Compare commits

..

No commits in common. "master" and "v0.5.0" have entirely different histories.

29 changed files with 283 additions and 380 deletions

View file

@ -22,9 +22,9 @@ build_script:
- mkdir build
- cd build
- cmake --version
- cmake %CMAKE_ARGS% ..
- cmake %CMAKE_ARGS% --parallel ..
- cmake ..
- cmake --build . --config %Configuration% --verbose
- cmake --build . --config %Configuration% --verbose --parallel
for:
-
@ -33,7 +33,7 @@ for:
- job_name: VS2019ARM
test_script:
- ctest --output-on-failure -C %Configuration% --verbose %CTEST_ARGS%
- ctest --output-on-failure -C %Configuration% --verbose %CTEST_ARGS% --parallel
clone_folder: c:\projects\fast_double_parser

View file

@ -1,6 +0,0 @@
FROM gcr.io/oss-fuzz-base/base-builder
RUN apt-get update && apt-get install -y make autoconf automake libtool
COPY . $SRC/fast_double_parser
COPY .clusterfuzzlite/build.sh $SRC/build.sh
WORKDIR $SRC/fast_double_parser

View file

@ -1,3 +0,0 @@
# ClusterFuzzLite set up
This folder contains a fuzzing set for [ClusterFuzzLite](https://google.github.io/clusterfuzzlite).

View file

@ -1,7 +0,0 @@
#!/bin/bash -eu
# Copy all fuzzer executable to $OUT/
$CXX $CFLAGS $LIB_FUZZING_ENGINE \
$SRC/fast_double_parser/.clusterfuzzlite/parse_number_fuzzer.cpp \
-o $OUT/parse_number_fuzzer \
-I$SRC/fast_double_parser/include

View file

@ -1,11 +0,0 @@
#include <fast_double_parser.h>
#include <string>
extern "C" int LLVMFuzzerTestOneInput(const uint8_t *data, size_t size) {
std::string fuzz_input(reinterpret_cast<const char *>(data), size);
double x;
fast_double_parser::parse_number(fuzz_input.c_str(), &x);
return 0;
}

View file

@ -1 +0,0 @@
language: c++

View file

@ -1,30 +0,0 @@
name: ClusterFuzzLite PR fuzzing
on:
workflow_dispatch:
pull_request:
branches: [ main ]
permissions: read-all
jobs:
PR:
runs-on: ubuntu-latest
strategy:
fail-fast: false
matrix:
sanitizer: [address]
steps:
- name: Build Fuzzers (${{ matrix.sanitizer }})
id: build
uses: google/clusterfuzzlite/actions/build_fuzzers@v1
with:
sanitizer: ${{ matrix.sanitizer }}
language: c++
bad-build-check: false
- name: Run Fuzzers (${{ matrix.sanitizer }})
id: run
uses: google/clusterfuzzlite/actions/run_fuzzers@v1
with:
github-token: ${{ secrets.GITHUB_TOKEN }}
fuzz-seconds: 100
mode: 'code-change'
report-unreproducible-crashes: false
sanitizer: ${{ matrix.sanitizer }}

View file

@ -1,41 +0,0 @@
name: "CodeQL"
on:
push:
branches: [ "master" ]
pull_request:
branches: [ "master" ]
schedule:
- cron: "1 8 * * 3"
jobs:
analyze:
name: Analyze
runs-on: ubuntu-latest
permissions:
actions: read
contents: read
security-events: write
strategy:
fail-fast: false
matrix:
language: [ cpp ]
steps:
- name: Checkout
uses: actions/checkout@v3
- name: Initialize CodeQL
uses: github/codeql-action/init@v2
with:
languages: ${{ matrix.language }}
queries: +security-and-quality
- name: Autobuild
uses: github/codeql-action/autobuild@v2
- name: Perform CodeQL Analysis
uses: github/codeql-action/analyze@v2
with:
category: "/language:${{ matrix.language }}"

49
.github/workflows/mingw-ci.yml vendored Normal file
View file

@ -0,0 +1,49 @@
name: MinGW32-CI
on: [push, pull_request]
# Important: scoop will either install 32-bit GCC or 64-bit GCC, not both.
# It is important to build static libraries because cmake is not smart enough under Windows/mingw to take care of the path. So
# with a dynamic library, you could get failures due to the fact that the EXE can't find its DLL.
jobs:
ci:
name: windows-gcc
runs-on: windows-2016
env:
CMAKE_GENERATOR: Ninja
CC: gcc
CXX: g++
steps: # To reproduce what is below, start a powershell with administrative rights, using scoop *is* a good idea
- uses: actions/checkout@v2
- uses: actions/cache@v2 # we cache the scoop setup with 32-bit GCC
id: cache
with:
path: |
C:\ProgramData\scoop
key: scoop32 # static key: should be good forever
- name: Setup Windows # This should almost never run if the cache works.
if: steps.cache.outputs.cache-hit != 'true'
shell: powershell
run: |
Invoke-Expression (New-Object System.Net.WebClient).DownloadString('https://get.scoop.sh')
scoop install sudo --global
sudo scoop install git --global
sudo scoop install ninja --global
sudo scoop install cmake --global
sudo scoop install gcc --arch 32bit --global
$env:path
Write-Host 'Everything has been installed, you are good!'
- name: Build and Test 32-bit x86
shell: powershell
run: |
$ENV:PATH="C:\ProgramData\scoop\shims;C:\ProgramData\scoop\apps\gcc\current\bin;C:\ProgramData\scoop\apps\ninja\current;$ENV:PATH"
mkdir build32
cd build32
cmake ..
cmake --build . --verbose
ctest -j4 --output-on-failure

