[GSProcessing] SagMaker image for v0.2.1

graphstorm-processing/docker/0.2.1/sagemaker/Dockerfile.cpu

@@ -0,0 +1,45 @@
+# syntax=docker/dockerfile:experimental
+FROM 153931337802.dkr.ecr.us-west-2.amazonaws.com/sagemaker-spark-processing:3.4-cpu-py39-v1.0 AS base
+
+# Python won’t try to write .pyc or .pyo files on the import of source modules
+# Force stdin, stdout and stderr to be totally unbuffered. Good for logging
+ENV PYTHONDONTWRITEBYTECODE=1
+ENV PYTHONUNBUFFERED=1
+ENV PYTHONIOENCODING=UTF-8
+ENV LANG=C.UTF-8
+ENV LC_ALL=C.UTF-8
+ENV LD_LIBRARY_PATH="${LD_LIBRARY_PATH}:/usr/local/lib"
+ENV LD_LIBRARY_PATH="${LD_LIBRARY_PATH}:/opt/conda/lib"
+ENV PATH=/opt/conda/bin:$PATH
+
+# GSProcessing requirements
+RUN pipenv install pip==23.1.2 setuptools wheel spacy==3.6.0 pyspark==3.4.1 \
+    pyarrow==13.0.0 joblib==1.3.1 psutil==5.9.5 pandas==1.3.5 \
+    boto3==1.28.38 protobuf==3.20.3 mock==5.1.0 \
+    && rm -rf /root/.cache
+# Do a pipenv sync so our base libs are independent from our editable code, making them cacheable
+RUN pipenv sync --system && python3 -m spacy download en_core_web_lg \
+    && rm -rf /root/.cache
+
+# Graphloader codebase
+COPY code/ /usr/lib/spark/code/
+WORKDIR /usr/lib/spark/code/
+
+# Base container assumes this is the workdir
+ENV SPARK_HOME /usr/lib/spark
+WORKDIR $SPARK_HOME
+
+# Ensure our python3 installation is the one used
+RUN echo 'alias python3=python3.9' >> ~/.bashrc
+
+# Starts framework
+ENTRYPOINT ["bash", "/usr/lib/spark/code/docker-entry.sh"]
+
+FROM base AS prod
+RUN python3 -m pip install /usr/lib/spark/code/graphstorm_processing-*.whl && \
+    rm /usr/lib/spark/code/graphstorm_processing-*.whl
+CMD ["gs-processing"]
+
+FROM base AS test
+RUN python3 -m pip install /usr/lib/spark/code/graphstorm-processing/
+CMD ["sh", "-c", "pytest ./code/tests/"]