Spaces:
Running
Running
File size: 14,361 Bytes
a8cf49b 294982a a8cf49b 1cc6912 a8cf49b 7078139 d569c38 a8cf49b 7078139 a8cf49b 294982a ace7446 a8cf49b 294982a a8cf49b ace7446 294982a ace7446 294982a ace7446 294982a a8cf49b 294982a a8cf49b d569c38 a8cf49b d569c38 a8cf49b 8f63119 a8cf49b | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 | #!/usr/bin/env python3
"""
DocAssistant - LangGraph agent that routes between responding and editing
Single node agent that decides whether to respond to user questions or edit the document
"""
import json
import logging
import traceback
from typing import Dict, Any, List, Optional
from langgraph.graph import StateGraph, END
from langchain_core.messages import SystemMessage, HumanMessage, AIMessage, ToolMessage
from agent_states.doc_assistant_state import DocAssistantState
from prompts.doc_assistant import get_router_system_prompt, get_router_user_prompt
logger = logging.getLogger(__name__)
class DocAssistant:
"""
Router agent that decides whether to respond to user questions or edit the document.
Workflow:
- Single node with optional tool calling
- Document is in user prompt (no inspect needed)
- Agent decides: respond directly OR call edit_document tool
- No looping at router level (single pass)
"""
def __init__(self, llm, tools: List[Any] = None, tools_facade: List[Any] = None):
"""
Initialize the router agent.
Args:
llm: LLM for routing decisions (with tool calling capability)
tools: Real tool implementations for lookup
tools_facade: Facade tools for LLM (minimal parameters)
"""
self.tools = tools
self.tools_facade = tools_facade if tools_facade else tools
self.llm= llm
# Bind facade tools to LLM - optional tool calling
self.llm_with_tools = self.llm.bind_tools(self.tools_facade)
logger.info("π§ DocAssistant initialized")
logger.info(f"π€ Using {type(llm).__name__} for routing decisions")
logger.info(f"π οΈ Tools available: {[t.name for t in self.tools_facade]}")
self.workflow = self._build_workflow()
def _build_workflow(self) -> StateGraph:
"""Build the LangGraph workflow for the router agent."""
workflow = StateGraph(DocAssistantState)
workflow.add_node("agent", self._agent_node)
workflow.add_node("tools", self._tools_node)
workflow.set_entry_point("agent")
# Conditional edge after agent: go to tools if tool calls, else END
workflow.add_conditional_edges(
"agent",
self._should_call_tools,
{"tools": "tools", "end": END}
)
# Conditional edge after tools: END if edit_document was called, else continue
workflow.add_conditional_edges(
"tools",
self._after_tools,
{"end": END, "continue": "agent"}
)
return workflow.compile()
def _should_call_tools(self, state: DocAssistantState) -> str:
"""
Decide whether to call tools after agent node.
Returns:
"tools" if agent made tool calls, "end" otherwise
"""
intermediate_steps = state.get("intermediate_steps", [])
last_message = intermediate_steps[-1]
# Check if last message has tool calls
if hasattr(last_message, 'tool_calls') and last_message.tool_calls:
tool_names = [tc['name'] for tc in last_message.tool_calls]
logger.info(f"π§ Agent calling tools: {tool_names}")
return "tools"
return "end"
async def _agent_node(self, state: DocAssistantState) -> DocAssistantState:
"""Agent node: Generate response or tool call based on user request."""
intermediate_steps = state.get("intermediate_steps", [])
logger.info("π― Router agent processing request...")
# Build messages
messages = [
SystemMessage(content=get_router_system_prompt()),
HumanMessage(content=get_router_user_prompt(
doc_text=state["doc_text"],
instruction=state["user_instruction"],
doc_summaries=state.get("doc_summaries", []),
conversation_history=state.get("conversation_history", [])
))
]
logger.info(f"π Document size: {len(state['doc_text'])} bytes")
logger.info(f"π Instruction: {state['user_instruction'][:100]}{'...' if len(state['user_instruction']) > 100 else ''}")
# Call LLM
response = await self.llm_with_tools.ainvoke(messages)
intermediate_steps.append(response)
logger.info(f"π Response type: {type(response).__name__}")
if hasattr(response, 'tool_calls') and response.tool_calls:
logger.info(f"π§ Tool calls: {[tc['name'] for tc in response.tool_calls]}")
else:
logger.info(f"π¬ Direct response (no tool calls):{response.content}")
state['message']= response.content
state["intermediate_steps"] = intermediate_steps
return state
async def _tools_node(self, state: DocAssistantState) -> DocAssistantState:
"""Tools node: Execute tool calls (edit_document or retrieve_lawyer_document)."""
