GH-42149: [C++] Use FetchContent for bundled ORC (#43011)

### Rationale for this change This also has a workaround for https://issues.apache.org/jira/browse/ORC-1732 . ### What changes are included in this PR? ORC 2.0.1 has a dependency detection problem. We can't override the detection with ExternalProject but can override the detection with FetchContent. ### Are these changes tested? Yes. ### Are there any user-facing changes? Yes. * GitHub Issue: #42149 Authored-by: Sutou Kouhei <[email protected]> Signed-off-by: Raúl Cumplido <[email protected]>
apache · Jul 8, 2024 · 12be569 · 12be569
1 parent 58d5142
commit 12be569
Show file tree

Hide file tree

Showing 4 changed files with 201 additions and 91 deletions.
diff --git a/cpp/CMakeLists.txt b/cpp/CMakeLists.txt
@@ -71,6 +71,14 @@ if(POLICY CMP0135)
   cmake_policy(SET CMP0135 NEW)
 endif()
 
+# https://cmake.org/cmake/help/latest/policy/CMP0170.html
+#
+# CMP0170 is for enforcing dependency populations by users with
+# FETCHCONTENT_FULLY_DISCONNECTED=ON.
+if(POLICY CMP0170)
+  cmake_policy(SET CMP0170 NEW)
+endif()
+
 set(ARROW_VERSION "17.0.0-SNAPSHOT")
 
 string(REGEX MATCH "^[0-9]+\\.[0-9]+\\.[0-9]+" ARROW_BASE_VERSION "${ARROW_VERSION}")

diff --git a/cpp/cmake_modules/ThirdpartyToolchain.cmake b/cpp/cmake_modules/ThirdpartyToolchain.cmake
@@ -2532,6 +2532,7 @@ macro(build_zlib)
     set_property(TARGET ZLIB::ZLIB
                  PROPERTY IMPORTED_LOCATION
                           "${EMSCRIPTEN_SYSROOT}/lib/wasm32-emscripten/pic/libz.a")
+    target_include_directories(ZLIB::ZLIB INTERFACE "${EMSCRIPTEN_SYSROOT}/include")
     list(APPEND ARROW_BUNDLED_STATIC_LIBS ZLIB::ZLIB)
   else()
     set(ZLIB_PREFIX "${CMAKE_CURRENT_BINARY_DIR}/zlib_ep/src/zlib_ep-install")
@@ -4490,116 +4491,216 @@ target_include_directories(arrow::hadoop INTERFACE "${HADOOP_HOME}/include")
 # ----------------------------------------------------------------------
 # Apache ORC
 
-macro(build_orc)
+function(build_orc)
   message(STATUS "Building Apache ORC from source")
 
