diff --git a/pyproject.toml b/pyproject.toml
index 44430e0..2795cdc 100644
--- a/pyproject.toml
+++ b/pyproject.toml
@@ -1,6 +1,6 @@
[project]
name = "uipath-llamaindex"
-version = "0.0.19"
+version = "0.0.20"
description = "UiPath LlamaIndex SDK"
readme = { file = "README.md", content-type = "text/markdown" }
requires-python = ">=3.10"
diff --git a/samples/action-center-hitl-agent/.env.example b/samples/action-center-hitl-agent/.env.example
new file mode 100644
index 0000000..36643c0
--- /dev/null
+++ b/samples/action-center-hitl-agent/.env.example
@@ -0,0 +1 @@
+OPENAI_API_KEY=xxx
diff --git a/samples/action-center-hitl-agent/agent.mermaid b/samples/action-center-hitl-agent/agent.mermaid
new file mode 100644
index 0000000..18ba266
--- /dev/null
+++ b/samples/action-center-hitl-agent/agent.mermaid
@@ -0,0 +1,35 @@
+flowchart TD
+ step__done["_done"]:::stepStyle
+ step_aggregate_tool_results["aggregate_tool_results"]:::stepStyle
+ step_call_tool["call_tool"]:::stepStyle
+ step_init_run["init_run"]:::stepStyle
+ step_parse_agent_output["parse_agent_output"]:::stepStyle
+ step_run_agent_step["run_agent_step"]:::stepStyle
+ step_setup_agent["setup_agent"]:::stepStyle
+ event_StopEvent([StopEvent
]):::stopEventStyle
+ event_ToolCallResult([ToolCallResult
]):::defaultEventStyle
+ event_AgentInput([AgentInput
]):::defaultEventStyle
+ event_ToolCall([ToolCall
]):::defaultEventStyle
+ event_AgentWorkflowStartEvent([AgentWorkflowStartEvent
]):::defaultEventStyle
+ event_AgentOutput([AgentOutput
]):::defaultEventStyle
+ event_AgentSetup([AgentSetup
]):::defaultEventStyle
+ event_StopEvent --> step__done
+ step_aggregate_tool_results --> event_AgentInput
+ step_aggregate_tool_results --> event_StopEvent
+ event_ToolCallResult --> step_aggregate_tool_results
+ step_call_tool --> event_ToolCallResult
+ event_ToolCall --> step_call_tool
+ step_init_run --> event_AgentInput
+ event_AgentWorkflowStartEvent --> step_init_run
+ step_parse_agent_output --> event_StopEvent
+ step_parse_agent_output --> event_ToolCall
+ event_AgentOutput --> step_parse_agent_output
+ step_run_agent_step --> event_AgentOutput
+ event_AgentSetup --> step_run_agent_step
+ step_setup_agent --> event_AgentSetup
+ event_AgentInput --> step_setup_agent
+ classDef stepStyle fill:#f2f0ff,line-height:1.2
+ classDef externalStyle fill:#f2f0ff,line-height:1.2
+ classDef defaultEventStyle fill-opacity:0
+ classDef stopEventStyle fill:#bfb6fc
+ classDef inputRequiredStyle fill:#f2f0ff,line-height:1.2
\ No newline at end of file
diff --git a/samples/action-center-hitl-agent/llama_index.json b/samples/action-center-hitl-agent/llama_index.json
new file mode 100644
index 0000000..c088db6
--- /dev/null
+++ b/samples/action-center-hitl-agent/llama_index.json
@@ -0,0 +1,7 @@
+{
+ "dependencies": ["."],
+ "workflows": {
+ "agent": "main.py:workflow"
+ },
+ "env": ".env"
+}
diff --git a/samples/action-center-hitl-agent/main.py b/samples/action-center-hitl-agent/main.py
new file mode 100644
index 0000000..be96195
--- /dev/null
+++ b/samples/action-center-hitl-agent/main.py
@@ -0,0 +1,58 @@
+import json
+
+from dotenv import load_dotenv
+from llama_index.core.agent.workflow import AgentWorkflow
+from llama_index.core.workflow import (
+ Context,
+ HumanResponseEvent,
+)
+from llama_index.llms.openai import OpenAI
+
+from uipath_llamaindex.models import CreateActionEvent
+
+load_dotenv()
+
+llm = OpenAI(model="gpt-4o-mini")
+
+
+async def may_research_company(ctx: Context, company_name: str) -> bool:
+ """Find whether a company may be researched.
+ Args:
+ ctx (Context): The context in which this function is called (autopopulated).
+ company_name (str): Name of the company to be researched.
+ Returns:
+ bool: True if the company can be researched, False otherwise.
