chore: optimize docker build

Author: bekmuradov

Dockerfile

@@ -1,17 +1,17 @@
+# syntax=docker/dockerfile:1.7-labs
 # =======================
 # Stage 1 - Builder
 # =======================
 FROM python:3.12-slim AS builder
 
 ARG ENABLE_CUDA=false
-
-WORKDIR /app
-
 ENV PYTHONDONTWRITEBYTECODE=1 \
     PYTHONUNBUFFERED=1 \
     PIP_NO_CACHE_DIR=1
 
-# System dependencies for building wheels
+WORKDIR /app
+
+# Install system dependencies required for building wheels
 RUN apt-get update && apt-get install -y --no-install-recommends \
     build-essential \
     curl \
@@ -21,79 +21,69 @@ RUN apt-get update && apt-get install -y --no-install-recommends \
 # Install Poetry
 RUN pip install --no-cache-dir poetry==2.2.1
 
-# Copy dependency files
+# Copy only the dependency descriptors first (cache friendly)
 COPY pyproject.toml poetry.lock* /app/
 
 # Configure Poetry
 RUN poetry config virtualenvs.create false
 
-# Install dependencies based on CUDA flag
-RUN --mount=type=cache,id=poetry-cache,target=/root/.cache/pypoetry \
-    --mount=type=cache,id=pip-cache,target=/root/.cache/pip \
+# Install packages (use BuildKit cache mounts so subsequent builds are fast)
+# This layer will be reused as long as pyproject/poetry.lock don't change.
+RUN --mount=type=cache,target=/root/.cache/pypoetry \
+    --mount=type=cache,target=/root/.cache/pip \
     if [ "$ENABLE_CUDA" = "true" ]; then \
         echo "Installing with CUDA support..."; \
         poetry install --without dev --no-interaction --no-ansi --no-root; \
     else \
         echo "Installing CPU-only packages..."; \
-        poetry install --without dev --no-interaction --no-ansi --no-root && \
-        pip uninstall -y torch torchvision && \
+        # install declared dependencies
+        poetry install --without dev --no-interaction --no-ansi --no-root; \
+        # replace torch with CPU wheel (if on amd64), keep torch separate to avoid invalidation
+        pip uninstall -y torch torchvision || true; \
         if [ "$TARGETARCH" = "amd64" ]; then \
             pip install --no-cache-dir \
-            torch==2.7.1+cpu \
-            torchvision==0.22.1+cpu \
-            --index-url https://download.pytorch.org/whl/cpu; \
+                torch==2.7.1+cpu \
+                torchvision==0.22.1+cpu \
+                --index-url https://download.pytorch.org/whl/cpu; \
         else \
             pip install --no-cache-dir torch torchvision; \
         fi \
     fi
 
-# Clean up unnecessary files
-RUN find /usr/local/lib/python3.12 -type d -name "tests" -exec rm -rf {} + 2>/dev/null || true && \
-    find /usr/local/lib/python3.12 -type d -name "test" -exec rm -rf {} + 2>/dev/null || true && \
-    find /usr/local/lib/python3.12 -type f -name "*.pyc" -delete && \
-    find /usr/local/lib/python3.12 -type d -name "__pycache__" -exec rm -rf {} + 2>/dev/null || true
+# Remove tests/caches from site-packages to reduce image size
+RUN find /usr/local/lib/python3.12 -type d \( -name "tests" -o -name "test" -o -name "__pycache__" \) -exec rm -rf {} + 2>/dev/null || true \
+    && find /usr/local/lib/python3.12 -type f -name "*.pyc" -delete
 
 # =======================
 # Stage 2 - Runtime
 # =======================
 FROM python:3.12-slim AS runtime
 
 ARG ENABLE_CUDA=false
-
-WORKDIR /app
-
 ENV PYTHONDONTWRITEBYTECODE=1 \
     PYTHONUNBUFFERED=1 \
     PYTHONPATH=/app \
     PORT=8080 \
     ENABLE_CUDA=${ENABLE_CUDA}
 
