Add antibiotic guidelines Excel files and PDF documentation

src/config.py

@@ -7,7 +7,10 @@ from typing import Literal, Optional
 from dotenv import load_dotenv
 from pydantic import BaseModel, Field
 
-load_dotenv()
+# Explicitly load .env from the project root so it works regardless of CWD
+# (e.g. when imported from a Kaggle notebook whose CWD is /kaggle/working/)
+_PROJECT_ROOT = Path(__file__).resolve().parents[1]
+load_dotenv(_PROJECT_ROOT / ".env")
 
 
 class Settings(BaseModel):

src/db/vector_store.py

@@ -9,17 +9,18 @@ import hashlib
 from langchain_text_splitters import RecursiveCharacterTextSplitter
 from pypdf import PdfReader
 
+from ..config import get_settings
+
 # Project paths
 PROJECT_ROOT = Path(__file__).parent.parent.parent
-DATA_DIR = PROJECT_ROOT / "data"
 DOCS_DIR = PROJECT_ROOT / "docs"
-CHROMA_DIR = DATA_DIR / "chroma"
 
 
 def get_chroma_client() -> chromadb.PersistentClient:
     """Get ChromaDB persistent client."""
-    CHROMA_DIR.mkdir(parents=True, exist_ok=True)
-    return chromadb.PersistentClient(path=str(CHROMA_DIR))
+    chroma_dir = get_settings().chroma_db_dir
+    chroma_dir.mkdir(parents=True, exist_ok=True)
+    return chromadb.PersistentClient(path=str(chroma_dir))
 
 
 def get_embedding_function():

src/rag.py

@@ -105,18 +105,31 @@ def search_drug_safety(
     n_results: int = 5,
     drug_name: Optional[str] = None,
 ) -> List[Dict[str, Any]]:
-    """Search the drug safety collection (interactions, warnings, contraindications)."""
-    collection = get_collection("drug_safety")
-    if collection is None:
+    """Search drug interactions from SQLite (drug_interactions table)."""
+    if not drug_name:
         return []
-    enhanced_query = f"{drug_name} {query}" if drug_name else query
     try:
-        results = collection.query(
-            query_texts=[enhanced_query],
-            n_results=n_results,
-            include=["documents", "metadatas", "distances"],
+        from .db.database import execute_query
+
+        rows = execute_query(
+            """SELECT drug_1, drug_2, interaction_description, severity
+               FROM drug_interactions
+               WHERE LOWER(drug_1) LIKE ? OR LOWER(drug_2) LIKE ?
+               LIMIT ?""",
+            (f"%{drug_name.lower()}%", f"%{drug_name.lower()}%", n_results),
         )
-        return _format_results(results)
+        return [
+            {
+                "content": (
+                    f"{r['drug_1']} + {r['drug_2']}: {r['interaction_description']}"
+                ),
+                "metadata": {"severity": r.get("severity", "unknown")},
+                "distance": None,
+                "source": "drug_interactions (SQLite)",
+                "relevance_score": 1.0,
+            }
+            for r in rows
+        ]
     except Exception as e:
         logger.error(f"Error querying drug safety: {e}")
         return []