-  set(ORC_PREFIX "${CMAKE_CURRENT_BINARY_DIR}/orc_ep-install")
-  set(ORC_HOME "${ORC_PREFIX}")
-  set(ORC_INCLUDE_DIR "${ORC_PREFIX}/include")
-  set(ORC_STATIC_LIB
-      "${ORC_PREFIX}/lib/${CMAKE_STATIC_LIBRARY_PREFIX}orc${CMAKE_STATIC_LIBRARY_SUFFIX}")
+  if(CMAKE_VERSION VERSION_GREATER_EQUAL 3.29)
+    fetchcontent_declare(orc
+                         ${FC_DECLARE_COMMON_OPTIONS}
+                         URL ${ORC_SOURCE_URL}
+                         URL_HASH "SHA256=${ARROW_ORC_BUILD_SHA256_CHECKSUM}")
+    prepare_fetchcontent()
+
+    set(CMAKE_UNITY_BUILD FALSE)
+
+    set(ORC_PREFER_STATIC_LZ4
+        OFF
+        CACHE BOOL "" FORCE)
+    get_target_property(LZ4_INCLUDE_DIR LZ4::lz4 INTERFACE_INCLUDE_DIRECTORIES)
+    get_filename_component(LZ4_ROOT "${LZ4_INCLUDE_DIR}" DIRECTORY)
+    set(LZ4_HOME
+        ${LZ4_ROOT}
+        CACHE STRING "" FORCE)
+    set(LZ4_LIBRARY
+        LZ4::lz4
+        CACHE STRING "" FORCE)
+
+    set(ORC_PREFER_STATIC_PROTOBUF
+        OFF
+        CACHE BOOL "" FORCE)
+    get_target_property(PROTOBUF_INCLUDE_DIR ${ARROW_PROTOBUF_LIBPROTOBUF}
+                        INTERFACE_INCLUDE_DIRECTORIES)
+    get_filename_component(Protobuf_ROOT "${PROTOBUF_INCLUDE_DIR}" DIRECTORY)
+    set(PROTOBUF_HOME
+        ${Protobuf_ROOT}
+        CACHE STRING "" FORCE)
+    # ORC uses this.
+    target_include_directories(${ARROW_PROTOBUF_LIBPROTOC}
+                               INTERFACE "${PROTOBUF_INCLUDE_DIR}")
+    set(PROTOBUF_EXECUTABLE ${ARROW_PROTOBUF_PROTOC})
+    set(PROTOBUF_LIBRARY ${ARROW_PROTOBUF_LIBPROTOBUF})
+    set(PROTOC_LIBRARY ${ARROW_PROTOBUF_LIBPROTOC})
+
+    set(ORC_PREFER_STATIC_SNAPPY
+        OFF
+        CACHE BOOL "" FORCE)
+    get_target_property(SNAPPY_INCLUDE_DIR ${Snappy_TARGET} INTERFACE_INCLUDE_DIRECTORIES)
+    get_filename_component(Snappy_ROOT "${SNAPPY_INCLUDE_DIR}" DIRECTORY)
+    set(SNAPPY_HOME
+        ${Snappy_ROOT}
+        CACHE STRING "" FORCE)
+    set(SNAPPY_LIBRARY
+        ${Snappy_TARGET}
+        CACHE STRING "" FORCE)
+
+    set(ORC_PREFER_STATIC_ZLIB
+        OFF
+        CACHE BOOL "" FORCE)
+    get_target_property(ZLIB_INCLUDE_DIR ZLIB::ZLIB INTERFACE_INCLUDE_DIRECTORIES)
+    get_filename_component(ZLIB_ROOT "${ZLIB_INCLUDE_DIR}" DIRECTORY)
+    set(ZLIB_HOME
+        ${ZLIB_ROOT}
+        CACHE STRING "" FORCE)
+    set(ZLIB_LIBRARY
+        ZLIB::ZLIB
+        CACHE STRING "" FORCE)
+
+    set(ORC_PREFER_STATIC_ZSTD
+        OFF
+        CACHE BOOL "" FORCE)
+    get_target_property(ZSTD_INCLUDE_DIR ${ARROW_ZSTD_LIBZSTD}
+                        INTERFACE_INCLUDE_DIRECTORIES)
+    get_filename_component(ZSTD_ROOT "${ZSTD_INCLUDE_DIR}" DIRECTORY)
+    set(ZSTD_HOME
+        ${ZSTD_ROOT}
+        CACHE STRING "" FORCE)
+    set(ZSTD_LIBRARY ${ARROW_ZSTD_LIBZSTD})
+
+    set(BUILD_CPP_TESTS
+        OFF
+        CACHE BOOL "" FORCE)
+    set(BUILD_JAVA
+        OFF
+        CACHE BOOL "" FORCE)
+    set(BUILD_LIBHDFSPP
+        OFF
+        CACHE BOOL "" FORCE)
+    set(BUILD_TOOLS
+        OFF
+        CACHE BOOL "" FORCE)
+    set(INSTALL_VENDORED_LIBS
+        OFF
+        CACHE BOOL "" FORCE)
+    set(STOP_BUILD_ON_WARNING
+        OFF
+        CACHE BOOL "" FORCE)
+
+    # We can remove this with ORC 2.0.2 or later.
+    list(PREPEND CMAKE_MODULE_PATH
+         ${CMAKE_CURRENT_BINARY_DIR}/_deps/orc-src/cmake_modules)
+
+    fetchcontent_makeavailable(orc)
+
+    add_library(orc::orc INTERFACE IMPORTED)
+    target_link_libraries(orc::orc INTERFACE orc)
+    target_include_directories(orc::orc INTERFACE "${orc_BINARY_DIR}/c++/include"
+                                                  "${orc_SOURCE_DIR}/c++/include")
+
+    list(APPEND ARROW_BUNDLED_STATIC_LIBS orc)
+  else()
+    set(ORC_PREFIX "${CMAKE_CURRENT_BINARY_DIR}/orc_ep-install")
+    set(ORC_HOME "${ORC_PREFIX}")
+    set(ORC_INCLUDE_DIR "${ORC_PREFIX}/include")
+    set(ORC_STATIC_LIB
+        "${ORC_PREFIX}/lib/${CMAKE_STATIC_LIBRARY_PREFIX}orc${CMAKE_STATIC_LIBRARY_SUFFIX}"
+    )
 
