Getting Started with PDF Oxide (Node.js)

PDF Oxide is the fastest Node.js PDF library — 0.8ms mean text extraction, 5× faster than PyMuPDF, 15× faster than pypdf, 100% pass rate on 3,830 PDFs. One package for extracting, creating, and editing PDFs, with TypeScript definitions included. MIT / Apache-2.0 licensed.

Running in a browser, Deno, Bun, or Cloudflare Workers? Use the WASM build instead — same API, no native binaries. The native addon on this page is for Node.js and Electron only.

Installation

npm install pdf-oxide

Requirements: Node.js 18 or newer. No system dependencies. No Rust toolchain required. Prebuilt .node addons for Linux (glibc + musl) x64/arm64, macOS x64/arm64, and Windows x64/arm64 are fetched automatically via platform-specific optionalDependencies — nothing compiles on install.

Opening a PDF

JavaScript

const { PdfDocument } = require("pdf-oxide");

const doc = new PdfDocument("research-paper.pdf");
console.log(`Pages: ${doc.getPageCount()}`);

const { major, minor } = doc.getVersion();
console.log(`PDF version: ${major}.${minor}`);

doc.close();

TypeScript

import { PdfDocument } from "pdf-oxide";

const doc: PdfDocument = new PdfDocument("research-paper.pdf");
const pageCount: number = doc.getPageCount();
const { major, minor }: { major: number; minor: number } = doc.getVersion();
console.log(`${pageCount} pages, PDF ${major}.${minor}`);
doc.close();

On Node.js 22+ use using for automatic cleanup:

{
  using doc = new PdfDocument("report.pdf");
  const text = doc.extractText(0);
} // doc.close() called automatically

Page API

Since v0.3.34 PdfDocument is iterable and doc.page(i) returns a PdfPage with cached width / height / rotation plus per-page extraction methods.

const { PdfDocument } = require("pdf-oxide");

const doc = new PdfDocument("paper.pdf");
for (const page of doc) {
  console.log(`Page ${page.index}: ${page.width}x${page.height} (rotation ${page.rotation})`);
  const md = page.markdown();
  const tables = page.tables();       // rows + cells with bboxes
}
doc.close();

Indexing: doc.page(0), doc.page(-1) (last page). Page methods: text(), markdown(), html(), plainText(), words(), lines(), tables(), images(), paths(), annotations(), fonts(), search(query, caseSensitive).

Text Extraction

Single Page

JavaScript

const { PdfDocument } = require("pdf-oxide");

const doc = new PdfDocument("report.pdf");
const text = doc.extractText(0);
console.log(text);
doc.close();

TypeScript

import { PdfDocument } from "pdf-oxide";

const doc: PdfDocument = new PdfDocument("report.pdf");
const text: string = doc.extractText(0);
console.log(text);
doc.close();

All Pages

const doc = new PdfDocument("book.pdf");
const pageCount = doc.getPageCount();

for (let i = 0; i < pageCount; i++) {
  console.log(`--- Page ${i + 1} ---`);
  console.log(doc.extractText(i));
}

doc.close();

Async Extraction

Every sync method has an *Async counterpart that runs on the libuv thread pool. Use these in HTTP handlers and other concurrent server code so extraction doesn’t block the event loop.

const { PdfDocument } = require("pdf-oxide");

async function extract(path) {
  const doc = new PdfDocument(path);
  try {
    return await doc.extractTextAsync(0);
  } finally {
    doc.close();
  }
}

See the async guide for patterns including Promise.all fan-out over pages.

Structured Extraction

Character-level and span-level data with positions and font metadata:

const chars = doc.extractChars(0);
for (const ch of chars.slice(0, 10)) {
  console.log(`'${ch.char}' at (${ch.x.toFixed(1)}, ${ch.y.toFixed(1)}) ` +
              `size=${ch.fontSize.toFixed(1)} font=${ch.fontName}`);
}

const spans = doc.extractSpans(0);
for (const span of spans) {
  console.log(`"${span.text}" font=${span.fontName} size=${span.fontSize}`);
}

Word- and line-level extraction with tunable segmentation:

const words = doc.extractWords(0);
const lines = doc.extractTextLines(0, { wordGapThreshold: 2.5, lineGapThreshold: 1.2 });

Markdown Conversion

JavaScript

const md = doc.toMarkdown(0, { detectHeadings: true });
console.log(md);

// All pages
const allMd = doc.toMarkdownAll();

TypeScript

const md: string = doc.toMarkdown(0, { detectHeadings: true });
const allMd: string = doc.toMarkdownAll();

HTML Conversion

const html = doc.toHtml(0);
const allHtml = doc.toHtmlAll();

Image Extraction

const { writeFileSync } = require("fs");

const doc = new PdfDocument("brochure.pdf");
const images = doc.extractImages(0);

for (const [i, img] of images.entries()) {
  console.log(`Image ${i}: ${img.width}x${img.height} ${img.format} (${img.data.length} bytes)`);
  writeFileSync(`image_${i}.${img.format}`, img.data);
}

doc.close();

Images extracted from Indexed-color PDFs are automatically expanded to RGB, including 1/2/4/8 bpc indexed palettes with RGB, Grayscale, or CMYK base colour spaces.