-# Install only runtime dependencies
+WORKDIR /app
+
+# Install small runtime system libs
 RUN apt-get update && apt-get install -y --no-install-recommends \
     curl \
     libgomp1 \
     && rm -rf /var/lib/apt/lists/*
 
-# Copy only site-packages (not entire lib)
+# Copy installed site-packages and runtime binaries from builder
 COPY --from=builder /usr/local/lib/python3.12/site-packages /usr/local/lib/python3.12/site-packages
 COPY --from=builder /usr/local/bin/uvicorn /usr/local/bin/uvicorn
 COPY --from=builder /usr/local/bin/gunicorn /usr/local/bin/gunicorn
 
-# Add non-root user
-# RUN groupadd -r appuser && useradd -r -g appuser appuser
-
-# Copy local spaCy model
-# COPY --chown=appuser:appuser spacy_models /app/spacy_models
-
-# Copy application code
-# COPY --chown=appuser:appuser app /app/app
+# Copy application source last so code changes do not bust earlier caches
+COPY . /app
 
 # Create runtime directories
 RUN mkdir -p /app/data/uploads /app/data/temp /app/logs
-# RUN mkdir -p /app/data/uploads /app/data/temp /app/logs \
-#     && chown -R appuser:appuser /app
-
-# USER appuser
 
 EXPOSE 8080
 

docker-compose.yml

@@ -1,58 +1,51 @@
+version: "3.9"
+
 services:
   braindrive-document-ai:
     build:
       context: .
       dockerfile: Dockerfile
       args:
         ENABLE_CUDA: ${ENABLE_CUDA:-false}
-      # cache_from:
-      #   - type=local,src=/tmp/docker-cache
-      # cache_to:
-      #   - type=local,dest=/tmp/docker-cache,mode=max
+      # BuildKit cache hints (requires BuildKit to be enabled)
+      cache_from:
+        - type=local,src=.buildx-cache
+      cache_to:
+        - type=local,dest=.buildx-cache,mode=max
     ports:
       - "8080:8080"
 
     environment:
-      # Authentication
       - DISABLE_AUTH=${DISABLE_AUTH:-true}
       - AUTH_METHOD=${AUTH_METHOD:-api_key}
       - AUTH_API_KEY=${AUTH_API_KEY:-}
       - JWT_SECRET=${JWT_SECRET:-}
       - JWT_ALGORITHM=${JWT_ALGORITHM:-HS256}
       - JWT_EXPIRE_MINUTES=${JWT_EXPIRE_MINUTES:-60}
-      
-      # Document processing
       - SPACY_MODEL=${SPACY_MODEL:-/app/spacy_models/en_core_web_sm/en_core_web_sm-3.8.0}
       - DEFAULT_CHUNKING_STRATEGY=${DEFAULT_CHUNKING_STRATEGY:-hierarchical}
       - DEFAULT_CHUNK_SIZE=${DEFAULT_CHUNK_SIZE:-1000}
       - DEFAULT_CHUNK_OVERLAP=${DEFAULT_CHUNK_OVERLAP:-200}
       - MIN_CHUNK_SIZE=${MIN_CHUNK_SIZE:-100}
       - MAX_CHUNK_SIZE=${MAX_CHUNK_SIZE:-2000}
-
-      # CUDA toggle (default CPU, enable with ENABLE_CUDA=true)
       - ENABLE_CUDA=${ENABLE_CUDA:-false}
-
-      # Logging
       - LOG_FORMAT=${LOG_FORMAT:-console}
       - LOG_FILE=${LOG_FILE:-/app/logs/app.log}
-    
+
     volumes:
       - ../../persistent/document_processing/data:/app/data
       - ../../persistent/document_processing/logs:/app/logs
       - .:/app  # Mount source code for development
-    
+
     healthcheck:
       test: ["CMD", "curl", "-f", "http://localhost:8080/health"]
       interval: 30s
       timeout: 10s
       retries: 3
       start_period: 40s
-    
+
     restart: unless-stopped
-    
-    # Development overrides
-    # command: ["uvicorn", "app.main:app", "--host", "0.0.0.0", "--port", "8080", "--reload"]
-    
+
     networks:
       - braindrive-network