fix(rate-limit): ignore hidden recaptcha (#244)

Author: stickerdaniel

linkedin_mcp_server/core/utils.py

@@ -15,8 +15,7 @@ async def detect_rate_limit(page: Page) -> None:
 
     Checks (in order):
     1. URL contains /checkpoint or /authwall (security challenge)
-    2. Page contains CAPTCHA iframe (bot detection)
-    3. Body text contains rate-limit phrases on error-shaped pages (throttling)
+    2. Body text contains rate-limit phrases on error-shaped pages (throttling)
 
     The body-text heuristic only runs on pages without a ``<main>`` element
     and with short body text (<2000 chars), since real rate-limit pages are
@@ -35,23 +34,6 @@ async def detect_rate_limit(page: Page) -> None:
             suggested_wait_time=30,
         )
 
-    # Check for CAPTCHA
-    try:
-        captcha = await page.locator(
-            'iframe[title*="captcha" i], iframe[src*="captcha" i]'
-        ).count()
-        if captcha > 0:
-            raise RateLimitError(
-                "CAPTCHA challenge detected. Manual intervention required.",
-                suggested_wait_time=30,
-            )
-    except RateLimitError:
-        raise
-    except PlaywrightTimeoutError:
-        pass
-    except Exception as e:
-        logger.debug("Error checking for CAPTCHA: %s", e)
-
     # Check for rate limit messages — only on error-shaped pages.
     # Real rate-limit pages have no <main> element and short body text.
     # Normal LinkedIn pages (profiles, jobs) have <main> and long content

tests/test_core_utils.py

@@ -32,39 +32,18 @@ async def test_authwall_url_raises(self, mock_page):
         with pytest.raises(RateLimitError, match="security checkpoint"):
             await detect_rate_limit(mock_page)
 
-    async def test_captcha_iframe_raises(self, mock_page):
-        captcha_locator = MagicMock()
-        captcha_locator.count = AsyncMock(return_value=1)
-
-        main_locator = MagicMock()
-        main_locator.count = AsyncMock(return_value=0)
-
-        def locator_side_effect(selector):
-            if "captcha" in selector:
-                return captcha_locator
-            return main_locator
-
-        mock_page.locator = MagicMock(side_effect=locator_side_effect)
-        with pytest.raises(RateLimitError, match="CAPTCHA"):
-            await detect_rate_limit(mock_page)
-
     async def test_normal_page_with_main_skips_body_heuristic(self, mock_page):
         """A normal page with <main> should NOT trigger body text checks."""
         main_locator = MagicMock()
         main_locator.count = AsyncMock(return_value=1)
 
-        captcha_locator = MagicMock()
-        captcha_locator.count = AsyncMock(return_value=0)
-
         body_locator = MagicMock()
         # Body contains a phrase that would false-positive
         body_locator.inner_text = AsyncMock(
             return_value="Helping SaaS teams slow down churn with data-driven retention"
         )
 
         def locator_side_effect(selector):
-            if "captcha" in selector:
-                return captcha_locator
             if selector == "main":
                 return main_locator
             if selector == "body":
@@ -80,17 +59,12 @@ async def test_error_page_without_main_triggers_heuristic(self, mock_page):
         main_locator = MagicMock()
         main_locator.count = AsyncMock(return_value=0)
 
-        captcha_locator = MagicMock()
-        captcha_locator.count = AsyncMock(return_value=0)
-
         body_locator = MagicMock()
         body_locator.inner_text = AsyncMock(
             return_value="Too many requests. Slow down."
         )
 
         def locator_side_effect(selector):
-            if "captcha" in selector:
-                return captcha_locator
             if selector == "main":
                 return main_locator
             if selector == "body":
@@ -106,18 +80,13 @@ async def test_long_body_without_main_does_not_trigger(self, mock_page):
         main_locator = MagicMock()
         main_locator.count = AsyncMock(return_value=0)
 
-        captcha_locator = MagicMock()
-        captcha_locator.count = AsyncMock(return_value=0)
-
         body_locator = MagicMock()
         # Long body with a matching phrase buried in content
         body_locator.inner_text = AsyncMock(
             return_value="x" * 2000 + " try again later"
         )
 
         def locator_side_effect(selector):
-            if "captcha" in selector:
-                return captcha_locator
             if selector == "main":
                 return main_locator
             if selector == "body":
@@ -128,17 +97,12 @@ def locator_side_effect(selector):
         # Should NOT raise — body is too long to be an error page
         await detect_rate_limit(mock_page)
 
-    async def test_normal_url_no_captcha_no_error_passes(self, mock_page):
+    async def test_normal_url_no_error_passes(self, mock_page):
         """A clean normal page passes all checks without raising."""
         main_locator = MagicMock()
         main_locator.count = AsyncMock(return_value=1)
 
-        captcha_locator = MagicMock()
-        captcha_locator.count = AsyncMock(return_value=0)
-
         def locator_side_effect(selector):
-            if "captcha" in selector:
-                return captcha_locator
             if selector == "main":
                 return main_locator
             return MagicMock(count=AsyncMock(return_value=0))