Tracer: support traceable attribute

Author: ionmincu

sdk/core/uipath_sdk/_cli/_runtime/_contracts.py

@@ -135,6 +135,7 @@ class UiPathTraceContext(BaseModel):
     folder_key: Optional[str] = None
     process_key: Optional[str] = None
     enabled: Union[bool, str] = False
+    reference_id: Optional[str] = None
 
 
 class UiPathRuntimeContext(BaseModel):

sdk/core/uipath_sdk/_cli/cli_run.py

@@ -3,6 +3,7 @@
 import logging
 import os
 import traceback
+import uuid
 from os import environ as env
 from typing import Optional
 
@@ -72,6 +73,7 @@ async def execute():
                 tenant_id=env.get("UIPATH_TENANT_ID"),
                 process_key=env.get("UIPATH_PROCESS_UUID"),
                 folder_key=env.get("UIPATH_FOLDER_KEY"),
+                reference_id=env.get("UIPATH_JOB_KEY") or str(uuid.uuid4()),
             )
             context.logs_min_level = env.get("LOG_LEVEL", "INFO")
 

sdk/langchain/uipath_langchain/_cli/_runtime/_runtime.py

@@ -14,6 +14,7 @@
     UiPathRuntimeResult,
 )
 
+from ..._utils import _instrument_traceable
 from ...tracers import AsyncUiPathTracer
 from ._context import LangGraphRuntimeContext
 from ._exception import LangGraphRuntimeError
@@ -43,6 +44,7 @@ async def execute(self) -> Optional[UiPathRuntimeResult]:
         Raises:
             LangGraphRuntimeError: If execution fails
         """
+        _instrument_traceable()
 
         await self.validate()
 
@@ -71,7 +73,7 @@ async def execute(self) -> Optional[UiPathRuntimeResult]:
                 callbacks: List[BaseCallbackHandler] = []
 
                 if self.context.job_id and self.context.tracing_enabled:
-                    tracer = AsyncUiPathTracer()
+                    tracer = AsyncUiPathTracer(context=self.context.trace_context)
                     await tracer.init_trace(
                         self.context.entrypoint, self.context.job_id
                     )

sdk/langchain/uipath_langchain/_utils/__init__.py

@@ -1,3 +1,4 @@
+from ..tracers._instrument_traceable import _instrument_traceable
 from ._request_mixin import UiPathRequestMixin
 
-__all__ = ["UiPathRequestMixin"]
+__all__ = ["UiPathRequestMixin", "_instrument_traceable"]

sdk/langchain/uipath_langchain/tracers/AsyncUiPathTracer.py

@@ -6,13 +6,15 @@
 import uuid
 import warnings
 from os import environ as env
-from typing import Any, Optional
+from typing import Any, Dict, Optional
 
 import httpx
 from langchain_core.tracers.base import AsyncBaseTracer
 from langchain_core.tracers.schemas import Run
 from pydantic import PydanticDeprecationWarning
+from uipath_sdk._cli._runtime._contracts import UiPathTraceContext
 
+from ._events import CustomTraceEvents, FunctionCallEventData
 from ._utils import _simple_serialize_defaults
 
 logger = logging.getLogger(__name__)
@@ -25,78 +27,87 @@ class Status:
 
 
 class AsyncUiPathTracer(AsyncBaseTracer):
-    def __init__(self, client=None, **kwargs):
+    def __init__(self, context: UiPathTraceContext = None, client=None, **kwargs):
         super().__init__(**kwargs)
 
         self.client = client or httpx.AsyncClient()
         self.retries = 3
         self.log_queue: queue.Queue[dict[str, Any]] = queue.Queue()
 
+        self.context = context or UiPathTraceContext()
+
         llm_ops_pattern = self._get_base_url() + "{orgId}/llmops_"
-        self.orgId = env.get(
-            "UIPATH_ORGANIZATION_ID", "00000000-0000-0000-0000-000000000000"
-        )
-        self.tenantId = env.get(
-            "UIPATH_TENANT_ID", "00000000-0000-0000-0000-000000000000"
-        )
-        self.url = llm_ops_pattern.format(orgId=self.orgId).rstrip("/")
 
-        self.auth_token = env.get("UNATTENDED_USER_ACCESS_TOKEN") or env.get(
+        self.url = llm_ops_pattern.format(orgId=self.context.org_id).rstrip("/")
+
+        auth_token = env.get("UNATTENDED_USER_ACCESS_TOKEN") or env.get(
             "UIPATH_ACCESS_TOKEN"
         )
 
