Harden Turnstile solve path for --fast/Xvfb runs

Add cf_sleep, multi-round shadow clicks, richer diagnostics, and longer profile/CF budgets so tokenLen=0 fails less on headless servers.

grok_register_ttk.py

@@ -1144,6 +1144,29 @@ def human_sleep(mean_seconds, cancel_callback=None):
     sleep_with_cancel(delay, cancel_callback)
 
 
+def cf_sleep(seconds, cancel_callback=None, *, floor: float = 0.6):
+    """Turnstile/Cloudflare 专用等待：fast 模式仍保留足够墙钟时间。
+
+    human_sleep 在 --fast 下会被压到 ~0.15x，导致 CF iframe 根本来不及加载/解题。
+    这里仅轻度缩放，并强制 floor，避免 token 恒为 0。
+    """
+    try:
+        sec = float(seconds)
+    except Exception:
+        sec = 1.0
+    scale = float(PERF_FLAGS.get("sleep_scale", 1.0) or 1.0)
+    if PERF_FLAGS.get("fast"):
+        # 最多压到 0.55x，且不低于 floor
+        scale = max(0.55, min(scale, 1.0))
+    delay = max(float(floor), sec * scale)
+    # 轻微抖动，避免机械节奏
+    try:
+        delay = max(float(floor), min(delay * 1.35, random.gauss(delay, delay * 0.12)))
+    except Exception:
+        pass
+    sleep_with_cancel(delay, cancel_callback)
+
+
 
 def get_domains(api_key=None):
     headers = {}
@@ -3713,32 +3736,11 @@ return 'clicked';
     raise Exception("验证码已获取，但自动填写/提交失败")
 
