Markdown to PDF

No. Parsing, layout and PDF writing all happen in your browser. Besides the tool's own code, the only other thing it may fetch is font data from pdf.xyz's own server, when your text or file name goes beyond the Western European set: a Noto face for the text, from 137.8 KB for Bengali to 10.1 MB for Chinese, plus the 145.0 KB monospace face for code when that face can draw every character, and another face only if the first cannot shape every character. Those requests name a font and carry none of your document.

Mostly. The converter lays out # headings and underlined (=== or ---) headings; - * + and numbered lists, with nested items indented under their parent; > quotes; ``` and ~~~ fences and four-space indented code; ---, *** or ___ rules, drawn as a thin line; and GitHub-flavored pipe tables, drawn as a ruled grid with a shaded header row and the column alignment you set. YAML front matter at the top prints as a code block, and images print as their alt text and address instead of being embedded. Strikethrough markers are removed without drawing a line, raw HTML prints literally, and task-list boxes print as [ ] or [x]. Underscores inside words are left alone, so snake_case_name prints as typed, and backtick code keeps every character. Numbered lists keep the numbers you typed.

Colour emoji such as ✅, ⭐ or 😀 are pictures, not letters, and no emoji font is bundled, so a document containing one is refused rather than converted with gaps. A few symbols that also exist as plain characters, such as ✓ and ☀, are drawn in black from a Noto face instead. The same refusal applies to a script pdf.xyz has no font for, right-to-left scripts it would otherwise draw backwards, and text mixing two scripts no single bundled font covers, such as Chinese with Bengali. The message names the characters or scripts involved; remove them and click Create PDF again.

The bundled fonts cover Bengali, Latin, Greek, Cyrillic, Devanagari, Hiragana, Katakana, Kanji, Hangul and Han. Western European text uses the PDF's built-in fonts, with bold, oblique and monospaced faces. One character outside that set anywhere in the text or the file name, even a Polish ż or an arrow such as →, moves the body text to a single regular-weight Noto face: headings then differ by size only and quotes are upright. Code stays in a monospaced face as long as that face can draw every character in the document, which it does for Western European, Central European, Baltic, Turkish, Russian, Ukrainian, Bulgarian and Serbian text, box drawing and ← → ↑ ↓ ↔ ↕; one character it cannot draw, such as Greek, ✓, Bengali, Chinese, Japanese or Korean, sets the code in the Noto face too.

The output is real PDF text, so you can select, copy and search it. Lines break at spaces where they can. A string too long for the line, such as a URL, a minified line of code or a Chinese or Japanese paragraph written without spaces, is broken between characters and continues on the next line instead of running past the margin.

No. Every document uses US Letter pages with fixed margins and type styles, and there is no title field. The title and the small header on page one come from the loaded file's name, so rename the file before loading it to change them; text pasted before any file was loaded is titled “document”.

Not fully. The page and its processing engine load over the network, and parts of the engine are fetched the first time a step needs them. The processing itself runs on your device. Drawing text in a script outside the Western European alphabet downloads a matching font from our site the first time it is needed, and that step needs a connection.

Up to 2 MB per file.

Headings, paragraphs, bulleted and numbered lists, block quotes, fenced code blocks, horizontal rules and tables. Bold and italics print as plain text, and a link keeps its address in parentheses.

US Letter (8.5 × 11 in) portrait pages with 0.75 in margins and a page number (1 / N) at the foot of each page. There is no page-size or margin setting.