+ """
+ # emit an event to the external stream to be captured
+ agent_name = "Company researcher"
+ ctx.write_event_to_stream(
+ CreateActionEvent(
+ prefix="hitl escalation to research company",
+ app_name="generic_escalation_app",
+ title=f"Action required for {agent_name}",
+ data={
+ "AgentOutput": (f"May I perform a research on company {company_name}?"),
+ "AgentName": agent_name,
+ },
+ app_version=1,
+ app_folder_path="Shared",
+ # assignee="(optional)"
+ )
+ )
+
+ # wait until we see a HumanResponseEvent
+ hitl_response = await ctx.wait_for_event(HumanResponseEvent)
+ feedback = json.loads(hitl_response.response)
+ # act on the input from the event
+ if isinstance(feedback["Answer"], bool) and feedback["Answer"] is True:
+ return True
+ else:
+ return False
+
+
+# example user input {"user_msg": "research Uipath company"}
+workflow = AgentWorkflow.from_tools_or_functions(
+ [may_research_company()],
+ llm=llm,
+ system_prompt="You are a helpful assistant that can use tools to perform actions requested by user",
+)
diff --git a/samples/action-center-hitl-agent/pyproject.toml b/samples/action-center-hitl-agent/pyproject.toml
new file mode 100644
index 0000000..fc4ef6a
--- /dev/null
+++ b/samples/action-center-hitl-agent/pyproject.toml
@@ -0,0 +1,11 @@
+[project]
+name = "llama-action-center"
+version = "0.0.1"
+description = "UiPath LlamaIndex Simple HITL Agent"
+authors = [{ name = "John Doe", email = "john.doe@myemail.com" }]
+readme = { file = "README.md", content-type = "text/markdown" }
+requires-python = ">=3.10"
+dependencies = [
+ "uipath-llamaindex==0.0.20",
+ "llama-index-llms-openai>=0.2.2"
+]
diff --git a/samples/action-center-hitl-agent/uipath.json b/samples/action-center-hitl-agent/uipath.json
new file mode 100644
index 0000000..f4aa963
--- /dev/null
+++ b/samples/action-center-hitl-agent/uipath.json
@@ -0,0 +1,27 @@
+{
+ "entryPoints": [
+ {
+ "filePath": "agent",
+ "uniqueId": "ca9074cb-d758-483c-a88d-4346b3214fb7",
+ "type": "agent",
+ "input": {
+ "type": "object",
+ "properties": {
+ "hack": {
+ "type": "object"
+ }
+ },
+ "required": []
+ },
+ "output": {
+ "type": "object",
+ "properties": {},
+ "required": []
+ }
+ }
+ ],
+ "bindings": {
+ "version": "2.0",
+ "resources": []
+ }
+}
diff --git a/samples/multi-agent/.env.example b/samples/multi-agent/.env.example
new file mode 100644
index 0000000..36643c0
--- /dev/null
+++ b/samples/multi-agent/.env.example
@@ -0,0 +1 @@
+OPENAI_API_KEY=xxx
diff --git a/samples/multi-agent/agent.mermaid b/samples/multi-agent/agent.mermaid
new file mode 100644
index 0000000..18ba266
--- /dev/null
+++ b/samples/multi-agent/agent.mermaid
@@ -0,0 +1,35 @@
+flowchart TD
+ step__done["_done"]:::stepStyle
+ step_aggregate_tool_results["aggregate_tool_results"]:::stepStyle
+ step_call_tool["call_tool"]:::stepStyle
+ step_init_run["init_run"]:::stepStyle
+ step_parse_agent_output["parse_agent_output"]:::stepStyle
+ step_run_agent_step["run_agent_step"]:::stepStyle
+ step_setup_agent["setup_agent"]:::stepStyle
+ event_StopEvent([StopEvent
]):::stopEventStyle
+ event_ToolCallResult([ToolCallResult
]):::defaultEventStyle
+ event_AgentInput([AgentInput
]):::defaultEventStyle
+ event_ToolCall([ToolCall
]):::defaultEventStyle
+ event_AgentWorkflowStartEvent([AgentWorkflowStartEvent
]):::defaultEventStyle
+ event_AgentOutput([AgentOutput
]):::defaultEventStyle
+ event_AgentSetup([AgentSetup
]):::defaultEventStyle
+ event_StopEvent --> step__done
+ step_aggregate_tool_results --> event_AgentInput
+ step_aggregate_tool_results --> event_StopEvent
+ event_ToolCallResult --> step_aggregate_tool_results
+ step_call_tool --> event_ToolCallResult
+ event_ToolCall --> step_call_tool
+ step_init_run --> event_AgentInput
+ event_AgentWorkflowStartEvent --> step_init_run
+ step_parse_agent_output --> event_StopEvent
+ step_parse_agent_output --> event_ToolCall
+ event_AgentOutput --> step_parse_agent_output
+ step_run_agent_step --> event_AgentOutput
+ event_AgentSetup --> step_run_agent_step
+ step_setup_agent --> event_AgentSetup
+ event_AgentInput --> step_setup_agent
+ classDef stepStyle fill:#f2f0ff,line-height:1.2
+ classDef externalStyle fill:#f2f0ff,line-height:1.2
+ classDef defaultEventStyle fill-opacity:0
+ classDef stopEventStyle fill:#bfb6fc
+ classDef inputRequiredStyle fill:#f2f0ff,line-height:1.2
\ No newline at end of file
diff --git a/samples/multi-agent/llama_index.json b/samples/multi-agent/llama_index.json
new file mode 100644
index 0000000..c088db6
--- /dev/null
+++ b/samples/multi-agent/llama_index.json
@@ -0,0 +1,7 @@
+{
+ "dependencies": ["."],
+ "workflows": {
+ "agent": "main.py:workflow"
+ },
+ "env": ".env"
+}
diff --git a/samples/multi-agent/main.py b/samples/multi-agent/main.py
new file mode 100644
index 0000000..d6702e1
--- /dev/null
+++ b/samples/multi-agent/main.py
@@ -0,0 +1,49 @@
+import json
+
+from dotenv import load_dotenv
+from llama_index.core.agent.workflow import AgentWorkflow
+from llama_index.core.workflow import (
+ Context,
+ HumanResponseEvent,
+)
+from llama_index.llms.openai import OpenAI
+
+from uipath_llamaindex.models import InvokeProcessEvent
+
+load_dotenv()
+
+llm = OpenAI(model="gpt-4o-mini")
+
+
+async def may_research_company(ctx: Context, company_name: str) -> str:
+ """Find whether a company may be researcher.