 
-def getTurnstileToken(log_callback=None, cancel_callback=None, max_rounds: int = 4):
-    """Try to obtain a Turnstile token.
-
-    max_rounds controls click/reset attempts; each round waits longer for CF.
-    CDP stealth should already be applied (webdriver hidden).
-    """
-    page = _get_page()
-    if page is None:
-        raise Exception("页面未就绪，无法执行 Turnstile")
-    apply_stealth_patches(page, log_callback=None)
-
+def _read_turnstile_token(page):
+    """Read cf-turnstile-response / turnstile.getResponse if present."""
     try:
-        page.run_js(
-            "try { if (window.turnstile && typeof turnstile.reset === 'function') turnstile.reset(); } catch(e) {}"
-        )
-    except Exception:
-        pass
-
-    rounds = max(2, int(max_rounds or 4))
-    # Give CF more time: ~1.2s * rounds, plus initial attach wait
-    human_sleep(0.8 if not PERF_FLAGS.get("fast") else 0.35, cancel_callback)
-    for round_i in range(0, rounds):
-        raise_if_cancelled(cancel_callback)
-        try:
-            token = page.run_js(
-                """
+        token = page.run_js(
+            """
 try {
   const byInput = String((document.querySelector('input[name="cf-turnstile-response"]') || {}).value || '').trim();
   if (byInput) return byInput;
@@ -3749,77 +3751,246 @@ try {
   }
   return '';
 } catch(e) { return ''; }
-                """
-            )
-            token = str(token or "").strip()
+            """
+        )
+        return str(token or "").strip()
+    except Exception:
+        return ""
+
+
+def _turnstile_diag(page) -> dict:
+    """Lightweight CF widget diagnostics for logs (no secrets)."""
+    try:
+        info = page.run_js(
+            """
+try {
+  const input = document.querySelector('input[name="cf-turnstile-response"], textarea[name="cf-turnstile-response"]');
+  const iframes = Array.from(document.querySelectorAll(
+    'iframe[src*="challenges.cloudflare.com"], iframe[src*="turnstile"]'
+  ));
+  const widgets = Array.from(document.querySelectorAll('div.cf-turnstile, [data-sitekey]'));
+  const srcs = iframes.slice(0, 3).map((f) => {
+    try { return String(f.src || '').slice(0, 120); } catch(e) { return ''; }
+  });
+  return {
+    hasInput: !!input,
+    tokenLen: input ? String(input.value || '').trim().length : 0,
+    iframeCount: iframes.length,
+    widgetCount: widgets.length,
+    sitekey: widgets[0] ? String(widgets[0].getAttribute('data-sitekey') || '').slice(0, 16) : '',
+    iframeSrc0: srcs[0] || '',
+    turnstileApi: !!(window.turnstile && typeof turnstile.getResponse === 'function'),
+    webdriver: !!navigator.webdriver,
+  };
+} catch(e) {
+  return { err: String(e && e.message || e) };
+}
+            """
+        )
+        return info if isinstance(info, dict) else {"raw": str(info)[:200]}
+    except Exception as exc:
+        return {"err": str(exc)[:160]}
+
+
+def _click_turnstile_widget(page, log_callback=None) -> str:
+    """Best-effort click path for interactive Turnstile checkbox.
+
+    Returns a short tag describing which path succeeded (or 'none').
+    """
+    # Path A: shadow-root checkbox under cf-turnstile-response parent
+    try:
+        challenge_input = page.ele("@name=cf-turnstile-response", timeout=0.6)
+    except Exception:
+        challenge_input = None
+    if challenge_input:
+        iframe = None
+        try:
+            wrapper = challenge_input.parent()
+            iframe = wrapper.shadow_root.ele("tag:iframe")
+        except Exception:
+            iframe = None
+        if iframe is None:
+            try:
+                iframe = page.ele(
+                    'css:iframe[src*="challenges.cloudflare.com"], iframe[src*="turnstile"]',
+                    timeout=0.6,
+                )
+            except Exception:
+                iframe = None
+        if iframe:
+            try:
+                iframe.run_js(
+                    """
+window.dtp = 1;
+function getRandomInt(min, max) { return Math.floor(Math.random() * (max - min + 1)) + min; }
+let sx = getRandomInt(800, 1200);
+let sy = getRandomInt(400, 700);
+try {
+  Object.defineProperty(MouseEvent.prototype, 'screenX', { get: function(){ return sx; } });
+  Object.defineProperty(MouseEvent.prototype, 'screenY', { get: function(){ return sy; } });
+} catch(e) {}
+                    """
+                )
+            except Exception:
+                pass
+            # A1: body.shadow_root input / .mark
+            try:
+                body_sr = iframe.ele("tag:body").shadow_root
+                btn = body_sr.ele("tag:input") or body_sr.ele("css:.mark") or body_sr.ele("css:label")
+                if btn:
+                    try:
+                        btn.click()
+                    except Exception:
+                        try:
+                            btn.click(by_js=True)
+                        except Exception:
+                            page.actions.click(btn)
+                    return "shadow-checkbox"
+            except Exception:
+                pass
+            # A2: click iframe element itself (center)
+            try:
+                iframe.click()
+                return "iframe-click"
+            except Exception:
+                pass
+            try:
+                page.actions.click(iframe)
+                return "iframe-actions"
+            except Exception:
+                pass
+
+    # Path B: DOM query + synthetic pointer events
+    try:
+        clicked = page.run_js(
+            """
+function firePointer(el) {
+  if (!el) return false;
+  try { el.scrollIntoView({block:'center', inline:'center'}); } catch(e) {}
+  const rect = el.getBoundingClientRect();
+  const x = rect.left + Math.max(8, Math.min(rect.width - 8, rect.width * 0.35));
+  const y = rect.top + Math.max(8, Math.min(rect.height - 8, rect.height * 0.5));
+  const opts = {bubbles:true, cancelable:true, view:window, clientX:x, clientY:y, button:0};
+  for (const t of ['pointerover','pointerenter','mouseover','mouseenter','mousemove','pointerdown','mousedown','pointerup','mouseup','click']) {
+    try {
+      const Ev = t.startsWith('pointer') ? PointerEvent : MouseEvent;
+      el.dispatchEvent(new Ev(t, opts));
+    } catch(e) {
+      try { el.dispatchEvent(new MouseEvent(t, opts)); } catch(e2) {}
+    }
+  }
+  try { if (typeof el.click === 'function') el.click(); } catch(e) {}
+  return true;
+}
+const nodes = Array.from(document.querySelectorAll(
+  'iframe[src*="challenges.cloudflare.com"], iframe[src*="turnstile"], div.cf-turnstile, [data-sitekey]'
+));
+let ok = false;
+for (const n of nodes) {
+  if (firePointer(n)) ok = true;
+}
+return ok;
+            """
+        )
+        if clicked:
+            return "js-pointer"
+    except Exception:
+        pass
+    return "none"
+
+
+def getTurnstileToken(log_callback=None, cancel_callback=None, max_rounds: int = 4):
+    """Try to obtain a Turnstile token.
+
+    max_rounds controls click/reset attempts; each round waits longer for CF.
+    CDP stealth should already be applied (webdriver hidden).
+
+    Note: --fast compresses human_sleep heavily; this function uses cf_sleep
+    so CF still has real wall-clock time to attach iframe + solve.
+    """
+    page = _get_page()
+    if page is None:
+        raise Exception("页面未就绪，无法执行 Turnstile")
+    apply_stealth_patches(page, log_callback=None)
+
+    rounds = max(3, int(max_rounds or 4))
+    # Initial attach wait — managed / auto modes often fill without click
+    cf_sleep(1.8 if not PERF_FLAGS.get("fast") else 1.2, cancel_callback, floor=1.0)
+
+    # Passive poll first (don't reset — reset often kills in-flight solve)
+    for _ in range(4):
+        raise_if_cancelled(cancel_callback)
+        token = _read_turnstile_token(page)
+        if len(token) >= 80:
+            if log_callback:
+                log_callback(f"[*] Turnstile 已通过(被动)，token长度={len(token)}")
+            return token
+        cf_sleep(0.9, cancel_callback, floor=0.7)
+
+    diag0 = _turnstile_diag(page)
+    if log_callback:
+        log_callback(
+            f"[Debug] Turnstile 初始态: tokenLen={diag0.get('tokenLen', 0)} "
+            f"iframe={diag0.get('iframeCount', 0)} widget={diag0.get('widgetCount', 0)} "
+            f"api={diag0.get('turnstileApi')} webdriver={diag0.get('webdriver')} "
+            f"sitekey={diag0.get('sitekey', '')!r}"
+        )
+
+    last_click = "none"
+    for round_i in range(0, rounds):
+        raise_if_cancelled(cancel_callback)
+        try:
+            token = _read_turnstile_token(page)
             if len(token) >= 80:
                 if log_callback:
                     log_callback(f"[*] Turnstile 已通过，token长度={len(token)}")
                 return token
 
