Update settings.py

settings.py

@@ -1,24 +1,27 @@
-from pathlib import Path
+# settings.py
+import os
 
-# Privacy / PHI
-PHI_SAFE_MODE: bool = True            # redact before indexing/logging
-PERSIST_CONTENT: bool = False         # never store message/upload contents to disk
+# Basic settings
+SNAPSHOT_PATH = os.getenv("SNAPSHOT_PATH", "snapshot.json")
+PERSIST_CONTENT = os.getenv("PERSIST_CONTENT", "false").lower() == "true"
 
-# Uploads
-MAX_UPLOAD_MB: int = 15
-ALLOWED_EXT = {".pdf", ".docx", ".txt", ".md", ".csv", ".png", ".jpg", ".jpeg", ".webp"}
-ALLOWED_MIME = {
-    "application/pdf",
-    "application/vnd.openxmlformats-officedocument.wordprocessingml.document",
-    "text/plain", "text/markdown", "text/csv",
-    "image/png", "image/jpeg", "image/webp",
+# Healthcare-specific settings
+HEALTHCARE_SETTINGS = {
+    "phi_detection_enabled": True,
+    "min_facility_count_for_aggregation": 10,
+    "default_occupancy_threshold": 85.0,  # Percentage
+    "bed_change_significance_threshold": 5.0,  # Percentage
+    "supported_file_types": [".csv", ".xlsx", ".pdf", ".docx", ".txt"],
+    "healthcare_keywords": [
+        "hospital", "patient", "bed", "care", "health", "medical", "clinical",
+        "facility", "nursing", "residential", "ambulatory", "healthcare", "occupancy"
+    ]
 }
 
-# Optional AV (clamd). If True and clamd is reachable, files will be scanned before parsing.
-ENABLE_AV_SCAN: bool = True
-CLAMD_UNIX_SOCKET: str | None = None        # e.g., "/var/run/clamav/clamd.ctl"
-CLAMD_NETWORK: tuple[str, int] | None = None  # e.g., ("localhost", 3310)
-
-# Paths
-SNAPSHOT_PATH = "snapshots/current.json"
-RAG_STORE_DIR = Path("rag_store")
+# Model settings
+MODEL_SETTINGS = {
+    "max_new_tokens": 2048,
+    "temperature": 0.3,
+    "top_p": 0.9,
+    "repetition_penalty": 1.15
+}