[env]: Change hard-code settings to be generated by py

.clangd

@@ -1,8 +1,11 @@
 CompileFlags:
   Add: [
-    "-std=c++17",
-    # -Wno-unused-value,
-  ] # ignore some specific warnings
-  CompilationDatabase: build/       # Search build/ directory for compile_commands.json
+      -DUNICODE,
+      -D_UNICODE,
+      "-std=c++17",
+      "-IC:/Users/19833/EDisk/CppCodes/ImeCodes/googlepinyinime-rev/build/vcpkg_installed/x64-windows/include",
+      # -Wno-unused-value,
+    ] # ignore some specific warnings
+  CompilationDatabase: build/ # Search build/ directory for compile_commands.json
 # Diagnostics:
 #   UnusedIncludes: None # ignore unused includes

CMakeLists.txt

@@ -4,13 +4,24 @@ project(pinyinime VERSION 1.0 LANGUAGES CXX)
 
 include_directories(${CMAKE_CURRENT_SOURCE_DIR}/src/include)
 
-set(CMAKE_CXX_STANDARD 11)
-set(CMAKE_CXX_STANDARD_REQUIRED True)
+set(CMAKE_CXX_STANDARD 17)
+set(CMAKE_CXX_STANDARD_REQUIRED ON)
 set(CMAKE_C_STANDARD 99)
-set(CMAKE_C_STANDARD_REQUIRED True)
+set(CMAKE_C_STANDARD_REQUIRED ON)
+
+add_definitions(
+    /D_UNICODE=1
+    /DUNICODE=1
+)
+
+# For utf8cpp, make it use C++11 and later
+add_compile_options(/Zc:__cplusplus)
 
 set(CMAKE_RUNTIME_OUTPUT_DIRECTORY ${CMAKE_BINARY_DIR}/bin)
 
+# For pure-header libs installed by vcpkg like utf8cpp
+include_directories("./build/vcpkg_installed/x64-windows/include")
+
 set(HEADERS
     ./src/include/atomdictbase.h
     ./src/include/dictbuilder.h

CMakePresets.json

@@ -0,0 +1,17 @@
+{
+  "version": 2,
+  "configurePresets": [
+    {
+      "name": "vcpkg",
+      "generator": "Visual Studio 17 2022",
+      "binaryDir": "${sourceDir}/build",
+      "environment": {
+        "VCPKG_ROOT": "C:/Users/19833/scoop/apps/vcpkg/current/"
+      },
+      "cacheVariables": {
+        "CMAKE_TOOLCHAIN_FILE": "C:/Users/19833/scoop/apps/vcpkg/current/scripts/buildsystems/vcpkg.cmake",
+        "CMAKE_BUILD_TYPE": "Debug"
+      }
+    }
+  ]
+}

CMakeUserPresets.json

@@ -0,0 +1,11 @@
+{
+  "version": 2,
+  "configurePresets": [
+    {
+      "name": "default",
+      "inherits": "vcpkg",
+      "generator": "Visual Studio 17 2022",
+      "binaryDir": "${sourceDir}/build"
+    }
+  ]
+}

scripts/config_files/.clangd

@@ -0,0 +1,11 @@
+CompileFlags:
+  Add: [
+      -DUNICODE,
+      -D_UNICODE,
+      "-std=c++17",
+      "-IC:/Users/19833/EDisk/CppCodes/ImeCodes/googlepinyinime-rev/build/vcpkg_installed/x64-windows/include",
+      # -Wno-unused-value,
+    ] # ignore some specific warnings
+  CompilationDatabase: build/ # Search build/ directory for compile_commands.json
+# Diagnostics:
+#   UnusedIncludes: None # ignore unused includes