49
.github/workflows/mingw64-ci.yml vendored Normal file
View file

@ -0,0 +1,49 @@
name: MinGW64-CI
on: [push, pull_request]
# Important: scoop will either install 32-bit GCC or 64-bit GCC, not both.
# It is important to build static libraries because cmake is not smart enough under Windows/mingw to take care of the path. So
# with a dynamic library, you could get failures due to the fact that the EXE can't find its DLL.
jobs:
ci:
name: windows-gcc
runs-on: windows-2016
env:
CMAKE_GENERATOR: Ninja
CC: gcc
CXX: g++
steps: # To reproduce what is below, start a powershell with administrative rights, using scoop *is* a good idea
- uses: actions/checkout@v2
- uses: actions/cache@v2 # we cache the scoop setup with 64-bit GCC
id: cache
with:
path: |
C:\ProgramData\scoop
key: scoop64 # static key: should be good forever
- name: Setup Windows # This should almost never run if the cache works.
if: steps.cache.outputs.cache-hit != 'true'
shell: powershell
run: |
Invoke-Expression (New-Object System.Net.WebClient).DownloadString('https://get.scoop.sh')
scoop install sudo --global
sudo scoop install git --global
sudo scoop install ninja --global
sudo scoop install cmake --global
sudo scoop install gcc --arch 64bit --global
$env:path
Write-Host 'Everything has been installed, you are good!'
- name: Build and Test 64-bit x64
shell: powershell
run: |
$ENV:PATH="C:\ProgramData\scoop\shims;C:\ProgramData\scoop\apps\gcc\current\bin;C:\ProgramData\scoop\apps\ninja\current;$ENV:PATH"
mkdir build64
cd build64
cmake ..
cmake --build . --verbose
ctest -j4 --output-on-failure

View file

@ -29,7 +29,7 @@ jobs:
CMAKE_GENERATOR: Ninja
steps:
- uses: actions/checkout@v3
- uses: actions/checkout@v2
- uses: msys2/setup-msys2@v2
with:
update: true

View file

@ -29,7 +29,7 @@ jobs:
CMAKE_GENERATOR: Ninja
steps:
- uses: actions/checkout@v3
- uses: actions/checkout@v2
- uses: msys2/setup-msys2@v2
with:
update: true

View file

@ -7,7 +7,7 @@ jobs:
ubuntu-build:
runs-on: ubuntu-latest
steps:
- uses: actions/checkout@v3
- uses: actions/checkout@v2
- name: start docker
run: |
docker run -w /src -dit --name rhub -v $PWD:/src rhub/rocker-gcc-san

20
.github/workflows/ubuntu18.yml vendored Normal file
View file

@ -0,0 +1,20 @@
name: Ubuntu 18.04 CI (GCC 7)
on: [push, pull_request]
jobs:
ubuntu-build:
runs-on: ubuntu-18.04
steps:
- uses: actions/checkout@v2
- name: Setup cmake
uses: jwlawson/actions-setup-cmake@v1.4
with:
cmake-version: '3.9.x'
- name: Use cmake
run: |
mkdir build &&
cd build &&
cmake -DFAST_DOUBLE_BENCHMARKS=ON .. &&
cmake --build . &&
ctest -j --output-on-failure

View file

@ -1,15 +1,16 @@
name: Ubuntu 22.04
name: Ubuntu 20.04 CI (GCC 9)
on: [push, pull_request]
jobs:
ubuntu-build:
runs-on: ubuntu-22.04
strategy:
matrix:
cxx: [g++-12, clang++-14]
runs-on: ubuntu-20.04
steps:
- uses: actions/checkout@0ad4b8fadaa221de15dcec353f45205ec38ea70b # v4.1.4
- uses: actions/checkout@v2
- name: Setup cmake
uses: jwlawson/actions-setup-cmake@v1.4
with:
cmake-version: '3.9.x'
- name: Use cmake
run: |
mkdir build &&

24
.github/workflows/vs16-ci.yml vendored Normal file
View file

@ -0,0 +1,24 @@
name: VS16-CI
on: [push, pull_request]
jobs:
ci:
name: windows-vs16
runs-on: windows-latest
steps:
- uses: actions/checkout@v2
- name: 'Run CMake with VS16'
uses: lukka/run-cmake@v2
with:
cmakeListsOrSettingsJson: CMakeListsTxtAdvanced
cmakeListsTxtPath: '${{ github.workspace }}/CMakeLists.txt'
buildDirectory: "${{ github.workspace }}/../../_temp/windows"
cmakeBuildType: Release
buildWithCMake: true
cmakeGenerator: VS16Win64
buildWithCMakeArgs: --config Release
- name: 'Run CTest'
run: ctest -C Release --output-on-failure
working-directory: "${{ github.workspace }}/../../_temp/windows"

