D2Snap

D2Snap is a first-of-its-kind DOM downsampling algorithm, designed for use with LLM-based web agents.

Integrate

D2Snap.d2Snap(
  dom: DOM,
  r_e: number, r_a: number, r_t: number,
  options?: Options
): Promise<{
  html: string;
  meta: {};
}>

D2Snap.adaptiveD2Snap(
  dom: DOM,
  maxTokens: number = 4096,
  maxIterations: number = 5,
  options?: Options
): Promise<{
  html: string;
  meta: {};
}>

type DOM = Document | Element | string;
type Options = {
  debug?: boolean;                // false
  groundTruth: object;            // see variables/ground-truth.json
  keepUnknownElements?: boolean;  // false
  skipMarkdown?: boolean;         // false
  uniqueIDs?: boolean;            // false
};

Browser

<script src="https://cdn.jsdelivr.net/gh/webfuse-com/D2Snap@main/dist.browser/D2Snap.js"></script>

Module

npm install webfuse-com/D2Snap

Install jsdom to use the library with Node.js:

npm install jsdom

import * as D2Snap from "@webfuse-com/d2snap";

Example

<main class="container" tabindex="3" required="true" type="example">
  <div class="mx-auto" data-topic="products" required="false">
    <h1>Our Pizza</h1>
    <div>
      <strong>Choose one</strong>
      <section class="shadow-lg">
        <h2>Margherita</h2>
        <p>
         A simple classic: mozzarela, tomatoes and basil.
         An everyday choice!
        </p>
        <button type="button">Add</button>
      </section>
      <section class="shadow-lg">
        <h2>Capricciosa</h2>
        <p>
          A rich taste: mozzarella, ham, mushrooms, artichokes, and olives.
          A true favourite!
        </p>
        <button type="button">Add</button>
      </section>
    </div>
  </div>
</main>

↓ D2Snap ↓

<main class="container" required="true">
  # Our Pizza
  <section class="shadow-lg">
    **Choose one**
    ## Margherita
    A simple classic mozzarela tomatoes and basil
    <button>
      Add
    </button>
    ## Capricciosa
    A rich taste
    A true favourite
    <button>
      Add
    </button>
  </section>
</main>

↓ D2Snap ↓

# Our Pizza
**Choose one**
## Margherita
A simple classic
<button>Add</button>
## Capricciosa
A rich taste
<button>Add</button>

Experiment

Setup

npm install
npm install jsdom

Build

npm run build

Test

npm run test

Evaluate

Provide LLM API provider key(s) to .env (compare example).

npm run eval:<snapshot>

<snapshot> ∈ { gui, dom, bu, D2Snap }

npm run eval:D2Snap -- --verbose --split 10,20 --provider openai --model gpt-4o

Re-create Snapshots

npm run snapshots:create

Beyond Pixels: Exploring DOM Downsampling for LLM-Based Web Agents
_{Thassilo M. Schiepanski   Nicholas Piël
Surfly BV}