Fix all cache directory permission issues for HuggingFace Spaces

app.py

@@ -6,10 +6,18 @@ import sys
 os.environ["STREAMLIT_CONFIG_DIR"] = "/tmp/.streamlit"
 os.environ["STREAMLIT_DATA_DIR"] = "/tmp/.streamlit"
 os.environ["NLTK_DATA"] = "/tmp/nltk_data"
+os.environ["HF_HOME"] = "/tmp/huggingface"
+os.environ["TRANSFORMERS_CACHE"] = "/tmp/huggingface/transformers"
+os.environ["SENTENCE_TRANSFORMERS_HOME"] = "/tmp/huggingface/sentence_transformers"
+os.environ["HF_HUB_CACHE"] = "/tmp/huggingface/hub"
 
 # Create directories
 os.makedirs("/tmp/.streamlit", exist_ok=True)
 os.makedirs("/tmp/nltk_data", exist_ok=True)
+os.makedirs("/tmp/huggingface", exist_ok=True)
+os.makedirs("/tmp/huggingface/transformers", exist_ok=True)
+os.makedirs("/tmp/huggingface/sentence_transformers", exist_ok=True)
+os.makedirs("/tmp/huggingface/hub", exist_ok=True)
 
 # Copy config if it exists
 try:

src/search_final.py

@@ -1,6 +1,34 @@
 
 import os
 from dotenv import load_dotenv
+
+# Set cache directories for HuggingFace Spaces compatibility BEFORE any imports
+if not os.getenv("HF_HOME"):
+    os.environ["HF_HOME"] = "/tmp/huggingface"
+if not os.getenv("TRANSFORMERS_CACHE"):
+    os.environ["TRANSFORMERS_CACHE"] = "/tmp/huggingface/transformers"
+if not os.getenv("SENTENCE_TRANSFORMERS_HOME"):
+    os.environ["SENTENCE_TRANSFORMERS_HOME"] = "/tmp/huggingface/sentence_transformers"
+if not os.getenv("HF_HUB_CACHE"):
+    os.environ["HF_HUB_CACHE"] = "/tmp/huggingface/hub"
+if not os.getenv("NLTK_DATA"):
+    os.environ["NLTK_DATA"] = "/tmp/nltk_data"
+if not os.getenv("TORCH_HOME"):
+    os.environ["TORCH_HOME"] = "/tmp/torch"
+if not os.getenv("HUGGINGFACE_HUB_CACHE"):
+    os.environ["HUGGINGFACE_HUB_CACHE"] = "/tmp/huggingface/hub"
+
+# Create cache directories
+try:
+    os.makedirs("/tmp/huggingface", exist_ok=True)
+    os.makedirs("/tmp/huggingface/transformers", exist_ok=True)
+    os.makedirs("/tmp/huggingface/sentence_transformers", exist_ok=True)
+    os.makedirs("/tmp/huggingface/hub", exist_ok=True)
+    os.makedirs("/tmp/nltk_data", exist_ok=True)
+    os.makedirs("/tmp/torch", exist_ok=True)
+except Exception as e:
+    print(f"Warning: Could not create cache directories: {e}")
+
 import re
 import pickle
 import faiss

src/streamlit_app.py

@@ -12,10 +12,22 @@ if not os.getenv("STREAMLIT_DATA_DIR"):
     os.environ["STREAMLIT_DATA_DIR"] = "/tmp/.streamlit"
 if not os.getenv("NLTK_DATA"):
     os.environ["NLTK_DATA"] = "/tmp/nltk_data"
+if not os.getenv("HF_HOME"):
+    os.environ["HF_HOME"] = "/tmp/huggingface"
+if not os.getenv("TRANSFORMERS_CACHE"):
+    os.environ["TRANSFORMERS_CACHE"] = "/tmp/huggingface/transformers"
+if not os.getenv("SENTENCE_TRANSFORMERS_HOME"):
+    os.environ["SENTENCE_TRANSFORMERS_HOME"] = "/tmp/huggingface/sentence_transformers"
+if not os.getenv("HF_HUB_CACHE"):
+    os.environ["HF_HUB_CACHE"] = "/tmp/huggingface/hub"
 
 # Create directories
 os.makedirs("/tmp/.streamlit", exist_ok=True)
 os.makedirs("/tmp/nltk_data", exist_ok=True)
+os.makedirs("/tmp/huggingface", exist_ok=True)
+os.makedirs("/tmp/huggingface/transformers", exist_ok=True)
+os.makedirs("/tmp/huggingface/sentence_transformers", exist_ok=True)
+os.makedirs("/tmp/huggingface/hub", exist_ok=True)
 
 try:
     from transformers import AutoModelForCausalLM, AutoTokenizer