intermediate_steps = state.get("intermediate_steps", [])
last_message = intermediate_steps[-1]
if not (hasattr(last_message, 'tool_calls') and last_message.tool_calls):
return state
for tool_call in last_message.tool_calls:
tool_name = tool_call['name']
# Get the tool function directly from self.tools (add underscore prefix)
tool_func = next((t for t in self.tools if t.name == "_" + tool_name), None)
if tool_func:
try:
args = tool_call['args'].copy()
logger.info(f"Launching tool: {tool_name} with args {json.dumps(args, default=str)}")
if tool_name == "edit_document":
logger.info("π edit_document tool called - invoking doc_editor_agent")
args["doc_text"] = state["doc_text"]
args["user_instruction"] = state["user_instruction"]
args["doc_summaries"] = state.get("doc_summaries", [])
args["conversation_history"] = state.get("conversation_history", [])
args["max_iterations"] = 10
args["document_id"] = state.get("document_id")
args["user_id"] = state.get("user_id")
elif tool_name == "retrieve_lawyer_document":
logger.info(f"π retrieve_lawyer_document tool called: {args.get('file_path')}")
if "user_id" not in args and state.get("user_id"):
args["user_id"] = state["user_id"]
result = await tool_func.ainvoke(args)
if tool_name == "edit_document":
doc_text = result['doc_text']
tool_result = f"Document was edited with this summary: {result['final_summary']}"
state['modified_document'] = doc_text
state['message'] = tool_result
logger.info(f"β
edit_document completed - ending router workflow")
else:
tool_result = result
intermediate_steps.append(
ToolMessage(
content=tool_result,
tool_call_id=tool_call['id'],
name=tool_name
)
)
except Exception as e:
logger.error(f"β Error executing {tool_name}: {str(e)}")
intermediate_steps.append(
ToolMessage(
content=f"Error: {str(e)}",
tool_call_id=tool_call['id'],
name=tool_name
)
)
else:
logger.warning(f"β οΈ Tool function not found for {tool_name}")
state["intermediate_steps"] = intermediate_steps
return state
def _after_tools(self, state: DocAssistantState) -> str:
"""
Decide whether to continue after tools node.
Returns:
"end" if edit_document was called (stops workflow),
"continue" if retrieve_lawyer_document or query_knowledge_graph were called (allows more tool calls or response)
"""
intermediate_steps = state.get("intermediate_steps", [])
# Check if edit_document was called
for msg in reversed(intermediate_steps):
if isinstance(msg, ToolMessage) and msg.name == "edit_document":
logger.info("β
edit_document called - ending router workflow")
return "end"
# If edit_document wasn't called, continue (allows agent to make more tool calls or respond)
logger.info("π Continuing router workflow (edit_document not yet called)")
return "continue"
async def process_request(
self,
doc_text: str,
user_instruction: str,
doc_summaries: List[str] = [],
conversation_history: List[Dict[str, str]] = [],
document_id: Optional[str] = None,
user_id: Optional[str] = None
) -> Dict[str, Any]:
"""
Process the user's request and decide whether to respond or edit.
Args:
doc_text: The HTML document content
user_instruction: User's instruction or question
doc_summaries: Optional list of document summaries for context
conversation_history: Optional conversation history
document_id: Optional UUID of the document for live updates
user_id: Optional user ID for authentication
Returns:
Dict with:
- message: Response message to the user
- modified_document: Modified document (if editing was done) or None
- success: Boolean indicating success
"""
logger.info("=" * 80)
logger.info("π― DOC ASSISTANT STARTING")
logger.info("=" * 80)
logger.info(f"π Document size: {len(doc_text)} bytes")
logger.info(f"π Instruction: {user_instruction[:100]}{'...' if len(user_instruction) > 100 else ''}")
logger.info(f"π Document summaries: {len(doc_summaries)}")
logger.info(f"π¬ Conversation history: {len(conversation_history)} messages")
try:
# Initialize state
initial_state = {
"doc_text": doc_text,
"doc_summaries": doc_summaries,
"conversation_history": conversation_history,
"user_instruction": user_instruction,
"intermediate_steps": [],
"document_id": document_id,
"user_id": user_id,
"modified_document": None
}
# Run workflow
logger.info("π Invoking router workflow...")
final_state = await self.workflow.ainvoke(initial_state)
modified_doc=final_state.get("modified_document")
message=final_state.get("message")
return {
"message": message,
"modified_document": modified_doc,
"success": True
}
except Exception as e:
logger.error("=" * 80)
logger.error("β DOC ASSISTANT FAILED")
logger.error("=" * 80)
logger.error(f"π Location: subagents/doc_assistant.py:{traceback.extract_tb(e.__traceback__)[-1].lineno} (process_request)")
logger.error("")
logger.error("π Input Parameters:")
logger.error(f" - User Instruction: {user_instruction[:100] if len(user_instruction) > 100 else user_instruction}")
logger.error(f" - Document Size: {len(doc_text):,} bytes")
if document_id:
logger.error(f" - Document ID: {document_id}")
if user_id:
logger.error(f" - User ID: {user_id}")
logger.error(f" - Document Summaries: {len(doc_summaries)}")
logger.error(f" - Conversation History: {len(conversation_history)} messages")
logger.error("")
logger.error("π€ LLM Configuration:")
logger.error(f" - LLM: {type(self.llm).__name__}")
logger.error(f" - Tools Available: {len(self.tools)}")
logger.error(f" - Tool Names: {', '.join([t.name for t in self.tools])}")
logger.error(f" - Facade Tools: {len(self.tools_facade)}")
logger.error("")
logger.error("π Workflow State:")
if final_state and "intermediate_steps" in final_state:
intermediate_steps = final_state["intermediate_steps"]
logger.error(f" - Intermediate Steps: {len(intermediate_steps)} messages")
if intermediate_steps:
last_msg = intermediate_steps[-1]
logger.error(f" - Last Message Type: {type(last_msg).__name__}")
if hasattr(last_msg, 'tool_calls') and last_msg.tool_calls:
logger.error(f" - Last Tool Calls: {', '.join([tc['name'] for tc in last_msg.tool_calls])}")
logger.error("")
logger.error("π Exception Details:")
logger.error(f" Type: {type(e).__name__}")
logger.error(f" Message: {str(e)}")
logger.error("")
logger.error("π Full Traceback:")
logger.error(traceback.format_exc())
logger.error("")
logger.error("πΎ Document Preview (first 200 chars):")
logger.error(f" {doc_text[:200]}")
logger.error("=" * 80)
return {
"message": f"Error processing request: {str(e)}",
"modified_document": None,
"success": False
}
|