-            challenge_input = page.ele("@name=cf-turnstile-response", timeout=0.5)
-            if challenge_input:
-                wrapper = challenge_input.parent()
-                iframe = None
+            # Only reset when previous click rounds clearly failed (not on first)
+            if round_i >= 2 and round_i % 2 == 0:
                 try:
-                    iframe = wrapper.shadow_root.ele("tag:iframe")
+                    page.run_js(
+                        "try { if (window.turnstile && typeof turnstile.reset === 'function') turnstile.reset(); } catch(e) {}"
+                    )
+                    if log_callback:
+                        log_callback(f"[Debug] Turnstile reset (round={round_i + 1})")
+                    cf_sleep(1.0, cancel_callback, floor=0.8)
                 except Exception:
-                    iframe = None
-                if iframe is None:
-                    try:
-                        iframe = page.ele(
-                            'css:iframe[src*="challenges.cloudflare.com"], iframe[src*="turnstile"]',
-                            timeout=0.5,
-                        )
-                    except Exception:
-                        iframe = None
-                if iframe:
-                    try:
-                        iframe.run_js(
-                            """
-window.dtp = 1;
-function getRandomInt(min, max) { return Math.floor(Math.random() * (max - min + 1)) + min; }
-let sx = getRandomInt(800, 1200);
-let sy = getRandomInt(400, 700);
-Object.defineProperty(MouseEvent.prototype, 'screenX', { value: sx });
-Object.defineProperty(MouseEvent.prototype, 'screenY', { value: sy });
-                            """
-                        )
-                    except Exception:
-                        pass
-                    clicked = False
-                    try:
-                        body_sr = iframe.ele("tag:body").shadow_root
-                        btn = body_sr.ele("tag:input") or body_sr.ele("css:.mark")
-                        if btn:
-                            btn.click()
-                            clicked = True
-                    except Exception:
-                        pass
-                    if not clicked:
-                        try:
-                            iframe.click()
-                        except Exception:
-                            pass
-            else:
-                # 兜底：尝试触发页面上可见的 Turnstile 容器 / iframe
-                page.run_js(
-                    """
-const nodes = Array.from(document.querySelectorAll(
-  'iframe[src*="challenges.cloudflare.com"], iframe[src*="turnstile"], div.cf-turnstile, [data-sitekey]'
-));
-for (const n of nodes) {
-  try { n.scrollIntoView({block:'center', inline:'center'}); } catch(e) {}
-  try { if (typeof n.click === 'function') n.click(); } catch(e) {}
-}
-                    """
-                )
-        except Exception:
-            pass
-        # Progressive wait: later rounds wait a bit longer for CF challenge
-        human_sleep(0.9 + 0.25 * round_i, cancel_callback)
+                    pass
 