-        self.jobKey = env.get("UIPATH_JOB_KEY")
-        self.folderKey = env.get("UIPATH_FOLDER_KEY")
-        self.processKey = env.get("UIPATH_PROCESS_UUID")
-        self.parent_span_id = env.get("UIPATH_PARENT_SPAN_ID")
-
-        self.referenceId = self.jobKey or str(uuid.uuid4())
-
-        self.headers = {
-            "Authorization": f"Bearer {self.auth_token}",
-        }
+        self.headers = {"Authorization": f"Bearer {auth_token}"}
 
         self.running = True
         self.worker_task = asyncio.create_task(self._worker())
+        self.function_call_run_map: Dict[str, Run] = {}
+
+    async def on_custom_event(
+        self,
+        name: str,
+        data: dict,
+        *,
+        run_id: uuid.UUID,
+        tags=None,
+        metadata=None,
+        **kwargs: Any,
+    ) -> None:
+        if name == CustomTraceEvents.UIPATH_TRACE_FUNCTION_CALL:
+            # only handle the function call event
+
+            if not isinstance(data, FunctionCallEventData):
+                logger.warning(
+                    f"Received unexpected data type for function call event: {type(data)}"
+                )
+                return
 
-    def _get_base_url(self) -> str:
-        uipath_url = (
-            env.get("UIPATH_URL") or "https://cloud.uipath.com/dummyOrg/dummyTennant/"
-        )
-        uipath_url = uipath_url.rstrip("/")
+            if data.event_type == "call":
+                run = self.run_map[str(run_id)]
+                child_run = run.create_child(name=data.function_name, run_type=data.run_type, tags=data.tags)
+                run.add_metadata(data.metadata)
 
-        # split by "//" to get ['', 'https:', 'alpha.uipath.com/ada/byoa']
-        parts = uipath_url.split("//")
+                call_uuid = data.call_uuid
+                self.function_call_run_map[call_uuid] = child_run
 
-        # after splitting by //, the base URL will be at index 1 along with the rest,
-        # hence split it again using "/" to get ['https:', 'alpha.uipath.com', 'ada', 'byoa']
-        base_url_parts = parts[1].split("/")
+                self._send_span(run)
 
-        # combine scheme and netloc to get the base URL
-        base_url = parts[0] + "//" + base_url_parts[0] + "/"
+            if data.event_type == "completion":
+                call_uuid = data.call_uuid
+                run = self.function_call_run_map.pop(call_uuid, None)
 
-        return base_url
+                if run:
+                    run.end(self._safe_dict_dump(data.result), data.error)
+                    self._send_span(run)
 
     async def init_trace(self, run_name, trace_id=None) -> None:
-        trace_id_env = env.get("UIPATH_TRACE_ID")
+        if self.context.trace_id:
+            # trace id already set no need to do anything
+            return
 
-        if trace_id_env:
-            self.trace_parent = trace_id_env
-        else:
-            await self.start_trace(run_name, trace_id)
+        # no trace id, start a new trace
+        await self.start_trace(run_name, trace_id)
 
     async def start_trace(self, run_name, trace_id=None) -> None:
-        self.trace_parent = trace_id or str(uuid.uuid4())
-        run_name = run_name or f"Job Run: {self.trace_parent}"
+        self.context.trace_id = str(uuid.uuid4())
+
+        run_name = run_name or f"Job Run: {self.context.trace_id}"
         trace_data = {
-            "id": self.trace_parent,
+            "id": self.context.trace_id,
             "name": re.sub(
                 "[!@#$<>\.]", "", run_name
             ),  # if we use these characters the Agents UI throws some error (but llmops backend seems fine)
-            "referenceId": self.referenceId,
+            "referenceId": self.context.reference_id,
             "attributes": "{}",
-            "organizationId": self.orgId,
-            "tenantId": self.tenantId,
+            "organizationId": self.context.org_id,
+            "tenantId": self.context.tenant_id,
         }
 
         for attempt in range(self.retries):
@@ -174,9 +185,9 @@ async def _worker(self):
 
     async def _persist_run(self, run: Run) -> None:
         # Determine if this is a start or end trace based on whether end_time is set
-        await self._send_span(run)
+        self._send_span(run)
 
-    async def _send_span(self, run: Run) -> None:
+    def _send_span(self, run: Run) -> None:
         """Send span data for a run to the API"""
         run_id = str(run.id)
 