+ Args:
+ ctx (Context): The context in which this function is called (autopopulated).
+ company_name (str): Name of the company to be researched.
+ Returns:
+ str: company report
+ """
+ # emit an event to the external stream to be captured
+ ctx.write_event_to_stream(
+ InvokeProcessEvent(
+ prefix="invoke langgraph researcher event",
+ name="my-first-uipath-agent",
+ # process_folder_path="(optional)",
+ input_arguments={
+ "topic": company_name,
+ },
+ )
+ )
+
+ # wait until we see a HumanResponseEvent
+ hitl_response = await ctx.wait_for_event(HumanResponseEvent)
+ feedback = json.loads(hitl_response.response)
+ # act on the input from the event
+ return feedback["report"]
+
+
+workflow = AgentWorkflow.from_tools_or_functions(
+ [may_research_company],
+ llm=llm,
+ system_prompt="You are a helpful assistant that can decide whether a company can be researched or not.",
+)
diff --git a/samples/multi-agent/pyproject.toml b/samples/multi-agent/pyproject.toml
new file mode 100644
index 0000000..2a1d3e3
--- /dev/null
+++ b/samples/multi-agent/pyproject.toml
@@ -0,0 +1,11 @@
+[project]
+name = "llama-multi-agent"
+version = "0.0.1"
+description = "UiPath LlamaIndex Simple HITL Agent"
+authors = [{ name = "John Doe", email = "john.doe@myemail.com" }]
+readme = { file = "README.md", content-type = "text/markdown" }
+requires-python = ">=3.10"
+dependencies = [
+ "uipath-llamaindex==0.0.20",
+ "llama-index-llms-openai>=0.2.2"
+]
diff --git a/samples/multi-agent/uipath.json b/samples/multi-agent/uipath.json
new file mode 100644
index 0000000..f078efc
--- /dev/null
+++ b/samples/multi-agent/uipath.json
@@ -0,0 +1,27 @@
+{
+ "entryPoints": [
+ {
+ "filePath": "agent",
+ "uniqueId": "ca9074cb-d758-483c-a88d-4346b3214fb8",
+ "type": "agent",
+ "input": {
+ "type": "object",
+ "properties": {
+ "hack": {
+ "type": "object"
+ }
+ },
+ "required": []
+ },
+ "output": {
+ "type": "object",
+ "properties": {},
+ "required": []
+ }
+ }
+ ],
+ "bindings": {
+ "version": "2.0",
+ "resources": []
+ }
+}
diff --git a/samples/simple-hitl-agent/main.py b/samples/simple-hitl-agent/main.py
index fabbbcc..25d535d 100644
--- a/samples/simple-hitl-agent/main.py
+++ b/samples/simple-hitl-agent/main.py
@@ -9,13 +9,21 @@
llm = OpenAI(model="gpt-4o-mini")
-async def research_company(ctx: Context) -> str:
- """Research a company."""
+async def may_research_company(ctx: Context, company_name: str) -> bool:
+ """Find whether a company may be researched.
+ Args:
+ ctx (Context): The context in which this function is called (autopopulated).
+ company_name (str): Name of the company to be researched.
+ Returns:
+ bool: True if the company can be researched, False otherwise.
+ """
print("Researching company...")
# emit an event to the external stream to be captured
ctx.write_event_to_stream(
- InputRequiredEvent(prefix="Are you sure you want to proceed?")
+ InputRequiredEvent(
+ prefix=f"May I perform a research on company {company_name}? \n (yes/no)"
+ )
)
# wait until we see a HumanResponseEvent
@@ -24,13 +32,13 @@ async def research_company(ctx: Context) -> str:
# act on the input from the event
if response.response.strip().lower() == "yes":
- return "Research completed successfully."
+ return True
else:
- return "Research task aborted."
