[GLUTEN-3953] [CH] Add contrib mmdb (#474)

(cherry picked from commit 83a1c9f)
(cherry picked from commit 4979b75)
(cherry picked from commit 80743ae)
(cherry picked from commit ccd8316)
(cherry picked from commit 60478da)
(cherry picked from commit 622d089)
(cherry picked from commit bf1015c)
(cherry picked from commit c368f36)
(cherry picked from commit 53e319a)
(cherry picked from commit 6e06c2a)
(cherry picked from commit 312397b)

.gitmodules

@@ -10,6 +10,9 @@
 [submodule "contrib/librdkafka"]
 	path = contrib/librdkafka
 	url = https://github.com/ClickHouse/librdkafka
+[submodule "contrib/libmaxminddb"]
+	path = contrib/libmaxminddb
+	url = https://github.com/maxmind/libmaxminddb.git	
 [submodule "contrib/cctz"]
 	path = contrib/cctz
 	url = https://github.com/ClickHouse/cctz

contrib/CMakeLists.txt

@@ -88,6 +88,8 @@ add_contrib (libarchive-cmake libarchive)
 
 add_contrib (corrosion-cmake corrosion)
 
+add_contrib (libmaxminddb-cmake libmaxminddb)
+
 if (ENABLE_FUZZING)
     add_contrib (libprotobuf-mutator-cmake libprotobuf-mutator)
 endif()

contrib/libmaxminddb-cmake/CMakeLists.txt

@@ -0,0 +1,94 @@
+option (ENABLE_MAXMINDDB "Enable maxminddb library" OFF)
+
+if (NOT ENABLE_MAXMINDDB)
+  message (STATUS "Not using maxminddb")
+  return()
+endif()
+
+set(LIBMAXMINDDB_VERSION "1.7.1")
+
+include(GNUInstallDirs)
+include(CheckTypeSize)
+check_type_size("unsigned __int128" UINT128)
+check_type_size("unsigned int __attribute__((mode(TI)))" UINT128_USING_MODE)
+if(HAVE_UINT128)
+  set(MMDB_UINT128_USING_MODE 0)
+  set(MMDB_UINT128_IS_BYTE_ARRAY 0)
+elseif(HAVE_UINT128_USING_MODE)
+  set(MMDB_UINT128_USING_MODE 1)
+  set(MMDB_UINT128_IS_BYTE_ARRAY 0)
+else()
+  set(MMDB_UINT128_USING_MODE 0)
+  set(MMDB_UINT128_IS_BYTE_ARRAY 1)
+endif()
+
+include (TestBigEndian)
+TEST_BIG_ENDIAN(IS_BIG_ENDIAN)
+
+if (${CMAKE_SYSTEM_NAME} MATCHES "Linux")
+  set(CMAKE_POSITION_INDEPENDENT_CODE ON)
+endif()
+
+set(LIBMAXMINDDB_SOURCE_DIR "${ClickHouse_SOURCE_DIR}/contrib/libmaxminddb")
+configure_file(${CMAKE_CURRENT_SOURCE_DIR}/maxminddb_config.h.in 
+               ${CMAKE_CURRENT_BINARY_DIR}/include/maxminddb_config.h)
+configure_file(
+    "${LIBMAXMINDDB_SOURCE_DIR}/include/maxminddb.h"
+    "${CMAKE_CURRENT_BINARY_DIR}/include/maxminddb.h"
+    COPYONLY
+)
+
+add_library(_maxminddb
+  ${LIBMAXMINDDB_SOURCE_DIR}/src/maxminddb.c
+  ${LIBMAXMINDDB_SOURCE_DIR}/src/data-pool.c
+)
+add_library(ch_contrib::maxminddb ALIAS _maxminddb)
+
+target_compile_definitions(_maxminddb PRIVATE PACKAGE_VERSION="${LIBMAXMINDDB_VERSION}")
+
+if(NOT IS_BIG_ENDIAN)
+  target_compile_definitions(_maxminddb PRIVATE MMDB_LITTLE_ENDIAN=1)
+endif()
+
+if(MSVC)
+  target_compile_definitions(_maxminddb PRIVATE _CRT_SECURE_NO_WARNINGS)
+endif()
+
+if(WIN32)
+  target_link_libraries(_maxminddb ws2_32)
+  if(MSVC_STATIC_RUNTIME)
+    # On MSVC, when MSVC_STATIC_RUNTIME is ON, MT (Release) and MTd (Debug)
+    # run-time libraries will be used instead of MD/MDd. The default is OFF so
+    # MD/MDd are used when nothing related is passed.
+    #
+    # Adapted from https://gitlab.kitware.com/cmake/community/-/wikis/FAQ#make-override-files
+    if(MSVC)
+      target_compile_options(my_target PRIVATE
+        $<$<CONFIG:Debug>:/D_DEBUG /MTd /Zi /Ob0 /Od /RTC1>
+        $<$<CONFIG:MinSizeRel>:/MT /O1 /Ob1 /D NDEBUG>
+        $<$<CONFIG:Release>:/MT /O2 /Ob2 /D NDEBUG>
+        $<$<CONFIG:RelWithDebInfo>:/MT /Zi /O2 /Ob1 /D NDEBUG>
+      )
+    endif()
+  endif()
+endif()
+
+target_include_directories(_maxminddb PUBLIC
+  ${LIBMAXMINDDB_SOURCE_DIR}
+  ${CMAKE_CURRENT_BINARY_DIR}/include
+)
+
+if(NOT MSVC)
+  add_executable(mmdblookup
+    ${LIBMAXMINDDB_SOURCE_DIR}/bin/mmdblookup.c
+  )
+  target_compile_definitions(mmdblookup PRIVATE PACKAGE_VERSION="${LIBMAXMINDDB_VERSION}")
+
+  # Otherwise 'undefined reference to WinMain' linker error happen due to wmain()
+  if(MINGW)
+    target_link_options(mmdblookup PRIVATE "-municode")
+  endif()
+
+  target_link_libraries(mmdblookup _maxminddb pthread)
+endif()
+