@@ -191,27 +202,27 @@ async def _send_span(self, run: Run) -> None:
             parent_id = (
                 str(run.parent_run_id)
                 if run.parent_run_id is not None
-                else self.parent_span_id
+                else self.context.parent_span_id
             )
-            attributes = self._safe_json_dump(self._run_to_dict(run))
+            attributes = self._safe_jsons_dump(self._run_to_dict(run))
             status = self._determine_status(run.error)
 
             span_data = {
                 "id": run_id,
                 "parentId": parent_id,
-                "traceId": self.trace_parent,
+                "traceId": self.context.trace_id,
                 "name": run.name,
                 "startTime": start_time,
                 "endTime": end_time,
-                "referenceId": self.referenceId,
+                "referenceId": self.context.reference_id,
                 "attributes": attributes,
-                "organizationId": self.orgId,
-                "tenantId": self.tenantId,
+                "organizationId": self.context.org_id,
+                "tenantId": self.context.tenant_id,
                 "spanType": "LangGraphRun",
                 "status": status,
-                "jobKey": self.jobKey,
-                "folderKey": self.folderKey,
-                "processKey": self.processKey,
+                "jobKey": self.context.job_id,
+                "folderKey": self.context.folder_key,
+                "processKey": self.context.folder_key,
             }
 
             self.log_queue.put(span_data)
@@ -235,13 +246,22 @@ def _determine_status(self, error: Optional[str]):
 
         return Status.SUCCESS
 
-    def _safe_json_dump(self, obj) -> str:
+    def _safe_jsons_dump(self, obj) -> str:
         try:
             json_str = json.dumps(obj, default=_simple_serialize_defaults)
             return json_str
         except Exception as e:
-            logger.warning(e)
+            logger.warning(f"Error serializing object to JSON: {e}")
             return "{ }"
+        
+    def _safe_dict_dump(self, obj) -> dict:
+        try:
+            serialized = json.loads(json.dumps(obj, default=_simple_serialize_defaults))
+            return serialized
+        except Exception as e:
+            # Last resort - string representation
+            logger.warning(f"Error serializing object to JSON: {e}")
+            return str(obj)
 
     def _run_to_dict(self, run: Run):
         with warnings.catch_warnings():
@@ -252,3 +272,21 @@ def _run_to_dict(self, run: Run):
                 "inputs": run.inputs.copy() if run.inputs is not None else None,
                 "outputs": run.outputs.copy() if run.outputs is not None else None,
             }
+
+    def _get_base_url(self) -> str:
+        uipath_url = (
+            env.get("UIPATH_URL") or "https://cloud.uipath.com/dummyOrg/dummyTennant/"
+        )
+        uipath_url = uipath_url.rstrip("/")
+
+        # split by "//" to get ['', 'https:', 'alpha.uipath.com/ada/byoa']
+        parts = uipath_url.split("//")
+
+        # after splitting by //, the base URL will be at index 1 along with the rest,
+        # hence split it again using "/" to get ['https:', 'alpha.uipath.com', 'ada', 'byoa']
+        base_url_parts = parts[1].split("/")
+
+        # combine scheme and netloc to get the base URL
+        base_url = parts[0] + "//" + base_url_parts[0] + "/"
+
+        return base_url

sdk/langchain/uipath_langchain/tracers/__init__.py

@@ -1,4 +1,5 @@
+from ._instrument_traceable import _instrument_traceable
 from .AsyncUiPathTracer import AsyncUiPathTracer
 from .UiPathTracer import UiPathTracer
 
-__all__ = ["AsyncUiPathTracer", "UiPathTracer"]
+__all__ = ["AsyncUiPathTracer", "UiPathTracer", "_instrument_traceable"]

sdk/langchain/uipath_langchain/tracers/_events.py

@@ -0,0 +1,29 @@
+from typing import Any, List, Mapping, Optional
+
+
+class CustomTraceEvents:
+    UIPATH_TRACE_FUNCTION_CALL = "__uipath_trace_function_call"
+
+
+class FunctionCallEventData:
+    def __init__(
+        self,
+        function_name: str,
+        event_type: str,
+        inputs: dict,
+        call_uuid: str,
+        output: Any,
+        error: Any,
+        run_type: Optional[str] = None,
+        tags: Optional[List[str]] = None,
+        metadata: Optional[Mapping[str, Any]] = None,
+    ):
+        self.function_name = function_name
+        self.event_type = event_type
+        self.inputs = inputs
+        self.call_uuid = call_uuid
+        self.output = output
+        self.error = str(error)
+        self.run_type = run_type
+        self.tags = tags
+        self.metadata = metadata