Dockerfile

FROM python:3.10-slim

# ── Non-root user ─────────────────────────────────────────────────────────────
RUN useradd -m -u 1000 user
USER user
ENV HOME=/home/user \
    PATH=/home/user/.local/bin:$PATH \
    RASA_TELEMETRY_ENABLED=false \
    PYTHONUNBUFFERED=1 \
    # Keep HuggingFace / torch caches writable by uid 1000
    HF_HOME=/home/user/.cache/huggingface \
    TRANSFORMERS_CACHE=/home/user/.cache/transformers

WORKDIR $HOME/app

# ── Python deps ───────────────────────────────────────────────────────────────
# Install order matters:
#   1. numpy pinned first — Rasa 3.6.x breaks on numpy ≥ 1.24
#   2. aiogram / jax pinned for compatibility
#   3. rasa itself (pulls in most heavy deps)
#   4. project-specific requirements last
COPY --chown=user requirements.txt .

RUN pip install --no-cache-dir --upgrade pip && \
    pip install --no-cache-dir \
        numpy==1.23.5 \
        aiogram==2.25.1 \
        jax==0.4.17 \
        jaxlib==0.4.17 && \
    pip install --no-cache-dir rasa==3.6.20 && \
    pip install --no-cache-dir -r requirements.txt

# ── Application code ──────────────────────────────────────────────────────────
COPY --chown=user . .

ENV PYTHONPATH=$HOME/app:$PYTHONPATH

# Ensure Components is a proper Python package
RUN touch Components/__init__.py

# ── Pre-download translation model weights ────────────────────────────────────
# Helsinki-NLP/opus-mt-bn-en is used by Components/translate_to_english.py.
# Baking weights into the image avoids a cold-start download at runtime.
RUN python -c "\
from transformers import MarianMTModel, MarianTokenizer; \
MarianTokenizer.from_pretrained('Helsinki-NLP/opus-mt-bn-en'); \
MarianMTModel.from_pretrained('Helsinki-NLP/opus-mt-bn-en'); \
print('[BUILD] opus-mt-bn-en downloaded.')"

# ── Sanity-check the custom component ────────────────────────────────────────
RUN python -c "\
from Components.multilingual_featurizer import MultilingualFeaturizer; \
print('[BUILD] MultilingualFeaturizer import OK.')"

# ── Train the Rasa model ──────────────────────────────────────────────────────
# --fixed-model-name keeps the artifact name predictable so the CMD can find it.
RUN rasa train --fixed-model-name model

# ── Runtime ───────────────────────────────────────────────────────────────────
EXPOSE 7860

CMD ["rasa", "run", \
     "--enable-api", \
     "--cors", "*", \
     "--port", "7860", \
     "--debug"]