refactor

multi_agent_chatbot.py

@@ -8,24 +8,26 @@ This system implements a 3-agent architecture:
 
 Each agent has specialized prompts and responsibilities.
 """
+import re
 import json
 import time
 import logging
+import traceback
 from pathlib import Path
 from datetime import datetime
 from dataclasses import dataclass
 from typing import Dict, List, Any, Optional, TypedDict
 
-
 from langchain_core.tools import tool
 from langgraph.graph import StateGraph, END
-from langchain_core.messages import HumanMessage, AIMessage, SystemMessage
 from langchain_core.prompts import ChatPromptTemplate
+from langchain_core.messages import HumanMessage, AIMessage, SystemMessage
 
 
 from src.pipeline import PipelineManager
-from src.config.loader import load_config
 from src.llm.adapters import get_llm_client
+from src.config.paths import PROJECT_DIR, CONVERSATIONS_DIR
+from src.config.loader import load_config, get_embedding_model_for_collection
 
 
 logging.basicConfig(level=logging.INFO, format='%(asctime)s - %(levelname)s - %(message)s')
@@ -46,6 +48,7 @@ class QueryContext:
     needs_follow_up: bool = False
     follow_up_question: Optional[str] = None
 
+
 class MultiAgentState(TypedDict):
     """State for the multi-agent conversation flow"""
     conversation_id: str
@@ -61,6 +64,7 @@ class MultiAgentState(TypedDict):
     session_start_time: float
     last_ai_message_time: float
 
+
 class MultiAgentRAGChatbot:
     """Multi-agent RAG chatbot with specialized agents"""
     
@@ -112,7 +116,6 @@ class MultiAgentRAGChatbot:
             logger.info("✅ Pipeline manager initialized and models loaded")
         except Exception as e:
             logger.error(f"❌ Failed to initialize pipeline manager: {e}")
-            import traceback
             traceback.print_exc()
             raise RuntimeError(f"Pipeline manager initialization failed: {e}")
         
@@ -129,7 +132,6 @@ class MultiAgentRAGChatbot:
             raise  # Re-raise RuntimeError as-is
         except Exception as e:
             logger.error(f"❌ Error during vector store connection: {e}")
-            import traceback
             traceback.print_exc()
             raise RuntimeError(f"Vector store connection failed: {e}")
         
@@ -139,8 +141,8 @@ class MultiAgentRAGChatbot:
         # Build the multi-agent graph
         self.graph = self._build_graph()
         
-        # Conversations directory - use absolute path in /app to ensure writability
-        self.conversations_dir = Path("/app/conversations")
+        # Conversations directory - use PROJECT_DIR for local vs deployed compatibility
+        self.conversations_dir = CONVERSATIONS_DIR
         try:
             # Use 777 permissions for maximum compatibility (HF Spaces runs as different user)
             self.conversations_dir.mkdir(parents=True, mode=0o777, exist_ok=True)
@@ -158,9 +160,9 @@ class MultiAgentRAGChatbot:
     
     def _load_dynamic_data(self):
         """Load dynamic data from filter_options.json and add_district_metadata.py"""
-        # Load filter options
+        # Load filter options - use PROJECT_DIR relative path
         try:
-            fo = Path("src/config/filter_options.json")
+            fo = PROJECT_DIR / "src" / "config" / "filter_options.json"
             if fo.exists():
                 with open(fo) as f:
                     data = json.load(f)
@@ -178,7 +180,7 @@ class MultiAgentRAGChatbot:
             self.source_whitelist = ['Consolidated', 'Local Government', 'Ministry, Department and Agency']
             self.district_whitelist = ['Kampala', 'Gulu', 'Kalangala']
         
-        # Enrich district list from add_district_metadata.py
+        # Enrich district list from add_district_metadata.py (if available)
         try:
             from add_district_metadata import DistrictMetadataProcessor
             proc = DistrictMetadataProcessor()
@@ -590,7 +592,6 @@ Analyze this query using ONLY the exact values provided above:""")
             # Clean and parse JSON with better error handling
             try:
                 # Remove comments (// and /* */) from JSON
-                import re
                 # Remove single-line comments
                 content = re.sub(r'//.*?$', '', content, flags=re.MULTILINE)
                 # Remove multi-line comments
@@ -603,7 +604,6 @@ Analyze this query using ONLY the exact values provided above:""")
                 logger.error(f"❌ Raw content: {content[:200]}...")
                 
                 # Try to extract JSON from text if embedded
-                import re
                 json_match = re.search(r'\{.*\}', content, re.DOTALL)
                 if json_match:
                     try:
@@ -1178,7 +1178,6 @@ Generate a conversational response based on your knowledge:""")
                 
         except Exception as e:
             logger.error(f"Could not save conversation: {e}")
-            import traceback
             logger.error(f"Traceback: {traceback.format_exc()}")