-    raise Exception("Turnstile 获取 token 失败")
+            last_click = _click_turnstile_widget(page, log_callback=log_callback)
+            if log_callback and (round_i == 0 or last_click != "none"):
+                log_callback(f"[Debug] Turnstile click path={last_click} round={round_i + 1}/{rounds}")
+
+            # After click: poll more densely for token
+            poll_budget = 3.2 + 0.7 * round_i
+            poll_deadline = time.time() + poll_budget
+            while time.time() < poll_deadline:
+                raise_if_cancelled(cancel_callback)
+                token = _read_turnstile_token(page)
+                if len(token) >= 80:
+                    if log_callback:
+                        log_callback(f"[*] Turnstile 已通过，token长度={len(token)}")
+                    return token
+                cf_sleep(0.55, cancel_callback, floor=0.45)
+        except Exception as exc:
+            if log_callback and round_i == 0:
+                log_callback(f"[Debug] Turnstile round error: {exc}")
+
+        # Progressive gap between rounds
+        cf_sleep(1.1 + 0.35 * round_i, cancel_callback, floor=0.9)
+
+    diag1 = _turnstile_diag(page)
+    if log_callback:
+        log_callback(
+            f"[Debug] Turnstile 失败态: tokenLen={diag1.get('tokenLen', 0)} "
+            f"iframe={diag1.get('iframeCount', 0)} last_click={last_click} "
+            f"src={str(diag1.get('iframeSrc0', ''))[:80]!r}"
+        )
+    raise Exception(
+        f"Turnstile 获取 token 失败 (tokenLen={diag1.get('tokenLen', 0)}, "
+        f"iframe={diag1.get('iframeCount', 0)}, click={last_click})"
+    )
 
 
 def build_profile():
@@ -3854,11 +4025,18 @@ def fill_profile_and_submit(timeout=120, log_callback=None, cancel_callback=None
     dump_state(page, "profile-form")
     take_screenshot(page, "profile-form")
     given_name, family_name, password = build_profile()
-    # 预热 Turnstile：等 iframe 初始化；CDP stealth 会尝试自动点 checkbox
+    # 预热 Turnstile：等 iframe 初始化；CDP stealth / extension 会尝试自动点 checkbox
     if log_callback:
         log_callback("[*] 预热 Turnstile...")
-    # fast mode scales human_sleep; keep enough time for turnstile iframe to attach
-    human_sleep(1.6 if not PERF_FLAGS.get("fast") else 0.8, cancel_callback)
+    # 必须用 cf_sleep：--fast 下 human_sleep 会把预热压到 <0.3s，iframe 来不及挂载
+    cf_sleep(2.4 if not PERF_FLAGS.get("fast") else 1.6, cancel_callback, floor=1.2)
+    # 被动预读一次，managed 模式常自动出 token
+    try:
+        pre = _read_turnstile_token(page)
+        if pre and len(pre) >= 80 and log_callback:
+            log_callback(f"[*] 预热阶段已有 token，长度={len(pre)}")
+    except Exception:
+        pass
     deadline = time.time() + timeout
     form_filled_once = False
     wait_cf_since = None
@@ -3866,22 +4044,22 @@ def fill_profile_and_submit(timeout=120, log_callback=None, cancel_callback=None
     last_cf_log_at = 0.0
     cf_token_fail_streak = 0
     try:
-        # Slightly more patient by default — Chrome CDP stealth still needs CF time
-        max_cf_token_fails = max(1, int(config.get("turnstile_fast_fail_count", 3) or 3))
+        # 服务器 Xvfb 上 CF 常需多轮；默认比桌面更耐心
+        max_cf_token_fails = max(2, int(config.get("turnstile_fast_fail_count", 4) or 4))
     except Exception:
-        max_cf_token_fails = 3
+        max_cf_token_fails = 4
     try:
-        cf_retry_after = float(config.get("turnstile_retry_after_sec", 10) or 10)
+        cf_retry_after = float(config.get("turnstile_retry_after_sec", 12) or 12)
     except Exception:
-        cf_retry_after = 10.0
+        cf_retry_after = 12.0
     try:
-        cf_retry_gap = float(config.get("turnstile_retry_gap_sec", 6) or 6)
+        cf_retry_gap = float(config.get("turnstile_retry_gap_sec", 8) or 8)
     except Exception:
-        cf_retry_gap = 6.0
+        cf_retry_gap = 8.0
     try:
-        turnstile_rounds = max(4, int(config.get("turnstile_max_rounds", 6) or 6))
+        turnstile_rounds = max(5, int(config.get("turnstile_max_rounds", 8) or 8))
     except Exception:
-        turnstile_rounds = 6
+        turnstile_rounds = 8
 
