route-web-api: check throttling on all gateways, not just dgw

router/route-web-api.py

@@ -196,31 +196,22 @@ THROTTLE_TIMEOUT = 8
 THROTTLE_SIZE_LIMIT = 32768  # 32KB — above typical TCP initial window
 
 
-def _detect_throttle(proxy, base_url):
-    """Detect TSPU throttling by downloading page and checking assets.
-
-    Downloads the HTML page via proxy.  If the page is small (< 32KB),
-    parses it for CSS/JS asset URLs and tries downloading them.
-    Throttling signature: download stalls at ~16KB after timeout.
-
-    Returns dict with throttle info or None.
-    """
-    # Download main page
+def _find_assets(proxy, base_url):
+    """Download HTML page via proxy and extract CSS/JS asset URLs."""
     try:
         result = subprocess.run(
             ["curl", "-4", "--proxy", proxy, "-s", "-m", "10", "-L", base_url],
             capture_output=True, timeout=15,
         )
         if result.returncode != 0:
-            return None
+            return []
         page = result.stdout.decode("utf-8", errors="ignore")
     except (subprocess.TimeoutExpired, OSError):
-        return None
+        return []
 
     if len(result.stdout) >= THROTTLE_SIZE_LIMIT:
-        return None  # page itself is large enough, no throttle detected
+        return []  # page itself is large, no need to check assets
 
-    # Parse asset URLs (CSS, JS) from HTML
     parsed_url = urllib.parse.urlparse(base_url)
     origin = "%s://%s" % (parsed_url.scheme, parsed_url.netloc)
     assets = []
@@ -235,58 +226,52 @@ def _detect_throttle(proxy, base_url):
         elif not href.startswith("http"):
             href = base_url.rstrip("/") + "/" + href
         assets.append(href)
+    return assets[:3]
 
-    if not assets:
-        return None
 
-    # Try downloading assets to detect throttle
-    for asset_url in assets[:3]:
-        try:
-            result = subprocess.run(
-                ["curl", "-4", "--proxy", proxy, "-o", "/dev/null", "-s",
-                 "-w", "%{size_download} %{time_total}",
-                 "-m", str(THROTTLE_TIMEOUT), "-L", asset_url],
-                capture_output=True, text=True, timeout=THROTTLE_TIMEOUT + 5,
-            )
-            parts = result.stdout.strip().split()
-            if len(parts) < 2:
-                continue
-            size = int(float(parts[0]))
-            elapsed = float(parts[1])
-            # Throttle: stuck near 16KB, timeout reached
-            if 0 < size < THROTTLE_SIZE_LIMIT and elapsed >= THROTTLE_TIMEOUT - 1:
-                fname = asset_url.split("/")[-1].split("?")[0]
-                return {
-                    "throttled": True,
-                    "asset": fname,
-                    "url": asset_url,
-                    "size": size,
-                    "time": round(elapsed, 1),
-                }
-            # Got large file — not throttled
-            if size >= THROTTLE_SIZE_LIMIT:
-                return {"throttled": False}
-        except (subprocess.TimeoutExpired, OSError, ValueError):
-            continue
+def _check_throttle(name, proxy, asset_url):
+    """Download a single asset via proxy and detect throttling.
 
-    return None
+    Returns (name, result_dict) or (name, None).
+    """
+    try:
+        result = subprocess.run(
+            ["curl", "-4", "--proxy", proxy, "-o", "/dev/null", "-s",
+             "-w", "%{size_download} %{time_total}",
+             "-m", str(THROTTLE_TIMEOUT), "-L", asset_url],
+            capture_output=True, text=True, timeout=THROTTLE_TIMEOUT + 5,
+        )
+        parts = result.stdout.strip().split()
+        if len(parts) < 2:
+            return (name, None)
+        size = int(float(parts[0]))
+        elapsed = float(parts[1])
+        fname = asset_url.split("/")[-1].split("?")[0]
+        if 0 < size < THROTTLE_SIZE_LIMIT and elapsed >= THROTTLE_TIMEOUT - 1:
+            return (name, {"throttled": True, "asset": fname,
+                           "size": size, "time": round(elapsed, 1)})
+        if size >= THROTTLE_SIZE_LIMIT:
+            return (name, {"throttled": False})
+    except (subprocess.TimeoutExpired, OSError, ValueError):
+        pass
+    return (name, None)
 
 
 def check_site(domain):
     """Check domain: resolve IPs, find in route lists, whois, test gateways."""
     url = "https://%s/" % domain
 
-    # Run gateway checks (IPv4 + IPv6), whois and throttle in parallel
+    # Phase 1: gateway checks (IPv4 + IPv6), whois, and fetch assets in parallel
     checks = {}
     workers = len(CHECK_GATEWAYS) * 2 + 2
-    dgw_proxy = CHECK_GATEWAYS[0][1]  # dgw — direct path
     with concurrent.futures.ThreadPoolExecutor(max_workers=workers) as pool:
         gw_futures = []
         for name, proxy in CHECK_GATEWAYS:
             gw_futures.append(pool.submit(_check_one, name, proxy, url, "-4"))
             gw_futures.append(pool.submit(_check_one, name, proxy, url, "-6"))
         whois_future = pool.submit(get_whois, domain)
-        throttle_future = pool.submit(_detect_throttle, dgw_proxy, url)
+        # Fetch HTML and parse assets via first available gateway
+        assets_future = pool.submit(_find_assets, CHECK_GATEWAYS[0][1], url)
 