-  get_target_property(ORC_PROTOBUF_ROOT ${ARROW_PROTOBUF_LIBPROTOBUF}
-                      INTERFACE_INCLUDE_DIRECTORIES)
-  get_filename_component(ORC_PROTOBUF_ROOT "${ORC_PROTOBUF_ROOT}" DIRECTORY)
+    get_target_property(ORC_PROTOBUF_ROOT ${ARROW_PROTOBUF_LIBPROTOBUF}
+                        INTERFACE_INCLUDE_DIRECTORIES)
+    get_filename_component(ORC_PROTOBUF_ROOT "${ORC_PROTOBUF_ROOT}" DIRECTORY)
 
-  get_target_property(ORC_SNAPPY_INCLUDE_DIR ${Snappy_TARGET}
-                      INTERFACE_INCLUDE_DIRECTORIES)
-  get_filename_component(ORC_SNAPPY_ROOT "${ORC_SNAPPY_INCLUDE_DIR}" DIRECTORY)
+    get_target_property(ORC_SNAPPY_INCLUDE_DIR ${Snappy_TARGET}
+                        INTERFACE_INCLUDE_DIRECTORIES)
+    get_filename_component(ORC_SNAPPY_ROOT "${ORC_SNAPPY_INCLUDE_DIR}" DIRECTORY)
 
-  get_target_property(ORC_LZ4_ROOT LZ4::lz4 INTERFACE_INCLUDE_DIRECTORIES)
-  get_filename_component(ORC_LZ4_ROOT "${ORC_LZ4_ROOT}" DIRECTORY)
+    get_target_property(ORC_LZ4_ROOT LZ4::lz4 INTERFACE_INCLUDE_DIRECTORIES)
+    get_filename_component(ORC_LZ4_ROOT "${ORC_LZ4_ROOT}" DIRECTORY)
 
-  get_target_property(ORC_ZSTD_ROOT ${ARROW_ZSTD_LIBZSTD} INTERFACE_INCLUDE_DIRECTORIES)
-  get_filename_component(ORC_ZSTD_ROOT "${ORC_ZSTD_ROOT}" DIRECTORY)
+    get_target_property(ORC_ZSTD_ROOT ${ARROW_ZSTD_LIBZSTD} INTERFACE_INCLUDE_DIRECTORIES)
+    get_filename_component(ORC_ZSTD_ROOT "${ORC_ZSTD_ROOT}" DIRECTORY)
 
