[vcpkg baseline][orc] Update, fix, cleanup (#49121)

ports/orc/external-project.diff

@@ -0,0 +1,31 @@
+diff --git a/CMakeLists.txt b/CMakeLists.txt
+index 341a4c3..1aaff3e 100644
+--- a/CMakeLists.txt
++++ b/CMakeLists.txt
+@@ -112,6 +112,13 @@ SET(CPACK_PACKAGE_CONTACT "Apache ORC <dev@orc.apache.org>")
+ 
+ INCLUDE(CPack)
+ INCLUDE(ExternalProject)
++function(ExternalProject_Add name)
++  z_vcpkg_function_arguments(args 1)
++  if(NOT "${name}" MATCHES "^orc-format")
++    message(FATAL_ERROR "Unexpected external project: ${name}")
++  endif()
++  _ExternalProject_Add(${name} "${args}")
++endfunction()
+ 
+ if (BUILD_POSITION_INDEPENDENT_LIB)
+   set(CMAKE_POSITION_INDEPENDENT_CODE ON)
+diff --git a/cmake_modules/ThirdpartyToolchain.cmake b/cmake_modules/ThirdpartyToolchain.cmake
+index 851cdc4..a23115e 100644
+--- a/cmake_modules/ThirdpartyToolchain.cmake
++++ b/cmake_modules/ThirdpartyToolchain.cmake
+@@ -397,7 +397,7 @@ add_library (orc::lz4 ALIAS orc_lz4)
+ # ----------------------------------------------------------------------
+ # IANA - Time Zone Database
+ 
+-if (WIN32)
++if (WIN32 AND BUILD_CPP_TESTS)
+   SET(CURRENT_TZDATA_FILE "")
+   SET(CURRENT_TZDATA_SHA512 "")
+   File(DOWNLOAD "https://ftp.osuosl.org/pub/cygwin/noarch/release/tzdata/sha512.sum" ${CMAKE_CURRENT_BINARY_DIR}/sha512.sum)

ports/orc/portfile.cmake

@@ -4,69 +4,54 @@ vcpkg_from_github(
   OUT_SOURCE_PATH SOURCE_PATH
   REPO apache/orc
   REF "v${VERSION}"
-  SHA512 eabee16a6e984452a8cb715d0524041b20dd1bd88d78bb32534db93e5dbdd786aa4df8c05975406cb0728241eb3025a506c4fefb8c334ef0d8a27e6cb920d44c
+  SHA512 b81e1c2cebfd97757aaaa8dc9bed913810b0a47e599556bc3e9989dadf6b7665c335c37cd29d8430cd9b1af9048f37efaa56869872eb5f50fe35a570ec109a40
   HEAD_REF master
+  PATCHES
+    external-project.diff
+    tools-build.diff
+)
+file(GLOB modules "${SOURCE_PATH}/cmake_modules/Find*.cmake")
+file(REMOVE ${modules} "${SOURCE_PATH}/c++/libs/libhdfspp/libhdfspp.tar.gz")
+
+set(orc_format_version 1.1.1)
+vcpkg_download_distfile(ORC_FORMAT_ARCHIVE
+  URLS "https://dlcdn.apache.org/orc/orc-format-${orc_format_version}/orc-format-${orc_format_version}.tar.gz"
+  FILENAME "apache-orc-format-${orc_format_version}.tar.gz"
+  SHA512 8aa0bcd3345ed8be836995d4347175526f4b0fc91f41e27f29279fad39b94ff157f5cd597bc2d9f3dc403f5ba405807675a283abe822f8a83758b7c3b8292c1c
+)
+set(ENV{ORC_FORMAT_URL} "file://${ORC_FORMAT_ARCHIVE}")
+
+vcpkg_check_features(OUT_FEATURE_OPTIONS options
+  FEATURES
+    tools   BUILD_TOOLS
 )
 
