vellum-workflow-server 1.3.8__py3-none-any.whl → 1.3.9.post2__py3-none-any.whl

{vellum_workflow_server-1.3.8.dist-info → vellum_workflow_server-1.3.9.post2.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: vellum-workflow-server
-Version: 1.3.8
+Version: 1.3.9.post2
 Summary: 
 License: AGPL
 Requires-Python: >=3.9.0,<4
@@ -29,7 +29,7 @@ Requires-Dist: pyjwt (==2.10.0)
 Requires-Dist: python-dotenv (==1.0.1)
 Requires-Dist: retrying (==1.3.4)
 Requires-Dist: sentry-sdk[flask] (==2.20.0)
-Requires-Dist: vellum-ai (==1.3.8)
+Requires-Dist: vellum-ai (==1.3.9)
 Description-Content-Type: text/markdown
 
 # Vellum Workflow Runner Server

{vellum_workflow_server-1.3.8.dist-info → vellum_workflow_server-1.3.9.post2.dist-info}/RECORD

@@ -4,19 +4,19 @@ workflow_server/api/auth_middleware.py,sha256=IlZaCiwZ5nwQqk5sYQorvOFj7lt0p1ZSSE
 workflow_server/api/healthz_view.py,sha256=itiRvBDBXncrw8Kbbc73UZLwqMAhgHOR3uSre_dAfgY,404
 workflow_server/api/tests/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 workflow_server/api/tests/test_input_display_mapping.py,sha256=drBZqMudFyB5wgiUOcMgRXz7E7ge-Qgxbstw4E4f0zE,2211
-workflow_server/api/tests/test_workflow_view.py,sha256=gwb53E44LBm06XHY1UwfHqKG-UfwwICh_IYPtowN_kE,19085
+workflow_server/api/tests/test_workflow_view.py,sha256=RlAw1tHeIlnOXGrFQN-w3EOLPZkhp6Dfy6d1r7kU5oc,22573
 workflow_server/api/tests/test_workflow_view_stream_workflow_route.py,sha256=Qo8u6mPyRCmE2jamY1yIh8l44hgo4-Nwlq03z61ND5g,27031
-workflow_server/api/workflow_view.py,sha256=gSKqlMR2r2urSBYm_jFlcpXHOk5jQsDMSDwpSahp5A8,21144
+workflow_server/api/workflow_view.py,sha256=Ufc49WJ4LON3FFxzY2MeP4q7uQyorMNMu9zLRuJvdp4,21523
 workflow_server/code_exec_runner.py,sha256=lBnMIorPZL8zZBye6TjeCIs06WTJM7P2HR07B1fjJJI,2533
 workflow_server/config.py,sha256=DyTty8NrAwvtx-esM3KthnpsNh-nKdWNlovWQOgiGpg,1417
 workflow_server/core/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 workflow_server/core/cancel_workflow.py,sha256=Ffkc3mzmrdMEUcD-sHfEhX4IwVrka-E--SxKA1dUfIU,2185
 workflow_server/core/events.py,sha256=24MA66DVQuaLJJcZrS8IL1Zq4Ohi9CoouKZ5VgoH3Cs,1402
