Dockerfile

# Include global ARGs at the dockerfile top
ARG ARCH="x86_64"
ARG LAMBDA_TASK_ROOT="/var/task"
ARG FASTAPI_STATIC="${LAMBDA_TASK_ROOT}/static"
ARG PYTHONPATH="${LAMBDA_TASK_ROOT}:${PYTHONPATH}:/usr/local/lib/python3/dist-packages"
ARG POETRY_NO_INTERACTION=1
ARG POETRY_VIRTUALENVS_IN_PROJECT=1
ARG POETRY_VIRTUALENVS_CREATE=1
ARG POETRY_CACHE_DIR=/tmp/poetry_cache
ARG DEPENDENCY_GROUP=fastapi


FROM pytorch/pytorch:2.2.2-cuda12.1-cudnn8-runtime as builder_global

LABEL authors="alessandro@trinca.tornidor.com"

ARG ARCH
ARG LAMBDA_TASK_ROOT
ARG PYTHONPATH
ARG POETRY_NO_INTERACTION
ARG POETRY_VIRTUALENVS_IN_PROJECT
ARG POETRY_VIRTUALENVS_CREATE
ARG POETRY_CACHE_DIR
ARG DEPENDENCY_GROUP

RUN echo "ARCH: $ARCH ..."

RUN echo "ARG POETRY_CACHE_DIR: ${POETRY_CACHE_DIR} ..."
RUN echo "ARG PYTHONPATH: $PYTHONPATH ..."
RUN test -n ${DEPENDENCY_GROUP:?}
RUN echo "python DEPENDENCY_GROUP: ${DEPENDENCY_GROUP} ..."
RUN echo "arg dep:"

# Set working directory to function root directory
WORKDIR ${LAMBDA_TASK_ROOT}
COPY requirements_poetry.txt pyproject.toml poetry.lock README.md ${LAMBDA_TASK_ROOT}/

RUN cat /etc/lsb-release
# avoid segment-geospatial exception caused by missing libGL.so.1 library
RUN echo "BUILDER: check libz.s* before start" && ls -l /usr/lib/${ARCH}-linux-gnu/libz.so*

RUN apt update && apt upgrade -y && apt install -y libgl1 curl python3-pip git-lfs && apt clean
COPY ./dockerfiles/apt_preferences_ubuntu /etc/apt/preferences
COPY ./dockerfiles/ubuntu.sources /etc/apt/sources.list.d/ubuntu.sources
#RUN echo "run update noble..."
#RUN apt update
#RUN apt update && apt install -t noble zlib1g -y
RUN git lfs install
RUN echo "BUILDER: check libz.s* after install from trixie" && ls -l /usr/lib/${ARCH}-linux-gnu/libz.so*

RUN ls -l /etc/apt/sources* /etc/apt/preferences*

# poetry installation path is NOT within ${LAMBDA_TASK_ROOT}: not needed for runtime docker image
RUN python -m pip install -r ${LAMBDA_TASK_ROOT}/requirements_poetry.txt

RUN which poetry && poetry --version && poetry config --list
RUN poetry config virtualenvs.path ${LAMBDA_TASK_ROOT}
RUN poetry config installer.max-workers 7
RUN echo "# poetry config --list #" && poetry config --list
RUN ls -ld ${LAMBDA_TASK_ROOT}/
#RUN . ${LAMBDA_TASK_ROOT}/.venv/bin/activate && ${LAMBDA_TASK_ROOT}/.venv/bin/python --version && ${LAMBDA_TASK_ROOT}/.venv/bin/python -m pip install pip wheel setuptools --upgrade
RUN poetry run python -m pip install pip wheel setuptools --upgrade
RUN poetry install --with ${DEPENDENCY_GROUP} --no-root

RUN git clone https://huggingface.co/aletrn/sam-quantized/ ${LAMBDA_TASK_ROOT}/sam-quantized
RUN git clone -n --depth=1 --filter=tree:0 https://huggingface.co/spaces/aletrn/lisa-on-cuda ${LAMBDA_TASK_ROOT}/lisa-on-cuda && \
    cd ${LAMBDA_TASK_ROOT}/lisa-on-cuda && \
    git sparse-checkout set --no-cone resources && \
    git checkout
WORKDIR ${LAMBDA_TASK_ROOT}
RUN ls -l ${LAMBDA_TASK_ROOT}//*

