Skip to content

Commit a2bac3b

Browse files
committed
Lint
1 parent 0e5aa00 commit a2bac3b

File tree

3 files changed

+57
-33
lines changed

3 files changed

+57
-33
lines changed

CMakeLists.txt

Lines changed: 34 additions & 21 deletions
Original file line numberDiff line numberDiff line change
@@ -19,7 +19,8 @@ project(Tokenizers)
1919
option(TOKENIZERS_BUILD_TEST "Build tests" OFF)
2020
option(TOKENIZERS_BUILD_TOOLS "Build tools" OFF)
2121
option(SUPPORT_REGEX_LOOKAHEAD
22-
"Support regex lookahead patterns (requires PCRE2)" OFF)
22+
"Support regex lookahead patterns (requires PCRE2)" OFF
23+
)
2324

2425
# Include CMakePackageConfigHelpers for configure_package_config_file
2526
include(CMakePackageConfigHelpers)
@@ -50,22 +51,27 @@ set(tokenizers_source_files
5051
${CMAKE_CURRENT_SOURCE_DIR}/src/regex.cpp
5152
${CMAKE_CURRENT_SOURCE_DIR}/src/sentencepiece.cpp
5253
${CMAKE_CURRENT_SOURCE_DIR}/src/tiktoken.cpp
53-
${CMAKE_CURRENT_SOURCE_DIR}/src/token_decoder.cpp)
54+
${CMAKE_CURRENT_SOURCE_DIR}/src/token_decoder.cpp
55+
)
5456

5557
file(GLOB unicode_source_files
56-
${CMAKE_CURRENT_SOURCE_DIR}/third-party/llama.cpp-unicode/src/*.cpp)
57-
add_library(tokenizers STATIC ${tokenizers_source_files}
58-
${unicode_source_files})
58+
${CMAKE_CURRENT_SOURCE_DIR}/third-party/llama.cpp-unicode/src/*.cpp
59+
)
60+
add_library(
61+
tokenizers STATIC ${tokenizers_source_files} ${unicode_source_files}
62+
)
5963

6064
# Using abseil from sentencepiece/third_party
6165
target_include_directories(
6266
tokenizers
63-
PUBLIC $<BUILD_INTERFACE:${CMAKE_CURRENT_SOURCE_DIR}/include>
64-
$<BUILD_INTERFACE:${CMAKE_CURRENT_SOURCE_DIR}/third-party/sentencepiece>
65-
$<BUILD_INTERFACE:${CMAKE_CURRENT_SOURCE_DIR}/third-party/sentencepiece/src>
66-
$<BUILD_INTERFACE:${CMAKE_CURRENT_SOURCE_DIR}/third-party/re2>
67-
$<BUILD_INTERFACE:${CMAKE_CURRENT_SOURCE_DIR}/third-party/json/single_include>
68-
$<BUILD_INTERFACE:${CMAKE_CURRENT_SOURCE_DIR}/third-party/llama.cpp-unicode/include>)
67+
PUBLIC
68+
$<BUILD_INTERFACE:${CMAKE_CURRENT_SOURCE_DIR}/include>
69+
$<BUILD_INTERFACE:${CMAKE_CURRENT_SOURCE_DIR}/third-party/sentencepiece>
70+
$<BUILD_INTERFACE:${CMAKE_CURRENT_SOURCE_DIR}/third-party/sentencepiece/src>
71+
$<BUILD_INTERFACE:${CMAKE_CURRENT_SOURCE_DIR}/third-party/re2>
72+
$<BUILD_INTERFACE:${CMAKE_CURRENT_SOURCE_DIR}/third-party/json/single_include>
73+
$<BUILD_INTERFACE:${CMAKE_CURRENT_SOURCE_DIR}/third-party/llama.cpp-unicode/include>
74+
)
6975
target_link_libraries(tokenizers PUBLIC sentencepiece-static re2::re2)
7076

7177
if(SUPPORT_REGEX_LOOKAHEAD)
@@ -82,18 +88,24 @@ if(SUPPORT_REGEX_LOOKAHEAD)
8288
add_subdirectory(${CMAKE_CURRENT_SOURCE_DIR}/third-party/pcre2)
8389