25
.github/workflows/vs16-clang-ci.yml vendored Normal file
View file

@ -0,0 +1,25 @@
name: VS16-CLANG-CI
on: [push, pull_request]
jobs:
ci:
name: windows-vs16
runs-on: windows-latest
steps:
- uses: actions/checkout@v2
- name: 'Run CMake with VS16'
uses: lukka/run-cmake@v2
with:
cmakeListsOrSettingsJson: CMakeListsTxtAdvanced
cmakeListsTxtPath: '${{ github.workspace }}/CMakeLists.txt'
buildDirectory: "${{ github.workspace }}/../../_temp/windows"
cmakeBuildType: Release
buildWithCMake: true
cmakeGenerator: VS16Win64
cmakeAppendedArgs: -T ClangCL
buildWithCMakeArgs: --config Release
- name: 'Run CTest'
run: ctest -C Release --output-on-failure
working-directory: "${{ github.workspace }}/../../_temp/windows"

25
.github/workflows/vs16-ninja-ci.yml vendored Normal file
View file

@ -0,0 +1,25 @@
name: VS16-Ninja-CI
on: [push, pull_request]
jobs:
ci:
name: windows-vs16
runs-on: windows-latest
steps:
- uses: actions/checkout@v2
- name: 'Run CMake with VS16'
uses: lukka/run-cmake@v2
with:
cmakeListsOrSettingsJson: CMakeListsTxtAdvanced
cmakeListsTxtPath: '${{ github.workspace }}/CMakeLists.txt'
buildDirectory: "${{ github.workspace }}/../../_temp/windows"
cmakeBuildType: Release
buildWithCMake: true
cmakeGenerator: VS16Win64
cmakeAppendedArgs: -G Ninja
buildWithCMakeArgs: --config Release
- name: 'Run CTest'
run: ctest -C Release --output-on-failure
working-directory: "${{ github.workspace }}/../../_temp/windows"

View file

@ -1,37 +0,0 @@
name: VS17-CI
on: [push, pull_request]
jobs:
ci:
if: >-
! contains(toJSON(github.event.commits.*.message), '[skip ci]') &&
! contains(toJSON(github.event.commits.*.message), '[skip github]')
name: windows-vs17
runs-on: windows-latest
strategy:
fail-fast: false
matrix:
include:
- {gen: Visual Studio 17 2022, arch: Win32}
- {gen: Visual Studio 17 2022, arch: Win32}
- {gen: Visual Studio 17 2022, arch: x64}
- {gen: Visual Studio 17 2022, arch: x64}
steps:
- name: checkout
uses: actions/checkout@v3
- name: Configure
run: |
cmake -G "${{matrix.gen}}" -A ${{matrix.arch}} -B build
- name: Build Debug
run: cmake --build build --config Debug --verbose
- name: Build Release
run: cmake --build build --config Release --verbose
- name: Run Release tests
run: |
cd build
ctest -C Release --output-on-failure
- name: Run Debug tests
run: |
cd build
ctest -C Debug --output-on-failure

View file

@ -1,37 +0,0 @@
name: VS17-CLANG-CI
on: [push, pull_request]
jobs:
ci:
if: >-
! contains(toJSON(github.event.commits.*.message), '[skip ci]') &&
! contains(toJSON(github.event.commits.*.message), '[skip github]')
name: windows-vs17
runs-on: windows-latest
strategy:
fail-fast: false
matrix:
include:
- {gen: Visual Studio 17 2022, arch: Win32}
- {gen: Visual Studio 17 2022, arch: Win32}
- {gen: Visual Studio 17 2022, arch: x64}
- {gen: Visual Studio 17 2022, arch: x64}
steps:
- name: checkout
uses: actions/checkout@v3
- name: Configure
run: |
cmake -G "${{matrix.gen}}" -A ${{matrix.arch}} -T ClangCL -B build
- name: Build Debug
run: cmake --build build --config Debug --verbose
- name: Build Release
run: cmake --build build --config Release --verbose
- name: Run Release tests
run: |
cd build
ctest -C Release --output-on-failure
- name: Run Debug tests
run: |
cd build
ctest -C Debug --output-on-failure

4
.gitignore vendored
View file

@ -44,7 +44,3 @@
/fast_double_parser.cxxflags
/fast_double_parser.files
/fast_double_parser.includes
# Visual Studio
/.vs
/out

View file

@ -1,22 +0,0 @@
cc_library(
name = "fast_double_parser",
hdrs = ["include/fast_double_parser.h"],
strip_include_prefix = "include",
visibility = ["//visibility:public"],
)
cc_test(
name = "unit",
srcs = ["tests/unit.cpp"],
deps = [":fast_double_parser"],
)
cc_binary(
name = "benchmark",
srcs = ["benchmarks/benchmark.cpp"],
deps = [
":fast_double_parser",
"@abseil-cpp//absl/strings",
"@double-conversion",
],
)