+ return False
workflow = AgentWorkflow.from_tools_or_functions(
- [research_company],
+ [may_research_company],
llm=llm,
- system_prompt="You are a helpful assistant that can research companies.",
+ system_prompt="You are a helpful assistant that can decide whether a company can be researched or not.",
)
diff --git a/samples/simple-hitl-agent/pyproject.toml b/samples/simple-hitl-agent/pyproject.toml
index 4461e02..7d5ed85 100644
--- a/samples/simple-hitl-agent/pyproject.toml
+++ b/samples/simple-hitl-agent/pyproject.toml
@@ -1,11 +1,11 @@
[project]
name = "llama-simple-hitl-agent"
-version = "0.0.8"
+version = "0.0.1"
description = "UiPath LlamaIndex Simple HITL Agent"
authors = [{ name = "John Doe", email = "john.doe@myemail.com" }]
readme = { file = "README.md", content-type = "text/markdown" }
requires-python = ">=3.10"
dependencies = [
- "uipath-llamaindex>=0.0.18",
+ "uipath-llamaindex==0.0.20",
"llama-index-llms-openai>=0.2.2"
]
diff --git a/samples/simple-hitl-agent/uipath.json b/samples/simple-hitl-agent/uipath.json
index 0632247..f4aa963 100644
--- a/samples/simple-hitl-agent/uipath.json
+++ b/samples/simple-hitl-agent/uipath.json
@@ -2,7 +2,7 @@
"entryPoints": [
{
"filePath": "agent",
- "uniqueId": "ca9074cb-d758-483c-a88d-4346b3214fb6",
+ "uniqueId": "ca9074cb-d758-483c-a88d-4346b3214fb7",
"type": "agent",
"input": {
"type": "object",
diff --git a/samples/simple-hitl-agent/uv.lock b/samples/simple-hitl-agent/uv.lock
index a0b095c..e223a57 100644
--- a/samples/simple-hitl-agent/uv.lock
+++ b/samples/simple-hitl-agent/uv.lock
@@ -1110,20 +1110,8 @@ wheels = [
]
[[package]]
-name = "llama-parse"
-version = "0.6.23"
-source = { registry = "https://pypi.org/simple" }
-dependencies = [
- { name = "llama-cloud-services" },
-]
-sdist = { url = "https://files.pythonhosted.org/packages/11/30/8170f31a139314479dafea5e2140405e5127207fbc95056aedee01d6b7fd/llama_parse-0.6.23.tar.gz", hash = "sha256:ee575d12660de57264900ae414e7c61646db9fbdd030fd795fb4ce4d28d83f85", size = 3537 }
-wheels = [
- { url = "https://files.pythonhosted.org/packages/cd/53/a7e06181f36eaff6c7176e33644db2f19a7761fc9110f0a5d13759ce94ac/llama_parse-0.6.23-py3-none-any.whl", hash = "sha256:8290b08c28fb6cec17b22df7bf37678bbf4f76e8864eccefdbe5d4c04b0e994a", size = 4944 },
-]
-
-[[package]]
-name = "llama-simple-hitl-agent"
-version = "0.0.7"
+name = "llama-new"
+version = "0.0.2"
source = { virtual = "." }
dependencies = [
{ name = "llama-index-llms-openai" },
@@ -1133,7 +1121,19 @@ dependencies = [
[package.metadata]
requires-dist = [
{ name = "llama-index-llms-openai", specifier = ">=0.2.2" },
- { name = "uipath-llamaindex", specifier = ">=0.0.17" },
+ { name = "uipath-llamaindex", specifier = "==0.0.20.dev1000220026", index = "https://test.pypi.org/simple/" },
+]
+
+[[package]]
+name = "llama-parse"
+version = "0.6.23"
+source = { registry = "https://pypi.org/simple" }
+dependencies = [
+ { name = "llama-cloud-services" },
+]
+sdist = { url = "https://files.pythonhosted.org/packages/11/30/8170f31a139314479dafea5e2140405e5127207fbc95056aedee01d6b7fd/llama_parse-0.6.23.tar.gz", hash = "sha256:ee575d12660de57264900ae414e7c61646db9fbdd030fd795fb4ce4d28d83f85", size = 3537 }
+wheels = [
+ { url = "https://files.pythonhosted.org/packages/cd/53/a7e06181f36eaff6c7176e33644db2f19a7761fc9110f0a5d13759ce94ac/llama_parse-0.6.23-py3-none-any.whl", hash = "sha256:8290b08c28fb6cec17b22df7bf37678bbf4f76e8864eccefdbe5d4c04b0e994a", size = 4944 },
]
[[package]]
@@ -2605,16 +2605,16 @@ wheels = [
[[package]]
name = "uipath-llamaindex"
-version = "0.0.17"
-source = { registry = "https://pypi.org/simple" }
+version = "0.0.20.dev1000220026"
+source = { registry = "https://test.pypi.org/simple/" }
dependencies = [
{ name = "llama-index" },
{ name = "openinference-instrumentation-llama-index" },
{ name = "uipath" },
]
-sdist = { url = "https://files.pythonhosted.org/packages/cf/e3/01e45a09b47cb8419c01146c99d71c806afa44612107dd504a2011f1be4c/uipath_llamaindex-0.0.17.tar.gz", hash = "sha256:ac646a2023ccef76f52bdd3e1f37323dc4cc5d7f16f891815859960c693d57ad", size = 589828 }
