From 225a7a5a077ba6f65865e638f678be737dfc0a6f Mon Sep 17 00:00:00 2001
From: James Lamb <jaylamb20@gmail.com>
Date: Mon, 30 Dec 2024 20:27:42 -0600
Subject: [PATCH 1/2] refactor Dockerfiles

---
 Dockerfile-cluster       | 15 +++++++------
 Dockerfile-cluster-base  | 46 +++++++++++++++++++++++++---------------
 Dockerfile-notebook      | 16 ++++++++------
 Dockerfile-notebook-base | 46 +++++++++++++++++++++++++---------------
 bin/install-cmake        |  9 ++++++--
 5 files changed, 84 insertions(+), 48 deletions(-)

diff --git a/Dockerfile-cluster b/Dockerfile-cluster
index fd657be..0a446cf 100644
--- a/Dockerfile-cluster
+++ b/Dockerfile-cluster
@@ -1,12 +1,15 @@
 ARG BASE_IMAGE=unset
-ARG INSTALL_DIR=/opt/LightGBM
 
 # hadolint ignore=DL3006
 FROM ${BASE_IMAGE}
 
-COPY LightGBM/ ${INSTALL_DIR}/
+RUN --mount=type=bind,source=LightGBM,target=/tmp/LightGBM \
+<<EOF
+cd /tmp/LightGBM
 
-RUN cd "${INSTALL_DIR}" && \
-    sh ./build-python.sh install --precompile && \
-    cd /opt && \
-    rm -rf "${INSTALL_DIR}"
+# build lightgbm package, install to site-packages/
+sh ./build-python.sh install --precompile
+
+# clear pip's cache
+rm -rf ~/.cache
+EOF
diff --git a/Dockerfile-cluster-base b/Dockerfile-cluster-base
index 1a4c301..7cb129f 100644
--- a/Dockerfile-cluster-base
+++ b/Dockerfile-cluster-base
@@ -39,23 +39,31 @@ find \
     /usr/local/lib/python${PYTHON_VERSION}/site-packages \
     -type f \
     \( \
-    -name '*.c' \
-    -o -name '*.cc' \
-    -o -name '*.cpp' \
-    -o -name '*.h' \
-    -o -name '*.hpp' \
-    -o -wholename '*bokeh/sampledata/*' \
-    -o -wholename '*dask/*tests/*' \
-    -o -wholename '*joblib/test/*' \
-    -o -wholename '*llvmlite/tests/*' \
-    -o -wholename '*numba/*tests/*' \
-    -o -wholename '*numpy/*tests/*' \
-    -o -wholename '*pandas/*tests/*' \
-    -o -wholename '*scikit-learn/tests*' \
-    -o -wholename '*scikit-learn/*/tests*' \
-    -o -wholename '*scipy/*/tests*' \
-    -o -wholename '*tornado/test/*' \
-    -o -wholename '*/__pycache__/*' \
+    -name '*.c'                                 \
+    -o -name '*.cc'                             \
+    -o -name '*.cpp'                            \
+    -o -name '*.h'                              \
+    -o -name '*.hpp'                            \
+    -o -wholename '*bokeh/sampledata/*'         \
+    -o -wholename '*dask/*tests/*'              \
+    -o -wholename '*joblib/test/*'              \
+    -o -wholename '*llvmlite/tests/*'           \
+    -o -wholename '*numba/*tests/*'             \
+    -o -wholename '*numpy/*tests/*'             \
+    -o -wholename '*pandas/tests*'              \
+    -o -wholename '*pandas/*/tests/*'           \
+    -o -wholename '*psutil/tests/*'             \
+    -o -wholename 'pyarrow/_pyarrow_cpp_tests*' \
+    -o -wholename '*scikit-learn/tests*'        \
+    -o -wholename '*scikit-learn/*/tests*'      \
+    -o -wholename '*sklearn/tests*'             \
+    -o -wholename '*sklearn/*/tests*'           \
+    -o -wholename '*scipy/*/tests*'             \
+    -o -wholename '*sparse/*/tests/*'           \
+    -o -wholename '*toolz/tests/*'              \
+    -o -wholename '*tornado/test/*'             \
+    -o -wholename '*zict/tests/*'               \
+    -o -wholename '*/__pycache__/*'             \
     \) \
     -exec rm '{}' '+'
 
@@ -67,5 +75,9 @@ find \
 
 # clean apt-get files
 apt-get clean
+apt-get purge -y --auto-remove
 rm -rf /var/lib/apt/lists/*
+
+# clean other files
+rm -rf ~/.cache
 EOF
diff --git a/Dockerfile-notebook b/Dockerfile-notebook
index a25d4fd..e15607d 100644
--- a/Dockerfile-notebook
+++ b/Dockerfile-notebook
@@ -1,15 +1,19 @@
 ARG BASE_IMAGE=unset
-ARG INSTALL_DIR=/root/testing/LightGBM
 
 # hadolint ignore=DL3006
 FROM ${BASE_IMAGE}
 
 COPY jupyter_notebook_config.py /root/.jupyter/jupyter_notebook_config.py
-COPY LightGBM/ ${INSTALL_DIR}/
 
