pdf-raster

Node.js and Bun

Convert PDFs from disk or from memory in scripts, workers and servers.

These examples run the same way in Node.js 24+ and Bun.

Save every page to disk

import { mkdir, writeFile } from "node:fs/promises";
import { convert } from "pdf-raster";

await mkdir("./out", { recursive: true });

const pages = await convert("./report.pdf", { outputFormat: "webp" });

for (const page of pages) {
  await writeFile(`./out/page-${page.pageIndex + 1}.webp`, page.data);
}

Make a thumbnail of the first page

A low DPI renders faster and gives a smaller file. At 72 DPI a US letter page is 612 × 792 pixels.

const [cover] = await convert("./report.pdf", {
  pages: [0],
  dpi: 72,
  outputFormat: "jpeg",
});

Convert bytes from memory

Pass a Buffer, Uint8Array or ArrayBuffer when the PDF comes from an upload, a queue message or a storage SDK.

import { GetObjectCommand, S3Client } from "@aws-sdk/client-s3";
import { convert } from "pdf-raster";

const s3 = new S3Client({});
const object = await s3.send(
  new GetObjectCommand({ Bucket: "documents", Key: "invoice.pdf" }),
);
const bytes = await object.Body!.transformToByteArray();

const pages = await convert(bytes, { dpi: 300 });

Process a long PDF in batches

convert() holds every result in memory until it resolves. For documents with hundreds of pages, request a batch at a time and write each batch out before the next one.

import { writeFile } from "node:fs/promises";
import { convert } from "pdf-raster";

const pageCount = 400; // from your own metadata
const batchSize = 20;

for (let start = 0; start < pageCount; start += batchSize) {
  const indices = Array.from(
    { length: Math.min(batchSize, pageCount - start) },
    (_, i) => start + i,
  );
  const pages = await convert("./archive.pdf", { pages: indices });

  for (const page of pages) {
    await writeFile(`./out/${page.pageIndex}.png`, page.data);
  }
}

pdf-raster has no page-count API, so get the count from your own records or from another PDF library.

Convert many files at once

Concurrent calls are safe. Because PDFium renders one page at a time across the process, running calls in parallel adds little speed. It helps most with JPEG, where encoding overlaps with the next call's rendering. See the throughput table.

const files = ["a.pdf", "b.pdf", "c.pdf", "d.pdf"];
const results = await Promise.all(
  files.map((file) => convert(file, { outputFormat: "jpeg" })),
);

On this page