+sdist = { url = "https://test-files.pythonhosted.org/packages/98/ce/6d3c35c57148a6cd9495c9311a7958cf0ab3a7c8e3a658d394d12c76fca7/uipath_llamaindex-0.0.20.dev1000220026.tar.gz", hash = "sha256:60644f798277b4008d4b00c9c8c885f6b8cc2d55c2d296ee5f82d763e60e7455", size = 590673 }
wheels = [
- { url = "https://files.pythonhosted.org/packages/97/ea/b62a4b9a4df06db0b8518365ca090acc990947250eb3b4c7fb0f89c2df25/uipath_llamaindex-0.0.17-py3-none-any.whl", hash = "sha256:7bd1227d5e0216c7349bdaf819cb7168cfe0f6cc332462535d4b098629615186", size = 16550 },
+ { url = "https://test-files.pythonhosted.org/packages/80/c7/47c37cc4162d90e7be67123af7d101680629003f27284ba8a6190aa87334/uipath_llamaindex-0.0.20.dev1000220026-py3-none-any.whl", hash = "sha256:9a3ba88bf8003a8aeb2a0440e34b7cc7c1a571b957c599927149c605e3aa58e6", size = 19396 },
]
[[package]]
diff --git a/src/uipath_llamaindex/_cli/_runtime/_hitl.py b/src/uipath_llamaindex/_cli/_runtime/_hitl.py
new file mode 100644
index 0000000..f2cce4e
--- /dev/null
+++ b/src/uipath_llamaindex/_cli/_runtime/_hitl.py
@@ -0,0 +1,194 @@
+import json
+import uuid
+from dataclasses import dataclass
+from functools import cached_property
+from typing import Any, Optional
+
+from llama_index.core.workflow import InputRequiredEvent
+from uipath import UiPath
+from uipath._cli._runtime._contracts import (
+ UiPathApiTrigger,
+ UiPathErrorCategory,
+ UiPathResumeTrigger,
+ UiPathResumeTriggerType,
+ UiPathRuntimeError,
+ UiPathRuntimeStatus,
+)
+from uipath.models import CreateAction, InvokeProcess, WaitAction, WaitJob
+
+uipath = UiPath()
+
+
+def _try_convert_to_json_format(value: str) -> str:
+ try:
+ return json.loads(value)
+ except json.decoder.JSONDecodeError:
+ return value
+
+
+async def _get_api_payload(inbox_id: str) -> Any:
+ """
+ Fetch payload data for API triggers.
+
+ Args:
+ inbox_id: The Id of the inbox to fetch the payload for.
+
+ Returns:
+ The value field from the API response payload, or None if an error occurs.
+ """
+ response = None
+ try:
+ response = uipath.api_client.request(
+ "GET",
+ f"/orchestrator_/api/JobTriggers/GetPayload/{inbox_id}",
+ include_folder_headers=True,
+ )
+ data = response.json()
+ return data.get("payload")
+ except Exception as e:
+ raise UiPathRuntimeError(
+ "API_CONNECTION_ERROR",
+ "Failed to get trigger payload",
+ f"Error fetching API trigger payload for inbox {inbox_id}: {str(e)}",
+ UiPathErrorCategory.SYSTEM,
+ response.status_code if response else None,
+ ) from e
+
+
+class HitlReader:
+ @classmethod
+ async def read(cls, resume_trigger: UiPathResumeTrigger) -> Optional[str]:
+ match resume_trigger.trigger_type:
+ case UiPathResumeTriggerType.ACTION:
+ if resume_trigger.item_key:
+ action = await uipath.actions.retrieve_async(
+ resume_trigger.item_key,
+ app_folder_key=resume_trigger.folder_key,
+ app_folder_path=resume_trigger.folder_path,
+ )
+ return action.data
+
+ case UiPathResumeTriggerType.JOB:
+ if resume_trigger.item_key:
+ job = await uipath.jobs.retrieve_async(
+ resume_trigger.item_key,
+ folder_key=resume_trigger.folder_key,
+ folder_path=resume_trigger.folder_path,
+ )
+ if (
+ job.state
+ and not job.state.lower()
+ == UiPathRuntimeStatus.SUCCESSFUL.value.lower()
+ ):
+ raise UiPathRuntimeError(
+ "INVOKED_PROCESS_FAILURE",
+ "Invoked process did not finish successfully.",
+ _try_convert_to_json_format(str(job.job_error or job.info)),
+ )
+ return job.output_arguments
+
+ case UiPathResumeTriggerType.API:
+ if resume_trigger.api_resume.inbox_id:
+ return await _get_api_payload(resume_trigger.api_resume.inbox_id)
+
+ case _:
+ raise UiPathRuntimeError(
+ "UNKNOWN_TRIGGER_TYPE",
+ "Unexpected trigger type received",
+ f"Trigger type :{type(resume_trigger.trigger_type)} is invalid",
+ UiPathErrorCategory.USER,
+ )
+
+ raise UiPathRuntimeError(
+ "HITL_FEEDBACK_FAILURE",
+ "Failed to receive payload from HITL action",
+ detail="Failed to receive payload from HITL action",
+ category=UiPathErrorCategory.SYSTEM,
+ )
+
+
+@dataclass
+class HitlProcessor:
+ """Processes events in a Human-(Robot/Agent)-In-The-Loop scenario."""