-workflow_server/core/executor.py,sha256=TTpMifrIZuzXxOUtX3E3CgLOb8eRXul2aGPpRCfWIco,18368
-workflow_server/core/utils.py,sha256=cmwHbKCfXqtUutBD3akGus0Ga7a1xG3zlOw-jEMx6mI,1795
+workflow_server/core/executor.py,sha256=qAAQcwN0BfNL--IVZAn9jWr_KeOYGqB8sPP7dm8cFQk,17548
+workflow_server/core/utils.py,sha256=aIpSINstLGslP2PIoDLM82_1GlJ1uC_0AIrP-V7Yobo,3230
 workflow_server/core/workflow_executor_context.py,sha256=w3OhV_AXpgh7AxpjEsc0vo-IJypgJcgr5DXJCqGptOU,1587
 workflow_server/server.py,sha256=QBU12AaAfAgLqfCDBd24qIJl_mbheiq0-hfcWV7rZM4,1234
-workflow_server/start.py,sha256=qpIg0SgIgz8RNyc8Cu9LxyzXdOXZRv9qq3M3uSBbgD0,2180
+workflow_server/start.py,sha256=pkwRcms6I4tkVHP06LdrZY6rG_DFHfBx4ioY5X91W5k,2264
 workflow_server/utils/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 workflow_server/utils/exit_handler.py,sha256=_FacDVi4zc3bfTA3D2mJsISePlJ8jpLrnGVo5-xZQFs,743
 workflow_server/utils/log_proxy.py,sha256=nugi6fOgAYKX2X9DIc39TG366rsmmDUPoEtG3gzma_Y,3088
@@ -28,7 +28,7 @@ workflow_server/utils/tests/test_sentry_integration.py,sha256=LGmWiaLhFrx-jslrRj
 workflow_server/utils/tests/test_system_utils.py,sha256=_4GwXvVvU5BrATxUEWwQIPg0bzQXMWBtiBmjP8MTxJM,4314
 workflow_server/utils/tests/test_utils.py,sha256=qwK5Rmy3RQyjtlUrYAuGuDlBeRzZKsf1yS-y2IpUizQ,6452
 workflow_server/utils/utils.py,sha256=zcjnY6tSPf90VZOBm8w0NfxYWPQYAL9TCNlza231Odg,4535
-vellum_workflow_server-1.3.8.dist-info/METADATA,sha256=XpStc7y5oUXAvWcPnz60GZCzjgXjbMITkuIUfFhfdSM,2267
-vellum_workflow_server-1.3.8.dist-info/WHEEL,sha256=sP946D7jFCHeNz5Iq4fL4Lu-PrWrFsgfLXbbkciIZwg,88
-vellum_workflow_server-1.3.8.dist-info/entry_points.txt,sha256=uB_0yPkr7YV6RhEXzvFReUM8P4OQBlVXD6TN6eb9-oc,277
-vellum_workflow_server-1.3.8.dist-info/RECORD,,
+vellum_workflow_server-1.3.9.post2.dist-info/METADATA,sha256=PsYb7bP17M8X4L_d4-3XTGOeL4vQQHU8CLGJ7hqF2uI,2273
+vellum_workflow_server-1.3.9.post2.dist-info/WHEEL,sha256=sP946D7jFCHeNz5Iq4fL4Lu-PrWrFsgfLXbbkciIZwg,88
+vellum_workflow_server-1.3.9.post2.dist-info/entry_points.txt,sha256=uB_0yPkr7YV6RhEXzvFReUM8P4OQBlVXD6TN6eb9-oc,277
+vellum_workflow_server-1.3.9.post2.dist-info/RECORD,,

