fix: Remove noisy sentry error on llm failure and upgrade deps (#396)

pyproject.toml

@@ -8,11 +8,11 @@ authors = [
 readme = "README.md"
 requires-python = ">=3.10"
 dependencies = [
-    "fastapi[standard]>=0.111.0",
+    "fastapi[standard]>=0.131.0",
     "groq>=0.31.0",
     "python-dotenv>=1.0.0",
     "sqlalchemy>=2.0.30",
-    "fastapi-pagination>=0.12.24",
+    "fastapi-pagination>=0.14.2",
     "pgvector>=0.2.5",
     "sentry-sdk[anthropic,fastapi,sqlalchemy]>=2.3.1",
     "greenlet>=3.0.3",

src/crud/document.py

@@ -1,9 +1,10 @@
 import datetime
 from collections.abc import Sequence
 from logging import getLogger
-from typing import Any
+from typing import Any, cast
 
 from sqlalchemy import delete, select, update
+from sqlalchemy.engine import CursorResult
 from sqlalchemy.exc import IntegrityError
 from sqlalchemy.ext.asyncio import AsyncSession
 from sqlalchemy.sql import Select
@@ -530,7 +531,7 @@ async def delete_document(
     update_stmt = (
         update(models.Document).where(*conditions).values(deleted_at=func.now())
     )
-    result = await db.execute(update_stmt)
+    result = cast(CursorResult[Any], await db.execute(update_stmt))
 
     if result.rowcount == 0:
         raise ResourceNotFoundException(
@@ -568,7 +569,7 @@ async def delete_document_by_id(
         )
         .values(deleted_at=func.now())
     )
-    result = await db.execute(update_stmt)
+    result = cast(CursorResult[Any], await db.execute(update_stmt))
 
     if result.rowcount == 0:
         raise ResourceNotFoundException(

src/crud/peer_card.py

@@ -1,9 +1,10 @@
 from __future__ import annotations
 
 import logging
-from typing import cast
+from typing import Any, cast
 
 from sqlalchemy import update
+from sqlalchemy.engine import CursorResult
 from sqlalchemy.ext.asyncio import AsyncSession
 
 from src import exceptions, models, schemas
@@ -84,7 +85,7 @@ async def set_peer_card(
             )
         )
     )
-    result = await db.execute(stmt)
+    result = cast(CursorResult[Any], await db.execute(stmt))
     if result.rowcount == 0:
         raise exceptions.ResourceNotFoundException(
             f"Peer {observer} not found in workspace {workspace_name}"

src/crud/session.py

@@ -1,11 +1,13 @@
 from dataclasses import dataclass
 from logging import getLogger
 from typing import Any
+from typing import cast as typing_cast
 
 from cashews import NOT_NONE
 from nanoid import generate as generate_nanoid
 from sqlalchemy import Select, and_, case, cast, delete, func, insert, select, update
 from sqlalchemy.dialects.postgresql import insert as pg_insert
+from sqlalchemy.engine import CursorResult
 from sqlalchemy.exc import IntegrityError
 from sqlalchemy.ext.asyncio import AsyncSession
 from sqlalchemy.types import BigInteger, Boolean
@@ -378,7 +380,7 @@ async def _batch_delete_matching(
             select(primary_key_column).where(and_(*filter_conditions)).limit(batch_size)
         )
         delete_stmt = delete(model).where(primary_key_column.in_(subquery))
-        delete_result = await db.execute(delete_stmt)
+        delete_result = typing_cast(CursorResult[Any], await db.execute(delete_stmt))
         batch_deleted = delete_result.rowcount or 0
         total_deleted += batch_deleted
 

src/deriver/queue_manager.py

@@ -4,7 +4,7 @@ from asyncio import Task
 from collections.abc import Sequence
 from datetime import datetime, timedelta, timezone
 from logging import getLogger
-from typing import NamedTuple
+from typing import Any, NamedTuple, cast
 
 import sentry_sdk
 from dotenv import load_dotenv
@@ -12,6 +12,7 @@ from nanoid import generate as generate_nanoid
 from sentry_sdk.integrations.asyncio import AsyncioIntegration
 from sqlalchemy import and_, delete, or_, select, update
 from sqlalchemy.dialects.postgresql import insert
+from sqlalchemy.engine import CursorResult
 from sqlalchemy.ext.asyncio import AsyncSession
 from sqlalchemy.sql import func
 
