Update src/enhanced_rag_system.py

src/enhanced_rag_system.py

@@ -1,4 +1,12 @@
 import os
+import tempfile
+
+# Fix HuggingFace cache directory issue for HuggingFace Spaces
+# Set cache directories to writable temporary directories
+os.environ['TRANSFORMERS_CACHE'] = tempfile.mkdtemp()
+os.environ['HF_HOME'] = tempfile.mkdtemp()
+os.environ['SENTENCE_TRANSFORMERS_HOME'] = tempfile.mkdtemp()
+
 import pandas as pd
 import plotly.express as px
 import plotly.graph_objects as go
@@ -108,17 +116,21 @@ class EnhancedRAGSystem:
         try:
             print("🔧 Initializing ChromaDB...")
             
+            # Create a writable directory for ChromaDB
+            chroma_dir = tempfile.mkdtemp(prefix="chroma_")
+            print(f"📁 Using ChromaDB directory: {chroma_dir}")
+            
             # Try different ChromaDB configurations for HuggingFace compatibility
             try:
                 # First try: PersistentClient (newer API)
-                self.chroma_client = chromadb.PersistentClient(path="./chroma_db")
+                self.chroma_client = chromadb.PersistentClient(path=chroma_dir)
                 print("✅ Using ChromaDB PersistentClient")
             except Exception as e1:
                 try:
                     # Second try: Client with settings (older API)
                     self.chroma_client = chromadb.Client(Settings(
                         chroma_db_impl="duckdb+parquet",
-                        persist_directory="./chroma_db"
+                        persist_directory=chroma_dir
                     ))
                     print("✅ Using ChromaDB Client with Settings")
                 except Exception as e2:
@@ -140,12 +152,22 @@ class EnhancedRAGSystem:
             
             # Initialize embedding model with smaller model for HuggingFace
             try:
-                self.embedding_model = SentenceTransformer("all-MiniLM-L6-v2")
+                # Set cache directory for sentence transformers
+                cache_dir = tempfile.mkdtemp(prefix="sentence_transformers_")
+                
+                self.embedding_model = SentenceTransformer(
+                    "all-MiniLM-L6-v2", 
+                    cache_folder=cache_dir
+                )
                 print("✅ Loaded sentence transformer model: all-MiniLM-L6-v2")
             except Exception as e:
                 # Fallback to even smaller model
                 try:
-                    self.embedding_model = SentenceTransformer("paraphrase-MiniLM-L3-v2")
+                    cache_dir = tempfile.mkdtemp(prefix="sentence_transformers_fallback_")
+                    self.embedding_model = SentenceTransformer(
+                        "paraphrase-MiniLM-L3-v2",
+                        cache_folder=cache_dir
+                    )
                     print("✅ Loaded fallback sentence transformer model: paraphrase-MiniLM-L3-v2")
                 except Exception as e2:
                     print(f"❌ Failed to load embedding model: {e2}")