scripts/config_files/CMakeLists.txt

@@ -0,0 +1,68 @@
+cmake_minimum_required(VERSION 3.15)
+
+project(pinyinime VERSION 1.0 LANGUAGES CXX)
+
+include_directories(${CMAKE_CURRENT_SOURCE_DIR}/src/include)
+
+set(CMAKE_CXX_STANDARD 17)
+set(CMAKE_CXX_STANDARD_REQUIRED ON)
+set(CMAKE_C_STANDARD 99)
+set(CMAKE_C_STANDARD_REQUIRED ON)
+
+add_definitions(
+    /D_UNICODE=1
+    /DUNICODE=1
+)
+
+# For utf8cpp, make it use C++11 and later
+add_compile_options(/Zc:__cplusplus)
+
+set(CMAKE_RUNTIME_OUTPUT_DIRECTORY ${CMAKE_BINARY_DIR}/bin)
+
+# For pure-header libs installed by vcpkg like utf8cpp
+include_directories("./build/vcpkg_installed/x64-windows/include")
+
+set(HEADERS
+    ./src/include/atomdictbase.h
+    ./src/include/dictbuilder.h
+    ./src/include/dictdef.h
+    ./src/include/dictlist.h
+    ./src/include/dicttrie.h
+    ./src/include/lpicache.h
+    ./src/include/matrixsearch.h
+    ./src/include/mystdlib.h
+    ./src/include/ngram.h
+    ./src/include/pinyinime.h
+    ./src/include/searchutility.h
+    ./src/include/spellingtable.h
+    ./src/include/spellingtrie.h
+    ./src/include/splparser.h
+    ./src/include/sync.h
+    ./src/include/userdict.h
+    ./src/include/utf16char.h
+    ./src/include/utf16reader.h
+)
+
+set(SOURCES
+    ./src/share/dictbuilder.cpp
+    ./src/share/dictlist.cpp
+    ./src/share/dicttrie.cpp
+    ./src/share/lpicache.cpp
+    ./src/share/matrixsearch.cpp
+    ./src/share/mystdlib.cpp
+    ./src/share/ngram.cpp
+    ./src/share/pinyinime.cpp
+    ./src/share/searchutility.cpp
+    ./src/share/spellingtable.cpp
+    ./src/share/spellingtrie.cpp
+    ./src/share/splparser.cpp
+    ./src/share/sync.cpp
+    ./src/share/userdict.cpp
+    ./src/share/utf16char.cpp
+    ./src/share/utf16reader.cpp
+    ./tests/main.cpp
+)
+
+set(MY_EXECUTABLE_NAME "pinyinime")
+
+add_executable(pinyinime ${SOURCES} ${HEADERS})

scripts/config_files/CMakePresets.json

@@ -0,0 +1,17 @@
+{
+  "version": 2,
+  "configurePresets": [
+    {
+      "name": "vcpkg",
+      "generator": "Visual Studio 17 2022",
+      "binaryDir": "${sourceDir}/build",
+      "environment": {
+        "VCPKG_ROOT": "C:/Users/19833/scoop/apps/vcpkg/current/"
+      },
+      "cacheVariables": {
+        "CMAKE_TOOLCHAIN_FILE": "C:/Users/19833/scoop/apps/vcpkg/current/scripts/buildsystems/vcpkg.cmake",
+        "CMAKE_BUILD_TYPE": "Debug"
+      }
+    }
+  ]
+}

scripts/lcompile.ps1

@@ -2,25 +2,23 @@
 $currentDirectory = Get-Location
 $cmakeListsPath = Join-Path -Path $currentDirectory -ChildPath "CMakeLists.txt"
 
-if (-not (Test-Path $cmakeListsPath))
-{
-  Write-Host("No CMakeLists.txt in current directory, please check.")
-  return
+if (-not (Test-Path $cmakeListsPath)) {
+    Write-Host("No CMakeLists.txt in current directory, please check.")
+    return
 }
 
 Write-Host "Start generating and compiling..."
 