@@ -832,10 +833,13 @@ class QueueManager:
         Clean up a specific work unit session by both work_unit_key and AQS ID.
         """
         async with tracked_db("cleanup_work_unit") as db:
-            result = await db.execute(
-                delete(models.ActiveQueueSession)
-                .where(models.ActiveQueueSession.id == aqs_id)
-                .where(models.ActiveQueueSession.work_unit_key == work_unit_key)
+            result = cast(
+                CursorResult[Any],
+                await db.execute(
+                    delete(models.ActiveQueueSession)
+                    .where(models.ActiveQueueSession.id == aqs_id)
+                    .where(models.ActiveQueueSession.work_unit_key == work_unit_key)
+                ),
             )
             await db.commit()
             return result.rowcount > 0

src/telemetry/logging.py

@@ -10,7 +10,7 @@ from collections.abc import Callable
 from typing import ParamSpec, TypeVar, overload
 
 from fastapi import Request
-from langfuse import observe  # pyright: ignore
+from langfuse import observe
 from rich import box
 from rich.console import Console, Group, RenderableType
 from rich.panel import Panel

src/telemetry/sentry.py

@@ -2,7 +2,7 @@
 
 from __future__ import annotations
 
-import asyncio
+import inspect
 import logging
 from collections.abc import Callable, Sequence
 from functools import wraps
@@ -70,7 +70,7 @@ def with_sentry_transaction(
     """
 
     def decorator(func: Callable[P, T]) -> Callable[P, T]:
-        if asyncio.iscoroutinefunction(func):
+        if inspect.iscoroutinefunction(func):
 
             @wraps(func)
             async def async_wrapper(*args: P.args, **kwargs: P.kwargs) -> T:

src/utils/json_parser.py

@@ -3,7 +3,7 @@ import logging
 import re
 from typing import Any
 
-from json_repair import repair_json  # pyright: ignore
+from json_repair import repair_json
 
 logger = logging.getLogger(__name__)
 # logging.getLogger("sqlalchemy.engine.Engine").disabled = True

tests/bench/explicit.py

@@ -430,11 +430,7 @@ class ExplicitJudge:
                 )
                 for block in resp.content:
                     if block.type == "tool_use":
-                        # block.input is typed as object, but we know it's a dict
-                        input_data = block.input
-                        if isinstance(input_data, dict):
-                            # Cast to dict[str, Any] for type checker
-                            return dict(input_data)  # pyright: ignore[reportUnknownArgumentType]
+                        return dict(block.input)
                 return {}
 
             else:  # AsyncOpenAI

tests/unified/runner.py

@@ -7,7 +7,7 @@ import threading
 import time
 from datetime import datetime, timezone
 from pathlib import Path
-from typing import Any, cast
+from typing import Any
 
 import httpx
 from anthropic import AsyncAnthropic
@@ -418,16 +418,10 @@ class UnifiedTestExecutor:
                     f"No tool use in judge response: {resp.content}"
                 )
 
-            data: object = tool_use.input
-            if not isinstance(data, dict):
-                raise TestExecutionError(f"Tool input is not a dict: {data}")
-
-            typed_data = cast(dict[str, Any], data)
-            passed: bool = typed_data.get("passed", False)
+            data = tool_use.input
+            passed = bool(data.get("passed", False))
             if passed != assertion.pass_if:
-                raise TestExecutionError(
-                    f"LLM Judge failed: {typed_data.get('reasoning')}"
-                )
+                raise TestExecutionError(f"LLM Judge failed: {data.get('reasoning')}")
 
         elif isinstance(assertion, ContainsAssertion):
             text = result_str if assertion.case_sensitive else result_str.lower()