FROM pytorch/pytorch:2.2.2-cuda12.1-cudnn8-runtime as runtime

ARG ARCH
ARG LAMBDA_TASK_ROOT

ENV VIRTUAL_ENV=${LAMBDA_TASK_ROOT}/.venv \
    PATH="${LAMBDA_TASK_ROOT}/.venv/bin:$PATH"

RUN echo "COPY --from=builder_global /usr/lib/${ARCH}-linux-gnu/libGL.so* /usr/lib/${ARCH}-linux-gnu/"
COPY --from=builder_global /usr/lib/${ARCH}-linux-gnu/libGL.so* /usr/lib/${ARCH}-linux-gnu/
RUN echo "RUNTIME: check libz.s* before upgrade" && ls -l /usr/lib/${ARCH}-linux-gnu/libz.so*
RUN echo "RUNTIME: remove libz.s* to force upgrade" && rm /usr/lib/${ARCH}-linux-gnu/libz.so*
COPY --from=builder_global /usr/lib/${ARCH}-linux-gnu/libz.so* /usr/lib/${ARCH}-linux-gnu/
RUN echo "RUNTIME: check libz.s* after copy" && ls -l /usr/lib/${ARCH}-linux-gnu/libz.so*
COPY --from=builder_global ${LAMBDA_TASK_ROOT}/.venv ${LAMBDA_TASK_ROOT}/.venv

RUN echo "new LAMBDA_TASK_ROOT after hidden venv copy => ${LAMBDA_TASK_ROOT}"
RUN ls -ld ${LAMBDA_TASK_ROOT}/
RUN ls -lA ${LAMBDA_TASK_ROOT}/
RUN echo "content of LAMBDA_TASK_ROOT/.venv => ${LAMBDA_TASK_ROOT}/.venv"
RUN ls -ld ${LAMBDA_TASK_ROOT}/.venv
RUN ls -lA ${LAMBDA_TASK_ROOT}/.venv


### conditional section
FROM node:20-slim AS node_fastapi

ARG DEPENDENCY_GROUP
ENV PNPM_HOME="/pnpm"
ENV PATH="$PNPM_HOME:$PATH"

RUN corepack enable

COPY ./static /appnode
WORKDIR /appnode
# RUN if [ "${DEPENDENCY_GROUP}" = "fastapi" ]; then echo "pnpm store path:" && pnpm store path; fi


FROM node_fastapi AS node_prod_deps

ARG DEPENDENCY_GROUP
RUN --mount=type=cache,id=pnpm,target=/pnpm/store if [ "${DEPENDENCY_GROUP}" = "fastapi" ]; then \
    pnpm install --prod --frozen-lockfile; else \
    echo "DEPENDENCY_GROUP 1: ${DEPENDENCY_GROUP} ..."; fi
# here multiple conditions concatenated to avoid failing on check
RUN if [ "${DEPENDENCY_GROUP}" = "fastapi" ]; then if [ ! -d /appnode/node_modules ]; then \
    echo "no node_modules folder" && exit 1; fi; fi


FROM node_fastapi AS node_build

ARG DEPENDENCY_GROUP
ARG VITE__MAP_DESCRIPTION
ARG VITE__SAMGIS_SPACE
RUN echo "VITE__MAP_DESCRIPTION:" ${VITE__MAP_DESCRIPTION}
RUN echo "VITE__SAMGIS_SPACE:" ${VITE__SAMGIS_SPACE}

RUN --mount=type=cache,id=pnpm,target=/pnpm/store if [ "${DEPENDENCY_GROUP}" = "fastapi" ]; then \
    pnpm install --frozen-lockfile; else \
    echo "DEPENDENCY_GROUP 2: ${DEPENDENCY_GROUP} ..."; fi
RUN --mount=type=cache,id=pnpm,target=/pnpm/store if [ "${DEPENDENCY_GROUP}" = "fastapi" ]; then pnpm build; fi
RUN --mount=type=cache,id=pnpm,target=/pnpm/store if [ "${DEPENDENCY_GROUP}" = "fastapi" ]; then \
    pnpm tailwindcss -i /appnode/src/input.css -o /appnode/dist/output.css; fi
RUN if [ "${DEPENDENCY_GROUP}" = "fastapi" ]; then if [ ! -d /appnode/dist ]; then echo "no dist folder" && exit 1; fi; fi


