Ticket PDFs in Node.js

Why tickets become PDFs

A ticket has to work at a gate with no signal, on a phone that may be offline, or on paper. Anything requiring a live request fails at exactly the wrong moment. That is the difference between a document and a view of a database, and it is why event organisers and travel operators keep asking for this.

What a ticket has to carry

Before any of the code below matters, the template has to produce a document that is actually a ticket. These are the fields that make it one, and the ones a reader or an auditor will look for first.

• the scannable code, which is the ticket; everything else is decoration
• the event or journey, with date and time including timezone
• the seat, zone or class where the entitlement is specific
• the holder name where the ticket is non-transferable
• the terms of entry, which are what a gate refusal is justified by

The Node.js implementation

fetch ships with the runtime, so this adds no dependency. The render happens in a route handler that returns the response body directly, and you pipe the response body straight to res, so the PDF never lands in a Buffer.

app.get("/ticket/:id.pdf", async (req, res) => {
  const html = await renderTicket(req.params.id);

  const upstream = await fetch("https://api.pdfpipe.xyz/v1/pdf", {
    method: "POST",
    headers: {
      Authorization: `Bearer ${process.env.PDFPIPE_KEY}`,
      "Content-Type": "application/json",
    },
    body: JSON.stringify({
      html,
      options: { format: "A4", printBackground: true },
    }),
  });

  if (!upstream.ok) throw new Error(`render failed: ${upstream.status}`);

  res.setHeader("Content-Type", "application/pdf");
  // Streamed, not buffered: a long ticket never sits in memory.
  upstream.body.pipe(res);
});

The CSS that makes a ticket page correctly

The layout problem specific to this document is that a barcode or QR code must render at a size and contrast a scanner can actually read after printing. These rules handle it.

/* The code has to survive printing. Keep it black on white at a
   size a scanner can resolve, and never let it break across a page. */
.barcode {
  break-inside: avoid;
  image-rendering: pixelated;
  min-width: 45mm;
  filter: none;
}

@page { size: A6 landscape; margin: 5mm; }

What goes wrong in Node.js

Node buffers the whole response if you call res.send(await res.arrayBuffer()). On a 40-page report that is tens of megabytes held per concurrent request, and it is the usual reason a PDF endpoint takes down an otherwise healthy service under load.

What people try first

Most Node.js projects reach for puppeteer, which pulls a 300 MB Chromium download into your image and needs its own memory headroom on every instance. That works until it is running on more than one machine, at which point the browser becomes the thing you operate rather than the thing you use.

Where the ticket lives afterwards

Rendering is the short part. A ticket is worthless the moment the event ends, so nothing needs to survive it except the record that it was scanned. That record lives in your database, not in the PDF.

Getting the document right

• Check the code itself. A ticket that will not scan is worthless no matter how it looks.
• Generation is triggered by a booking being confirmed, so size the timeout for that path rather than for a health check.
• Volume arrives in bursts, so queue the render rather than doing it inside the request that triggered it.
• Backgrounds are painted by default here, so a design that uses colour needs nothing set. Only an explicit print_background of false turns them off.

Frequently asked

Related

Other Node.js documents, and the same ticket in other stacks.