Dockerfile

# Restore full pre-retirement app + patch frontend + backend models + auto-continue V17 + auto-approve + document uploads
FROM node:20-alpine AS frontend-builder
RUN apk add --no-cache git python3
RUN git clone --depth 1 https://huggingface.co/spaces/smolagents/ml-intern /source

# Restore the pre-retirement full app (upstream retired it: App.tsx -> RetirementPage,
# backend/start.sh -> retired_main:app). This puts back the full UI + real backend
# BEFORE our model patches run, so they take effect again.
COPY patch_restore_full_app.py /tmp/patch_restore_full_app.py
RUN python3 /tmp/patch_restore_full_app.py

# Patch frontend: add OpenRouter models
COPY patch_frontend.py /tmp/patch_frontend.py
RUN python3 /tmp/patch_frontend.py
# Patch frontend: add stealth/space-bunny-alpha
COPY patch_space_bunny.py /tmp/patch_space_bunny.py
RUN python3 /tmp/patch_space_bunny.py

# Patch frontend: multi-format file uploads (PDF/images/Excel/Word/PPT/ODF/text)
COPY patch_frontend_uploads.py /tmp/patch_frontend_uploads.py
RUN python3 /tmp/patch_frontend_uploads.py
# Patch frontend: multi-file uploads (many images/files at once)
COPY patch_frontend_multi.py /tmp/patch_frontend_multi.py
RUN python3 /tmp/patch_frontend_multi.py
# Patch frontend: add 3 new OpenRouter models
COPY patch_frontend_new_models.py /tmp/patch_frontend_new_models.py
RUN python3 /tmp/patch_frontend_new_models.py

# Patch auto-continue V17
COPY patch_auto_continue_v17.py /tmp/patch_auto_continue_v17.py
RUN python3 /tmp/patch_auto_continue_v17.py

WORKDIR /source/frontend
RUN npm config set fetch-timeout 120000 && \
    npm config set fetch-retries 3 && \
    npm install && \
    npm run build

# Stage 2: Production
FROM python:3.12-slim
COPY --from=ghcr.io/astral-sh/uv:latest /uv /uvx /bin/
RUN useradd -m -u 1000 user
WORKDIR /app
# System deps: git/curl/ca-certificates plus document parsing tools.
#   tesseract-ocr        -> OCR for image uploads (with eng + vie language packs)
#   libreoffice-writer + libreoffice-calc + libreoffice-impress
#                        -> legacy/OpenDocument formats (.xls, .doc, .odt, .ods, .odp, .rtf)
RUN apt-get update && apt-get install -y --no-install-recommends \
    git curl ca-certificates \
    tesseract-ocr tesseract-ocr-eng tesseract-ocr-vie \
    libreoffice-writer libreoffice-calc libreoffice-impress \
    && rm -rf /var/lib/apt/lists/*

RUN git clone --depth 1 https://huggingface.co/spaces/smolagents/ml-intern /tmp/source && \
    cp /tmp/source/pyproject.toml /tmp/source/uv.lock ./ && \
    cp -r /tmp/source/agent ./agent && \
    cp -r /tmp/source/backend ./backend && \
    cp -r /tmp/source/configs ./configs && \
    cp -r /tmp/source/scripts ./scripts && \
    rm -rf /tmp/source

# Restore pre-retirement backend entrypoint (start.sh -> uvicorn main:app instead of retired_main:app)
COPY patch_restore_full_app.py /tmp/patch_restore_full_app.py
RUN python3 /tmp/patch_restore_full_app.py

RUN uv sync --no-dev --frozen

# Document parsing Python deps (PDF, images/OCR, Excel, Word, PPT, OpenDocument (ODT/ODS/ODP, RTF)
RUN uv pip install --python /app/.venv/bin/python \
    pypdf \
    python-docx \
    openpyxl \
    python-pptx \
    pytesseract \
    Pillow

COPY --from=frontend-builder /source/frontend/dist ./static/
COPY configs/frontend_agent_config.json ./configs/frontend_agent_config.json
# Document text extractor module used by the backend upload route
COPY backend/document_parser.py /app/backend/document_parser.py

# Patch backend: accept new model IDs
COPY patch_models.py /tmp/patch_models.py
RUN python /tmp/patch_models.py
# Patch backend: add 3 new OpenRouter models
COPY patch_new_models.py /tmp/patch_new_models.py
RUN python /tmp/patch_new_models.py
# Patch backend: fix OpenRouter provider resolution (custom_llm_provider)
COPY patch_openrouter_fix.py /tmp/patch_openrouter_fix.py
RUN python /tmp/patch_openrouter_fix.py

# Patch backend: auto-continue detection + auto-approve
COPY patch_agent_backend.py /tmp/patch_agent_backend.py
RUN python /tmp/patch_agent_backend.py

# Patch backend: multi-format document uploads + text extraction
COPY patch_document_upload.py /tmp/patch_document_upload.py
RUN python /tmp/patch_document_upload.py
# Patch backend: multi-file uploads (many files per request)
COPY patch_multi_upload.py /tmp/patch_multi_upload.py
RUN python /tmp/patch_multi_upload.py

RUN mkdir -p /app/session_logs && chown -R user:user /app

USER user
ENV HOME=/home/user \
    PYTHONUNBUFFERED=1 \
    PYTHONPATH=/app \
    PATH="/app/.venv/bin:$PATH"

EXPOSE 7860
WORKDIR /app/backend
CMD ["sh", "start.sh"]