FROM runtime
ARG FASTAPI_STATIC

RUN echo "show disk space, df -h ..."
RUN df -h
RUN echo "creating FASTAPI_STATIC folder: ${FASTAPI_STATIC}, use `mkdir -p` to avoid failure if missing parent folder ..."
RUN mkdir -p ${FASTAPI_STATIC}

COPY --from=builder_global  ${LAMBDA_TASK_ROOT}/sam-quantized/machine_learning_models \
    ${LAMBDA_TASK_ROOT}/machine_learning_models
RUN ls -ld ${LAMBDA_TASK_ROOT}/machine_learning_models
RUN ls -lh ${LAMBDA_TASK_ROOT}/machine_learning_models
COPY --from=builder_global ${LAMBDA_TASK_ROOT}/lisa-on-cuda/resources ${LAMBDA_TASK_ROOT}/resources
COPY --from=node_prod_deps /appnode/node_modules* ${FASTAPI_STATIC}/node_modules
COPY --from=node_build /appnode/dist* ${FASTAPI_STATIC}/dist


# Include global arg in this stage of the build
ARG LAMBDA_TASK_ROOT="/var/task"
ARG PYTHONPATH="${LAMBDA_TASK_ROOT}:${PYTHONPATH}:/usr/local/lib/python3/dist-packages"
ENV VIRTUAL_ENV=${LAMBDA_TASK_ROOT}/.venv \
    PATH="${LAMBDA_TASK_ROOT}/.venv/bin:$PATH"
ENV IS_AWS_LAMBDA=""

# Set working directory to function root directory
WORKDIR ${LAMBDA_TASK_ROOT}

COPY samgis_lisa_on_cuda ${LAMBDA_TASK_ROOT}/samgis_lisa_on_cuda
COPY wrappers ${LAMBDA_TASK_ROOT}/wrappers
COPY scripts ${LAMBDA_TASK_ROOT}/scripts
RUN chmod +x ${LAMBDA_TASK_ROOT}/scripts/entrypoint.sh
RUN chmod +x  ${LAMBDA_TASK_ROOT}/scripts/docker_entrypoint.sh
RUN ls -l ${LAMBDA_TASK_ROOT}/scripts/entrypoint.sh ${LAMBDA_TASK_ROOT}/scripts/docker_entrypoint.sh

RUN ls -l /usr/bin/which
RUN /usr/bin/which python
RUN python -v
RUN echo "PYTHONPATH: ${PYTHONPATH}."
RUN echo "PATH: ${PATH}."
RUN echo "LAMBDA_TASK_ROOT: ${LAMBDA_TASK_ROOT}."
RUN ls -l ${LAMBDA_TASK_ROOT}
RUN ls -ld ${LAMBDA_TASK_ROOT}
RUN ls -l ${LAMBDA_TASK_ROOT}/machine_learning_models
RUN python -c "import sys; print(sys.path)"
RUN python -c "import cv2"
RUN python -c "import fastapi"
RUN python -c "import geopandas"
RUN python -c "import loguru"
RUN python -c "import onnxruntime"
RUN python -c "import rasterio"
RUN python -c "import uvicorn"
RUN df -h
RUN ls -l ${LAMBDA_TASK_ROOT}/samgis_lisa_on_cuda/
RUN ls -l ${LAMBDA_TASK_ROOT}/wrappers/
RUN echo "LAMBDA_TASK_ROOT /static/:"
RUN ls -l ${LAMBDA_TASK_ROOT}/static/ || true
RUN ls -l ${LAMBDA_TASK_ROOT}/static/dist || true
RUN ls -l ${LAMBDA_TASK_ROOT}/static/node_modules || true
RUN echo "FASTAPI_STATIC:"
RUN ls -l ${FASTAPI_STATIC}/ || true
RUN ls -l ${FASTAPI_STATIC}/dist || true
RUN ls -l ${FASTAPI_STATIC}/node_modules || true

CMD ["/var/task/scripts/docker_entrypoint.sh"]