+
+ value: Any
+
+ @cached_property
+ def type(self) -> Optional[UiPathResumeTriggerType]:
+ """Returns the type of the interrupt value."""
+ if isinstance(self.value, CreateAction) or isinstance(self.value, WaitAction):
+ return UiPathResumeTriggerType.ACTION
+ if isinstance(self.value, InvokeProcess) or isinstance(self.value, WaitJob):
+ return UiPathResumeTriggerType.JOB
+ if isinstance(self.value, InputRequiredEvent):
+ return UiPathResumeTriggerType.API
+ return UiPathResumeTriggerType.NONE
+
+ async def create_resume_trigger(self) -> Optional[UiPathResumeTrigger]:
+ """Returns the resume trigger."""
+ try:
+ hitl_input = self.value
+ resume_trigger = UiPathResumeTrigger(
+ triggerType=self.type, interruptObject=hitl_input.model_dump_json()
+ )
+ match self.type:
+ case UiPathResumeTriggerType.ACTION:
+ resume_trigger.folder_path = hitl_input.app_folder_path
+ resume_trigger.folder_key = hitl_input.app_folder_key
+ if isinstance(hitl_input, WaitAction):
+ resume_trigger.item_key = hitl_input.action.key
+ elif isinstance(hitl_input, CreateAction):
+ action = await uipath.actions.create_async(
+ title=hitl_input.title,
+ app_name=hitl_input.app_name if hitl_input.app_name else "",
+ app_folder_path=hitl_input.app_folder_path
+ if hitl_input.app_folder_path
+ else "",
+ app_folder_key=hitl_input.app_folder_key
+ if hitl_input.app_folder_key
+ else "",
+ app_key=hitl_input.app_key if hitl_input.app_key else "",
+ app_version=hitl_input.app_version
+ if hitl_input.app_version
+ else 1,
+ assignee=hitl_input.assignee if hitl_input.assignee else "",
+ data=hitl_input.data,
+ )
+ if action:
+ resume_trigger.item_key = action.key
+
+ case UiPathResumeTriggerType.JOB:
+ resume_trigger.folder_path = hitl_input.process_folder_path
+ resume_trigger.folder_key = hitl_input.process_folder_key
+ if isinstance(hitl_input, WaitJob):
+ resume_trigger.item_key = hitl_input.job.key
+ elif isinstance(hitl_input, InvokeProcess):
+ job = await uipath.processes.invoke_async(
+ name=hitl_input.name,
+ input_arguments=hitl_input.input_arguments,
+ folder_path=hitl_input.process_folder_path,
+ folder_key=hitl_input.process_folder_key,
+ )
+ if job:
+ resume_trigger.item_key = job.key
+
+ case UiPathResumeTriggerType.API:
+ resume_trigger.api_resume = UiPathApiTrigger(
+ inboxId=str(uuid.uuid4()), request=hitl_input.prefix
+ )
+ case _:
+ raise UiPathRuntimeError(
+ "UNKNOWN_HITL_MODEL",
+ "Unexpected model received",
+ f"{type(hitl_input)} is not a valid Human(Robot/Agent)-In-The-Loop model",
+ UiPathErrorCategory.USER,
+ )
+ except Exception as e:
+ raise UiPathRuntimeError(
+ "HITL_ACTION_CREATION_FAILED",
+ "Failed to create HITL action",
+ f"{str(e)}",
+ UiPathErrorCategory.SYSTEM,
+ ) from e
+
+ return resume_trigger
diff --git a/src/uipath_llamaindex/_cli/_runtime/_runtime.py b/src/uipath_llamaindex/_cli/_runtime/_runtime.py
index 93024cd..4236b22 100644
--- a/src/uipath_llamaindex/_cli/_runtime/_runtime.py
+++ b/src/uipath_llamaindex/_cli/_runtime/_runtime.py
@@ -2,9 +2,8 @@
import logging
import os
import pickle
-import uuid
from contextlib import suppress
-from typing import Any, Optional, cast
+from typing import Optional, cast
from llama_index.core.workflow import (
Context,
@@ -19,7 +18,6 @@
from opentelemetry.sdk.trace.export import BatchSpanProcessor
from uipath import UiPath
from uipath._cli._runtime._contracts import (
- UiPathApiTrigger,
UiPathBaseRuntime,
UiPathErrorCategory,
UiPathResumeTrigger,
@@ -30,6 +28,7 @@
from .._tracing._oteladapter import LlamaIndexExporter
from ._context import UiPathLlamaIndexRuntimeContext
from ._exception import UiPathLlamaIndexRuntimeError
+from ._hitl import HitlProcessor, HitlReader
logger = logging.getLogger(__name__)
@@ -83,11 +82,13 @@ async def execute(self) -> Optional[UiPathRuntimeResult]:
**self.context.input_json,
)
- resume_trigger: UiPathResumeTrigger = None
+ resume_trigger: Optional[UiPathResumeTrigger] = None
response_applied = False
async for event in handler.stream_events():
+ # log the received event on trace level
if isinstance(event, InputRequiredEvent):
+ hitl_processor = HitlProcessor(value=event)
if self.context.resume and not response_applied:
# If we are resuming, we need to apply the response to the event stream.
response_applied = True
@@ -95,13 +96,8 @@ async def execute(self) -> Optional[UiPathRuntimeResult]:
await self.get_response_event()
)
else:
- resume_trigger = UiPathResumeTrigger(
- api_resume=UiPathApiTrigger(
- inbox_id=str(uuid.uuid4()), request=event.prefix
- )
- )
+ resume_trigger = await hitl_processor.create_resume_trigger()
break
- print(event)
if resume_trigger is None:
output = await handler
@@ -250,7 +246,7 @@ async def load_workflow_context(self):
loaded_ctx_dict,
serializer=serializer,
)
-
+ # TODO check multiple HITL same agent
resumed_trigger_data = loaded_ctx_dict["uipath_resume_trigger"]
if resumed_trigger_data:
self.context.resumed_trigger = cast(
@@ -267,41 +263,15 @@ async def get_response_event(self) -> Optional[HumanResponseEvent]:
if self.context.input_json:
# If input_json is provided, use it to create a HumanResponseEvent
return HumanResponseEvent(**self.context.input_json)
- # If resumed_trigger is set, fetch the payload from the API
+ # If resumed_trigger is set, fetch the feedback
if self.context.resumed_trigger:
- inbox_id = self.context.resumed_trigger.api_resume.inbox_id
- payload = await self._get_api_payload(inbox_id)
- if payload:
- return HumanResponseEvent(response=payload)
+ feedback = await HitlReader.read(self.context.resumed_trigger)
+ if feedback:
+ if isinstance(feedback, dict):
+ feedback = json.dumps(feedback)
+ return HumanResponseEvent(response=feedback)
return None
- async def _get_api_payload(self, inbox_id: str) -> Any:
- """
- Fetch payload data for API triggers.
-
- Args:
- inbox_id: The Id of the inbox to fetch the payload for.
-
- Returns:
- The value field from the API response payload, or None if an error occurs.
- """
- try:
- response = self._uipath.api_client.request(
- "GET",
- f"/orchestrator_/api/JobTriggers/GetPayload/{inbox_id}",
- include_folder_headers=True,
- )
- data = response.json()
- return data.get("payload")
- except Exception as e:
- raise UiPathLlamaIndexRuntimeError(
- "API_CONNECTION_ERROR",
- "Failed to get trigger payload",
- f"Error fetching API trigger payload for inbox {inbox_id}: {str(e)}",
- UiPathErrorCategory.SYSTEM,
- response.status_code,
- ) from e
-
def _serialize_object(self, obj):
"""Recursively serializes an object and all its nested components."""
# Handle Pydantic models
diff --git a/src/uipath_llamaindex/_cli/_tracing/_oteladapter.py b/src/uipath_llamaindex/_cli/_tracing/_oteladapter.py
index af92be4..6aa4a05 100644
--- a/src/uipath_llamaindex/_cli/_tracing/_oteladapter.py
+++ b/src/uipath_llamaindex/_cli/_tracing/_oteladapter.py
@@ -9,15 +9,15 @@
logger = logging.getLogger(__name__)
-class LlamaIndexExporter(LlmOpsHttpExporter):
+class LlamaIndexExporter(LlmOpsHttpExporter):
# Mapping of old attribute names to new attribute names or (new name, function)
ATTRIBUTE_MAPPING = {
"input.value": ("input", lambda s: json.loads(s)),
"output.value": ("output", lambda s: json.loads(s)),
"llm.model_name": "model",
}
-
+
# Mapping of span types
SPAN_TYPE_MAPPING = {
"LLM": "completion",
@@ -41,9 +41,11 @@ def _send_with_retries(
# Remove the span kind attribute
span_type = attributes["openinference.span.kind"]
# Map span type using SPAN_TYPE_MAPPING
- span_data["SpanType"] = self.SPAN_TYPE_MAPPING.get(span_type, span_type)
+ span_data["SpanType"] = self.SPAN_TYPE_MAPPING.get(
+ span_type, span_type
+ )
del attributes["openinference.span.kind"]
-
+
# Apply the transformation logic
for old_key, mapping in self.ATTRIBUTE_MAPPING.items():
if old_key in attributes:
@@ -57,14 +59,14 @@ def _send_with_retries(
new_key = mapping
attributes[new_key] = attributes[old_key]
del attributes[old_key]
-
+
# Transform token usage data if present
token_keys = {
"llm.token_count.prompt": "promptTokens",
- "llm.token_count.completion": "completionTokens",
- "llm.token_count.total": "totalTokens"
+ "llm.token_count.completion": "completionTokens",
+ "llm.token_count.total": "totalTokens",
}
-
+
# Check if any token count keys exist
if any(key in attributes for key in token_keys):
usage = {}
@@ -72,27 +74,28 @@ def _send_with_retries(
if old_key in attributes:
usage[new_key] = attributes[old_key]
del attributes[old_key]
-
+
# Add default values for BYO execution fields
usage["isByoExecution"] = False
# usage["executionDeploymentType"] = "PAYGO"
-
+
# Add usage to attributes
attributes["usage"] = usage
-
+
# Clean up any other token count attributes
- keys_to_remove = [k for k in attributes if k.startswith("llm.token_count.")]
