on_Snapshot__23_ssl.bg.js 6.0 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197
  1. #!/usr/bin/env node
  2. /**
  3. * Extract SSL/TLS certificate details from a URL.
  4. *
  5. * This hook sets up CDP listeners BEFORE chrome_navigate loads the page,
  6. * then waits for navigation to complete. The listener captures SSL details
  7. * during the navigation request.
  8. *
  9. * Usage: on_Snapshot__23_ssl.js --url=<url> --snapshot-id=<uuid>
  10. * Output: Writes ssl.jsonl
  11. */
  12. const fs = require('fs');
  13. const path = require('path');
  14. // Add NODE_MODULES_DIR to module resolution paths if set
  15. if (process.env.NODE_MODULES_DIR) module.paths.unshift(process.env.NODE_MODULES_DIR);
  16. const puppeteer = require('puppeteer-core');
  17. // Import shared utilities from chrome_utils.js
  18. const {
  19. getEnvBool,
  20. getEnvInt,
  21. parseArgs,
  22. connectToPage,
  23. waitForPageLoaded,
  24. } = require('../chrome/chrome_utils.js');
  25. const PLUGIN_NAME = 'ssl';
  26. const OUTPUT_DIR = '.';
  27. const OUTPUT_FILE = 'ssl.jsonl';
  28. const CHROME_SESSION_DIR = '../chrome';
  29. let browser = null;
  30. let page = null;
  31. let client = null;
  32. let sslCaptured = false;
  33. let shuttingDown = false;
  34. async function setupListener(url) {
  35. const outputPath = path.join(OUTPUT_DIR, OUTPUT_FILE);
  36. const timeout = getEnvInt('SSL_TIMEOUT', 30) * 1000;
  37. let targetHost = null;
  38. // Only extract SSL for HTTPS URLs
  39. if (!url.startsWith('https://')) {
  40. throw new Error('URL is not HTTPS');
  41. }
  42. try {
  43. targetHost = new URL(url).host;
  44. } catch (e) {
  45. targetHost = null;
  46. }
  47. // Connect to Chrome page using shared utility
  48. const { browser, page } = await connectToPage({
  49. chromeSessionDir: CHROME_SESSION_DIR,
  50. timeoutMs: timeout,
  51. puppeteer,
  52. });
  53. client = await page.target().createCDPSession();
  54. await client.send('Network.enable');
  55. client.on('Network.responseReceived', (params) => {
  56. try {
  57. if (sslCaptured) return;
  58. if (params.type && params.type !== 'Document') return;
  59. const response = params.response || {};
  60. const responseUrl = response.url || '';
  61. if (!responseUrl.startsWith('http')) return;
  62. if (targetHost) {
  63. try {
  64. const responseHost = new URL(responseUrl).host;
  65. if (responseHost !== targetHost) return;
  66. } catch (e) {
  67. // Ignore URL parse errors, fall through
  68. }
  69. }
  70. const securityDetails = response.securityDetails || null;
  71. let sslInfo = { url: responseUrl };
  72. if (securityDetails) {
  73. sslInfo.protocol = securityDetails.protocol;
  74. sslInfo.subjectName = securityDetails.subjectName;
  75. sslInfo.issuer = securityDetails.issuer;
  76. sslInfo.validFrom = securityDetails.validFrom;
  77. sslInfo.validTo = securityDetails.validTo;
  78. sslInfo.certificateId = securityDetails.subjectName;
  79. sslInfo.securityState = response.securityState || 'secure';
  80. sslInfo.schemeIsCryptographic = true;
  81. const sanList = securityDetails.sanList;
  82. if (sanList && sanList.length > 0) {
  83. sslInfo.subjectAlternativeNames = sanList;
  84. }
  85. } else if (responseUrl.startsWith('https://')) {
  86. sslInfo.securityState = response.securityState || 'unknown';
  87. sslInfo.schemeIsCryptographic = true;
  88. sslInfo.error = 'No security details available';
  89. } else {
  90. sslInfo.securityState = 'insecure';
  91. sslInfo.schemeIsCryptographic = false;
  92. }
  93. fs.writeFileSync(outputPath, JSON.stringify(sslInfo, null, 2));
  94. sslCaptured = true;
  95. } catch (e) {
  96. // Ignore errors
  97. }
  98. });
  99. return { browser, page };
  100. }
  101. function emitResult(status = 'succeeded') {
  102. if (shuttingDown) return;
  103. shuttingDown = true;
  104. const outputStr = sslCaptured ? OUTPUT_FILE : OUTPUT_FILE;
  105. console.log(JSON.stringify({
  106. type: 'ArchiveResult',
  107. status,
  108. output_str: outputStr,
  109. }));
  110. }
  111. async function handleShutdown(signal) {
  112. console.error(`\nReceived ${signal}, emitting final results...`);
  113. emitResult('succeeded');
  114. if (browser) {
  115. try {
  116. browser.disconnect();
  117. } catch (e) {}
  118. }
  119. process.exit(0);
  120. }
  121. async function main() {
  122. const args = parseArgs();
  123. const url = args.url;
  124. const snapshotId = args.snapshot_id;
  125. if (!url || !snapshotId) {
  126. console.error('Usage: on_Snapshot__23_ssl.js --url=<url> --snapshot-id=<uuid>');
  127. process.exit(1);
  128. }
  129. if (!getEnvBool('SSL_ENABLED', true)) {
  130. console.error('Skipping (SSL_ENABLED=False)');
  131. console.log(JSON.stringify({type: 'ArchiveResult', status: 'skipped', output_str: 'SSL_ENABLED=False'}));
  132. process.exit(0);
  133. }
  134. try {
  135. // Set up listener BEFORE navigation
  136. const connection = await setupListener(url);
  137. browser = connection.browser;
  138. page = connection.page;
  139. // Register signal handlers for graceful shutdown
  140. process.on('SIGTERM', () => handleShutdown('SIGTERM'));
  141. process.on('SIGINT', () => handleShutdown('SIGINT'));
  142. // Wait for chrome_navigate to complete (non-fatal)
  143. try {
  144. const timeout = getEnvInt('SSL_TIMEOUT', 30) * 1000;
  145. await waitForPageLoaded(CHROME_SESSION_DIR, timeout * 4);
  146. } catch (e) {
  147. console.error(`WARN: ${e.message}`);
  148. }
  149. // console.error('SSL listener active, waiting for cleanup signal...');
  150. await new Promise(() => {}); // Keep alive until SIGTERM
  151. return;
  152. } catch (e) {
  153. const error = `${e.name}: ${e.message}`;
  154. console.error(`ERROR: ${error}`);
  155. console.log(JSON.stringify({
  156. type: 'ArchiveResult',
  157. status: 'failed',
  158. output_str: error,
  159. }));
  160. process.exit(1);
  161. }
  162. }
  163. main().catch(e => {
  164. console.error(`Fatal error: ${e.message}`);
  165. process.exit(1);
  166. });