workflow_server/api/tests/test_workflow_view.py

@@ -392,6 +392,7 @@ class MyAdditionNode(BaseNode):
         "ports": [{"id": "9a9e4ef6-febf-4093-a515-217bbb1373db", "name": "default", "type": "DEFAULT"}],
         "trigger": {"id": "a5298668-d808-4a45-a62e-790943948e8a", "merge_behavior": "AWAIT_ATTRIBUTES"},
         "type": "GENERIC",
+        "should_file_merge": True,
     }
 
 
@@ -472,3 +473,64 @@ def test_serialize_route__with__workflow():
 
     # AND at least some of the expected nodes should be present
     assert not DeepDiff(node_labels, expected_nodes, ignore_order=True)
+
+
+def test_serialize_route__with_workspace_api_key():
+    """
+    Tests that the serialize route accepts workspace_api_key and passes it through serialization.
+    """
+    # GIVEN a Flask application
+    flask_app = create_app()
+
+    # AND a complete workflow with multiple files
+    workflow_files = {
+        "__init__.py": "# flake8: noqa: F401, F403\n\n",
+        "inputs.py": "from typing import Any, Optional\n\nfrom vellum.workflows.inputs import BaseInputs\n\n\nclass Inputs(BaseInputs):\n    text: str\n    var_1: Optional[Any]\n",  # noqa: E501
+        "workflow.py": "from vellum.workflows import BaseWorkflow\nfrom vellum.workflows.state import BaseState\n\nfrom .inputs import Inputs\nfrom .nodes.final_output import FinalOutput\nfrom .nodes.templating_node import TemplatingNode\n\n\nclass Workflow(BaseWorkflow[Inputs, BaseState]):\n    graph = TemplatingNode >> FinalOutput\n\n    class Outputs(BaseWorkflow.Outputs):\n        final_output = FinalOutput.Outputs.value\n",  # noqa: E501
+        "nodes/__init__.py": 'from .final_output import FinalOutput\nfrom .templating_node import TemplatingNode\n\n__all__ = [\n    "FinalOutput",\n    "TemplatingNode",\n]\n',  # noqa: E501
+        "nodes/templating_node.py": 'from vellum.workflows.nodes.displayable import TemplatingNode as BaseTemplatingNode\nfrom vellum.workflows.state import BaseState\n\nfrom ..inputs import Inputs\n\n\nclass TemplatingNode(BaseTemplatingNode[BaseState, str]):\n    template = """{{ text }}"""\n    inputs = {\n        "text": Inputs.text,\n    }\n',  # noqa: E501
+        "nodes/final_output.py": "from vellum.workflows.nodes.displayable import FinalOutputNode\nfrom vellum.workflows.state import BaseState\n\nfrom .templating_node import TemplatingNode\n\n\nclass FinalOutput(FinalOutputNode[BaseState, str]):\n    class Outputs(FinalOutputNode.Outputs):\n        value = TemplatingNode.Outputs.result\n",  # noqa: E501
+    }
+
+    # WHEN we make a request to the serialize route with workspace_api_key
+    with flask_app.test_client() as test_client:
+        response = test_client.post(
+            "/workflow/serialize", json={"files": workflow_files, "workspace_api_key": "test_workspace_key"}
+        )
+
+    # THEN we should get a successful response
+    assert response.status_code == 200
+
+    # AND the response should contain the full WorkflowSerializationResult
+    assert "exec_config" in response.json
+    assert "errors" in response.json
+
+
+def test_serialize_route__with_invalid_workspace_api_key():
+    """
+    Tests that the serialize route handles invalid workspace_api_key gracefully.
+    """
+    # GIVEN a Flask application
+    flask_app = create_app()
+
+    workflow_files = {
+        "__init__.py": "",
+        "workflow.py": (
+            "from vellum.workflows import BaseWorkflow\n\n"
+            "class Workflow(BaseWorkflow):\n"
+            "    class Outputs(BaseWorkflow.Outputs):\n"
+            "        foo = 'hello'\n"
+        ),
+    }
+
+    # WHEN we make a request with an invalid workspace_api_key
+    with flask_app.test_client() as test_client:
+        response = test_client.post(
+            "/workflow/serialize", json={"files": workflow_files, "workspace_api_key": ""}  # Invalid empty key
+        )
+
+    # THEN we should still get a successful response (graceful degradation)
+    assert response.status_code == 200
+
+    # AND the response should contain the serialization result
+    assert "exec_config" in response.json

workflow_server/api/workflow_view.py

@@ -37,6 +37,7 @@ from workflow_server.core.events import (
 )
 from workflow_server.core.executor import stream_node_pebble_timeout, stream_workflow, stream_workflow_process_timeout
 from workflow_server.core.utils import (
+    create_vellum_client,
     create_vembda_rejected_event,
     is_events_emitting_enabled,
     serialize_vembda_rejected_event,
@@ -55,11 +56,10 @@ from workflow_server.utils.system_utils import (
 from workflow_server.utils.utils import convert_json_inputs_to_vellum, get_version
 
 bp = Blueprint("exec", __name__)
+logger = logging.getLogger(__name__)
 
 set_start_method("fork", force=True)
 
