothing to download. They are not the fonts of any particular brand, and they will not be. What you get instead is a file that opens identically everywhere, on anything, forever.

Sinhala and Tamil are set properly, not approximated

Those fourteen typefaces stop at the Latin alphabet — which is why, until now, a Sinhala page would have come out as rows of question marks. Sinhala and Tamil now get a real font: Noto Sans Sinhala or Noto Sans Tamil, the same family the major operating systems use, in both regular and bold so that headings look like headings.

Embedding a font is the easy half. These two scripts are shaped: the characters are not drawn in the order they are typed. "ශ්‍රී" is four characters that print as one consonant with a mark hanging beneath it, and a reader that simply places one glyph after another produces something that looks almost right and is wrong in a way that is hard to name. So the text is run through the same shaping engine a browser uses, and the result was checked against HarfBuzz — the reference implementation — glyph for glyph before any of it was put on this page.

One consequence: the Typeface box on this page covers Latin text only. A Sinhala document is set in Noto Sans Sinhala whatever it says, because that is the choice that renders correctly.

Questions

How do I make a word bold?

Put two asterisks around it: **like this**. One asterisk gives *italic* and two underscores give __underline__. They can be combined, and they work inside headings and bullets as well as paragraphs.

Can I bring in text from an image?

Yes, and that is the usual way. Read the words with the Image to Text tool, then press “Download as document” on the result — it comes through here with the same layout rules already applied.

Does it add a font or a template?

No template, no cover page, no logo — the document contains your words and nothing else. For English it uses the fourteen typefaces every PDF reader has built in, so nothing is downloaded at all. For Sinhala and Tamil a font has to be embedded, because those scripts are not in the standard set; that font is about 250 KB and is fetched once, only when the text actually needs it.

Will justified text look wrong?

Justification is on by default and is deliberately cautious: short lines and the last line of every paragraph are left alone, and a line is only stretched when the slack is small. If a line looks odd, turning the checkbox off gives you ordinary left-aligned text, which never has that problem.

Can I control where the page breaks?

No. Pages break where the text runs out of room. That is a real limitation of a tool this size, and the alternative — guessing at page breaks and getting them wrong — is worse.

What if my text is in another language?

English, Sinhala and Tamil are supported, and they can be mixed in one document — an English address under a Sinhala letter is a normal thing to want. A word in a script that is neither is replaced by a question mark rather than silently dropped, so you can see it happened. Chinese, Arabic, Cyrillic and the rest need a font this page does not carry.

Why is my Sinhala document not in the typeface I chose?

Because Sinhala and Tamil are not in the standard typefaces at all, so the choice has nothing to apply to. They are set in Noto Sans Sinhala or Noto Sans Tamil, in regular and bold; the message under the result says so every time. The Typeface box still governs any Latin text in the same document.

Related guides

Other tools