View file

@ -1,7 +1,6 @@
cmake_minimum_required(VERSION 3.11)
cmake_policy(SET CMP0048 NEW)
project(fast_double_parser LANGUAGES CXX VERSION 0.0.0.0)
cmake_minimum_required(VERSION 3.9)
set(CMAKE_CXX_STANDARD 11)
set(CMAKE_CXX_STANDARD_REQUIRED ON)
if (NOT CMAKE_BUILD_TYPE)
@ -20,82 +19,53 @@ set(benchmark_src benchmarks/benchmark.cpp)
add_library(fast_double_parser INTERFACE)
target_include_directories(fast_double_parser
INTERFACE
$<BUILD_INTERFACE:${CMAKE_CURRENT_LIST_DIR}/include>
$<INSTALL_INTERFACE:include>
)
target_include_directories(fast_double_parser INTERFACE include/)
include(GNUInstallDirs)
install(FILES ${headers} DESTINATION "${CMAKE_INSTALL_INCLUDEDIR}")
install(TARGETS fast_double_parser EXPORT fast_double_parser-targets)
install(
EXPORT fast_double_parser-targets
DESTINATION "share/fast_double_parser"
NAMESPACE fast_double_parser::
)
include(CMakePackageConfigHelpers)
configure_package_config_file(
"${CMAKE_CURRENT_SOURCE_DIR}/fast_double_parser-config.cmake.in"
"${CMAKE_CURRENT_BINARY_DIR}/fast_double_parser-config.cmake"
INSTALL_DESTINATION "share/fast_double_parser"
)
install(
FILES "${CMAKE_CURRENT_BINARY_DIR}/fast_double_parser-config.cmake"
DESTINATION "share/fast_double_parser"
)
option(BUILD_TESTING "Build unit tests" ON)
if(BUILD_TESTING)
add_executable(unit ${unit_src} ${bogus_src} ${rebogus_src})
if(FAST_DOUBLE_PARSER_SANITIZE)
target_compile_options(unit PUBLIC -fsanitize=address -fno-omit-frame-pointer -fsanitize=undefined -fno-sanitize-recover=all)
target_link_options(unit PUBLIC -fsanitize=address -fno-omit-frame-pointer -fsanitize=undefined -fno-sanitize-recover=all)
# Ubuntu bug for GCC 5.0+ (safe for all versions)
if (CMAKE_COMPILER_IS_GNUCC AND NOT APPLE)
target_link_libraries(unit PUBLIC -fuse-ld=gold)
endif()
endif()
target_link_libraries(unit PRIVATE fast_double_parser)
enable_testing()
add_test(unit unit)
add_executable(unit ${unit_src} ${bogus_src} ${rebogus_src})
if(FAST_DOUBLE_PARSER_SANITIZE)
target_compile_options(unit PUBLIC -fsanitize=address -fno-omit-frame-pointer -fsanitize=undefined -fno-sanitize-recover=all)
target_link_options(unit PUBLIC -fsanitize=address -fno-omit-frame-pointer -fsanitize=undefined -fno-sanitize-recover=all)
# Ubuntu bug for GCC 5.0+ (safe for all versions)
if (CMAKE_COMPILER_IS_GNUCC)
target_link_libraries(unit PUBLIC -fuse-ld=gold)
endif()
endif()
option(FAST_DOUBLE_BENCHMARKS "include benchmarks" OFF)
target_link_libraries(unit PUBLIC fast_double_parser)
enable_testing()
add_test(unit unit)
option(FAST_DOUBLE_BENCHMARKS "Sanitize addresses" OFF)
function(initialize_submodule DIRECTORY)
if(NOT EXISTS ${CMAKE_CURRENT_SOURCE_DIR}/${DIRECTORY}/.git)
find_package(Git QUIET REQUIRED)
message(STATUS "${CMAKE_CURRENT_SOURCE_DIR}/${DIRECTORY}/.git does not exist. Initializing ${DIRECTORY} submodule ...")
execute_process(COMMAND ${GIT_EXECUTABLE} submodule update --init ${CMAKE_CURRENT_SOURCE_DIR}/${DIRECTORY}
WORKING_DIRECTORY ${CMAKE_SOURCE_DIR}
RESULT_VARIABLE GIT_EXIT_CODE)
if(NOT GIT_EXIT_CODE EQUAL "0")
message(FATAL_ERROR "${GIT_EXECUTABLE} submodule update --init dependencies/${DIRECTORY} failed with exit code ${GIT_EXIT_CODE}, please checkout submodules")
endif()
endif()
endfunction(initialize_submodule)
if(FAST_DOUBLE_BENCHMARKS)
initialize_submodule(benchmarks/dependencies/abseil-cpp)
initialize_submodule(benchmarks/dependencies/double-conversion)
include(FetchContent)
include(ExternalProject)
add_subdirectory(benchmarks/dependencies/abseil-cpp)
add_subdirectory(benchmarks/dependencies/double-conversion)
set(ABSL_ENABLE_INSTALL ON)
set(ABSL_RUN_TEST OFF CACHE INTERNAL "")
set(ABSL_USE_GOOGLETEST_HEAD OFF CACHE INTERNAL "")
FetchContent_Declare(abseil
GIT_REPOSITORY https://github.com/abseil/abseil-cpp.git
GIT_TAG "20210324.2")
FetchContent_GetProperties(abseil)
if(NOT abseil_POPULATED)
set(BUILD_TESTING OFF)
FetchContent_Populate(abseil)
add_subdirectory(${abseil_SOURCE_DIR} ${abseil_BINARY_DIR})
set(CMAKE_MODULE_PATH ${CMAKE_MODULE_PATH} ${abseil_SOURCE_DIR}/absl/copts)
include(${abseil_SOURCE_DIR}/absl/copts/AbseilConfigureCopts.cmake)
endif()
FetchContent_Declare(doubleconversion
GIT_REPOSITORY https://github.com/google/double-conversion.git
GIT_TAG "v3.1.5")
FetchContent_GetProperties(doubleconversion)
FetchContent_MakeAvailable(doubleconversion)
add_executable(benchmark ${benchmark_src})
target_link_libraries(benchmark PUBLIC double-conversion absl::strings)
target_include_directories(benchmark PUBLIC include)
endif(FAST_DOUBLE_BENCHMARKS)
add_executable(benchmark ${benchmark_src})
target_link_libraries(benchmark PUBLIC double-conversion absl_strings)
target_include_directories(benchmark PUBLIC include)
endif(FAST_DOUBLE_BENCHMARKS)