8490
# Set the INTERFACE_INCLUDE_DIRECTORIES property for pcre2-8-static
85-
set_target_properties(pcre2-8-static PROPERTIES
86-
INTERFACE_INCLUDE_DIRECTORIES $<BUILD_INTERFACE:${CMAKE_CURRENT_SOURCE_DIR}/third-party/pcre2/src>
91+
set_target_properties(
92+
pcre2-8-static
93+
PROPERTIES
94+
INTERFACE_INCLUDE_DIRECTORIES
95+
$<BUILD_INTERFACE:${CMAKE_CURRENT_SOURCE_DIR}/third-party/pcre2/src>
8796
)
8897
add_library(
8998
regex_lookahead STATIC
9099
${CMAKE_CURRENT_SOURCE_DIR}/src/pcre2_regex.cpp
91100
${CMAKE_CURRENT_SOURCE_DIR}/src/regex_lookahead.cpp
92-
${CMAKE_CURRENT_SOURCE_DIR}/src/std_regex.cpp)
101+
${CMAKE_CURRENT_SOURCE_DIR}/src/std_regex.cpp
102+
)
93103
target_link_libraries(regex_lookahead PUBLIC pcre2-8)
94104
target_include_directories(
95-
regex_lookahead PUBLIC $<BUILD_INTERFACE:${CMAKE_CURRENT_SOURCE_DIR}/include>
96-
$<BUILD_INTERFACE:${CMAKE_CURRENT_SOURCE_DIR}/third-party/pcre2/src>)
105+
regex_lookahead
106+
PUBLIC $<BUILD_INTERFACE:${CMAKE_CURRENT_SOURCE_DIR}/include>
107+
$<BUILD_INTERFACE:${CMAKE_CURRENT_SOURCE_DIR}/third-party/pcre2/src>
108+
)
97109
target_link_options_shared_lib(regex_lookahead)
98110
target_link_libraries(tokenizers PUBLIC regex_lookahead)
99111
install(
@@ -126,13 +138,15 @@ install(
126138
install(
127139
DIRECTORY ${CMAKE_CURRENT_SOURCE_DIR}/include/
128140
DESTINATION ${CMAKE_INSTALL_INCLUDEDIR}
129-
FILES_MATCHING PATTERN "*.h"
141+
FILES_MATCHING
142+
PATTERN "*.h"
130143
)
131144

132145
install(
133146
DIRECTORY ${CMAKE_CURRENT_SOURCE_DIR}/third-party/json/single_include/
134147
DESTINATION ${CMAKE_INSTALL_INCLUDEDIR}
135-
FILES_MATCHING PATTERN "*.hpp"
148+
FILES_MATCHING
149+
PATTERN "*.hpp"
136150
)
137151

138152
# Install the CMake config files
@@ -151,7 +165,6 @@ configure_package_config_file(
151165
PATH_VARS CMAKE_INSTALL_INCLUDEDIR
152166
)
153167

154-
install(
155-
FILES ${CMAKE_CURRENT_BINARY_DIR}/tokenizers-config.cmake
156-
DESTINATION ${CMAKE_INSTALL_LIBDIR}/cmake/tokenizers
168+
install(FILES ${CMAKE_CURRENT_BINARY_DIR}/tokenizers-config.cmake
169+
DESTINATION ${CMAKE_INSTALL_LIBDIR}/cmake/tokenizers
157170
)

cmake/tokenizers-config.cmake.in

Lines changed: 18 additions & 5 deletions
Original file line numberDiff line numberDiff line change
@@ -14,7 +14,10 @@ find_dependency(absl REQUIRED)
1414
# Directly include sentencepiece library
1515
set(SENTENCEPIECE_LIBRARY "${CMAKE_INSTALL_PREFIX}/lib64/libsentencepiece.a")
1616
if(NOT EXISTS "${SENTENCEPIECE_LIBRARY}")
17-
message(FATAL_ERROR "Could not find sentencepiece library at ${SENTENCEPIECE_LIBRARY}")
17+
message(
18+
FATAL_ERROR
19+
"Could not find sentencepiece library at ${SENTENCEPIECE_LIBRARY}"
20+
)
1821
endif()
1922

