pdfplumber-wasm

WebAssembly/JavaScript bindings for pdfplumber-rs — extract text, words, characters, and tables from PDFs in the browser or Node.js.

Features

• Text extraction with optional layout detection
• Word extraction with configurable tolerance
• Table detection and extraction (lattice + stream strategies)
• Character-level access with font, size, and position data
• Regex search across page content
• Zero native dependencies — runs entirely in WebAssembly

Installation

npm install pdfplumber-wasm

Browser Usage

<script type="module">
import init, { WasmPdf } from './pdfplumber_wasm.js';

await init();

const response = await fetch('document.pdf');
const bytes = new Uint8Array(await response.arrayBuffer());
const pdf = WasmPdf.open(bytes);

console.log(`Pages: ${pdf.pageCount}`);

const page = pdf.page(0);
console.log(page.extractText());

// Extract tables
const tables = page.extractTables();
for (const table of tables) {
  console.table(table);
}
</script>

With a Bundler (Webpack, Vite, etc.)

import { WasmPdf } from 'pdfplumber-wasm';

const response = await fetch('/document.pdf');
const bytes = new Uint8Array(await response.arrayBuffer());
const pdf = WasmPdf.open(bytes);

const page = pdf.page(0);
const text = page.extractText();

Node.js Usage

import { readFileSync } from 'fs';
import { WasmPdf } from 'pdfplumber-wasm';

const bytes = readFileSync('document.pdf');
const pdf = WasmPdf.open(bytes);

console.log(`Pages: ${pdf.pageCount}`);

const page = pdf.page(0);

// Extract text
console.log(page.extractText());

// Extract words with bounding boxes
const words = page.extractWords();
for (const word of words) {
  console.log(`"${word.text}" at (${word.x0}, ${word.top})`);
}

// Extract tables as 2D arrays
const tables = page.extractTables();
for (const table of tables) {
  for (const row of table) {
    console.log(row.join(' | '));
  }
}

// Search for text
const matches = page.search('hello', false, false);
console.log(`Found ${matches.length} matches`);

API Reference

WasmPdf

| Method / Property | Description | |---|---| | WasmPdf.open(data: Uint8Array) | Open a PDF from raw bytes | | .pageCount | Number of pages | | .page(index: number) | Get a page by 0-based index | | .metadata | Document metadata (title, author, etc.) |

WasmPage

| Method / Property | Description | |---|---| | .pageNumber | Page index (0-based) | | .width | Page width in points | | .height | Page height in points | | .extractText(layout?) | Extract text (optional layout detection) | | .extractWords(xTol?, yTol?) | Extract words with bounding boxes | | .chars() | Get all characters with font/position data | | .findTables() | Detect tables with cell structure | | .extractTables() | Extract tables as 2D text arrays | | .search(pattern, regex?, case?) | Search for text patterns |

TypeScript Types

Import type definitions for rich typing:

import type {
  PdfChar,
  PdfWord,
  PdfTable,
  PdfTableData,
  PdfSearchMatch,
  PdfMetadata,
  BBox,
} from 'pdfplumber-wasm';

Building from Source

# Install wasm-pack
cargo install wasm-pack

# Build for bundlers (Webpack, Vite, Rollup)
wasm-pack build --target bundler crates/pdfplumber-wasm

# Build for Node.js
wasm-pack build --target nodejs crates/pdfplumber-wasm

# Build for browser (no bundler)
wasm-pack build --target web crates/pdfplumber-wasm

Comparison with Other Tools

| Feature | pdfplumber-wasm | pdf.js | pdf-lib | |---|---|---|---| | Text extraction | Yes | Yes | No | | Table detection | Yes | No | No | | Word grouping | Yes | Partial | No | | Character positions | Yes | Yes | No | | Regex search | Yes | No | No | | Runs in browser | Yes | Yes | Yes | | Runs in Node.js | Yes | Yes | Yes |

Performance

pdfplumber-wasm is compiled from Rust to WebAssembly, offering near-native performance for PDF extraction tasks. Benchmarks show 2-5x speedup over pure JavaScript PDF processing libraries for text and table extraction.

License

MIT OR Apache-2.0