Files
2026-08-20 08:01:06 +08:00

237 lines
10 KiB
CMake

cmake_minimum_required(VERSION 3.26)
cmake_policy(SET CMP0077 NEW)
project(zvec)
set(CC_CXX_STANDARD 17)
if(MSVC)
set(INTTYPES_FORMAT VC7)
add_compile_options(/utf-8)
# Use /Z7 (embed debug info in .obj) instead of /Zi (separate .pdb) to enable sccache/ccache.
# Use CACHE FORCE so it propagates into thirdparty subdirectories that call project().
foreach(_cfg DEBUG RELWITHDEBINFO)
foreach(_lang C CXX)
string(REPLACE "/Zi" "/Z7" _flags "${CMAKE_${_lang}_FLAGS_${_cfg}}")
set(CMAKE_${_lang}_FLAGS_${_cfg} "${_flags}" CACHE STRING "" FORCE)
endforeach()
endforeach()
add_compile_options(/EHsc) # def c++ exception behavior
add_compile_options(/Zc:preprocessor /Zc:__cplusplus)
add_compile_options(/we4716) # -Werror=return-type
###### reduce output length to make vibe coding work better :) should be removed or solved later
# TODO(windows): fix&cleanup
add_compile_options(/wd4267 /wd4244 /wd4305 /wd4242) # conversion/truncation from 'size_t' to 'type', possible loss of data
add_compile_options(/wd4310) # cast truncates constant value
add_compile_options(/wd4146) # unary minus operator applied to unsigned type #usage: uint32_t seg_id_{-1U};
add_compile_options(/wd4245) # conversion from 'type1' to 'type2', signed/unsigned mismatch
add_compile_options(/wd4251) # class 'type' needs to have dll-interface to be used by clients of class 'type2'
add_compile_options(/wd4099) # type name first seen using 'object_type1' now seen using 'object_type2'
add_compile_options(/wd4200) # nonstandard extension used : zero-sized array in struct/union
add_compile_options(/wd4324) # structure was padded due to alignment specifier
add_compile_options(/wd4702) # unreachable code
add_compile_options(/wd4996) # wstring_convert are deprecated in C++17 # usage: in antlr4
add_compile_options(/wd4101) # unreferenced local variable
add_compile_options(/wd4996) # The POSIX name for this item is deprecated. Instead, use the ISO C and C++ conformant name: _lseek
# result of 32-bit shift implicitly converted to 64 bits (was 64-bit shift intended?) # usage: keeps |= (1 << k);
add_compile_options(/wd4334)
######
else()
set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -Wall -Werror=return-type")
set(CMAKE_C_FLAGS "${CMAKE_C_FLAGS} -Wall -Werror=return-type")
# Fine-grained dead code elimination: place each function/data in its own section
set(CMAKE_C_FLAGS "${CMAKE_C_FLAGS} -ffunction-sections -fdata-sections")
set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -ffunction-sections -fdata-sections")
endif()
if(CMAKE_CXX_COMPILER_ID STREQUAL "GNU" AND NOT IOS)
set(CMAKE_EXE_LINKER_FLAGS "${CMAKE_EXE_LINKER_FLAGS} -Wl,--no-as-needed")
set(CMAKE_SHARED_LINKER_FLAGS "${CMAKE_SHARED_LINKER_FLAGS} -Wl,--no-as-needed")
endif()
# Detect musl libc. Some code paths need to avoid large stack buffers on musl
# because its default thread stack is much smaller than glibc's (128KB vs 8MB).
# Detection: musl's ldd prints "musl libc" to stderr.
set(ZVEC_ON_MUSL FALSE CACHE BOOL "Building on musl libc" FORCE)
if(CMAKE_SYSTEM_NAME STREQUAL "Linux")
execute_process(
COMMAND ldd --version
OUTPUT_VARIABLE _zvec_ldd_stdout
ERROR_VARIABLE _zvec_ldd_stderr
)
if("${_zvec_ldd_stdout}${_zvec_ldd_stderr}" MATCHES "musl")
set(ZVEC_ON_MUSL TRUE CACHE BOOL "Building on musl libc" FORCE)
message(STATUS "musl libc detected")
endif()
endif()
if(NOT DEFINED PROJECT_ROOT_DIR OR NOT PROJECT_ROOT_DIR)
set(PROJECT_ROOT_DIR ${CMAKE_CURRENT_SOURCE_DIR} CACHE PATH "Root directory of the project" FORCE)
endif()