View file

@ -1,10 +0,0 @@
"""fast_double_parser: 4x faster than strtod."""
module(
name = "fast_double_parser",
version = "0.8.0",
compatibility_level = 0,
)
bazel_dep(name = "abseil-cpp", version = "20240722.0", dev_dependency = True)
bazel_dep(name = "double-conversion", version = "3.3.0", dev_dependency = True)

View file

@ -1,20 +1,17 @@
# fast_double_parser: 4x faster than strtod
![MSYS2-CI](https://github.com/lemire/fast_double_parser/workflows/MSYS2-CI/badge.svg)[![Ubuntu 22.04](https://github.com/lemire/fast_double_parser/actions/workflows/ubuntu.yml/badge.svg)](https://github.com/lemire/fast_double_parser/actions/workflows/ubuntu.yml)
Unless you need support for [RFC 7159](https://tools.ietf.org/html/rfc7159) (JSON standard), we encourage users to adopt [fast_float](https://github.com/fastfloat/fast_float) library instead. It has more functionality.
[![Build Status](https://cloud.drone.io/api/badges/lemire/fast_double_parser/status.svg)](https://cloud.drone.io/lemire/fast_double_parser) [![Build status](https://ci.appveyor.com/api/projects/status/y7215jgem4ggswnj/branch/master?svg=true)](https://ci.appveyor.com/project/lemire/fast-double-parser/branch/master)![VS16-CI](https://github.com/lemire/fast_double_parser/workflows/VS16-CI/badge.svg)![Ubuntu 18.04 CI (GCC 7)](https://github.com/lemire/fast_double_parser/workflows/Ubuntu%2018.04%20CI%20(GCC%207)/badge.svg)![VS16-Ninja-CI](https://github.com/lemire/fast_double_parser/workflows/VS16-Ninja-CI/badge.svg)![MSYS2-CI](https://github.com/lemire/fast_double_parser/workflows/MSYS2-CI/badge.svg)![VS16-CLANG-CI](https://github.com/lemire/fast_double_parser/workflows/VS16-CLANG-CI/badge.svg)![MinGW32-CI](https://github.com/lemire/fast_double_parser/workflows/MinGW32-CI/badge.svg)![MinGW64-CI](https://github.com/lemire/fast_double_parser/workflows/MinGW64-CI/badge.svg)![Ubuntu 20.04 CI (GCC 9)](https://github.com/lemire/fast_double_parser/workflows/Ubuntu%2020.04%20CI%20(GCC%209)/badge.svg)[![Build Status](https://api.cirrus-ci.com/github/lemire/fast_double_parser.svg)](https://cirrus-ci.com/github/lemire/fast_double_parser)
Fast function to parse ASCII strings containing decimal numbers into double-precision (binary64) floating-point values. That is, given the string "1.0e10", it should return a 64-bit floating-point value equal to 10000000000. We do not sacrifice accuracy. The function will match exactly (down the smallest bit) the result of a standard function like `strtod`.
Fast function to parse strings containing decimal numbers into double-precision (binary64) floating-point values. That is, given the string "1.0e10", it should return a 64-bit floating-point value equal to 10000000000. We do not sacrifice accuracy. The function will match exactly (down the smallest bit) the result of a standard function like `strtod`.
We support all major compilers: Visual Studio, GNU GCC, LLVM Clang. We require C++11.
The core of this library was ported to Go by Nigel Tao and is now a standard float-parsing routine in Go (`strconv.ParseFloat`).
The core of this library was ported to Go by Nigel Tao and is now a standard float-parsing routine in Go (`strconv.ParseFloat`).
## Reference
- Daniel Lemire, [Number Parsing at a Gigabyte per Second](https://arxiv.org/abs/2101.11408), Software: Practice and Experience 51 (8), 2021.
- Daniel Lemire, [Number Parsing at a Gigabyte per Second](https://arxiv.org/abs/2101.11408), arXiv:2101.11408
## Usage
@ -30,13 +27,13 @@ The current API is simple enough:
double x;
char * string = ...
const char * endptr = fast_double_parser::parse_number(string, &x);
bool isok = fast_double_parser::parse_number(string, &x);
```
You must check the value returned (`endptr`): if it is `nullptr`, then the function refused to parse the input.
You must check the value returned (`isok`): if it is `nullptr`, then the function refused to parse the input.
Otherwise, we return a pointer (`const char *`) to the end of the parsed string. The provided
pointer (`string`) should point at the beginning of the number: if you must skip whitespace characters,
it is your responsibility to do so.
it is your responsability to do so.
We expect string numbers to follow [RFC 7159](https://tools.ietf.org/html/rfc7159) (JSON standard). In particular,