-  set(ORC_CMAKE_ARGS
-      ${EP_COMMON_CMAKE_ARGS}
-      "-DCMAKE_INSTALL_PREFIX=${ORC_PREFIX}"
-      -DSTOP_BUILD_ON_WARNING=OFF
-      -DBUILD_LIBHDFSPP=OFF
-      -DBUILD_JAVA=OFF
-      -DBUILD_TOOLS=OFF
-      -DBUILD_CPP_TESTS=OFF
-      -DINSTALL_VENDORED_LIBS=OFF
-      "-DLZ4_HOME=${ORC_LZ4_ROOT}"
-      "-DPROTOBUF_EXECUTABLE=$<TARGET_FILE:${ARROW_PROTOBUF_PROTOC}>"
-      "-DPROTOBUF_HOME=${ORC_PROTOBUF_ROOT}"
-      "-DPROTOBUF_INCLUDE_DIR=$<TARGET_PROPERTY:${ARROW_PROTOBUF_LIBPROTOBUF},INTERFACE_INCLUDE_DIRECTORIES>"
-      "-DPROTOBUF_LIBRARY=$<TARGET_FILE:${ARROW_PROTOBUF_LIBPROTOBUF}>"
-      "-DPROTOC_LIBRARY=$<TARGET_FILE:${ARROW_PROTOBUF_LIBPROTOC}>"
-      "-DSNAPPY_HOME=${ORC_SNAPPY_ROOT}"
-      "-DSNAPPY_LIBRARY=$<TARGET_FILE:${Snappy_TARGET}>"
-      "-DLZ4_LIBRARY=$<TARGET_FILE:LZ4::lz4>"
-      "-DLZ4_STATIC_LIB=$<TARGET_FILE:LZ4::lz4>"
-      "-DLZ4_INCLUDE_DIR=${ORC_LZ4_ROOT}/include"
-      "-DSNAPPY_INCLUDE_DIR=${ORC_SNAPPY_INCLUDE_DIR}"
-      "-DZSTD_HOME=${ORC_ZSTD_ROOT}"
-      "-DZSTD_INCLUDE_DIR=$<TARGET_PROPERTY:${ARROW_ZSTD_LIBZSTD},INTERFACE_INCLUDE_DIRECTORIES>"
-      "-DZSTD_LIBRARY=$<TARGET_FILE:${ARROW_ZSTD_LIBZSTD}>")
-  if(ZLIB_ROOT)
-    set(ORC_CMAKE_ARGS ${ORC_CMAKE_ARGS} "-DZLIB_HOME=${ZLIB_ROOT}")
-  endif()
+    set(ORC_CMAKE_ARGS
+        ${EP_COMMON_CMAKE_ARGS}
+        "-DCMAKE_INSTALL_PREFIX=${ORC_PREFIX}"
+        -DSTOP_BUILD_ON_WARNING=OFF
+        -DBUILD_LIBHDFSPP=OFF
+        -DBUILD_JAVA=OFF
+        -DBUILD_TOOLS=OFF
+        -DBUILD_CPP_TESTS=OFF
+        -DINSTALL_VENDORED_LIBS=OFF
+        "-DLZ4_HOME=${ORC_LZ4_ROOT}"
+        "-DPROTOBUF_EXECUTABLE=$<TARGET_FILE:${ARROW_PROTOBUF_PROTOC}>"
+        "-DPROTOBUF_HOME=${ORC_PROTOBUF_ROOT}"
+        "-DPROTOBUF_INCLUDE_DIR=$<TARGET_PROPERTY:${ARROW_PROTOBUF_LIBPROTOBUF},INTERFACE_INCLUDE_DIRECTORIES>"
+        "-DPROTOBUF_LIBRARY=$<TARGET_FILE:${ARROW_PROTOBUF_LIBPROTOBUF}>"
+        "-DPROTOC_LIBRARY=$<TARGET_FILE:${ARROW_PROTOBUF_LIBPROTOC}>"
+        "-DSNAPPY_HOME=${ORC_SNAPPY_ROOT}"
+        "-DSNAPPY_LIBRARY=$<TARGET_FILE:${Snappy_TARGET}>"
+        "-DLZ4_LIBRARY=$<TARGET_FILE:LZ4::lz4>"
+        "-DLZ4_STATIC_LIB=$<TARGET_FILE:LZ4::lz4>"
+        "-DLZ4_INCLUDE_DIR=${ORC_LZ4_ROOT}/include"
+        "-DSNAPPY_INCLUDE_DIR=${ORC_SNAPPY_INCLUDE_DIR}"
+        "-DZSTD_HOME=${ORC_ZSTD_ROOT}"
+        "-DZSTD_INCLUDE_DIR=$<TARGET_PROPERTY:${ARROW_ZSTD_LIBZSTD},INTERFACE_INCLUDE_DIRECTORIES>"
+        "-DZSTD_LIBRARY=$<TARGET_FILE:${ARROW_ZSTD_LIBZSTD}>")
+    if(ZLIB_ROOT)
+      set(ORC_CMAKE_ARGS ${ORC_CMAKE_ARGS} "-DZLIB_HOME=${ZLIB_ROOT}")
+    endif()
 