2023
# Include the exported targets file
@@ -25,16 +28,26 @@ set_and_check(TOKENIZERS_INCLUDE_DIRS "@PACKAGE_CMAKE_INSTALL_INCLUDEDIR@")
2528

2629
# Add --whole-archive linker flag for tokenizers library
2730
if(APPLE)
28-
set(TOKENIZERS_LINK_OPTIONS "SHELL:-force_load,$<TARGET_FILE:tokenizers::regex_lookahead>")
31+
set(TOKENIZERS_LINK_OPTIONS
32+
"SHELL:-force_load,$<TARGET_FILE:tokenizers::regex_lookahead>"
33+
)
2934
elseif(MSVC)
30-
set(TOKENIZERS_LINK_OPTIONS "SHELL:/WHOLEARCHIVE:$<TARGET_FILE:tokenizers::regex_lookahead>")
35+
set(TOKENIZERS_LINK_OPTIONS
36+
"SHELL:/WHOLEARCHIVE:$<TARGET_FILE:tokenizers::regex_lookahead>"
37+
)
3138
else()
32-
set(TOKENIZERS_LINK_OPTIONS "SHELL:LINKER:--whole-archive $<TARGET_FILE:tokenizers::regex_lookahead> LINKER:--no-whole-archive")
39+
set(TOKENIZERS_LINK_OPTIONS
40+
"SHELL:LINKER:--whole-archive $<TARGET_FILE:tokenizers::regex_lookahead> LINKER:--no-whole-archive"
41+
)
3342
endif()
3443

3544
# Set the libraries and link options
3645
set(TOKENIZERS_LIBRARIES tokenizers::tokenizers)
37-
set_property(TARGET tokenizers::tokenizers APPEND PROPERTY INTERFACE_LINK_OPTIONS "${TOKENIZERS_LINK_OPTIONS}")
46+
set_property(
47+
TARGET tokenizers::tokenizers
48+
APPEND
49+
PROPERTY INTERFACE_LINK_OPTIONS "${TOKENIZERS_LINK_OPTIONS}"
50+
)
3851

3952
# Check if the library was found
4053
check_required_components(tokenizers)

test/CMakeLists.txt

Lines changed: 5 additions & 7 deletions
Original file line numberDiff line numberDiff line change
@@ -13,9 +13,7 @@ include(FindPackageHandleStandardArgs)
1313
include(FetchContent)
1414
include(ExternalProject)
1515
FetchContent_Declare(
16-
tokenizers
17-
SOURCE_DIR ${CMAKE_CURRENT_SOURCE_DIR}/..
18-
BUILD_ALWAYS ON
16+
tokenizers SOURCE_DIR ${CMAKE_CURRENT_SOURCE_DIR}/.. BUILD_ALWAYS ON
1917
)
2018
set(SUPPORT_REGEX_LOOKAHEAD ON)
2119
FetchContent_MakeAvailable(tokenizers)
@@ -29,7 +27,8 @@ FetchContent_Declare(
2927
)
3028
set(gtest_force_shared_crt
3129
ON
32-
CACHE BOOL "" FORCE)
30+
CACHE BOOL "" FORCE
31+
)
3332
FetchContent_MakeAvailable(googletest)
3433

3534
file(GLOB test_source_files test_*.cpp)
@@ -41,9 +40,8 @@ foreach(test_source_file ${test_source_files})
4140
message(STATUS "Configuring unit test ${test_name}")
4241
add_executable(${test_name} ${test_source_file})
4342
target_include_directories(
44-
${test_name}
45-
PRIVATE GTEST_INCLUDE_PATH
46-
${TOKENIZERS_INCLUDE_DIRS})
43+
${test_name} PRIVATE GTEST_INCLUDE_PATH ${TOKENIZERS_INCLUDE_DIRS}
44+
)
4745
target_link_libraries(${test_name} PUBLIC gtest_main GTest::gmock tokenizers)
4846
add_test(${test_name} "${test_name}")
4947
set_tests_properties(${test_name} PROPERTIES ENVIRONMENT ${test_env})

0 commit comments

Comments
 (0)