fix(twitter): 进主页后关 cookie 横幅 + 点掉「敏感内容」确认页

敏感内容账号的主页会挡一层 "Yes, view profile" 确认,不点就不加载 timeline,导致抓到 0 条。

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

core/scraper_twitter.py

@@ -266,6 +266,8 @@ class XTwitterScraper:
             await page.goto(target, wait_until="domcontentloaded", timeout=45000)
             # 等首批 GraphQL 响应落地
             await self._wait_settle(page, 2.0)
+            # 关 cookie 横幅 + 点掉「敏感内容」确认页(否则 timeline 不会加载)
+            await self._dismiss_gates(page)
 
             await self._scroll_loop(page, deadline)
         except ScrapeCancelled:
@@ -320,6 +322,48 @@ class XTwitterScraper:
         except Exception:  # noqa: BLE001
             pass
 
+    async def _dismiss_gates(self, page) -> None:
+        """关掉 cookie 同意横幅;点掉「此账号可能含敏感内容 → 是的,查看」确认页。"""
+        # cookie 横幅:点「拒绝非必要」/「Accept all」任一即可消失
+        for sel in (
+            "button:has-text('Refuse non-essential cookies')",
+            "button:has-text('拒绝非必要 Cookie')",
+            "button:has-text('Accept all cookies')",
+            "button:has-text('接受所有 Cookie')",
+        ):
+            try:
+                btn = page.locator(sel).first
+                if await btn.count() and await btn.is_visible():
+                    await btn.click(timeout=3000)
+                    await asyncio.sleep(0.6)
+                    break
+            except Exception:  # noqa: BLE001
+                pass
+        # 敏感内容确认:多试几次(账号 UI 语言可能是英文或中文)
+        clicked = False
+        for _ in range(3):
+            for sel in (
+                "button:has-text('Yes, view profile')",
+                "button:has-text('是的，查看个人资料')",
+                "button:has-text('是的,查看个人资料')",
+                "[data-testid='empty_state_button_text']",
+                "a[href='/settings/content_you_see'] ~ button",
+            ):
+                try:
+                    btn = page.locator(sel).first
+                    if await btn.count() and await btn.is_visible():
+                        await btn.click(timeout=3000)
+                        clicked = True
+                        self.log("已点掉「敏感内容」确认页")
+                        break
+                except Exception:  # noqa: BLE001
+                    pass
+            if clicked:
+                break
+            await asyncio.sleep(1.0)
+        if clicked:
+            await self._wait_settle(page, 3.0)
+
     async def _wait_settle(self, page, seconds: float) -> None:
         try:
             await page.wait_for_load_state("networkidle", timeout=int(seconds * 1000))