-  # Work around CMake bug
-  file(MAKE_DIRECTORY ${ORC_INCLUDE_DIR})
+    # Work around CMake bug
+    file(MAKE_DIRECTORY ${ORC_INCLUDE_DIR})
 
-  externalproject_add(orc_ep
-                      ${EP_COMMON_OPTIONS}
-                      URL ${ORC_SOURCE_URL}
-                      URL_HASH "SHA256=${ARROW_ORC_BUILD_SHA256_CHECKSUM}"
-                      BUILD_BYPRODUCTS ${ORC_STATIC_LIB}
-                      CMAKE_ARGS ${ORC_CMAKE_ARGS}
-                      DEPENDS ${ARROW_PROTOBUF_LIBPROTOBUF}
-                              ${ARROW_PROTOBUF_PROTOC}
-                              ${ARROW_ZSTD_LIBZSTD}
-                              ${Snappy_TARGET}
-                              LZ4::lz4
-                              ZLIB::ZLIB)
-
-  set(ORC_VENDORED 1)
-
-  add_library(orc::orc STATIC IMPORTED)
-  set_target_properties(orc::orc PROPERTIES IMPORTED_LOCATION "${ORC_STATIC_LIB}")
-  target_include_directories(orc::orc BEFORE INTERFACE "${ORC_INCLUDE_DIR}")
-  target_link_libraries(orc::orc INTERFACE LZ4::lz4 ZLIB::ZLIB ${ARROW_ZSTD_LIBZSTD}
-                                           ${Snappy_TARGET})
-  # Protobuf generated files may use ABSL_DCHECK*() and
-  # absl::log_internal_check_op is needed for them.
-  if(TARGET absl::log_internal_check_op)
-    target_link_libraries(orc::orc INTERFACE absl::log_internal_check_op)
-  endif()
-  if(NOT MSVC)
-    if(NOT APPLE AND ARROW_ENABLE_THREADING)
-      target_link_libraries(orc::orc INTERFACE Threads::Threads)
-    endif()
-    target_link_libraries(orc::orc INTERFACE ${CMAKE_DL_LIBS})
-  endif()
-  if(CMAKE_CXX_COMPILER_ID STREQUAL "GNU")
-    if(CMAKE_CXX_COMPILER_VERSION VERSION_LESS "9")
-      target_link_libraries(orc::orc INTERFACE stdc++fs)
+    externalproject_add(orc_ep
+                        ${EP_COMMON_OPTIONS}
+                        URL ${ORC_SOURCE_URL}
+                        URL_HASH "SHA256=${ARROW_ORC_BUILD_SHA256_CHECKSUM}"
+                        BUILD_BYPRODUCTS ${ORC_STATIC_LIB}
+                        CMAKE_ARGS ${ORC_CMAKE_ARGS}
+                        DEPENDS ${ARROW_PROTOBUF_LIBPROTOBUF}
+                                ${ARROW_PROTOBUF_PROTOC}
+                                ${ARROW_ZSTD_LIBZSTD}
+                                ${Snappy_TARGET}
+                                LZ4::lz4
+                                ZLIB::ZLIB)
+    add_library(orc::orc STATIC IMPORTED)
+    set_target_properties(orc::orc PROPERTIES IMPORTED_LOCATION "${ORC_STATIC_LIB}")
+    target_include_directories(orc::orc BEFORE INTERFACE "${ORC_INCLUDE_DIR}")
+    target_link_libraries(orc::orc INTERFACE LZ4::lz4 ZLIB::ZLIB ${ARROW_ZSTD_LIBZSTD}
+                                             ${Snappy_TARGET})
+    # Protobuf generated files may use ABSL_DCHECK*() and
+    # absl::log_internal_check_op is needed for them.
+    if(TARGET absl::log_internal_check_op)
+      target_link_libraries(orc::orc INTERFACE absl::log_internal_check_op)
     endif()
-  elseif(CMAKE_CXX_COMPILER_ID STREQUAL "Clang")
-    if(CMAKE_CXX_COMPILER_VERSION VERSION_LESS "8")
-      target_link_libraries(orc::orc INTERFACE c++fs)
+    if(NOT MSVC)
+      if(NOT APPLE AND ARROW_ENABLE_THREADING)
+        target_link_libraries(orc::orc INTERFACE Threads::Threads)
+      endif()
+      target_link_libraries(orc::orc INTERFACE ${CMAKE_DL_LIBS})
     endif()
+    target_link_libraries(orc::orc INTERFACE ${ARROW_PROTOBUF_LIBPROTOBUF})
+    add_dependencies(orc::orc orc_ep)
+    list(APPEND ARROW_BUNDLED_STATIC_LIBS orc::orc)
   endif()
 