@ -47,13 +44,10 @@ It works much like the C standard function `strtod` expect that the parsing is l
1/2 even if you are under a French system. Locale independence is by design (it is not a limitation). Like the standard C functions, it expects that the string
representation of your number ends with a non-number character (e.g., a null character, a space, a colon, etc.). If you wish the specify the end point of the string, as is common in C++, please consider the [fast_float](https://github.com/lemire/fast_float) C++ library instead.
We assume that the rounding mode is set to nearest, the default setting (`std::fegetround() == FE_TONEAREST`). It is uncommon to have a different setting.
## What if I prefer another API?
The [fast_float](https://github.com/lemire/fast_float) offers an API resembling that of the C++17 `std::from_chars` functions. In particular, you can specify the beginning and the end of the string.
Furthermore [fast_float](https://github.com/lemire/fast_float) supports both 32-bit and 64-bit floating-point numbers. The [fast_float](https://github.com/lemire/fast_float) library is part of Apache Arrow, GCC 12, Safari/WebKit and other important systems.
Furthermore [fast_float](https://github.com/lemire/fast_float) supports both 32-bit and 64-bit floating-point numbers. The [fast_float](https://github.com/lemire/fast_float) library is part of Apache Arrow.
## Why should I expect this function to be faster?
@ -160,7 +154,6 @@ double-conv 243.90 MB/s
## Ports and users
- The algorithm is part of the [LLVM standard libraries](https://github.com/llvm/llvm-project/commit/87c016078ad72c46505461e4ff8bfa04819fe7ba).
- This library has been ported to Go and integrated in the Go standard library.
- It is part of the [Microsoft LightGBM machine-learning framework](https://github.com/microsoft/LightGBM).
- It is part of the database engine [noisepage](https://github.com/cmu-db/noisepage).
@ -169,7 +162,6 @@ double-conv 243.90 MB/s
- [There is a Rust port](https://github.com/ezrosent/frawk/tree/master/src/runtime/float_parse).
- [There is a Java port](https://github.com/wrandelshofer/FastDoubleParser).
- [There is a C# port](https://github.com/CarlVerret/csFastFloat).
- [Bazel Central Registry](https://registry.bazel.build/modules/fast_double_parser).
## Credit

View file

@ -20,7 +20,7 @@
#include <vector>
#include "double-conversion/ieee.h"
#include "double-conversion/double-conversion.h"
#include "double-conversion/string-to-double.h"
double findmax_fast_double_parser(const std::vector<std::string>& s) {
double answer = 0;

View file

@ -1,3 +0,0 @@
@PACKAGE_INIT@
include("${CMAKE_CURRENT_LIST_DIR}/fast_double_parser-targets.cmake")

View file

@ -190,26 +190,28 @@ really_inline value128 full_multiplication(uint64_t value1, uint64_t value2) {
return answer;
}
/* result might be undefined when input_num is zero */
inline int leading_zeroes(uint64_t input_num) {
#ifdef _MSC_VER
unsigned long leading_zero = 0;
// Search the mask data from most significant bit (MSB)
// to least significant bit (LSB) for a set bit (1).
#ifdef _WIN64
if (_BitScanReverse64(&leading_zero, input_num))
return (int)(63 - leading_zero);
#else
if (_BitScanReverse(&leading_zero, (uint32_t)(input_num >> 32)))
return (int)(63 - (leading_zero + 32));
if (_BitScanReverse(&leading_zero, (uint32_t)input_num))
return (int)(63 - leading_zero);
#endif // _WIN64
else
return 64;
#else
return __builtin_clzll(input_num);
#endif // _MSC_VER
}
// Precomputed powers of ten from 10^0 to 10^22. These
// can be represented exactly using the double type.
static const double power_of_ten[] = {
1e0, 1e1, 1e2, 1e3, 1e4, 1e5, 1e6, 1e7, 1e8, 1e9, 1e10, 1e11,
1e12, 1e13, 1e14, 1e15, 1e16, 1e17, 1e18, 1e19, 1e20, 1e21, 1e22};
static inline bool is_integer(char c) {
return (c >= '0' && c <= '9');
// this gets compiled to (uint8_t)(c - '0') <= 9 on all decent compilers
@ -227,28 +229,13 @@ static inline bool is_integer(char c) {
* affect the binary significand.
*/
// Attempts to compute i * 10^(power) exactly; and if "negative" is
// true, negate the result.
// This function will only work in some cases, when it does not work, success is
// set to false. This should work *most of the time* (like 99% of the time).
// We assume that power is in the [FASTFLOAT_SMALLEST_POWER,
// FASTFLOAT_LARGEST_POWER] interval: the caller is responsible for this check.