     while time.time() < deadline:
         raise_if_cancelled(cancel_callback)
@@ -3972,6 +4150,14 @@ return 'filled-no-submit';
                     waited = now - wait_cf_since if wait_cf_since else 0
                     log_callback(f"[*] 资料已填写，等待 Cloudflare... token长度={token_len} waited={waited:.0f}s")
                     last_cf_log_at = now
+                # 卡住前先轻点一次 widget（不 reset），给 managed/interactive 更多时间
+                if wait_cf_since and now - wait_cf_since >= 3.0 and now - last_cf_retry_at >= 4.0:
+                    try:
+                        path = _click_turnstile_widget(page)
+                        if path != "none" and log_callback and (now - last_cf_log_at >= 4):
+                            log_callback(f"[Debug] 等待期轻点 Turnstile path={path}")
+                    except Exception:
+                        pass
                 # 卡住后自动二次复用 Turnstile 组件
                 if now - wait_cf_since >= cf_retry_after and now - last_cf_retry_at >= cf_retry_gap:
                     if log_callback:
@@ -4003,7 +4189,7 @@ return String(cfInput.value || '').trim().length;
                         if cf_token_fail_streak >= max_cf_token_fails:
                             raise Exception(f"Turnstile 连续失败，快速放弃资料页: {cf_exc}")
                     last_cf_retry_at = now
-                human_sleep(0.8, cancel_callback)
+                cf_sleep(1.0, cancel_callback, floor=0.7)
                 continue
 
             if filled in ("ready-to-submit", "filled-no-submit"):
@@ -4058,6 +4244,11 @@ return 'submitted';
                 waited = now - wait_cf_since if wait_cf_since else 0
                 log_callback(f"[*] 等待 Cloudflare 后再提交... token长度={token_len} waited={waited:.0f}s")
                 last_cf_log_at = now
+            if wait_cf_since and now - wait_cf_since >= 3.0 and now - last_cf_retry_at >= 4.0:
+                try:
+                    _click_turnstile_widget(page)
+                except Exception:
+                    pass
             if now - wait_cf_since >= cf_retry_after and now - last_cf_retry_at >= cf_retry_gap:
                 if log_callback:
                     log_callback("[*] 提交前仍卡住，自动再次复用 Turnstile...")
@@ -4088,7 +4279,7 @@ return String(cfInput.value || '').trim().length;
                     if cf_token_fail_streak >= max_cf_token_fails:
                         raise Exception(f"Turnstile 连续失败，快速放弃资料页: {cf_exc}")
                 last_cf_retry_at = now
-            human_sleep(0.8, cancel_callback)
+            cf_sleep(1.0, cancel_callback, floor=0.7)
             continue
 
         if submit_state == "submitted":