-file(REMOVE "${SOURCE_PATH}/cmake_modules/FindGTest.cmake")
-file(REMOVE "${SOURCE_PATH}/cmake_modules/FindLZ4.cmake")
-file(REMOVE "${SOURCE_PATH}/cmake_modules/FindZSTD.cmake")
-file(REMOVE "${SOURCE_PATH}/cmake_modules/FindProtobuf.cmake")
-file(REMOVE "${SOURCE_PATH}/cmake_modules/FindSnappy.cmake")
-file(REMOVE "${SOURCE_PATH}/cmake_modules/FindZLIB.cmake")
-
 if(VCPKG_TARGET_IS_WINDOWS)
-  set(BUILD_TOOLS OFF)
-  # when cross compiling, we can't run their test. however:
-  #  - Windows doesn't support time_t < 0 => HAS_PRE_1970 test returns false
-  #  - Windows doesn't support setenv => HAS_POST_2038 test fails to compile
-  set(time_t_checks "-DHAS_PRE_1970=OFF" "-DHAS_POST_2038=OFF")
-else()
-  set(BUILD_TOOLS ON)
-  set(time_t_checks "")
-endif()
-
-if(VCPKG_TARGET_IS_UWP)
-    set(configure_opts WINDOWS_USE_MSBUILD)
+  list(APPEND options "-DHAS_PRE_1970=OFF" "-DHAS_POST_2038=OFF")
 endif()
 
 vcpkg_cmake_configure(
   SOURCE_PATH "${SOURCE_PATH}"
-  ${configure_opts}
   OPTIONS
-    ${time_t_checks}
-    -DBUILD_TOOLS=${BUILD_TOOLS}
+    ${options}
     -DBUILD_CPP_TESTS=OFF
     -DBUILD_JAVA=OFF
     -DINSTALL_VENDORED_LIBS=OFF
-    -DBUILD_LIBHDFSPP=OFF
-    -DSTOP_BUILD_ON_WARNING=OFF
-    -DENABLE_TEST=OFF
     -DORC_PACKAGE_KIND=vcpkg
-  MAYBE_UNUSED_VARIABLES
-    ENABLE_TEST
+    -DSTOP_BUILD_ON_WARNING=OFF
+  OPTIONS_DEBUG
+    -DBUILD_TOOLS=OFF
 )
-
 vcpkg_cmake_install()
 vcpkg_copy_pdbs()
+vcpkg_cmake_config_fixup()
 
-file(GLOB TOOLS ${CURRENT_PACKAGES_DIR}/bin/orc-*)
-if(TOOLS)
-  file(COPY ${TOOLS} DESTINATION "${CURRENT_PACKAGES_DIR}/tools/orc")
-  file(REMOVE ${TOOLS})
-endif()
-
-file(GLOB BINS "${CURRENT_PACKAGES_DIR}/bin/*")
-if(NOT BINS)
-  file(REMOVE_RECURSE "${CURRENT_PACKAGES_DIR}/bin" "${CURRENT_PACKAGES_DIR}/debug/bin")
-endif()
-
-if(VCPKG_LIBRARY_LINKAGE STREQUAL "static")
-    file(REMOVE_RECURSE "${CURRENT_PACKAGES_DIR}/bin" "${CURRENT_PACKAGES_DIR}/debug/bin")
+if("tools" IN_LIST FEATURES)
+  vcpkg_copy_tools(TOOL_NAMES csv-import orc-contents orc-memory orc-metadata orc-scan orc-statistics timezone-dump AUTO_CLEAN)
 endif()
 
 file(REMOVE_RECURSE "${CURRENT_PACKAGES_DIR}/debug/include")
 file(REMOVE_RECURSE "${CURRENT_PACKAGES_DIR}/debug/share")
+file(REMOVE_RECURSE "${CURRENT_PACKAGES_DIR}/share/doc")
 
-vcpkg_install_copyright(FILE_LIST "${SOURCE_PATH}/LICENSE")
-file(COPY "${CMAKE_CURRENT_LIST_DIR}/usage" DESTINATION "${CURRENT_PACKAGES_DIR}/share/${PORT}")
+vcpkg_install_copyright(FILE_LIST "${SOURCE_PATH}/NOTICE" "${SOURCE_PATH}/LICENSE")