contrib/libmaxminddb-cmake/maxminddb_config.h.in

@@ -0,0 +1,8 @@
+#pragma once
+
+/* Define as 1 if we use unsigned int __atribute__ ((__mode__(TI))) for uint128 values */
+#cmakedefine01 MMDB_UINT128_USING_MODE
+
+/* Define as 1 if we don't have an unsigned __int128 type */
+#cmakedefine01 MMDB_UINT128_IS_BYTE_ARRAY
+

src/CMakeLists.txt

@@ -552,6 +552,10 @@ if (TARGET ch_contrib::rocksdb)
     dbms_target_link_libraries(PUBLIC ch_contrib::rocksdb)
 endif()
 
+if (TARGET ch_contrib::maxminddb)
+    dbms_target_link_libraries(PUBLIC ch_contrib::maxminddb)
+endif()
+
 if (TARGET ch_contrib::libpqxx)
     dbms_target_link_libraries(PUBLIC ch_contrib::libpqxx)
 endif()

src/Common/config.h.in

@@ -65,6 +65,7 @@
 #cmakedefine01 USE_BCRYPT
 #cmakedefine01 USE_LIBARCHIVE
 #cmakedefine01 USE_POCKETFFT
+#cmakedefine01 USE_MAXMINDDB
 
 /// This is needed for .incbin in assembly. For some reason, include paths don't work there in presence of LTO.
 /// That's why we use absolute paths.

src/configure_config.cmake

@@ -176,5 +176,8 @@ endif()
 if (TARGET ch_contrib::pocketfft)
     set(USE_POCKETFFT 1)
 endif()
+if (TARGET ch_contrib::maxminddb)
+    set(USE_MAXMINDDB 1)
+endif()
 
 set(SOURCE_DIR ${PROJECT_SOURCE_DIR})