really_inline double compute_float_64(int64_t power, uint64_t i, bool negative,
bool *success) {
// Precomputed powers of ten from 10^0 to 10^22. These
// can be represented exactly using the double type.
static const double power_of_ten[] = {
1e0, 1e1, 1e2, 1e3, 1e4, 1e5, 1e6, 1e7, 1e8, 1e9, 1e10, 1e11,
1e12, 1e13, 1e14, 1e15, 1e16, 1e17, 1e18, 1e19, 1e20, 1e21, 1e22};
// The mantissas of powers of ten from -308 to 308, extended out to sixty four
// bits. The array contains the powers of ten approximated
// as a 64-bit mantissa. It goes from 10^FASTFLOAT_SMALLEST_POWER to
// 10^FASTFLOAT_LARGEST_POWER (inclusively).
// The mantissa is truncated, and
// never rounded up. Uses about 5KB.
static const uint64_t mantissa_64[] = {
// The mantissas of powers of ten from -308 to 308, extended out to sixty four
// bits. The array contains the powers of ten approximated
// as a 64-bit mantissa. It goes from 10^FASTFLOAT_SMALLEST_POWER to
// 10^FASTFLOAT_LARGEST_POWER (inclusively).
// The mantissa is truncated, and
// never rounded up. Uses about 5KB.
static const uint64_t mantissa_64[] = {
0xa5ced43b7e3e9188, 0xcf42894a5dce35ea,
0x818995ce7aa0e1b2, 0xa1ebfb4219491a1f,
0xca66fa129f9b60a6, 0xfd00b897478238d0,
@ -566,10 +553,10 @@ really_inline double compute_float_64(int64_t power, uint64_t i, bool negative,
0xbaa718e68396cffd, 0xe950df20247c83fd,
0x91d28b7416cdd27e, 0xb6472e511c81471d,
0xe3d8f9e563a198e5, 0x8e679c2f5e44ff8f};
// A complement to mantissa_64
// complete to a 128-bit mantissa.
// Uses about 5KB but is rarely accessed.
static const uint64_t mantissa_128[] = {
// A complement to mantissa_64
// complete to a 128-bit mantissa.
// Uses about 5KB but is rarely accessed.
const uint64_t mantissa_128[] = {
0x419ea3bd35385e2d, 0x52064cac828675b9,
0x7343efebd1940993, 0x1014ebe6c5f90bf8,
0xd41a26e077774ef6, 0x8920b098955522b4,
@ -888,6 +875,16 @@ really_inline double compute_float_64(int64_t power, uint64_t i, bool negative,
0x4cdc331d57fa5441, 0xe0133fe4adf8e952,
0x58180fddd97723a6, 0x570f09eaa7ea7648,};
// Attempts to compute i * 10^(power) exactly; and if "negative" is
// true, negate the result.
// This function will only work in some cases, when it does not work, success is
// set to false. This should work *most of the time* (like 99% of the time).
// We assume that power is in the [FASTFLOAT_SMALLEST_POWER,
// FASTFLOAT_LARGEST_POWER] interval: the caller is responsible for this check.
really_inline double compute_float_64(int64_t power, uint64_t i, bool negative,
bool *success) {
// we start with a fast path
// It was described in
// Clinger WD. How to read floating point numbers accurately.
@ -1104,8 +1101,6 @@ static const char * parse_float_strtod(const char *ptr, double *outDouble) {
#elif defined(_WIN32)
static _locale_t c_locale = _create_locale(LC_ALL, "C");
*outDouble = _strtod_l(ptr, &endptr, c_locale);
#elif defined(__PASE__)
*outDouble = strtod(ptr, &endptr);
#else
static locale_t c_locale = newlocale(LC_ALL_MASK, "C", NULL);
*outDouble = strtod_l(ptr, &endptr, c_locale);

View file

@ -226,16 +226,6 @@ void issue32() {
std::cout << "zero maps to zero" << std::endl;
}
void negative_subsubnormal_to_negative_zero() {
std::string a = "-1e-999";
double x;
bool ok = fast_double_parser::parse_number(a.c_str(), &x);
if(!ok) throw std::runtime_error("could not parse -1e-999");
if(!std::signbit(x)) throw std::runtime_error("-1e-999 signbit is positive");
if(x != -0) throw std::runtime_error("-1e-999 is not -0");
std::cout << "-1e-999 parses to -0" << std::endl;
}
void issue50_fastpath() {
std::string a = "-0.0";
double x;
@ -245,7 +235,6 @@ void issue50_fastpath() {
std::cout << "-0.0 signbit is negative" << std::endl;
}
void issue50_off_fastpath() {
std::string a = "-0.0e-22";
double x;
@ -274,27 +263,6 @@ void issue23_2() {
std::cout << "can parse 5e0012" << std::endl;
}
void issue63() {
std::string a = "1-4-abc";
double x;
const char * ok = fast_double_parser::parse_number(a.c_str(), &x);
if(!ok) throw std::runtime_error("1-4-abc");
if(ok != a.c_str() + 1) throw std::runtime_error("does not point at the next number");
if(x != 1) throw std::runtime_error("cannot parse 1.");
std::cout << "1-4-abc" << std::endl;
}
void issue2093() {
std::string a = "0.95000000000000000000";
double x;