 $buildFolderPath = ".\build"
 
-if (-not (Test-Path $buildFolderPath))
-{
-  New-Item -ItemType Directory -Path $buildFolderPath | Out-Null
-  Write-Host "build folder created."
+if (-not (Test-Path $buildFolderPath)) {
+    New-Item -ItemType Directory -Path $buildFolderPath | Out-Null
+    Write-Host "build folder created."
 }
 
-cmake -G "Visual Studio 17 2022" -A x64 -S . -B ./build/
+# cmake -G "Visual Studio 17 2022" -A x64 -S . -B ./build/
+cmake --preset=default
 
-if ($LASTEXITCODE -eq 0)
-{
-  cmake --build ./build/ --config DEBUG
+if ($LASTEXITCODE -eq 0) {
+    cmake --build build
 }

scripts/llaunch.ps1

@@ -31,10 +31,11 @@ if (-not (Test-Path $buildFolderPath)) {
     Write-Host "build folder created."
 }
 
-cmake -G "Visual Studio 17 2022" -A x64 -S . -B ./build/
+# cmake -G -A x64 -DCMAKE_TOOLCHAIN_FILE=C:/Users/SonnyCalcr/scoop/apps/vcpkg/current/scripts/buildsystems/vcpkg.cmake -S . -B ./build/
+cmake --preset=default
 
 if ($LASTEXITCODE -eq 0) {
-    cmake --build ./build/ --config DEBUG
+    cmake --build build
     if ($LASTEXITCODE -eq 0) {
         $exePath = getExePathFromCMakeLists
         Write-Host "start running as follows..."
@@ -42,4 +43,3 @@ if ($LASTEXITCODE -eq 0) {
         Invoke-Expression $exePath
     }
 }
-

scripts/prepare_env.py

@@ -0,0 +1,96 @@
+import os
+
+
+def normpath(path):
+    return path.replace("\\", "/")
+
+
+local_app_data_dir = os.environ.get("LOCALAPPDATA")
+home_dir = os.path.expanduser("~")
+
+cur_file_path = os.path.dirname(os.path.abspath(__file__))
+project_root_path = os.path.dirname(cur_file_path)
+
+user_home = os.path.expanduser("~")
+
+googlepinyinime_rev_root_path = normpath(project_root_path)
+googlepinyinime_rev_src_path = normpath(
+    os.path.join(googlepinyinime_rev_root_path, "src")
+)
+vcpkg_include_path = normpath(
+    os.path.join(
+        googlepinyinime_rev_root_path,
+        "build",
+        "vcpkg_installed",
+        "x64-windows",
+        "include",
+    )
+)
+webview2_path = normpath(
+    os.path.join(
+        user_home,
+        ".nuget",
+        "packages",
+        "microsoft.web.webview2",
+        "1.0.3240.44",
+        "build",
+        "native",
+        "include",
+    )
+)
+wim_path = normpath(
+    os.path.join(
+        user_home,
+        ".nuget",
+        "packages",
+        "microsoft.windows.implementationlibrary",
+        "1.0.240803.1",
+        "include",
+    )
+)
+boost_path = normpath(os.path.join(user_home, "scoop", "apps", "boost", "current"))
+
+#
+# project_root/.clangd
+#
+dot_clangd_file = os.path.join(
+    googlepinyinime_rev_root_path, "scripts", "config_files", ".clangd"
+)
+dot_clangd_output_file = os.path.join(googlepinyinime_rev_root_path, ".clangd")
+with open(dot_clangd_file, "r", encoding="utf-8") as f:
+    lines = f.readlines()
+lines[5] = (
+    f'      "-I{googlepinyinime_rev_root_path}/build/vcpkg_installed/x64-windows/include",\n'
+)
+with open(dot_clangd_output_file, "w", encoding="utf-8") as f:
+    f.writelines(lines)
+
+#
+# project_root/tests/CMakeLists.txt
+#
+CMakeLists_file = os.path.join(
+    googlepinyinime_rev_root_path, "scripts", "config_files", "CMakeLists.txt"
+)
+CMakeLists_output_file = os.path.join(googlepinyinime_rev_root_path, "CMakeLists.txt")
+with open(CMakeLists_file, "r", encoding="utf-8") as f:
+    lines = f.readlines()
+with open(CMakeLists_output_file, "w", encoding="utf-8") as f:
+    f.writelines(lines)
+
+#
+# CMakePresets.json
+#
+CMakePresets_file = os.path.join(
+    googlepinyinime_rev_root_path, "scripts", "config_files", "CMakePresets.json"
+)
+CMakePresets_file_output_file = os.path.join(
+    googlepinyinime_rev_root_path, "CMakePresets.json"
+)
+with open(CMakePresets_file, "r", encoding="utf-8") as f:
+    lines = f.readlines()
+lines[8] = f'        "VCPKG_ROOT": "{normpath(user_home)}/scoop/apps/vcpkg/current/"\n'
+lines[11] = (
+    f'        "CMAKE_TOOLCHAIN_FILE": "{normpath(user_home)}/scoop/apps/vcpkg/current/scripts/buildsystems/vcpkg.cmake",\n'
+)
+with open(CMakePresets_file_output_file, "w", encoding="utf-8") as f:
+    f.writelines(lines)

