Dockerfile

# Dockerfile
FROM python:3.11-slim

# INSTALA DEPENDÊNCIAS DO SISTEMA
RUN apt-get update && \
    apt-get install -y --no-install-recommends \
        curl \
        ca-certificates \
    && rm -rf /var/lib/apt/lists/*

# BAIXA E INSTALA OLLAMA
RUN curl -fsSL https://ollama.com/install.sh | sh

# CRIA DIRETÓRIO
WORKDIR /app

# COPIA ARQUIVOS
COPY requirements.txt .
COPY app.py .

# INSTALA PYTHON DEPENDÊNCIAS
RUN pip install --no-cache-dir -r requirements.txt

# PORTAS
EXPOSE 7860
EXPOSE 11434

# === PUXA MODELO CORRETO: qwen2.5:3b-instruct-q4_0 ===
RUN ollama serve & \
    OLLAMA_PID=$! && \
    sleep 25 && \
    ollama pull qwen2.5:3b-instruct-q4_0 && \
    kill $OLLAMA_PID || true

# HEALTHCHECK
HEALTHCHECK --interval=30s --timeout=10s --start-period=60s --retries=3 \
    CMD curl -f http://localhost:7860/health || exit 1

# VARIÁVEIS DE AMBIENTE
ENV OLLAMA_NUM_PARALLEL=3
ENV OLLAMA_MAX_QUEUE=10
ENV OLLAMA_KEEP_ALIVE=10m
ENV OLLAMA_MAX_LOADED_MODELS=1

# INICIA app.py
CMD ["python", "app.py"]