From 274e70aa392291e5ff4ddbe17f83890407341d38 Mon Sep 17 00:00:00 2001
From: Gaetan Lepage <gaetan@glepage.com>
Date: Fri, 28 Feb 2025 23:21:52 +0100
Subject: [PATCH] [python312Packages.]lightgbm: 4.5.0 -> 4.6.0

Diff: https://github.com/microsoft/lightgbm/compare/v4.5.0...v4.6.0

Changelog: https://github.com/microsoft/LightGBM/releases/tag/v4.6.0
---
 pkgs/by-name/li/lightgbm/package.nix          | 129 +++++++++++-------
 .../python-modules/lightgbm/default.nix       |  46 ++++---
 2 files changed, 105 insertions(+), 70 deletions(-)

diff --git a/pkgs/by-name/li/lightgbm/package.nix b/pkgs/by-name/li/lightgbm/package.nix
index c5d0238724d5..ee829c4e9cb8 100644
--- a/pkgs/by-name/li/lightgbm/package.nix
+++ b/pkgs/by-name/li/lightgbm/package.nix
@@ -1,8 +1,9 @@
 {
-  config,
-  stdenv,
   lib,
+  stdenv,
+  config,
   fetchFromGitHub,
+  fetchpatch,
   cmake,
   gtest,
   doCheck ? true,
@@ -25,14 +26,14 @@
   R,
   rPackages,
   pandoc,
+  python3Packages,
 }:
 
 assert doCheck -> !mpiSupport;
 assert openclSupport -> !cudaSupport;
 assert cudaSupport -> !openclSupport;
 
-stdenv.mkDerivation rec {
-  pnameBase = "lightgbm";
+stdenv.mkDerivation (finalAttrs: {
   # prefix with r when building the R library
   # The R package build results in a special binary file
   # that contains a subset of the .so file use for the CLI
@@ -45,17 +46,52 @@ stdenv.mkDerivation rec {
   #     lgbm = lightgbm.override{rLibrary = true; doCheck = false;}; \
   #   in \
   #   rWrapper.override{ packages = [ lgbm ]; }"
-  pname = lib.optionalString rLibrary "r-" + pnameBase;
-  version = "4.5.0";
+  pname = lib.optionalString rLibrary "r-" + "lightgbm";
+  version = "4.6.0";
 
   src = fetchFromGitHub {
     owner = "microsoft";
-    repo = pnameBase;
-    rev = "v${version}";
+    repo = "lightgbm";
+    tag = "v${finalAttrs.version}";
     fetchSubmodules = true;
-    hash = "sha256-nST6+/c3Y4/hqwgEUhx03gWtjxhlmUu1XKDCy2pSsvU=";
+    hash = "sha256-vq/TlM87i1GNq0Rpy0OTulT9LF+uvi4PhOUz7ZNeceA=";
   };
 
+  patches = [
+    # Fix boost 1.83+ compatibility
+    # https://github.com/microsoft/LightGBM/issues/6786
+    # Patch taken from https://github.com/conda-forge/lightgbm-feedstock/pull/69
+    (fetchpatch {
+      name = "fix-boost-sha1";
+      url = "https://raw.githubusercontent.com/conda-forge/lightgbm-feedstock/68ca96d25d8a6f7281310a4ad3b8d5cdd01f067b/recipe/patches/0003-boost-sha1.patch";
+      hash = "sha256-Hw2YmoduOPri7O1XV2p/3Ny4hC8xq7Jq4zoSuKhVeVQ=";
+    })
+  ];
+
+  # Skip APPLE in favor of linux build for .so files
+  postPatch = ''
+    export PROJECT_SOURCE_DIR=./
+    substituteInPlace CMakeLists.txt \
+      --replace-fail "find_package(GTest CONFIG)" "find_package(GTest REQUIRED)" \
+      --replace-fail "OpenCL_INCLUDE_DIRS}" "OpenCL_INCLUDE_DIRS}" \
+      --replace-fail "elseif(APPLE)" "elseif(APPLESKIP)"
+    substituteInPlace \
+      external_libs/compute/include/boost/compute/cl.hpp \
+      external_libs/compute/include/boost/compute/cl_ext.hpp \
+      --replace-fail "include <OpenCL/" "include <CL/"
+    substituteInPlace build_r.R \
+      --replace-fail "shQuote(normalizePath" "shQuote(type = 'cmd', string = normalizePath" \
+      --replace-fail "file.path(getwd(), \"lightgbm_r\")" "'$out/tmp'" \
+      --replace-fail \
+        "install_args <- c(\"CMD\", \"INSTALL\", \"--no-multiarch\", \"--with-keep.source\", tarball)" \
+        "install_args <- c(\"CMD\", \"INSTALL\", \"--no-multiarch\", \"--with-keep.source\", \"-l $out/library\", tarball)"
+
+    # Retry this test in next release. Something fails in the setup, so GTEST_FILTER is not enough
+    substituteInPlace CMakeLists.txt \
+      --replace-fail "tests/cpp_tests/test_arrow.cpp" ""
+    rm tests/cpp_tests/test_arrow.cpp
+  '';
+
   nativeBuildInputs =
     [ cmake ]
     ++ lib.optionals stdenv.hostPlatform.isDarwin [ llvmPackages.openmp ]
@@ -84,48 +120,36 @@ stdenv.mkDerivation rec {
     rPackages.R6
   ];
 