ports/orc/tools-build.diff

@@ -0,0 +1,66 @@
+diff --git a/tools/src/CMakeLists.txt b/tools/src/CMakeLists.txt
+index d247f90..498d369 100644
+--- a/tools/src/CMakeLists.txt
++++ b/tools/src/CMakeLists.txt
+@@ -36,8 +36,12 @@
+ 
+ set (CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -g ${CXX17_FLAGS} ${WARN_FLAGS}")
+ 
+-add_library (orc-tools-common INTERFACE)
+-target_include_directories (orc-tools-common INTERFACE
++add_library (orc-tools-common STATIC ToolsHelper.cc)
++if(WIN32 AND NOT MINGW)
++  find_package(getopt CONFIG REQUIRED)
++  target_link_libraries(orc-tools-common PUBLIC $<IF:$<TARGET_EXISTS:getopt::getopt_shared>,getopt::getopt_shared,getopt::getopt_static>)
++endif()
++target_include_directories (orc-tools-common PUBLIC
+   ${PROJECT_BINARY_DIR}/c++/include
+   ${PROJECT_BINARY_DIR}/c++/src
+   ${PROJECT_SOURCE_DIR}/c++/include
+@@ -50,7 +54,6 @@ target_link_libraries (orc-tools-common INTERFACE
+ 
+ add_executable (orc-contents
+   FileContents.cc
+-  ToolsHelper.cc
+   )
+ 
+ target_link_libraries (orc-contents
+@@ -59,7 +62,6 @@ target_link_libraries (orc-contents
+ 
+ add_executable (orc-scan
+   FileScan.cc
+-  ToolsHelper.cc
+   )
+ 
+ target_link_libraries (orc-scan
+@@ -68,7 +70,6 @@ target_link_libraries (orc-scan
+ 
+ add_executable (orc-metadata
+   FileMetadata.cc
+-  ToolsHelper.cc
+   )
+ 
+ target_link_libraries (orc-metadata
+@@ -86,7 +87,6 @@ target_link_libraries (orc-statistics
+ 
+ add_executable (orc-memory
+   FileMemory.cc
+-  ToolsHelper.cc
+   )
+ 
+ target_link_libraries (orc-memory
+@@ -108,6 +108,14 @@ add_executable (csv-import
+ target_link_libraries (csv-import
+   orc-tools-common
+   )
++if(WIN32 AND NOT MINGW)
++  find_path(GETTIMEOFDAY_INCLUDE_DIR gettimeofday.h REQUIRED)
++  target_include_directories(csv-import PRIVATE "${GETTIMEOFDAY_INCLUDE_DIR}")
++  file(MAKE_DIRECTORY "${PROJECT_BINARY_DIR}/c++/include/sys")
++  file(WRITE "${PROJECT_BINARY_DIR}/c++/include/sys/time.h" "extern \"C\" {\n#include <gettimeofday.h>\n}\n")
++  find_library(GETTIMEOFDAY_LIBRARY gettimeofday PATH_SUFFIXES manual-link REQUIRED)
++  target_link_libraries(csv-import "${GETTIMEOFDAY_LIBRARY}")
++endif()
+ 
+ set(CPP_TOOL_NAMES
+   orc-contents

ports/orc/usage

@@ -1,4 +0,0 @@
-orc provides CMake targets:
-
-    find_package(orc CONFIG REQUIRED)
-    target_link_libraries(main PRIVATE orc::orc)

ports/orc/vcpkg.json

@@ -1,6 +1,6 @@
 {
   "name": "orc",
-  "version": "2.1.0",
+  "version": "2.2.1",
   "description": "The smallest, fastest columnar storage for Hadoop workloads.",
   "homepage": "https://orc.apache.org/",
   "license": "Apache-2.0",
@@ -23,5 +23,20 @@
     },
     "zlib",
     "zstd"
-  ]
+  ],
+  "features": {
+    "tools": {
+      "description": "Build tools",
+      "dependencies": [
+        {
+          "name": "getopt-win32",
+          "platform": "windows & !mingw"
+        },
+        {
+          "name": "gettimeofday",
+          "platform": "windows & !mingw"
+        }
+      ]
+    }
+  }
 }

scripts/ci.feature.baseline.txt

@@ -199,6 +199,7 @@ openzl:x64-windows-release=fail
 openzl:x64-windows-static-md=fail
 openzl:x64-windows-static=fail
 openzl:x64-windows=fail
+orc(android)=fail  # try_run for time_t properties
 paho-mqtt(uwp)=fail
 paraview(arm64 & windows & !uwp)=fail # in-build host tools
 pfring:arm64-osx=fail

versions/baseline.json

@@ -7369,7 +7369,7 @@
       "port-version": 0
     },
     "orc": {
-      "baseline": "2.1.0",
+      "baseline": "2.2.1",
       "port-version": 0
     },
     "orefkov-simstr": {

versions/o-/orc.json

@@ -1,5 +1,10 @@
 {
   "versions": [
+    {
+      "git-tree": "8257ee1040f79b5efc737713fbd6b524518e0845",
+      "version": "2.2.1",
+      "port-version": 0
+    },
     {
       "git-tree": "0be0ad4bd945bf213667aa0d682b91b498073eb0",
       "version": "2.1.0",