         for future in concurrent.futures.as_completed(gw_futures):
             name, ipver, status, code = future.result()
@@ -300,7 +285,21 @@ def check_site(domain):
                 checks[name]["http_code_v6"] = code
 
         whois_info = whois_future.result()
-        throttle = throttle_future.result()
+        assets = assets_future.result()
+
+    # Phase 2: throttle check — download first asset via all gateways in parallel
+    throttle = {}
+    if assets:
+        asset_url = assets[0]
+        with concurrent.futures.ThreadPoolExecutor(max_workers=len(CHECK_GATEWAYS)) as pool:
+            throttle_futures = [
+                pool.submit(_check_throttle, name, proxy, asset_url)
+                for name, proxy in CHECK_GATEWAYS
+            ]
+            for future in concurrent.futures.as_completed(throttle_futures):
+                gw_name, result = future.result()
+                if result:
+                    throttle[gw_name] = result
 
     ips = resolve_domain(domain)
     routes = find_in_routes(domain, ips)
@@ -907,7 +906,7 @@ function renderCheck(data) {
   container.appendChild(rsec);
 
   // Gateways (IPv4 + IPv6)
-  const throttled = data.throttle && data.throttle.throttled;
+  const thr = data.throttle || {};
   const gsec = mkDiv('check-section');
   gsec.appendChild(mkDiv('check-section-title', '\\u0414\\u043e\\u0441\\u0442\\u0443\\u043f\\u043d\\u043e\\u0441\\u0442\\u044c \\u0447\\u0435\\u0440\\u0435\\u0437 \\u0448\\u043b\\u044e\\u0437\\u044b'));
   const gwrap = document.createElement('div');
@@ -916,13 +915,13 @@ function renderCheck(data) {
     const info = data.checks[name];
     if (!info) continue;
     let st4 = info.status;
-    // Override dgw status to SLOW when throttling detected
-    if (name === 'dgw' && throttled && st4 === 'OK') st4 = 'SLOW';
+    const gwThr = thr[name];
+    if (gwThr && gwThr.throttled && st4 === 'OK') st4 = 'SLOW';
     const cls4 = st4 === 'OK' ? 'ok' : st4 === 'SLOW' ? 'slow' : st4 === 'BLOCK' ? 'block' : st4 === 'PROXY?' ? 'proxy' : 'other';
     const el4 = document.createElement('span');
     el4.className = 'check-gw ' + cls4;
     el4.textContent = name + ': ' + st4;
-    if (name === 'dgw' && throttled) el4.title = data.throttle.asset + ': ' + data.throttle.size + ' \\u0431\\u0430\\u0439\\u0442 \\u0437\\u0430 ' + data.throttle.time + '\\u0441';
+    if (gwThr && gwThr.throttled) el4.title = gwThr.asset + ': ' + gwThr.size + ' \\u0431\\u0430\\u0439\\u0442 \\u0437\\u0430 ' + gwThr.time + '\\u0441';
     gwrap.appendChild(el4);
     if (info.status_v6 && info.status_v6 !== 'PROXY?') {
       const cls6 = info.status_v6 === 'OK' ? 'ok' : info.status_v6 === 'BLOCK' ? 'block' : 'other';
@@ -936,15 +935,18 @@ function renderCheck(data) {
   container.appendChild(gsec);
 
   // Throttle detection
-  if (data.throttle) {
+  if (Object.keys(thr).length) {
     const tsec = mkDiv('check-section');
-    if (data.throttle.throttled) {
+    const slowGws = Object.entries(thr).filter(([,v]) => v.throttled);
+    if (slowGws.length) {
       tsec.appendChild(mkDiv('check-section-title', '\\u0417\\u0430\\u043c\\u0435\\u0434\\u043b\\u0435\\u043d\\u0438\\u0435 (\\u0422\\u0421\\u041f\\u0423)'));
-      const msg = data.throttle.asset + ': ' + data.throttle.size + ' \\u0431\\u0430\\u0439\\u0442 \\u0437\\u0430 ' + data.throttle.time + '\\u0441 (\\u043e\\u0436\\u0438\\u0434\\u0430\\u043b\\u043e\\u0441\\u044c \\u0431\\u043e\\u043b\\u044c\\u0448\\u0435)';
-      const el = mkDiv('check-route', msg);
-      el.style.background = '#f8d7da';
-      el.style.color = '#721c24';
-      tsec.appendChild(el);
+      for (const [gw, info] of slowGws) {
+        const msg = gw + ': ' + info.asset + ' \\u2014 ' + info.size + ' \\u0431\\u0430\\u0439\\u0442 \\u0437\\u0430 ' + info.time + '\\u0441';
+        const el = mkDiv('check-route', msg);
+        el.style.background = '#f8d7da';
+        el.style.color = '#721c24';
+        tsec.appendChild(el);
+      }
     } else {
       tsec.appendChild(mkDiv('check-section-title', '\\u0417\\u0430\\u043c\\u0435\\u0434\\u043b\\u0435\\u043d\\u0438\\u0435'));
       tsec.appendChild(mkDiv('check-no-route', '\\u041d\\u0435 \\u043e\\u0431\\u043d\\u0430\\u0440\\u0443\\u0436\\u0435\\u043d\\u043e'));