-  # Skip APPLE in favor of linux build for .so files
-  postPatch = ''
-    export PROJECT_SOURCE_DIR=./
-    substituteInPlace CMakeLists.txt \
-      --replace "find_package(GTest CONFIG)" "find_package(GTest REQUIRED)" \
-      --replace "OpenCL_INCLUDE_DIRS}" "OpenCL_INCLUDE_DIRS}" \
-      --replace "elseif(APPLE)" "elseif(APPLESKIP)"
-    substituteInPlace \
-      external_libs/compute/include/boost/compute/cl.hpp \
-      external_libs/compute/include/boost/compute/cl_ext.hpp \
-      --replace "include <OpenCL/" "include <CL/"
-    substituteInPlace build_r.R \
-      --replace "shQuote(normalizePath" "shQuote(type = 'cmd', string = normalizePath" \
-      --replace "file.path(getwd(), \"lightgbm_r\")" "'$out/tmp'" \
-      --replace \
-        "install_args <- c(\"CMD\", \"INSTALL\", \"--no-multiarch\", \"--with-keep.source\", tarball)" \
-        "install_args <- c(\"CMD\", \"INSTALL\", \"--no-multiarch\", \"--with-keep.source\", \"-l $out/library\", tarball)"
-
-    # Retry this test in next release. Something fails in the setup, so GTEST_FILTER is not enough
-    rm tests/cpp_tests/test_arrow.cpp
-  '';
-
   cmakeFlags =
-    lib.optionals doCheck [ "-DBUILD_CPP_TEST=ON" ]
+    lib.optionals doCheck [
+      (lib.cmakeBool "BUILD_CPP_TEST" true)
+    ]
     ++ lib.optionals cudaSupport [
-      "-DUSE_CUDA=1"
-      "-DCMAKE_CXX_COMPILER=${cudaPackages.backendStdenv.cc}/bin/cc"
+      (lib.cmakeBool "USE_CUDA" true)
+      (lib.cmakeFeature "CMAKE_CXX_COMPILER" (lib.getExe cudaPackages.backendStdenv.cc))
+    ]
+    ++ lib.optionals openclSupport [
+      (lib.cmakeBool "USE_GPU" true)
+    ]
+    ++ lib.optionals mpiSupport [
+      (lib.cmakeBool "USE_MPI" true)
     ]
-    ++ lib.optionals openclSupport [ "-DUSE_GPU=ON" ]
-    ++ lib.optionals mpiSupport [ "-DUSE_MPI=ON" ]
     ++ lib.optionals hdfsSupport [
-      "-DUSE_HDFS=ON"
-      "-DHDFS_LIB=${hadoop}/lib/hadoop-${hadoop.version}/lib/native/libhdfs.so"
-      "-DHDFS_INCLUDE_DIR=${hadoop}/lib/hadoop-${hadoop.version}/include"
+      (lib.cmakeBool "USE_HDFS" true)
+      (lib.cmakeFeature "HDFS_LIB" "${hadoop}/lib/hadoop-${hadoop.version}/lib/native/libhdfs.so")
+      (lib.cmakeFeature "HDFS_INCLUDE_DIR" "${hadoop}/lib/hadoop-${hadoop.version}/include")
     ]
     ++ lib.optionals javaWrapper [
-      "-DUSE_SWIG=ON"
+      (lib.cmakeBool "USE_SWIG" true)
       # RPATH of binary /nix/store/.../bin/... contains a forbidden reference to /build/
-      "-DCMAKE_SKIP_BUILD_RPATH=ON"
+      (lib.cmakeBool "CMAKE_SKIP_BUILD_RPATH" true)
     ]
-    ++ lib.optionals rLibrary [ "-D__BUILD_FOR_R=ON" ]
-    ++ lib.optionals pythonLibrary [ "-D__BUILD_FOR_PYTHON=ON" ];
+    ++ lib.optionals rLibrary [
+      (lib.cmakeBool "__BUILD_FOR_R" true)
+    ]
+    ++ lib.optionals pythonLibrary [
+      (lib.cmakeBool "__BUILD_FOR_PYTHON" true)
+    ];
 
   configurePhase = lib.optionals rLibrary ''
     export R_LIBS_SITE="$out/library:$R_LIBS_SITE''${R_LIBS_SITE:+:}"
@@ -182,16 +206,23 @@ stdenv.mkDerivation rec {
 
   postFixup = lib.optionalString rLibrary ''
     if test -e $out/nix-support/propagated-build-inputs; then
-        ln -s $out/nix-support/propagated-build-inputs $out/nix-support/propagated-user-env-packages
+      ln -s $out/nix-support/propagated-build-inputs $out/nix-support/propagated-user-env-packages
     fi
   '';
 
