@pinkpixel/prysm-mcp/prysm-llm/scraper/scrape.js

MCP server for the Prysm web scraper - enabling AI assistants to scrape web content

#!/usr/bin/env node

/**
 * Simple helper script for Prysm scraper
 * 
 * This script forwards all arguments to the main scraper
 * and provides a more intuitive way to run the scraper.
 */

// Just a wrapper around main_scraper.js to make the command more intuitive
const mainScraper = require('./main_scraper');
const args = process.argv.slice(2);

// Minimal logging function that uses stderr
function log(message) {
  process.stderr.write(`${message}\n`);
}

// Check if help is needed
if (args.length === 0 || args.includes('--help')) {
  log("Prysm - Structure-Aware Web Scraper");
  log("\nUsage: node scrape.js [url] [options]");
  log("\nOptions:");
  log("  --pages <number>          Number of links to follow from initial URL (default: 1)");
  log("  --images                  Download images from the page");
  log("  --output <path>           Custom output path for results");
  log("  --image-output <path>     Custom output path for images");
  log("  --help                    Show this help message");
  log("\nExamples:");
  log('  node scrape.js "https://example.com"');
  log('  node scrape.js "https://example.com" --pages 5 --images');
  process.exit(0);
}

// Pass all arguments to the main scraper
try {
  mainScraper.main(args);
} catch (error) {
  log(`Error: ${error.message}`);
  process.exit(1);
}