PDF Generation

Producing PDFs like invoices and reports.

Backend Engineer · Mid-level Frontend Engineer · Mid-level

Also known as: pdf generation, generating pdfs, server-side pdf

PDF generation creates a PDF document on the server — an invoice, a receipt, a report, a shipping label, a ticket. Because PDFs render identically everywhere and are printable and archivable, they’re the standard format for these documents.

The usual approaches:

Template → PDF — render an HTML/other template (with a headless browser or a templating engine) and convert to PDF. Familiar to web developers, good for rich layouts.
PDF library — build the document programmatically with a PDF library, placing text and elements. More control, less HTML/CSS convenience.
Document templating — fill a pre-designed template (with placeholders) with data; good when the layout is fixed.

data → template → render → PDF → store → serve / attach

The classic mistakes:

Generating in the request synchronously. PDF rendering is CPU-heavy; doing it inline can block the request and time out. For anything non-trivial, generate in a background job (see job-queue) and hand back a link.
Font and script issues. PDFs embed fonts; missing fonts or non-Latin scripts (CJK, Arabic, emoji) render as boxes unless the font is included and the text shaping supported. Test with real content.
Page-break and layout surprises. Long tables, headers/footers and page breaks are where HTML-to-PDF gets fiddly; design templates with pagination in mind.
Not versioning/archiving the generated file. Financial documents (invoices) should be stored immutably, not regenerated on the fly each time — the rendered file is the record (see invoices and receipts).
Assuming pixel-perfect across engines. Different renderers/libraries produce subtly different output; pin your renderer and test.
Security of the renderer. A headless browser rendering untrusted HTML can be an attack surface; don’t render user-provided HTML unsandboxed.
Ignoring accessibility and text. Scanned or image-only PDFs aren’t searchable or accessible; generate real text where possible.
Large batches in memory. Generating thousands of PDFs at once exhausts memory; stream/store each and process in batches.

How to generate: choose a template-based or library approach, render in a background job for anything heavy, embed fonts, store the result in object storage, and serve or attach it. For documents that are records (invoices), archive the exact PDF rather than regenerating. See file uploads and data export.