config.ts

gistfile1.txt

export function modifyConfig(config: Config): Config {
  // Extract allowed domains from config.docs[].startUrl
  const allowedDomains = new Set<string>();
  (config.docs || []).forEach((doc) => {
    if (doc.startUrl) {
      try {
        const parsedUrl = new URL(doc.startUrl);
        allowedDomains.add(parsedUrl.hostname);
      } catch (e) {
        console.log(`Failed to parse startUrl: ${doc.startUrl}`, e);
      }
    }
  });

  // Helper function to check if a domain is an AWS Bedrock domain
  function isAWSBedrockDomain(hostname: string) {
    // Adjust this check as needed if your AWS Bedrock domain pattern is different
    return hostname.endsWith("bedrock.amazonaws.com");
  }

  // Save the original fetch
  const originalFetch = global.fetch || fetch;

  // Replace the global fetch with our whitelist version
  (global as any).fetch = async (url: string, init?: RequestInit) => {
    let allowed = false;

    try {
      // Parse the URL to extract its hostname
      const { hostname } = new URL(url);

      // Allow request if it’s either an AWS Bedrock domain or in the config’s docs list
      if (allowedDomains.has(hostname) || isAWSBedrockDomain(hostname)) {
        allowed = true;
      }
    } catch (e) {
      // If parsing failed, default to blocking; you can adjust this as needed
      console.log(`Failed to parse URL: ${url}`, e);
    }

    if (!allowed) {
      console.log(`Blocked request to ${url}`);
      return new Response(`Request blocked by whitelist for ${url}`, {
        status: 500,
      });
    }

    // Otherwise, pass along to the real fetch
    return originalFetch(url, init);
  };

  return config;
}