fix: truncate transaction and vertex messages (#7893)

* add pyinstrument and fastapi wrapper

* Truncate vertex build responses

* Truncate vertex response and transactions

* [autofix.ci] apply automated fixes

* Update test for vertex response structure to reflect new truncation logic

* Refactor profiling middleware to use environment variable for configuration

* Update test for vertex response structure to correct truncation logic

* move pyinstrument to dev dependencies

* Update serialization test to include max_items parameter for list truncation

* remove profiling from this PR, fix constants

* [autofix.ci] apply automated fixes

* Add asyncio configuration options to pyproject.toml

---------

Co-authored-by: autofix-ci[bot] <114827586+autofix-ci[bot]@users.noreply.github.com>
Co-authored-by: Gabriel Luiz Freitas Almeida <gabriel@langflow.org>

src/backend/base/langflow/api/v1/schemas.py

@@ -26,7 +26,6 @@ from langflow.services.database.models.flow import FlowCreate, FlowRead
 from langflow.services.database.models.user import UserRead
 from langflow.services.settings.feature_flags import FeatureFlags
 from langflow.services.tracing.schema import Log
-from langflow.utils.util_strings import truncate_long_strings
 
 
 class BuildStatus(Enum):
@@ -310,8 +309,7 @@ class VertexBuildResponse(BaseModel):
 
     @field_serializer("data")
     def serialize_data(self, data: ResultDataResponse) -> dict:
-        data_dict = data.model_dump() if isinstance(data, BaseModel) else data
+        return serialize(data, max_length=MAX_TEXT_LENGTH, max_items=MAX_ITEMS_LENGTH)
-        return truncate_long_strings(data_dict)
 
 
 class VerticesBuiltResponse(BaseModel):

src/backend/base/langflow/graph/utils.py

@@ -12,6 +12,7 @@ from langflow.interface.utils import extract_input_variables_from_prompt
 from langflow.schema.data import Data
 from langflow.schema.message import Message
 from langflow.serialization import serialize
+from langflow.serialization.constants import MAX_ITEMS_LENGTH, MAX_TEXT_LENGTH
 from langflow.services.database.models.transactions.crud import log_transaction as crud_log_transaction
 from langflow.services.database.models.transactions.model import TransactionBase
 from langflow.services.database.models.vertex_builds.crud import log_vertex_build as crud_log_vertex_build
@@ -142,8 +143,8 @@ async def log_transaction(
         transaction = TransactionBase(
             vertex_id=source.id,
             target_id=target.id if target else None,
-            inputs=inputs,
+            inputs=serialize(inputs, max_length=MAX_TEXT_LENGTH, max_items=MAX_ITEMS_LENGTH),
-            outputs=outputs,
+            outputs=serialize(outputs, max_length=MAX_TEXT_LENGTH, max_items=MAX_ITEMS_LENGTH),
             status=status,
             error=error,
             flow_id=flow_id if isinstance(flow_id, UUID) else UUID(flow_id),
@@ -175,10 +176,8 @@ async def log_vertex_build(
             id=vertex_id,
             valid=valid,
             params=str(params) if params else None,
-            # Serialize data using our custom serializer
+            data=serialize(data, max_length=MAX_TEXT_LENGTH, max_items=MAX_ITEMS_LENGTH),
-            data=serialize(data),
+            artifacts=serialize(artifacts, max_length=MAX_TEXT_LENGTH, max_items=MAX_ITEMS_LENGTH),
-            # Serialize artifacts using our custom serializer
-            artifacts=serialize(artifacts) if artifacts else None,
         )
         async with session_getter(get_db_service()) as session:
             inserted = await crud_log_vertex_build(session, vertex_build)

src/backend/base/langflow/main.py

@@ -207,7 +207,11 @@ def create_app():
     __version__ = get_version_info()["version"]
     configure()
     lifespan = get_lifespan(version=__version__)
-    app = FastAPI(lifespan=lifespan, title="Langflow", version=__version__)
+    app = FastAPI(
+        title="Langflow",
+        version=__version__,
+        lifespan=lifespan,
+    )
     app.add_middleware(
         ContentSizeLimitMiddleware,
     )

src/backend/base/langflow/serialization/constants.py

@@ -1,2 +1,2 @@
-MAX_TEXT_LENGTH = 20000
+MAX_TEXT_LENGTH = 2000
-MAX_ITEMS_LENGTH = 1000
+MAX_ITEMS_LENGTH = 100

src/backend/tests/unit/api/v1/test_schemas.py

@@ -63,7 +63,7 @@ def test_vertex_response_structure_when_truncate_applies():
     ("size", "expected"),
     [
         (0, 0),
-        (42, 42),
+        (8, 8),
         (MAX_ITEMS_LENGTH, MAX_ITEMS_LENGTH),
         (MAX_ITEMS_LENGTH + 1000, MAX_ITEMS_LENGTH + 1),
         (MAX_ITEMS_LENGTH + 2000, MAX_ITEMS_LENGTH + 1),

src/backend/tests/unit/serialization/test_serialization.py

@@ -92,7 +92,7 @@ class TestSerializationHypothesis:
     @settings(max_examples=100)
     @given(lst=list_strategy)
     def test_list_truncation(self, lst: list) -> None:
-        result: list = serialize(lst)
+        result: list = serialize(lst, max_items=MAX_ITEMS_LENGTH)
         if len(lst) > MAX_ITEMS_LENGTH:
             assert len(result) == MAX_ITEMS_LENGTH + 1
             assert f"... [truncated {len(lst) - MAX_ITEMS_LENGTH} items]" in result