const char * ok = fast_double_parser::parse_number(a.c_str(), &x);
if(!ok) throw std::runtime_error("0.95000000000000000000");
if(x != 0.95) throw std::runtime_error("cannot parse 0.95000000000000000000.");
std::cout << "0.95000000000000000000" << std::endl;
}
inline void Assert(bool Assertion) {
if (!Assertion)
throw std::runtime_error("bug");
@ -339,13 +307,11 @@ int main() {
printf("Aborting further tests.");
return EXIT_SUCCESS;
}
issue2093();
Assert(basic_test_64bit("1090544144181609348835077142190",0x1.b8779f2474dfbp+99));
Assert(basic_test_64bit("4503599627370496.5", 4503599627370496.5));
Assert(basic_test_64bit("4503599627370497.5", 4503599627370497.5));
Assert(basic_test_64bit("0.000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000044501477170144022721148195934182639518696390927032912960468522194496444440421538910330590478162701758282983178260792422137401728773891892910553144148156412434867599762821265346585071045737627442980259622449029037796981144446145705102663115100318287949527959668236039986479250965780342141637013812613333119898765515451440315261253813266652951306000184917766328660755595837392240989947807556594098101021612198814605258742579179000071675999344145086087205681577915435923018910334964869420614052182892431445797605163650903606514140377217442262561590244668525767372446430075513332450079650686719491377688478005309963967709758965844137894433796621993967316936280457084866613206797017728916080020698679408551343728867675409720757232455434770912461317493580281734466552734375", 0.000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000044501477170144022721148195934182639518696390927032912960468522194496444440421538910330590478162701758282983178260792422137401728773891892910553144148156412434867599762821265346585071045737627442980259622449029037796981144446145705102663115100318287949527959668236039986479250965780342141637013812613333119898765515451440315261253813266652951306000184917766328660755595837392240989947807556594098101021612198814605258742579179000071675999344145086087205681577915435923018910334964869420614052182892431445797605163650903606514140377217442262561590244668525767372446430075513332450079650686719491377688478005309963967709758965844137894433796621993967316936280457084866613206797017728916080020698679408551343728867675409720757232455434770912461317493580281734466552734375));
Assert(basic_test_64bit("0.000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000022250738585072008890245868760858598876504231122409594654935248025624400092282356951787758888037591552642309780950434312085877387158357291821993020294379224223559819827501242041788969571311791082261043971979604000454897391938079198936081525613113376149842043271751033627391549782731594143828136275113838604094249464942286316695429105080201815926642134996606517803095075913058719846423906068637102005108723282784678843631944515866135041223479014792369585208321597621066375401613736583044193603714778355306682834535634005074073040135602968046375918583163124224521599262546494300836851861719422417646455137135420132217031370496583210154654068035397417906022589503023501937519773030945763173210852507299305089761582519159720757232455434770912461317493580281734466552734375", 0.000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000022250738585072008890245868760858598876504231122409594654935248025624400092282356951787758888037591552642309780950434312085877387158357291821993020294379224223559819827501242041788969571311791082261043971979604000454897391938079198936081525613113376149842043271751033627391549782731594143828136275113838604094249464942286316695429105080201815926642134996606517803095075913058719846423906068637102005108723282784678843631944515866135041223479014792369585208321597621066375401613736583044193603714778355306682834535634005074073040135602968046375918583163124224521599262546494300836851861719422417646455137135420132217031370496583210154654068035397417906022589503023501937519773030945763173210852507299305089761582519159720757232455434770912461317493580281734466552734375));
issue63();
issue40();
issue23();
issue32();
@ -394,7 +360,6 @@ int main() {
throw std::runtime_error("fast_double_parser disagrees");
}
}
negative_subsubnormal_to_negative_zero();
std::cout << std::endl;
std::cout << "All ok" << std::endl;
printf("Good!\n");