# LTO (Link-Time Optimization) — helps on Linux with --gc-sections but may
# increase size on macOS due to aggressive inlining. Default OFF; enable
# explicitly with -DZVEC_ENABLE_LTO=ON when targeting Linux.
option(ZVEC_ENABLE_LTO "Enable Link-Time Optimization for Release builds" OFF)
if(ZVEC_ENABLE_LTO AND CMAKE_BUILD_TYPE STREQUAL "Release")
include(CheckIPOSupported)
check_ipo_supported(RESULT IPO_SUPPORTED OUTPUT IPO_OUTPUT)
if(IPO_SUPPORTED)
set(CMAKE_INTERPROCEDURAL_OPTIMIZATION ON)
message(STATUS "LTO enabled for Release build")
else()
message(STATUS "LTO not supported: ${IPO_OUTPUT}")
endif()
endif()
message(STATUS "PROJECT_ROOT_DIR = ${PROJECT_ROOT_DIR}")
include(${PROJECT_ROOT_DIR}/cmake/bazel.cmake)
include(${PROJECT_ROOT_DIR}/cmake/option.cmake)
include(${PROJECT_ROOT_DIR}/cmake/version.cmake)
# iOS platform detection
if(NOT ANDROID AND NOT IOS AND CMAKE_SYSTEM_NAME STREQUAL "iOS")
set(IOS TRUE)
endif()
# iOS bundle properties for test executables
if(IOS)
set(MACOSX_BUNDLE_BUNDLE_VERSION "1")
set(MACOSX_BUNDLE_SHORT_VERSION_STRING "1.0")
set(CMAKE_MACOSX_BUNDLE_INFO_PLIST "${PROJECT_ROOT_DIR}/cmake/iOSBundleInfo.plist.in")
endif()
if (NOT ANDROID AND NOT IOS AND AUTO_DETECT_ARCH AND HOST_ARCH MATCHES "^(x86|x64)$")
setup_compiler_march_for_x86(MATH_MARCH_FLAG_SSE MATH_MARCH_FLAG_AVX2 MATH_MARCH_FLAG_AVX512 MATH_MARCH_FLAG_AVX512FP16)
message(STATUS "best compiler march, sse: " ${MATH_MARCH_FLAG_SSE} ", avx2: " ${MATH_MARCH_FLAG_AVX2} ", avx512: " ${MATH_MARCH_FLAG_AVX512} ", avx512fp16: " ${MATH_MARCH_FLAG_AVX512FP16})
endif()
include_directories(${PROJECT_ROOT_DIR}/src/include)
include_directories(${PROJECT_ROOT_DIR}/src)
option(BUILD_ZVEC_SHARED "Build all-in-one C++ shared library libzvec" ON)
option(BUILD_ZVEC_AILEGO_SHARED "Build all-in-one zvec-ailego shared library libzvec_ailego" ON)
option(BUILD_ZVEC_CORE_SHARED "Build all-in-one zvec-core shared library libzvec_core" ON)
option(BUILD_PYTHON_BINDINGS "Build Python bindings using pybind11" OFF)
option(BUILD_C_BINDINGS "Build C bindings" ON)
option(BUILD_TOOLS "Build tools" ON)
message(STATUS "BUILD_ZVEC_SHARED:${BUILD_ZVEC_SHARED}")
message(STATUS "BUILD_ZVEC_AILEGO_SHARED:${BUILD_ZVEC_AILEGO_SHARED}")
message(STATUS "BUILD_ZVEC_CORE_SHARED:${BUILD_ZVEC_CORE_SHARED}")
message(STATUS "BUILD_PYTHON_BINDINGS:${BUILD_PYTHON_BINDINGS}")
message(STATUS "BUILD_C_BINDINGS:${BUILD_C_BINDINGS}")
message(STATUS "BUILD_TOOLS:${BUILD_TOOLS}")
if(CMAKE_SYSTEM_NAME STREQUAL "Linux" AND CMAKE_SYSTEM_PROCESSOR MATCHES "x86_64|amd64|AMD64" AND NOT ANDROID AND NOT IOS)
include(CheckCXXCompilerFlag)
check_cxx_compiler_flag("-mavx2 -mfma" COMPILER_SUPPORTS_RABITQ_AVX2)
check_cxx_compiler_flag(
"-mavx512f -mavx512bw -mavx512dq -mfma"
COMPILER_SUPPORTS_RABITQ_AVX512)
check_cxx_compiler_flag(
"-mavx512f -mavx512bw -mavx512dq -mavx512vpopcntdq -mfma"
COMPILER_SUPPORTS_RABITQ_AVX512_POPCNT)
if(COMPILER_SUPPORTS_RABITQ_AVX2 AND
COMPILER_SUPPORTS_RABITQ_AVX512 AND
COMPILER_SUPPORTS_RABITQ_AVX512_POPCNT)
set(RABITQ_SUPPORTED ON)
add_definitions(-DRABITQ_SUPPORTED=1)
else()
set(RABITQ_SUPPORTED OFF)
add_definitions(-DRABITQ_SUPPORTED=0)
message(STATUS "RaBitQ support disabled - compiler cannot build all runtime dispatch variants")
endif()
elseif(IOS)
set(RABITQ_SUPPORTED OFF)
add_definitions(-DRABITQ_SUPPORTED=0)