-RUN cd "${INSTALL_DIR}" && \
-    sh ./build-python.sh install --precompile && \
-    cd /root/ && \
-    rm -rf "${INSTALL_DIR}"
+RUN --mount=type=bind,source=LightGBM,target=/tmp/LightGBM \
+<<EOF
+cd /tmp/LightGBM
+
+# build lightgbm package, install to site-packages/
+sh ./build-python.sh install --precompile
+
+# clear pip's cache
+rm -rf ~/.cache
+EOF
 
 WORKDIR /root/testing/notebooks
diff --git a/Dockerfile-notebook-base b/Dockerfile-notebook-base
index 87bd2f5..2d548fb 100644
--- a/Dockerfile-notebook-base
+++ b/Dockerfile-notebook-base
@@ -43,23 +43,31 @@ find \
     /usr/local/lib/python${PYTHON_VERSION}/site-packages \
     -type f \
     \( \
-    -name '*.c' \
-    -o -name '*.cc' \
-    -o -name '*.cpp' \
-    -o -name '*.h' \
-    -o -name '*.hpp' \
-    -o -wholename '*bokeh/sampledata/*' \
-    -o -wholename '*dask/*tests/*' \
-    -o -wholename '*joblib/test/*' \
-    -o -wholename '*llvmlite/tests/*' \
-    -o -wholename '*numba/*tests/*' \
-    -o -wholename '*numpy/*tests/*' \
-    -o -wholename '*pandas/*tests/*' \
-    -o -wholename '*scikit-learn/tests*' \
-    -o -wholename '*scikit-learn/*/tests*' \
-    -o -wholename '*scipy/*/tests*' \
-    -o -wholename '*tornado/test/*' \
-    -o -wholename '*/__pycache__/*' \
+    -name '*.c'                                 \
+    -o -name '*.cc'                             \
+    -o -name '*.cpp'                            \
+    -o -name '*.h'                              \
+    -o -name '*.hpp'                            \
+    -o -wholename '*bokeh/sampledata/*'         \
+    -o -wholename '*dask/*tests/*'              \
+    -o -wholename '*joblib/test/*'              \
+    -o -wholename '*llvmlite/tests/*'           \
+    -o -wholename '*numba/*tests/*'             \
+    -o -wholename '*numpy/*tests/*'             \
+    -o -wholename '*pandas/tests*'              \
+    -o -wholename '*pandas/*/tests/*'           \
+    -o -wholename '*psutil/tests/*'             \
+    -o -wholename 'pyarrow/_pyarrow_cpp_tests*' \
+    -o -wholename '*scikit-learn/tests*'        \
+    -o -wholename '*scikit-learn/*/tests*'      \
+    -o -wholename '*sklearn/tests*'             \
+    -o -wholename '*sklearn/*/tests*'           \
+    -o -wholename '*scipy/*/tests*'             \
+    -o -wholename '*sparse/*/tests/*'           \
+    -o -wholename '*toolz/tests/*'              \
+    -o -wholename '*tornado/test/*'             \
+    -o -wholename '*zict/tests/*'               \
+    -o -wholename '*/__pycache__/*'             \
     \) \
     -exec rm '{}' '+'
 
@@ -71,7 +79,11 @@ find \
 
 # clean apt-get files
 apt-get clean
+apt-get purge -y --auto-remove
 rm -rf /var/lib/apt/lists/*
+
+# clean other files
+rm -rf ~/.cache
 EOF
 
 ENTRYPOINT ["jupyter", "lab", "--ip=0.0.0.0", "--allow-root", "--port=8888"]
diff --git a/bin/install-cmake b/bin/install-cmake
index 2129290..f46f7ad 100755
--- a/bin/install-cmake
+++ b/bin/install-cmake
@@ -4,12 +4,17 @@ set -e -u -o pipefail
 
 CMAKE_VERSION=${1}
 
+declare -r install_script
+install_script="cmake-${CMAKE_VERSION}-linux-$(arch).sh"
+
 curl -O -L \
-    "https://github.com/Kitware/CMake/releases/download/v${CMAKE_VERSION}/cmake-${CMAKE_VERSION}-linux-$(arch).sh"
+    "https://github.com/Kitware/CMake/releases/download/v${CMAKE_VERSION}/${install_script}"
 
 mkdir /opt/cmake
-sh "cmake-${CMAKE_VERSION}-linux-$(arch).sh" \
+sh "${install_script}" \
     --skip-license \
     --prefix=/opt/cmake
 
+rm "./${install_script}"
+
 ln -sf /opt/cmake/bin/cmake /usr/local/bin/cmake

From df5224ebc9c0e1035513cfb21d977dfd8187da60 Mon Sep 17 00:00:00 2001
From: James Lamb <jaylamb20@gmail.com>
Date: Mon, 30 Dec 2024 20:34:12 -0600
Subject: [PATCH 2/2] fix install-cmake script

---
 bin/install-cmake | 1 -
 1 file changed, 1 deletion(-)

diff --git a/bin/install-cmake b/bin/install-cmake
index f46f7ad..fee15a0 100755
--- a/bin/install-cmake
+++ b/bin/install-cmake
@@ -4,7 +4,6 @@ set -e -u -o pipefail
 
 CMAKE_VERSION=${1}
 
-declare -r install_script
 install_script="cmake-${CMAKE_VERSION}-linux-$(arch).sh"
 
 curl -O -L \