-  meta = with lib; {
+  passthru = {
+    tests = {
+      pythonPackage = python3Packages.lightgbm;
+    };
+  };
+
+  meta = {
     description = "LightGBM is a gradient boosting framework that uses tree based learning algorithms.";
     mainProgram = "lightgbm";
     homepage = "https://github.com/microsoft/LightGBM";
-    license = licenses.mit;
-    platforms = platforms.unix;
-    maintainers = with maintainers; [ nviets ];
+    changelog = "https://github.com/microsoft/LightGBM/releases/tag/v${finalAttrs.version}";
+    license = lib.licenses.mit;
+    platforms = lib.platforms.unix;
+    maintainers = with lib.maintainers; [ nviets ];
   };
-}
+})
diff --git a/pkgs/development/python-modules/lightgbm/default.nix b/pkgs/development/python-modules/lightgbm/default.nix
index 231b9e5c258c..09808cafb11b 100644
--- a/pkgs/development/python-modules/lightgbm/default.nix
+++ b/pkgs/development/python-modules/lightgbm/default.nix
@@ -2,23 +2,31 @@
   lib,
   config,
   stdenv,
+  pkgs,
   buildPythonPackage,
   fetchPypi,
 
   # build-system
+  scikit-build-core,
+
+  # nativeBuildInputs
   cmake,
   ninja,
   pathspec,
   pyproject-metadata,
-  scikit-build-core,
+  writableTmpDirAsHomeHook,
+
+  # buildInputs
+  llvmPackages,
+  boost,
+  ocl-icd,
+  opencl-headers,
 
   # dependencies
-  llvmPackages,
   numpy,
   scipy,
-  pythonOlder,
 
-  # optionals
+  # optional-dependencies
   cffi,
   dask,
   pandas,
@@ -26,9 +34,6 @@
   scikit-learn,
 
   # optionals: gpu
-  boost,
-  ocl-icd,
-  opencl-headers,
   gpuSupport ? stdenv.hostPlatform.isLinux && !cudaSupport,
   cudaSupport ? config.cudaSupport,
   cudaPackages,
@@ -38,23 +43,28 @@ assert gpuSupport -> !cudaSupport;
 assert cudaSupport -> !gpuSupport;
 
 buildPythonPackage rec {
-  pname = "lightgbm";
-  version = "4.5.0";
+  inherit (pkgs.lightgbm)
+    pname
+    version
+    patches
+    ;
   pyproject = true;
 
-  disabled = pythonOlder "3.7";
-
   src = fetchPypi {
     inherit pname version;
-    hash = "sha256-4c17rwMY1OMIomV1pjpGNfCN+GatNiKp2OPXHZY3obo=";
+    hash = "sha256-yxxZcg61aTicC6dNFPUjUbVzr0ifIwAyocnzFPi6t/4=";
   };
 
+  build-system = [
+    scikit-build-core
+  ];
+
   nativeBuildInputs = [
     cmake
     ninja
     pathspec
     pyproject-metadata
-    scikit-build-core
+    writableTmpDirAsHomeHook
   ] ++ lib.optionals cudaSupport [ cudaPackages.cuda_nvcc ];
 
   dontUseCmakeConfigure = true;
@@ -71,7 +81,7 @@ buildPythonPackage rec {
       cudaPackages.cuda_cudart
     ];
 
-  propagatedBuildInputs = [
+  dependencies = [
     numpy
     scipy
   ];
@@ -80,10 +90,6 @@ buildPythonPackage rec {
     lib.optionals gpuSupport [ "--config-setting=cmake.define.USE_GPU=ON" ]
     ++ lib.optionals cudaSupport [ "--config-setting=cmake.define.USE_CUDA=ON" ];
 
-  postConfigure = ''
-    export HOME=$(mktemp -d)
-  '';
-
   optional-dependencies = {
     arrow = [
       cffi
@@ -101,9 +107,7 @@ buildPythonPackage rec {
     scikit-learn = [ scikit-learn ];
   };
 
-  # The pypi package doesn't distribute the tests from the GitHub
-  # repository. It contains c++ tests which don't seem to wired up to
-  # `make check`.
+  # No python tests
   doCheck = false;
 
   pythonImportsCheck = [ "lightgbm" ];