pdf-lib

pdf-lib is a pure JavaScript/TypeScript library for creating and modifying PDF documents without relying on native dependencies or a headless browser. Unlike most JavaScript PDF tools that can only generate new documents from scratch, pdf-lib can load and edit existing PDFs — adding pages, filling and flattening forms, embedding images and fonts, and drawing text or vector graphics directly onto pages.

Because it has zero native bindings, pdf-lib runs identically in Node.js, browsers, Deno, and React Native, making it a common choice for client-side PDF generation and for serverless environments where installing tools like wkhtmltopdf or a headless Chrome instance is impractical.

What You Get

• A PDFDocument API to create new PDFs or load and modify existing ones
• Full AcroForm support: create, fill, and flatten form fields (text, checkboxes, radio groups, dropdowns, option lists)
• Page manipulation: add, insert, remove, and copy pages between documents
• Drawing primitives for text, images (PNG/JPEG), vector graphics, and SVG paths
• Standard and custom font embedding with Unicode (UTF-8/UTF-16) support
• Document metadata and viewer-preference reading/writing, plus attachment embedding
• Runs with zero native dependencies in Node, browsers, Deno, and React Native

Common Use Cases

• Filling and flattening PDF forms server-side for contracts, applications, or tax documents
• Generating invoices, receipts, or reports entirely in the browser without a backend
• Merging or splitting existing PDFs by copying pages between documents
• Watermarking or stamping existing PDFs with text, images, or SVG graphics
• Programmatically embedding attachments or updating document metadata in a document pipeline

Under The Hood

Architecture — pdf-lib is organized around a layered API in src/: api/ exposes the high-level, user-facing classes (PDFDocument, PDFPage, PDFForm and field types under api/form/, PDFFont, PDFImage), while core/ implements the low-level PDF object model (dictionaries, streams, xref tables, the document parser/writer) that the API layer manipulates. PDFDocument acts as the central mutable in-memory representation: PDFDocument.load() parses an existing file’s object graph into this model, create() builds one from scratch, and save() serializes it back to bytes — the same representation drives editing and creation through one code path.

Tech Stack — Written in TypeScript, built with Rollup into CJS, ESM, and UMD bundles (plus a downlevel-dts pass for older TypeScript consumers) so it can be imported directly in Node, bundled for the browser, or loaded via <script> tag. It has no runtime dependencies on native binaries or headless browsers — the entire PDF object model and rendering-primitive logic is implemented in pure JS/TS, which is what allows identical behavior across Node, browsers, Deno, and React Native.

Code Quality — Tests live under tests/ as Jest specs (tests/api/*.spec.ts, plus dedicated suites for forms and text handling) run via jest --config jest.json --runInBand, with fixture PDFs used to assert against known-good outputs. The codebase enforces tslint and prettier in its lint scripts, and TypeScript’s tsc --noEmit is run as a standalone typecheck step. The project has been effectively feature-complete since ~2021, with recent commit activity dropping to near zero, so open issues around edge-case encodings/newer PDF spec features can persist longer than in an actively maintained project.

API Design — The PDFDocument/PDFPage/PDFForm object-oriented surface reads close to a document-editing DSL (pdfDoc.addPage(), page.drawText(...), form.getTextField(...).setText(...)), keeping common operations to a handful of chained calls. Every public method and class is documented with TSDoc comments that ship in the published types, and the README’s example-driven structure (create/modify/fill/flatten/copy, each with a live JSFiddle demo) makes the learning curve shallow for common tasks, though deeper PDF-spec concepts (encryption, low-level object references) require reading core/ directly.