+ keys_to_remove = [
+ k for k in attributes if k.startswith("llm.token_count.")
+ ]
for key in keys_to_remove:
del attributes[key]
-
+
# Convert back to JSON string
span_data["attributes"] = json.dumps(attributes)
except json.JSONDecodeError as e:
logger.warning(f"Failed to parse attributes JSON: {e}")
-
return super()._send_with_retries(
url=url,
payload=payload,
max_retries=max_retries,
- )
\ No newline at end of file
+ )
diff --git a/src/uipath_llamaindex/_cli/cli_init.py b/src/uipath_llamaindex/_cli/cli_init.py
index e4ff548..36ba9a1 100644
--- a/src/uipath_llamaindex/_cli/cli_init.py
+++ b/src/uipath_llamaindex/_cli/cli_init.py
@@ -124,7 +124,7 @@ def draw_all_possible_flows_mermaid(
# Only one kind of `StopEvent` is allowed in a `Workflow`.
current_stop_event = None
for _, step_func in steps.items():
- step_config : StepConfig = getattr(step_func, "__step_config", None)
+ step_config: StepConfig = getattr(step_func, "__step_config", None)
if step_config is None:
continue
@@ -176,7 +176,7 @@ def draw_all_possible_flows_mermaid(
if event_id not in nodes:
nodes.add(event_id)
style = get_event_style(event_type)
- mermaid_diagram.append(f' {event_id}([{event_name}
]):::{style}')
+ mermaid_diagram.append(f" {event_id}([{event_name}
]):::{style}")
if issubclass(event_type, InputRequiredEvent):
# Add node for conceptual external step
@@ -241,7 +241,9 @@ def draw_all_possible_flows_mermaid(
mermaid_diagram.append(" classDef externalStyle fill:#f2f0ff,line-height:1.2")
mermaid_diagram.append(" classDef defaultEventStyle fill-opacity:0")
mermaid_diagram.append(" classDef stopEventStyle fill:#bfb6fc")
- mermaid_diagram.append(" classDef inputRequiredStyle fill:#f2f0ff,line-height:1.2")
+ mermaid_diagram.append(
+ " classDef inputRequiredStyle fill:#f2f0ff,line-height:1.2"
+ )
# Join all lines
mermaid_string = "\n".join(mermaid_diagram)
diff --git a/src/uipath_llamaindex/models/__init__.py b/src/uipath_llamaindex/models/__init__.py
new file mode 100644
index 0000000..1fce6a0
--- /dev/null
+++ b/src/uipath_llamaindex/models/__init__.py
@@ -0,0 +1,13 @@
+from .hitl_models import (
+ CreateActionEvent,
+ InvokeProcessEvent,
+ WaitActionEvent,
+ WaitJobEvent,
+)
+
+__all__ = [
+ "CreateActionEvent",
+ "WaitActionEvent",
+ "InvokeProcessEvent",
+ "WaitJobEvent",
+]
diff --git a/src/uipath_llamaindex/models/hitl_models.py b/src/uipath_llamaindex/models/hitl_models.py
new file mode 100644
index 0000000..5c75812
--- /dev/null
+++ b/src/uipath_llamaindex/models/hitl_models.py
@@ -0,0 +1,18 @@
+from llama_index.core.workflow import InputRequiredEvent
+from uipath.models import CreateAction, InvokeProcess, WaitAction, WaitJob
+
+
+class InvokeProcessEvent(InvokeProcess, InputRequiredEvent):
+ pass
+
+
+class WaitJobEvent(WaitJob, InputRequiredEvent):
+ pass
+
+
+class CreateActionEvent(CreateAction, InputRequiredEvent):
+ pass
+
+
+class WaitActionEvent(WaitAction, InputRequiredEvent):
+ pass
diff --git a/uv.lock b/uv.lock
index a612099..bbdaf96 100644
--- a/uv.lock
+++ b/uv.lock
@@ -2840,7 +2840,7 @@ wheels = [
[[package]]
name = "uipath-llamaindex"
-version = "0.0.11"
+version = "0.0.20"
source = { editable = "." }
dependencies = [
{ name = "llama-index" },