message(STATUS "RaBitQ support disabled - not supported on iOS")
else()
set(RABITQ_SUPPORTED OFF)
add_definitions(-DRABITQ_SUPPORTED=0)
message(STATUS "RaBitQ support disabled - only supported on Linux x86_64")
endif()
message(STATUS "RABITQ_SUPPORTED: ${RABITQ_SUPPORTED}")
# DiskAnn support:
# - Linux x86_64 and ARM64 with io_uring, libaio, or pread
# - macOS ARM64 (Apple Silicon) with synchronous pread
if((CMAKE_SYSTEM_NAME STREQUAL "Linux" AND CMAKE_SYSTEM_PROCESSOR MATCHES "^(x86_64|amd64|AMD64|aarch64|arm64)$" AND NOT ANDROID AND NOT IOS)
OR (CMAKE_SYSTEM_NAME STREQUAL "Darwin" AND CMAKE_SYSTEM_PROCESSOR MATCHES "^(arm64|aarch64)$" AND NOT IOS))
set(DISKANN_SUPPORTED ON)
add_definitions(-DDISKANN_SUPPORTED=1)
else()
set(DISKANN_SUPPORTED OFF)
add_definitions(-DDISKANN_SUPPORTED=0)
message(STATUS "DiskAnn support disabled for ${CMAKE_SYSTEM_NAME}/${CMAKE_SYSTEM_PROCESSOR}")
endif()
message(STATUS "DISKANN_SUPPORTED: ${DISKANN_SUPPORTED}")
option(USE_OSS_MIRROR "Use OSS mirror for faster third-party downloads" OFF)
if(DEFINED ENV{USE_OSS_MIRROR} AND NOT "$ENV{USE_OSS_MIRROR}" STREQUAL "")
set(USE_OSS_MIRROR "$ENV{USE_OSS_MIRROR}" CACHE BOOL "Use OSS mirror for faster third-party downloads" FORCE)
endif()
message(STATUS "USE_OSS_MIRROR:${USE_OSS_MIRROR}")
cc_directory(thirdparty)
cc_directories(src)
cc_directories(tests)
add_custom_target(clang_tidy_deps DEPENDS ARROW.BUILD glog gflags Lz4.BUILD)
if(BUILD_TOOLS)
cc_directories(tools)
endif()
set(CPACK_PACKAGE_VERSION ${ZVEC_VERSION_NUMBER})
set(CPACK_PACKAGE_VERSION_MAJOR ${ZVEC_VERSION_MAJOR})
set(CPACK_PACKAGE_VERSION_MINOR ${ZVEC_VERSION_MINOR})
set(CPACK_PACKAGE_VERSION_PATCH ${ZVEC_VERSION_PATCH})
set(CPACK_PACKAGE_NAME zvec)
include(CPack)
if(BUILD_PYTHON_BINDINGS)
if(APPLE)
set(CMAKE_STRIP "")
message(STATUS "Disabled strip on macOS to preserve code signature")
endif()
include(GNUInstallDirs)
if(DEFINED SKBUILD_PLATLIB_DIR)
set(ZVEC_PY_INSTALL_DIR "${SKBUILD_PLATLIB_DIR}")
elseif(DEFINED Python_SITEARCH)
set(ZVEC_PY_INSTALL_DIR "${Python_SITEARCH}")
else()
set(ZVEC_PY_INSTALL_DIR "${CMAKE_INSTALL_LIBDIR}")
endif()
message(STATUS "Zvec install path: ${ZVEC_PY_INSTALL_DIR}")
# COMPONENT python: only these runtime artifacts are pulled into the wheel
# (see install.components in pyproject.toml). The cc_library(PACKED) SDK
# install rules use the default component and are excluded from the wheel.
# Install the extension inside the zvec package (zvec/_zvec*.so) rather than
# at the site-packages root, so it does not pollute the top-level namespace.
# The Python code imports it as `zvec._zvec` accordingly.
install(TARGETS _zvec LIBRARY DESTINATION ${ZVEC_PY_INSTALL_DIR}/zvec
COMPONENT python)
# DiskAnn is now statically linked into _zvec.so via --whole-archive
# (see src/binding/python/CMakeLists.txt -> core_knn_diskann_static),
# so no separate runtime .so needs to be installed in the wheel.
# Bundle cppjieba's dictionary files so the `jieba` FTS tokenizer works
# out of the box. python/zvec/__init__.py resolves this directory via
# importlib.resources and registers it with set_default_jieba_dict_dir().
set(ZVEC_JIEBA_DICT_SRC
"${PROJECT_SOURCE_DIR}/thirdparty/cppjieba/cppjieba-5.6.7/dict")
install(FILES
"${ZVEC_JIEBA_DICT_SRC}/jieba.dict.utf8"
"${ZVEC_JIEBA_DICT_SRC}/hmm_model.utf8"
DESTINATION ${ZVEC_PY_INSTALL_DIR}/zvec/data/jieba_dict
COMPONENT python)
endif()