tests/main.cpp

@@ -1,16 +1,22 @@
 #include "../src/include/pinyinime.h"
-#include <codecvt>
 #include <chrono>
 #include <iostream>
-#include <locale>
 #include <string>
 #include <vector>
 #include <cstring>
+#include <utf8cpp/utf8.h>
+#include <Windows.h>
 
 std::string fromUtf16(const ime_pinyin::char16 *buf, size_t len) {
-    std::u16string utf16Str(reinterpret_cast<const char16_t *>(buf), len);
-    std::wstring_convert<std::codecvt_utf8_utf16<char16_t>, char16_t> convert;
-    return convert.to_bytes(utf16Str);
+    // 先将输入的 UTF-16 buffer 转成 std::u16string（等价于 char16_t）
+    const char16_t *utf16Data = reinterpret_cast<const char16_t *>(buf);
+    std::u16string utf16Str(utf16Data, len);
+
+    std::string utf8Result;
+    // 使用 utfcpp 提供的 utf16to8 转换迭代器，将 UTF-16 转为 UTF-8
+    utf8::utf16to8(utf16Str.begin(), utf16Str.end(), std::back_inserter(utf8Result));
+
+    return utf8Result;
 }
 
 void test_pinyin_search_and_segment(const std::string &user_pinyin) {
@@ -88,6 +94,9 @@ void test_pinyin_search_when_retriving_first_element(const std::string &user_pin
 }
 
 int main() {
+#ifdef _WIN32
+    SetConsoleOutputCP(CP_UTF8);
+#endif
     ime_pinyin::im_set_max_lens(64, 32);
     if (!ime_pinyin::im_open_decoder("./data/dict_pinyin.dat", "./data/user_dict.dat")) {
         std::cout << "fany bug.\n";
@@ -107,5 +116,6 @@ int main() {
     test_pinyin_search_and_segment("ni'shuo'ni'ma'ne");
     test_pinyin_search_when_retriving_first_element("qunimadegouridequsibawonengzenmeban");
     test_pinyin_search_when_retriving_first_element("zh'ta'ma'an'jing");
+    test_pinyin_search_when_retriving_first_element("ni'shuo'ne'ma'de");
     return 0;
 }

vcpkg-configuration.json

@@ -0,0 +1,14 @@
+{
+  "default-registry": {
+    "kind": "git",
+    "baseline": "b02e341c927f16d991edbd915d8ea43eac52096c",
+    "repository": "https://github.com/microsoft/vcpkg"
+  },
+  "registries": [
+    {
+      "kind": "artifact",
+      "location": "https://github.com/microsoft/vcpkg-ce-catalog/archive/refs/heads/main.zip",
+      "name": "microsoft"
+    }
+  ]
+}

vcpkg.json

@@ -0,0 +1,5 @@
+{
+  "dependencies": [
+    "utfcpp"
+  ]
+}