-logger = logging.getLogger(__name__)
-
 CUSTOM_NODES_DIRECTORY = "vellum_custom_nodes"
 WORKFLOW_INITIATION_TIMEOUT_SECONDS = 60
 
@@ -425,6 +425,7 @@ def serialize_route() -> Response:
     data = request.get_json()
 
     files = data.get("files", {})
+    workspace_api_key = data.get("workspace_api_key")
 
     if not files:
         return Response(
@@ -434,6 +435,13 @@ def serialize_route() -> Response:
         )
 
     try:
+        client = None
+        if workspace_api_key:
+            try:
+                client = create_vellum_client(api_key=workspace_api_key)
+            except Exception as e:
+                logger.warning(f"Failed to create VellumClient with provided workspace_api_key: {str(e)}")
+
         # Generate a unique namespace for this serialization request
         namespace = "".join(random.choice(string.ascii_letters + string.digits) for i in range(14))
         virtual_finder = VirtualFileFinder(files, namespace)
@@ -441,7 +449,7 @@ def serialize_route() -> Response:
         try:
             sys.meta_path.append(virtual_finder)
             try:
-                result = BaseWorkflowDisplay.serialize_module(namespace)
+                result = BaseWorkflowDisplay.serialize_module(namespace, client=client)
             except Exception as e:
                 raise WorkflowInitializationException(str(e)) from e
 

workflow_server/core/executor.py

@@ -20,7 +20,6 @@ from vellum_ee.workflows.display.utils.events import event_enricher
 from vellum_ee.workflows.display.workflows import BaseWorkflowDisplay
 from vellum_ee.workflows.server.virtual_file_loader import VirtualFileFinder
 
-from vellum import Vellum, VellumEnvironment
 from vellum.workflows import BaseWorkflow
 from vellum.workflows.emitters.base import BaseWorkflowEmitter
 from vellum.workflows.emitters.vellum_emitter import VellumEmitter
@@ -36,7 +35,6 @@ from vellum.workflows.state.base import BaseState, StateMeta
 from vellum.workflows.state.context import WorkflowContext
 from vellum.workflows.state.store import EmptyStore
 from vellum.workflows.workflows.event_filters import all_workflow_event_filter
-from workflow_server.config import IS_VPC, VELLUM_API_URL_HOST, VELLUM_API_URL_PORT
 from workflow_server.core.cancel_workflow import CancelWorkflowWatcherThread
 from workflow_server.core.events import (
     SPAN_ID_EVENT,
@@ -44,7 +42,7 @@ from workflow_server.core.events import (
     VembdaExecutionFulfilledBody,
     VembdaExecutionFulfilledEvent,
 )
-from workflow_server.core.utils import is_events_emitting_enabled, serialize_vembda_rejected_event
+from workflow_server.core.utils import create_vellum_client, is_events_emitting_enabled, serialize_vembda_rejected_event
 from workflow_server.core.workflow_executor_context import (
     DEFAULT_TIMEOUT_SECONDS,
     BaseExecutorContext,
@@ -374,21 +372,10 @@ def _create_workflow(executor_context: WorkflowExecutorContext, namespace: str)
 
 
 def _create_workflow_context(executor_context: BaseExecutorContext) -> WorkflowContext:
-    if IS_VPC:
-        environment = VellumEnvironment(
-            default=os.getenv("VELLUM_DEFAULT_API_URL", VellumEnvironment.PRODUCTION.default),
-            documents=os.getenv("VELLUM_DOCUMENTS_API_URL", VellumEnvironment.PRODUCTION.documents),
-            predict=os.getenv("VELLUM_PREDICT_API_URL", VellumEnvironment.PRODUCTION.predict),
-        )
-    elif os.getenv("USE_LOCAL_VELLUM_API") == "true":
-        VELLUM_API_URL = f"http://{VELLUM_API_URL_HOST}:{VELLUM_API_URL_PORT}"
-        environment = VellumEnvironment(
-            default=VELLUM_API_URL,
-            documents=VELLUM_API_URL,
-            predict=VELLUM_API_URL,
-        )
-    else:
-        environment = VellumEnvironment.PRODUCTION
+    vellum_client = create_vellum_client(
+        api_key=executor_context.environment_api_key,
+        api_version=executor_context.api_version,
+    )
 
