About Text to PDF

Two ways in, one way out. Paste or type text and the PDF is rebuilt as you go, so the preview on screen is the file the Download button hands over. Or switch to the file side and drop up to 20 plain-text files at 3 MB each — .txt is the obvious one, but a server log, a Markdown note, a CSV export, a subtitle file and a SQL dump are all just characters, and all convert the same way.

What comes out is a PDF, not a picture of one. Pages can be A4, US Letter or US Legal, portrait or landscape; text can be sans-serif, serif or monospaced from 8 to 16 pt, single, one-and-a-half or double spaced, inside margins of half an inch, one inch or an inch and a half. Line numbers, page numbers and a header carrying the file name are each one checkbox. A batch can stay as separate PDFs or be merged into one document where every file starts on its own page.

Plain text carries no layout, so all of it has to be decided here, in four steps that all run in the browser tab:

1. Read the bytes. A dropped file announces no encoding, so the byte order mark decides when there is one; otherwise a strict UTF-8 pass is tried, and files that fail it are read as Windows-1252 — the encoding behind decades of Notepad documents. UTF-16 without a mark is caught by the zero bytes it leaves behind.
2. Normalise. Windows and classic Mac line endings become plain line feeds, invisible characters that no font can draw are dropped, tabs stay tabs, and a form feed stays a page break.
3. Lay out. Tabs expand to the tab stop, then each line is measured against the metrics of the font it will be drawn in. A line that fits is placed as it is; one that does not breaks at a space and carries the original indentation onto the continuation, and a single unbroken run — a URL, a hash, a base64 blob — is split by character rather than pushed off the page.
4. Draw. Each page is written as one PDF text object holding real text operators, which is what keeps the output searchable, selectable and small.

Fonts are the interesting part. Every PDF reader already carries a standard set of faces, so text that fits it — English, most of Western Europe, curly quotes, the euro sign — is drawn with no font embedded at all and nothing to download; that is why a page of English prose lands under two kilobytes. The moment a character falls outside that set, the matching face — Noto Sans, Noto Serif or Noto Sans Mono — is fetched from this site and embedded as a subset holding only the glyphs the document uses, which covers Latin diacritics, Greek and Cyrillic. Box drawing, arrows and mathematical symbols live in the monospaced face only, so the tool counts them and offers the switch instead of printing empty rectangles.

Chinese, Japanese, Korean, Arabic, Hebrew, Thai and emoji are not covered: those families are multi-megabyte downloads on their own, which does not belong in a tool that starts converting the moment you paste. Those characters become a question mark and the count is reported, so nothing disappears quietly.

A document may run to 1,000 pages; past that the converter stops and says so, because a browser tab spending a minute laying out thousands of pages is worse than an honest error. Files are capped at 3 MB and batches at 20 files for the same reason.

And it is a converter, not a word processor: plain text has no bold, no headings and no tables, so nothing here invents them. Formatted documents belong in a tool that understands their markup.

The four steps above all happen on your device. Your text is not uploaded, not logged and not stored: it lives in the tab's memory while you work, and goes when the tab goes. The finished PDF is held the same way, as a blob the browser keeps until you close the page — which is also why reloading loses your work.

While converting, the only request this site can make is for one of its own font files, and only when your text needs one. Nothing about the request describes your document. There are no accounts, so there is nothing for a document to be attached to.

This covers what Text to PDF does. It cannot speak for a browser extension, a corporate proxy or the device itself — only for the fact that the converter sends your text nowhere.