A headless PDF viewer for React. Compose pages, text selection, search, and annotations into your own reading experience. Lector handles PDF.js rendering and page virtualization; you control the layout and UI.
Documentation · Live demo · npm · Contributing
Use React 19+. The current source expects pdfjs-dist ^5.5.207. These docs track main; check the releases against your installed package version.
npm install @anaralabs/lector pdfjs-dist@^5.5.207Copy the PDF.js worker from your app's installed dependency to its public assets. Recopy it whenever you update PDF.js so the runtime and worker versions match.
mkdir -p public/pdfjs
cp node_modules/pdfjs-dist/legacy/build/pdf.worker.min.mjs public/pdfjs/Put a PDF at public/sample.pdf, then add this browser-rendered component:
import { CanvasLayer, Page, Pages, Root, TextLayer } from "@anaralabs/lector";
import { GlobalWorkerOptions } from "pdfjs-dist/legacy/build/pdf.mjs";
import "pdfjs-dist/web/pdf_viewer.css";
GlobalWorkerOptions.workerSrc = "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/pdfjs/pdf.worker.min.mjs";
export default function PDFViewer() {
return (
<Root
source="/sample.pdf"
style={{ height: 600 }}
loader={<p role="status">Loading PDF…</p>}
>
<Pages>
<Page>
<CanvasLayer />
<TextLayer />
</Page>
</Pages>
</Root>
);
}For Next.js, load the viewer through a Client Component using dynamic(..., { ssr: false }); keep PDF.js setup inside the dynamically loaded module. See installation for the full wrapper, Vite worker setup, and deployments below a path prefix.
Root loads the document and provides its state. Pages owns scrolling and clones one Page template for visible pages. CanvasLayer paints the PDF; TextLayer adds selectable text. Give the viewer a definite height and import the PDF.js stylesheet so its layers align.
| Feature | Start here |
|---|---|
| Toolbar, page input, and layout | Your first viewer |
| Authenticated URLs, local files, errors, and self-hosted assets | Loading documents |
| Page navigation and fit width | Navigation, zoom |
| Page previews | Thumbnails |
| Text search and highlighted results | Search |
| Selection and citation regions | Selection, highlights |
| PDF links and editable form fields | Links, forms |
| Dark page rendering | Dark mode |
| Props, hooks, and defaults | API reference |
Lector is a toolkit rather than a finished toolbar or a PDF editor. Your app supplies accessible controls, error UI, and storage for user annotations. Search requires embedded text; it does not perform OCR. Custom highlight overlays do not automatically modify the PDF file. See troubleshooting for worker errors, blank pages, and layout issues.
Connect an MCP client to https://anara.com/lector/mcp using Streamable HTTP, with no API key. It can search and read every guide through tools and resources. See AI agents and MCP for setup and example requests.
For direct fetching, start with llms.txt, read individual Markdown guides, or use the complete documentation. These exports are generated from the same source as the website.
From the repository root, with Node.js 22.13+ and pnpm 9.5.0:
pnpm install --frozen-lockfile
pnpm --filter @anaralabs/lector build
pnpm --filter docs devOpen localhost:3000/docs. The docs app uses the local workspace package. For watch mode, validation commands, and testing in another application, read CONTRIBUTING.md.
Inspired by react-pdf-headless and pdfreader. Built on PDF.js and React.