deploy Nous v21 update

api/browser_investigator.py

@@ -7,10 +7,11 @@ CAPTCHAs, rotate proxies, or continue after an access challenge.
 from __future__ import annotations
 
 import asyncio
+import inspect
 import json
 import os
 import re
-from typing import Any
+from typing import Any, Callable
 from urllib.parse import urlparse
 
 
@@ -96,6 +97,7 @@ def investigate_websites(
     api_key: str,
     model: str,
     user_agent: str,
+    log_callback: Callable[[str, str, dict[str, Any]], Any] | None = None,
 ) -> list[dict[str, Any]]:
     """Run one bounded Browser Use investigation and return candidates."""
     if os.environ.get("BROWSER_USE_ENABLED", "1").strip().lower() in {"0", "false", "no", "off"}:
@@ -132,6 +134,52 @@ Return ONLY JSON in this shape:
 Confidence must be an integer from 0 to 100. Include only plausible official website URLs and never invent a URL.
 """.strip()
 
+    async def emit(kind: str, event: str, details: dict[str, Any]) -> None:
+        if not log_callback:
+            return
+        result = log_callback(kind, event, details)
+        if inspect.isawaitable(result):
+            await result
+
+    def output_actions(model_output: Any) -> list[Any]:
+        try:
+            actions = getattr(model_output, "action", []) or []
+            values: list[Any] = []
+            for action in list(actions)[:8]:
+                if hasattr(action, "model_dump"):
+                    values.append(action.model_dump(mode="json"))
+                else:
+                    values.append(str(action)[:2_000])
+            return values
+        except Exception:
+            return [str(model_output)[:2_000]] if model_output is not None else []
+
+    async def on_step(browser_state: Any, model_output: Any, step_number: int) -> None:
+        details = {
+            "step": step_number,
+            "url": str(getattr(browser_state, "url", "") or "")[:700],
+            "title": str(getattr(browser_state, "title", "") or "")[:300],
+            "actions": output_actions(model_output),
+            "model": model,
+        }
+        await emit("browser", "browser_use_step", details)
+        await emit("ai", "browser_use_model_step", {"step": step_number, "model": model, "actions": details["actions"]})
+
+    class LoggedChatOpenAI(ChatOpenAI):
+        """Capture Browser Use LLM turns without recording provider secrets."""
+
+        async def ainvoke(self, messages: list[Any], output_format: Any = None, **kwargs: Any) -> Any:
+            started = asyncio.get_running_loop().time()
+            serialised = [{"role": getattr(message, "type", "message"), "content": str(getattr(message, "content", ""))[:4_000]} for message in list(messages)[-20:]]
+            await emit("ai", "browser_use_llm_request", {"model": self.model, "messages": serialised, "structured_output": bool(output_format)})
+            try:
+                result = await super().ainvoke(messages, output_format=output_format, **kwargs)
+                await emit("ai", "browser_use_llm_response", {"model": self.model, "completion": str(getattr(result, "completion", result))[:4_000], "duration_ms": int((asyncio.get_running_loop().time() - started) * 1000)})
+                return result
+            except Exception as exc:
+                await emit("ai", "browser_use_llm_error", {"model": self.model, "duration_ms": int((asyncio.get_running_loop().time() - started) * 1000)},)
+                raise
+
     async def run() -> Any:
         browser_profile = BrowserProfile(
             headless=True,
@@ -141,7 +189,7 @@ Confidence must be an integer from 0 to 100. Include only plausible official web
             enable_default_extensions=False,
             captcha_solver=False,
         )
-        llm = ChatOpenAI(
+        llm = LoggedChatOpenAI(
             model=model,
             api_key=api_key,
             base_url=base_url,
@@ -161,10 +209,19 @@ Confidence must be an integer from 0 to 100. Include only plausible official web
             use_judge=False,
             enable_planning=False,
             enable_signal_handler=False,
+            register_new_step_callback=on_step,
         )
-        return await agent.run(max_steps=max(3, min(15, int(os.environ.get("BROWSER_USE_MAX_STEPS", "10")))))
+        try:
+            max_steps = int(os.environ.get("BROWSER_USE_MAX_STEPS", "10"))
+        except (TypeError, ValueError):
+            max_steps = 10
+        return await agent.run(max_steps=max(3, min(15, max_steps)))
 
-    timeout = max(30, min(300, int(os.environ.get("BROWSER_USE_TIMEOUT_SECONDS", "120"))))
+    try:
+        timeout = int(os.environ.get("BROWSER_USE_TIMEOUT_SECONDS", "120"))
+    except (TypeError, ValueError):
+        timeout = 120
+    timeout = max(30, min(300, timeout))
     try:
         history = asyncio.run(asyncio.wait_for(run(), timeout=timeout))
     except asyncio.TimeoutError as exc: