Dockerfile

# Base Image
FROM python:3.10-slim

# Build argument for Hugging Face token
ARG HF_TOKEN

ENV DEBIAN_FRONTEND=noninteractive \
    PYTHONUNBUFFERED=1 \
    PYTHONDONTWRITEBYTECODE=1 \
    HF_TOKEN=${HF_TOKEN}

WORKDIR /code

# System Dependencies
RUN apt-get update && apt-get install -y --no-install-recommends \
    build-essential \
    git \
    curl \
    libopenblas-dev \
    libomp-dev \
    ffmpeg \
    && rm -rf /var/lib/apt/lists/*

# Copy requirements and install Python dependencies
COPY requirements.txt .
RUN pip install --no-cache-dir -r requirements.txt

# Hugging Face + model tools
RUN pip install --no-cache-dir huggingface-hub sentencepiece accelerate fasttext

# Hugging Face cache environment
ENV HF_HOME=/models/huggingface \
    TRANSFORMERS_CACHE=/models/huggingface \
    HUGGINGFACE_HUB_CACHE=/models/huggingface \
    HF_HUB_CACHE=/models/huggingface

# Created cache dir and set permissions
RUN mkdir -p /models/huggingface && chmod -R 777 /models/huggingface

RUN python -c "from huggingface_hub import snapshot_download; snapshot_download(repo_id='facebook/mms-tts-hau')" \
 && python -c "from huggingface_hub import snapshot_download; snapshot_download(repo_id='facebook/mms-tts-eng')" \
 && python -c "from huggingface_hub import snapshot_download; snapshot_download(repo_id='facebook/mms-tts-yor')" \
 && find /models/huggingface -name '*.lock' -delete

RUN python -c "from transformers import pipeline; pipeline('text-to-speech', model='facebook/mms-tts-hau')" \
 && python -c "from transformers import pipeline; pipeline('text-to-speech', model='facebook/mms-tts-eng')" \
 && python -c "from transformers import pipeline; pipeline('text-to-speech', model='facebook/mms-tts-yor')"

# Pre-load N-ATLaS model during build
RUN python -c "from huggingface_hub import snapshot_download; snapshot_download(repo_id='NCAIR1/N-ATLaS', token='$HF_TOKEN')" \
 && python -c "from transformers import AutoTokenizer, AutoModelForCausalLM; import torch; tokenizer = AutoTokenizer.from_pretrained('NCAIR1/N-ATLaS', token='$HF_TOKEN'); model = AutoModelForCausalLM.from_pretrained('NCAIR1/N-ATLaS', torch_dtype=torch.float16 if torch.cuda.is_available() else torch.float32, device_map='auto' if torch.cuda.is_available() else None, trust_remote_code=True, low_cpu_mem_usage=True, use_cache=True, token='$HF_TOKEN'); print('N-ATLaS model loaded successfully')"

# Pre-download ASR models (will be lazy-loaded at runtime)
RUN python -c "from huggingface_hub import snapshot_download; snapshot_download(repo_id='NCAIR1/Hausa-ASR', token='$HF_TOKEN')" \
 && python -c "from huggingface_hub import snapshot_download; snapshot_download(repo_id='NCAIR1/Yoruba-ASR', token='$HF_TOKEN')" \
 && python -c "from huggingface_hub import snapshot_download; snapshot_download(repo_id='NCAIR1/Igbo-ASR', token='$HF_TOKEN')" \
 && python -c "from huggingface_hub import snapshot_download; snapshot_download(repo_id='NCAIR1/NigerianAccentedEnglish', token='$HF_TOKEN')" \
 && find /models/huggingface -name '*.lock' -delete

# Copy project files
COPY . .

EXPOSE 7860

CMD ["uvicorn", "app:app", "--host", "0.0.0.0", "--port", "7860", "--workers", "1"]