Merge pull request #196 from SentienceAPI/scaffold_captcha

Pagecontrol hook for evaluateJs

Author: rcholic

sentience/agent_runtime.py

@@ -70,7 +70,13 @@
 from dataclasses import dataclass
 from typing import TYPE_CHECKING, Any
 
-from .captcha import CaptchaContext, CaptchaHandlingError, CaptchaOptions, CaptchaResolution
+from .captcha import (
+    CaptchaContext,
+    CaptchaHandlingError,
+    CaptchaOptions,
+    CaptchaResolution,
+    PageControlHook,
+)
 from .failure_artifacts import FailureArtifactBuffer, FailureArtifactsOptions
 from .models import (
     EvaluateJsRequest,
@@ -479,8 +485,18 @@ def _build_captcha_context(self, snapshot: Snapshot, source: str) -> CaptchaCont
             url=snapshot.url,
             source=source,  # type: ignore[arg-type]
             captcha=captcha,
+            page_control=self._create_captcha_page_control(),
         )
 
+    def _create_captcha_page_control(self) -> PageControlHook:
+        async def _eval(code: str) -> Any:
+            result = await self.evaluate_js(EvaluateJsRequest(code=code))
+            if not result.ok:
+                raise RuntimeError(result.error or "evaluate_js failed")
+            return result.value
+
+        return PageControlHook(evaluate_js=_eval)
+
     def _emit_captcha_event(self, reason_code: str, details: dict[str, Any] | None = None) -> None:
         payload = {
             "kind": "captcha",

sentience/captcha.py

@@ -2,7 +2,7 @@
 
 from collections.abc import Awaitable, Callable
 from dataclasses import dataclass
-from typing import Literal, Optional
+from typing import Any, Literal, Optional
 
 from .models import CaptchaDiagnostics
 
@@ -11,6 +11,11 @@
 CaptchaSource = Literal["extension", "gateway", "runtime"]
 
 
+@dataclass
+class PageControlHook:
+    evaluate_js: Callable[[str], Awaitable[Any]]
+
+
 @dataclass
 class CaptchaContext:
     run_id: str
@@ -23,6 +28,7 @@ class CaptchaContext:
     snapshot_path: str | None = None
     live_session_url: str | None = None
     meta: dict[str, str] | None = None
+    page_control: PageControlHook | None = None
 
 
 @dataclass

tests/unit/test_captcha_context_page_control.py

@@ -0,0 +1,56 @@
+from __future__ import annotations
+
+import pytest
+
+from sentience.agent_runtime import AgentRuntime
+from sentience.captcha import PageControlHook
+from sentience.models import CaptchaDiagnostics, CaptchaEvidence, Snapshot, SnapshotDiagnostics
+
+
+class EvalBackend:
+    async def eval(self, code: str):
+        _ = code
+        return "ok"
+
+
+class MockTracer:
+    def __init__(self) -> None:
+        self.events: list[dict] = []
+        self.run_id = "test-run"
+
+    def emit(self, event_type: str, data: dict, step_id: str | None = None) -> None:
+        self.events.append({"type": event_type, "data": data, "step_id": step_id})
+
+
+def make_captcha_snapshot() -> Snapshot:
+    evidence = CaptchaEvidence(
+        iframe_src_hits=["https://www.google.com/recaptcha/api2/anchor"],
+        text_hits=["captcha"],
+        selector_hits=[],
+        url_hits=[],
+    )
+    captcha = CaptchaDiagnostics(
+        detected=True,
+        provider_hint="recaptcha",
+        confidence=0.9,
+        evidence=evidence,
+    )
+    diagnostics = SnapshotDiagnostics(captcha=captcha)
+    return Snapshot(
+        status="success",
+        url="https://example.com",
+        elements=[],
+        diagnostics=diagnostics,
+    )
+
+
+@pytest.mark.asyncio
+async def test_captcha_context_page_control_evaluate_js() -> None:
+    runtime = AgentRuntime(backend=EvalBackend(), tracer=MockTracer())
+    runtime.begin_step("captcha_test")
+
+    ctx = runtime._build_captcha_context(make_captcha_snapshot(), source="gateway")
+    assert isinstance(ctx.page_control, PageControlHook)
+
+    result = await ctx.page_control.evaluate_js("1+1")
+    assert result == "ok"