Skip to content

Automate browser based workflows with AI

added raw dada logs (#2736)

Prakash Maheshwaran3e3522fc4c37
diff --git a/skyvern/config.py b/skyvern/config.py
index 9cf08a3b7..22acabd86 100644
--- a/skyvern/config.py
+++ b/skyvern/config.py
@@ -39,6 +39,7 @@ class Settings(BaseSettings):
ENV: str = "local"
EXECUTE_ALL_STEPS: bool = True
JSON_LOGGING: bool = False
+ LOG_RAW_API_REQUESTS: bool = True
LOG_LEVEL: str = "INFO"
PORT: int = 8000
ALLOWED_ORIGINS: list[str] = ["*"]
diff --git a/skyvern/forge/api_app.py b/skyvern/forge/api_app.py
index c07a7a9d1..cb97ffcb7 100644
--- a/skyvern/forge/api_app.py
+++ b/skyvern/forge/api_app.py
@@ -15,6 +15,7 @@ from starlette_context.plugins.base import Plugin
from skyvern.config import settings
from skyvern.exceptions import SkyvernHTTPException
from skyvern.forge import app as forge_app
+from skyvern.forge.request_logging import log_raw_request_middleware
from skyvern.forge.sdk.core import skyvern_context
from skyvern.forge.sdk.core.skyvern_context import SkyvernContext
from skyvern.forge.sdk.db.exceptions import NotFoundError
@@ -113,6 +114,10 @@ def get_agent_app() -> FastAPI:
finally:
skyvern_context.reset()
+ @app.middleware("http")
+ async def raw_request_logging(request: Request, call_next: Callable[[Request], Awaitable[Response]]) -> Response:
+ return await log_raw_request_middleware(request, call_next)
+
if settings.ADDITIONAL_MODULES:
for module in settings.ADDITIONAL_MODULES:
LOG.info("Loading additional module to set up api app", module=module)
diff --git a/skyvern/forge/request_logging.py b/skyvern/forge/request_logging.py
new file mode 100644
index 000000000..c8d259fd8
--- /dev/null
+++ b/skyvern/forge/request_logging.py
@@ -0,0 +1,66 @@
+from __future__ import annotations
+
+import typing
+
+import structlog
+
+from skyvern.config import settings
+
+if typing.TYPE_CHECKING: # pragma: no cover - import only for type hints
+ from typing import Awaitable, Callable
+
+ from fastapi import Response
+ from starlette.requests import Request
+
+LOG = structlog.get_logger()
+
+_SENSITIVE_HEADERS = {"authorization", "cookie"}
+_MAX_BODY_LENGTH = 1000
+_BINARY_PLACEHOLDER = "<binary>"
+
+
+def _sanitize_headers(headers: typing.Mapping[str, str]) -> dict[str, str]:
+ sanitized: dict[str, str] = {}
+ for key, value in headers.items():
+ if key.lower() in _SENSITIVE_HEADERS:
+ continue
+ sanitized[key] = value
+ return sanitized
+
+
+def _sanitize_body(body: bytes, content_type: str | None) -> str:
+ if not body:
+ return ""
+ if content_type and not (content_type.startswith("text/") or content_type.startswith("application/json")):
+ return _BINARY_PLACEHOLDER
+ try:
+ text = body.decode("utf-8", errors="replace")
+ except Exception:
+ return _BINARY_PLACEHOLDER
+ if len(text) > _MAX_BODY_LENGTH:
+ return text[:_MAX_BODY_LENGTH] + "...[truncated]"
+ return text
+
+
+async def log_raw_request_middleware(request: Request, call_next: Callable[[Request], Awaitable[Response]]) -> Response:
+ if not settings.LOG_RAW_API_REQUESTS:
+ return await call_next(request)
+
+ body_bytes = await request.body()
+ # ensure downstream handlers can access body again
+ try:
+ request._body = body_bytes # type: ignore[attr-defined]
+ except Exception:
+ pass
+
+ sanitized_headers = _sanitize_headers(dict(request.headers))
+ body_text = _sanitize_body(body_bytes, request.headers.get("content-type"))
+
+ LOG.info(
+ "api.raw_request",
+ method=request.method,
+ path=request.url.path,
+ headers=sanitized_headers,
+ body=body_text,
+ )
+ return await call_next(request)