ok

Dockerfile

@@ -1,31 +1,45 @@
 # Use the official Python 3.11 slim image
 FROM python:3.11-slim
 
-# Install curl for the keep-alive script (and clean up after)
-RUN apt-get update && apt-get install -y curl && rm -rf /var/lib/apt/lists/*
-
-# Set the working directory inside the container
+# Set environment variables
+ENV PYTHONDONTWRITEBYTECODE=1 \
+    PYTHONUNBUFFERED=1 \
+    PYTHONIOENCODING=UTF-8 \
+    # Point HF cache to a writable directory inside /app
+    HF_HOME=/app/cache \
+    TRANSFORMERS_CACHE=/app/cache
+
+# Install system dependencies (curl) and clean up in the same layer to reduce image size
+RUN apt-get update && apt-get install -y --no-install-recommends curl \
+    && rm -rf /var/lib/apt/lists/* \
+    && useradd -m -u 1000 user
+
+# Set working directory
 WORKDIR /app
 
-# Environment variables for optimization and logging
-ENV PYTHONUNBUFFERED=1
-ENV PYTHONIOENCODING=UTF-8
-ENV HF_HOME=/tmp/cache
+# --- LAYER 1: Dependencies (Cached unless requirements.txt changes) ---
+COPY --chown=user:user requirements.txt .
+RUN pip install --no-cache-dir -r requirements.txt
 
-# Copy the requirements file first
-COPY requirements.txt .
+# --- LAYER 2: Model Files (Cached unless model weights change) ---
+# We explicitly copy the models folder before the rest of the code.
+# This prevents the heavy model layer from breaking cache when you just edit main.py.
+# Note: Ensure you have a 'models' folder locally. If not, you can remove this line.
+COPY --chown=user:user models ./models
 
-# Install dependencies
-RUN pip install --no-cache-dir -r requirements.txt
+# --- LAYER 3: Application Code (Cached until you edit code) ---
+COPY --chown=user:user . .
 
-# Copy the rest of the application code
-COPY . .
+# Ensure the cache directory exists and the user owns it
+RUN mkdir -p $HF_HOME && chown -R user:user /app/cache
 
-# Create cache directory
-RUN mkdir -p ${HF_HOME} && chmod 777 ${HF_HOME}
+# Switch to the non-root user
+USER user
 
-# Expose port 7860 (required by Hugging Face Spaces)
+# Expose port 7860
 EXPOSE 7860
 
-# Keep-alive script + start Uvicorn with optimized workers
+# Start script
+# 1. Background loop pings the HF Space URL to keep it alive
+# 2. Uvicorn runs the app
 CMD bash -c "while true; do curl -s https://sasasas635-database-chat.hf.space/ping >/dev/null && sleep 300; done & uvicorn main:app --host 0.0.0.0 --port 7860 --workers 4 --loop asyncio"