Opening from Bytes

Open a PDF from in-memory bytes — useful when downloading from S3, HTTP, or databases:

const { PdfDocument } = require("pdf-oxide");
const { readFileSync } = require("fs");

const bytes = readFileSync("document.pdf");
const doc = PdfDocument.openFromBytes(bytes);
const text = doc.extractText(0);
doc.close();

Password-Protected PDFs

const doc = PdfDocument.openWithPassword("confidential.pdf", "secret");
const text = doc.extractText(0);
doc.close();

You can also authenticate after opening:

const doc = new PdfDocument("confidential.pdf");
if (doc.authenticate("secret")) {
  console.log(doc.extractText(0));
}
doc.close();

AES-256 (V=5, R=6) PDFs are fully supported — including push-button widget captions and correctly-invalidated object caches after late authentication.

PDF Creation

The Pdf class provides factory methods to create PDFs from various source formats.

From Markdown

const { Pdf } = require("pdf-oxide");
const { writeFileSync } = require("fs");

const pdf = Pdf.fromMarkdown("# Hello World\n\nThis is a PDF.");
writeFileSync("output.pdf", pdf.toBytes());

From HTML

const pdf = Pdf.fromHtml("<h1>Invoice</h1><p>Amount due: $42.00</p>");
writeFileSync("invoice.pdf", pdf.toBytes());

From Plain Text

const pdf = Pdf.fromText("Plain text document.\n\nSecond paragraph.");
writeFileSync("notes.pdf", pdf.toBytes());

From Images

const pdf = Pdf.fromImage("scan.jpg");
writeFileSync("scan.pdf", pdf.toBytes());

Search

const doc = new PdfDocument("manual.pdf");

// Search all pages
const results = doc.searchAll("configuration", { caseSensitive: false });
for (const r of results) {
  console.log(`Page ${r.page}: "${r.text}" at (${r.x.toFixed(0)}, ${r.y.toFixed(0)})`);
}

// Search a single page
const pageResults = doc.searchPage(0, "configuration");
doc.close();

For streaming search over large documents, use SearchStream:

const { PdfDocument, SearchStream, SearchManager } = require("pdf-oxide");

const doc = new PdfDocument("large.pdf");
const manager = new SearchManager(doc);
const stream = new SearchStream(manager, "invoice");

stream.on("data", (r) => console.log(`page ${r.pageIndex + 1}: ${r.text}`));
stream.on("end", () => doc.close());

See the Node.js streams guide for details.

Editing

Use DocumentEditor for metadata, page operations, annotations, and form fields:

const { DocumentEditor } = require("pdf-oxide");

const editor = DocumentEditor.open("document.pdf");

// Metadata
editor.setTitle("Updated Title");
editor.setAuthor("Jane Doe");

// Page operations
editor.rotatePage(0, 90);
editor.deletePage(5);
editor.movePage(2, 0);

// Forms
editor.setFormFieldValue("employee.name", "Jane Doe");
editor.flattenForms();

editor.save("edited.pdf");
editor.close();

OCR

Opt into the ocr feature at install time to enable OCR on scanned pages:

npm install pdf-oxide --build-from-source -- --features ocr

const { PdfDocument, OcrEngine } = require("pdf-oxide");

const doc = new PdfDocument("scanned.pdf");
const ocr = new OcrEngine();

if (ocr.pageNeedsOcr(doc, 0)) {
  const text = ocr.extractText(doc, 0);
  console.log(text);
}

ocr.close();
doc.close();

See the OCR guide for end-to-end recipes.

Thread Safety

PdfDocument is Send + Sync — you can share a single document across worker threads for parallel page reads. The *Async method family performs this automatically using the libuv thread pool; see concurrency for manual worker patterns.

Error Handling

All methods throw on failure:

const { PdfDocument } = require("pdf-oxide");

try {
  const doc = new PdfDocument("document.pdf");
  const text = doc.extractText(0);
  doc.close();
} catch (err) {
  console.error(`Extraction failed: ${err.message}`);
}

Next Steps

• Python Getting Started — using PDF Oxide from Python
• WASM Getting Started — browser / Deno / Bun / edge runtimes
• Node.js API Reference — full native API documentation
• Async Guide — *Async methods and Promise.all patterns
• Node.js Streams — SearchStream and friends
• Text Extraction — detailed extraction options