-  add_dependencies(orc::orc orc_ep)
-
-  list(APPEND ARROW_BUNDLED_STATIC_LIBS orc::orc)
-endmacro()
+  set(ORC_VENDORED
+      TRUE
+      PARENT_SCOPE)
+  set(ARROW_BUNDLED_STATIC_LIBS
+      ${ARROW_BUNDLED_STATIC_LIBS}
+      PARENT_SCOPE)
+endfunction()
 
 if(ARROW_ORC)
   resolve_dependency(orc HAVE_ALT TRUE)
-  target_link_libraries(orc::orc INTERFACE ${ARROW_PROTOBUF_LIBPROTOBUF})
   if(ORC_VENDORED)
     set(ARROW_ORC_VERSION ${ARROW_ORC_BUILD_VERSION})
   else()
+    target_link_libraries(orc::orc INTERFACE ${ARROW_PROTOBUF_LIBPROTOBUF})
     set(ARROW_ORC_VERSION ${orcAlt_VERSION})
+    message(STATUS "Found ORC static library: ${ORC_STATIC_LIB}")
+    message(STATUS "Found ORC headers: ${ORC_INCLUDE_DIR}")
   endif()
-  message(STATUS "Found ORC static library: ${ORC_STATIC_LIB}")
-  message(STATUS "Found ORC headers: ${ORC_INCLUDE_DIR}")
 endif()
 
 # ----------------------------------------------------------------------

diff --git a/cpp/thirdparty/versions.txt b/cpp/thirdparty/versions.txt
@@ -90,8 +90,8 @@ ARROW_OPENTELEMETRY_BUILD_VERSION=v1.13.0
 ARROW_OPENTELEMETRY_BUILD_SHA256_CHECKSUM=7735cc56507149686e6019e06f588317099d4522480be5f38a2a09ec69af1706
 ARROW_OPENTELEMETRY_PROTO_BUILD_VERSION=v0.17.0
 ARROW_OPENTELEMETRY_PROTO_BUILD_SHA256_CHECKSUM=f269fbcb30e17b03caa1decd231ce826e59d7651c0f71c3b28eb5140b4bb5412
-ARROW_ORC_BUILD_VERSION=2.0.0
-ARROW_ORC_BUILD_SHA256_CHECKSUM=9107730919c29eb39efaff1b9e36166634d1d4d9477e5fee76bfd6a8fec317df
+ARROW_ORC_BUILD_VERSION=2.0.1
+ARROW_ORC_BUILD_SHA256_CHECKSUM=1ffac0228aa83f04a1b1cf2788a3af5953e82587ae3a77c41900e99f2557132d
 ARROW_PROTOBUF_BUILD_VERSION=v21.3
 ARROW_PROTOBUF_BUILD_SHA256_CHECKSUM=2f723218f6cb709ae4cdc4fb5ed56a5951fc5d466f0128ce4c946b8c78c8c49f
 # Because of https://github.com/Tencent/rapidjson/pull/1323, we require

diff --git a/dev/tasks/linux-packages/apache-arrow/debian/rules b/dev/tasks/linux-packages/apache-arrow/debian/rules
@@ -51,6 +51,7 @@ override_dh_auto_configure:
 	  -DARROW_WITH_ZSTD=ON					\
 	  -DCMAKE_BUILD_TYPE=$(BUILD_TYPE)			\
 	  -DCUDAToolkit_ROOT=/usr				\
+	  -DFETCHCONTENT_FULLY_DISCONNECTED=OFF			\
 	  -DPARQUET_BUILD_EXECUTABLES=ON			\
 	  -DPARQUET_REQUIRE_ENCRYPTION=ON