Markdown to PDF API

The whole thing

One call makes the document. A second, later, fetches the PDF — rendered at the moment you ask, so it always matches the document as it stands.

curl -s https://docubend.com/api/v1/pdfs \
  -H "Authorization: Bearer $TOKEN" \
  -H "Content-Type: application/json" \
  -d '{
    "title": "Nightly build",
    "markdown": "# Nightly build\n\n**All green.** 412 tests, 0 failures.\n\n| Suite | Time |\n| --- | --- |\n| unit | 41s |\n| e2e | 2m 03s |"
  }'

const BASE = "https://docubend.com";
const head = { Authorization: `Bearer ${TOKEN}` };

const report = `# Nightly build — 14 August

**All green.** 412 tests, 0 failures, 4m 06s.

| Suite | Tests | Time |
| --- | --- | --- |
| unit | 288 | 41s |
| end to end | 28 | 2m 03s |

> Deployed to production at 03:12 UTC. No rollback.`;

const r = await fetch(`${BASE}/api/v1/pdfs`, {
  method: "POST",
  headers: { ...head, "Content-Type": "application/json" },
  body: JSON.stringify({ title: "Nightly build", markdown: report }),
});
const { document: doc } = await r.json();

import requests

BASE, head = "https://docubend.com", {"Authorization": f"Bearer {TOKEN}"}

report = """# Nightly build — 14 August

**All green.** 412 tests, 0 failures, 4m 06s.

| Suite | Tests | Time |
| --- | --- | --- |
| unit | 288 | 41s |
| integration | 96 | 1m 22s |
| end to end | 28 | 2m 03s |

## What changed

- api: documents can be replaced in place, keeping their URL
- editor: connectors re-route around shapes added after them

> Deployed to production at 03:12 UTC. No rollback.
"""

doc = requests.post(f"{BASE}/api/v1/pdfs", headers=head,
                    json={"title": "Nightly build", "markdown": report}
                    ).json()["document"]

open("report.pdf", "wb").write(requests.get(doc["links"]["pdf"], headers=head).content)

Real text, not a picture of text

The words are written into the file as text in the standard PDF fonts, so they can be selected, searched and copied, the file is a few kilobytes rather than a few megabytes, and it prints at full resolution on anything. Line breaking uses the same metrics the editor draws with, so a line that fits on screen fits on the page.

Supported: headings, bold, italic, strike-through, inline and fenced code, ordered and unordered lists with nesting, tables, links, images, blockquotes and rules. Anything outside that vocabulary is left as the words you wrote rather than half-applied — a half-supported extension that silently eats a paragraph is worse than one that does nothing.

What it does not do, so you can decide before you build

There is no CSS. One column, sensible typography, and no control over page breaks, floats, fonts or margins. If what you have is a styled HTML invoice and you need it to come out looking exactly like the stylesheet says, you want a service that runs a headless browser, and you should use one — this is not that, and a page telling you otherwise would be wasting your afternoon.

Characters outside the standard fonts become a question mark — CJK and emoji, mostly — and the response says so in an X-Docubend-Warning header rather than leaving you to find them in a report somebody has already sent on.

Pictures, and where they go

Send images alongside the Markdown as base64, keyed by the name you use in ![](chart.png), and each one is placed where it appears in the text. PNGs and JPEGs go into the PDF as themselves rather than being decoded and re-encoded, so a chart does not double in size on the way in. A whole screenshot with no Markdown at all is a single call of its own.

One URL, regenerated for ever

The usual reason people wire this up is a report that runs every night. Make the document once, keep its id, and PUT to it on a schedule: the contents are replaced, the URL does not change, and the version it replaced stays in the document's history. The link you handed round in March still opens tonight's numbers, and anybody can open it in the editor to annotate or comment on what they see.

The rest of it

The full API reference — every call, its parameters and what it answers. Or turn mermaid into a diagram the same way, and the sixteen free PDF tools that need no account at all.