     if executor_context.environment_variables:
         os.environ.update(executor_context.environment_variables)
@@ -396,11 +383,7 @@ def _create_workflow_context(executor_context: BaseExecutorContext) -> WorkflowC
     namespace = _get_file_namespace(executor_context)
 
     return WorkflowContext(
-        vellum_client=Vellum(
-            api_key=executor_context.environment_api_key,
-            environment=environment,
-            api_version=executor_context.api_version,
-        ),
+        vellum_client=vellum_client,
         execution_context=executor_context.execution_context,
         generated_files=executor_context.files,
         namespace=namespace,

workflow_server/core/utils.py

@@ -1,7 +1,10 @@
 from datetime import datetime
+import os
 from uuid import uuid4
 from typing import Optional
 
+from vellum import ApiVersionEnum, Vellum, VellumEnvironment
+from workflow_server.config import IS_VPC, VELLUM_API_URL_HOST, VELLUM_API_URL_PORT
 from workflow_server.core.events import VembdaExecutionFulfilledBody, VembdaExecutionFulfilledEvent
 from workflow_server.core.workflow_executor_context import BaseExecutorContext
 
@@ -48,3 +51,42 @@ def is_events_emitting_enabled(executor_context: Optional[BaseExecutorContext])
         return False
 
     return executor_context.feature_flags.get("vembda-event-emitting-enabled") or False
+
+
+def create_vellum_client(
+    api_key: str,
+    api_version: Optional[ApiVersionEnum] = None,
+) -> Vellum:
+    """
+    Create a VellumClient with proper environment configuration.
+
+    Args:
+        api_key: The API key for the Vellum client
+        api_version: Optional API version to use
+
+    Returns:
+        Configured Vellum client instance
+
+    Note: Ideally we replace this with `vellum.workflows.vellum_client.create_vellum_client`
+    """
+    if IS_VPC:
+        environment = VellumEnvironment(
+            default=os.getenv("VELLUM_DEFAULT_API_URL", VellumEnvironment.PRODUCTION.default),
+            documents=os.getenv("VELLUM_DOCUMENTS_API_URL", VellumEnvironment.PRODUCTION.documents),
+            predict=os.getenv("VELLUM_PREDICT_API_URL", VellumEnvironment.PRODUCTION.predict),
+        )
+    elif os.getenv("USE_LOCAL_VELLUM_API") == "true":
+        VELLUM_API_URL = f"http://{VELLUM_API_URL_HOST}:{VELLUM_API_URL_PORT}"
+        environment = VellumEnvironment(
+            default=VELLUM_API_URL,
+            documents=VELLUM_API_URL,
+            predict=VELLUM_API_URL,
+        )
+    else:
+        environment = VellumEnvironment.PRODUCTION
+
+    return Vellum(
+        api_key=api_key,
+        environment=environment,
+        api_version=api_version,
+    )

workflow_server/start.py

@@ -49,8 +49,9 @@ def start() -> None:
         "bind": f"0.0.0.0:{PORT}",
         "workers": int(os.getenv("GUNICORN_WORKERS", 2)),
         "threads": int(os.getenv("GUNICORN_THREADS", 9 if ENABLE_PROCESS_WRAPPER else 6)),
-        # Try to avoid memory leaks when using non process mode
-        "max_requests": 0 if ENABLE_PROCESS_WRAPPER else 20,
+        # Aggressively try to avoid memory leaks when using non process mode
+        "max_requests": 120 if ENABLE_PROCESS_WRAPPER else 20,
+        "max_requests_jitter": 30 if ENABLE_PROCESS_WRAPPER else 10,
         "worker_class": "gthread",
         "timeout": int(os.getenv("GUNICORN_TIMEOUT", 1800)),
         "logger_class": CustomGunicornLogger,