SCRAPE-22: Implement rate limiting for trigger endpoint (#26)

Co-authored-by: Stephen Minakian <stephenminakian@gmail.com>
Co-committed-by: Stephen Minakian <stephenminakian@gmail.com>

routes/admin.js

@@ -1162,6 +1162,53 @@ const scraperConfig = require('../config/scraper');
 // Logger for scraper admin routes
 const scraperRouteLogger = createLogger('scraper-admin');
 
+// Rate limiting for manual scrape trigger (per-user, in-memory)
+const RATE_LIMIT_MAX_REQUESTS = 5;
+const RATE_LIMIT_WINDOW_MS = 60 * 60 * 1000; // 1 hour in milliseconds
+const rateLimitStore = new Map();
+
+/**
+ * Check rate limit for a given user ID.
+ * Returns an object indicating whether the request is allowed.
+ *
+ * @param {string} userId - The user ID to check
+ * @returns {{ allowed: boolean, retryAfterSeconds: number|null }}
+ */
+function checkRateLimit(userId) {
+  const now = Date.now();
+  const userKey = userId.toString();
+
+  if (!rateLimitStore.has(userKey)) {
+    rateLimitStore.set(userKey, []);
+  }
+
+  const timestamps = rateLimitStore.get(userKey);
+
+  // Remove timestamps outside the current window
+  const windowStart = now - RATE_LIMIT_WINDOW_MS;
+  const validTimestamps = timestamps.filter(ts => ts > windowStart);
+  rateLimitStore.set(userKey, validTimestamps);
+
+  if (validTimestamps.length >= RATE_LIMIT_MAX_REQUESTS) {
+    // Calculate when the oldest request in the window will expire
+    const oldestTimestamp = validTimestamps[0];
+    const retryAfterMs = (oldestTimestamp + RATE_LIMIT_WINDOW_MS) - now;
+    const retryAfterSeconds = Math.ceil(retryAfterMs / 1000);
+    return { allowed: false, retryAfterSeconds };
+  }
+
+  // Record this request
+  validTimestamps.push(now);
+  return { allowed: true, retryAfterSeconds: null };
+}
+
+/**
+ * Reset the rate limiter (for testing)
+ */
+function resetRateLimiter() {
+  rateLimitStore.clear();
+}
+
 /**
  * POST /api/admin/scraper/run
  * Trigger a manual scrape
@@ -1175,6 +1222,15 @@ router.post('/scraper/run', async (req, res) => {
     const db = req.app.locals.db;
     const { dryRun = false, htmlContent = null } = req.body || {};
 
+    // Check rate limit (per-user, before mutex check)
+    const rateLimitResult = checkRateLimit(req.user._id);
+    if (!rateLimitResult.allowed) {
+      res.setHeader('Retry-After', rateLimitResult.retryAfterSeconds.toString());
+      return res.status(429).json({
+        error: 'Rate limit exceeded. Maximum 5 trigger requests per hour.'
+      });
+    }
+
     // Check if scraper is already running
     if (isScraperRunning()) {
       return res.status(409).json({ error: 'Scrape already in progress' });
@@ -1318,3 +1374,4 @@ router.get('/scraper/history', async (req, res) => {
 
 module.exports = router;
 module.exports.clearStatsCache = clearStatsCache;
+module.exports.resetRateLimiter = resetRateLimiter;