Function htmlToAiReady

Converts HTML content to a format optimized for AI processing.

This function performs several transformations:

1. Removes non-content HTML elements like scripts, styles, and navigation
2. Cleans up remaining HTML content
3. Optionally converts to simplified markdown (with QUALITY preset)
4. Removes remaining HTML tags
5. Normalizes whitespace and line breaks

Example

import fs from "node:fs";
import { PRESET_FAST, PRESET_QUALITY, htmlToAiReady } from "@mrspartak/html-ai-ready";

// Load an HTML file
const htmlContent = fs.readFileSync("./path/to/your/file.html", "utf-8");

// Example 1: Using PRESET_FAST for maximum token reduction
// This is ideal when you need to minimize token usage and don't need formatting
const fastResult = htmlToAiReady(htmlContent, PRESET_FAST);
console.log("FAST MODE RESULT:");
console.log(fastResult);
console.log(`Token reduction: ${((fastResult.length / htmlContent.length) * 100).toFixed(2)}% of original size`);

// Example 2: Using PRESET_QUALITY for better readability
// This preserves more formatting like bold, italic, lists, and headings
const qualityResult = htmlToAiReady(htmlContent, PRESET_QUALITY);
console.log("\nQUALITY MODE RESULT:");
console.log(qualityResult);
console.log(`Token reduction: ${((qualityResult.length / htmlContent.length) * 100).toFixed(2)}% of original size`);

// Example 3: Custom configuration, beta! no proper type support yet
// You can create your own configuration to fine-tune the behavior
const customConfig = {
  // Tags to completely remove with their content
  tags: ["script", "style", "noscript", "iframe", "svg", "video", "audio", "canvas", "footer"],

  // Whether to use simplified markdown formatting
  useMarkdown: true,

  // Additional cleanup options
  cleanup: {
    removeExtraWhitespace: true,
    collapseNewlines: true,
    removeEmptyLines: true,
  },
};

const customResult = htmlToAiReady(htmlContent, customConfig);
console.log("\nCUSTOM CONFIG RESULT:");
console.log(customResult);
console.log(`Token reduction: ${((customResult.length / htmlContent.length) * 100).toFixed(2)}% of original size`);

// Example 4: Comparing token usage for AI context
function estimateTokens(text: string): number {
  // Rough estimate: ~4 characters per token for English text
  return Math.ceil(text.length / 4);
}

console.log("\nTOKEN USAGE COMPARISON:");
console.log(`Original HTML: ~${estimateTokens(htmlContent)} tokens`);
console.log(`FAST mode: ~${estimateTokens(fastResult)} tokens`);
console.log(`QUALITY mode: ~${estimateTokens(qualityResult)} tokens`);
console.log(`Custom config: ~${estimateTokens(customResult)} tokens`);

Copy