wip from windows

Author: Christian Parpart

.gitignore

@@ -4,6 +4,7 @@
 /_ucd/
 /.clangd/
 /compile_commands.json
+/.vs/
 /.vscode/
 /sandbox/
 /target/

CMakeLists.txt

@@ -32,7 +32,8 @@ if(("${CMAKE_CXX_COMPILER_ID}" MATCHES "GNU") OR ("${CMAKE_CXX_COMPILER_ID}" MAT
     endif()
 elseif(DEFINED MSVC)
     add_definitions(-DNOMINMAX)
-    add_compile_options(/utf-8)
+    add_compile_options(/nologo)
+    set(CMAKE_STATIC_LINKER_FLAGS "${CMAKE_STATIC_LINKER_FLAGS} /nologo")
 endif()
 
 include(EnableCcache)

cmake/presets/os-windows.json

@@ -16,8 +16,8 @@
             "cacheVariables": {
                 "VCPKG_TARGET_TRIPLET": "x64-windows",
                 "CMAKE_INSTALL_PREFIX": "${sourceDir}/out/install/${presetName}",
-                "CMAKE_VERBOSE_MAKEFILE": "ON",
-                "CMAKE_TOOLCHAIN_FILE": "${sourceDir}/../vcpkg/scripts/buildsystems/vcpkg.cmake"
+                "CMAKE_TOOLCHAIN_FILE": "$env{VCPKG_ROOT}/scripts/buildsystems/vcpkg.cmake",
+                "CMAKE_VERBOSE_MAKEFILE": "ON"
             }
         },
         { "name": "windows-cl-debug", "inherits": ["windows-common", "debug"], "displayName": "Windows (MSVC) Debug", "description": "Using MSVC compiler (64-bit)" },

src/libunicode/CMakeLists.txt

@@ -86,6 +86,9 @@ add_custom_command(
 
 add_library(unicode_loader ${LIBUNICODE_LIB_MODE} codepoint_properties_loader.h codepoint_properties_loader.cpp)
 add_library(unicode::loader ALIAS unicode_loader)
+if(DEFINED MSVC)
+    target_compile_options(unicode_loader PUBLIC /EHsc) # We currently `throw` in the loader, so we need this.
+endif()
 if(LIBUNICODE_TABLEGEN_FASTBUILD)
     target_compile_definitions(unicode_loader PRIVATE LIBUNICODE_TABLEGEN_FASTBUILD)
 endif()
@@ -229,24 +232,26 @@ if(LIBUNICODE_TESTING)
         run_segmenter_test.cpp
         script_segmenter_test.cpp
         test_main.cpp
-        unicode_test.cpp
         utf8_grapheme_segmenter_test.cpp
         utf8_test.cpp
         width_test.cpp
         word_segmenter_test.cpp
     )
 
+    if(DEFINED MSVC)
+        target_compile_options(unicode_test PRIVATE /utf-8)
+    endif()
+
     if(NOT Catch2_FOUND)
         # supress conversion warnings for Catch2
         # https://github.com/catchorg/Catch2/issues/2583
         # https://github.com/SFML/SFML/blob/e45628e2ebc5843baa3739781276fa85a54d4653/test/CMakeLists.txt#L18-L22
         set_target_properties(Catch2 PROPERTIES COMPILE_OPTIONS "" EXPORT_COMPILE_COMMANDS OFF)
-        set_target_properties(Catch2WithMain PROPERTIES EXPORT_COMPILE_COMMANDS OFF)
         get_target_property(CATCH2_INCLUDE_DIRS Catch2 INTERFACE_INCLUDE_DIRECTORIES)
         target_include_directories(Catch2 SYSTEM INTERFACE ${CATCH2_INCLUDE_DIRS})
     endif()
 
-    target_link_libraries(unicode_test unicode Catch2::Catch2WithMain fmt::fmt-header-only)
+    target_link_libraries(unicode_test unicode Catch2::Catch2 fmt::fmt-header-only)
     add_test(unicode_test unicode_test)
 endif()
 # }}}

src/libunicode/grapheme_line_segmenter.h

@@ -696,11 +696,16 @@ class grapheme_line_segmenter<OptionalEventListener>
                     assert(processedTotalWidth + consumedWidth <= maxWidth);
                     auto const asciiTextChunk = std::string_view(next(), consumedWidth);
                     _eventHandler.on_ascii(asciiTextChunk);
+                    assert(!asciiTextChunk.empty());
+                    // We do not store the last ASCII character as last codepoint, because
+                    // we only need to remember the last codepoint to distinguish
+                    // breakability in the grapheme cluster segmentation algorithm.
+                    // We've already reported all ASCII and start fresh after that again.
+                    _complexUnicodeState.lastCodepointHint = 0;
+                    _complexUnicodeState.currentClusterWidth = 0;
                     _complexUnicodeState.utf8DecodeNext = consumedEnd;
                     _complexUnicodeState.currentClusterStart = consumedEnd;
                     _complexUnicodeState.currentCodepointStart = consumedEnd;
-                    assert(!asciiTextChunk.empty());
-                    _complexUnicodeState.lastCodepointHint = asciiTextChunk.back();
                     processedTotalWidth += consumedWidth;
                     assert(processedTotalWidth <= maxWidth);
                     break;

src/libunicode/grapheme_line_segmenter_test.cpp

@@ -492,7 +492,7 @@ TEST_CASE("grapheme_line_segmenter.invalid_char.1")
 // {{{ mixed primitive ASCII and complex unicode
 TEST_CASE("grapheme_line_segmenter.mixed.1")
 {
-    auto const text = "0123456789{\xE2\x94\x80}ABCDEF"sv;
+    auto const text = "0123456789{\xE2\x94\x80}ABCDEF"sv; // \xE2 \x94 \x80 == U+2500
     auto recorder = event_recorder { text };
     auto segmenter = grapheme_line_segmenter { recorder, text };
     auto const result = segmenter.process(80);

src/libunicode/test_main.cpp

@@ -1,6 +1,6 @@
 /**
  * This file is part of the "libunicode" project
- *   Copyright (c) 2020 Christian Parpart <[email protected]>
+ *   Copyright (c) 2024 Christian Parpart <[email protected]>
  *
  * Licensed under the Apache License, Version 2.0 (the "License");
  * you may not use this file except in compliance with the License.
@@ -11,15 +11,39 @@
  * See the License for the specific language governing permissions and
  * limitations under the License.
  */
-#define CATCH_CONFIG_RUNNER
 #include <catch2/catch_session.hpp>
 
-int main(int argc, char const* argv[])
+#if defined(_WIN32)
+    #include <Windows.h>
+#endif
+
+namespace
+{
+
+struct SetupTeardown
+{
+    SetupTeardown()
+    {
+#if defined(_WIN32)
+        const auto stdoutHandle = GetStdHandle(STD_OUTPUT_HANDLE);
+        const auto stdoutCP = GetConsoleOutputCP();
+        DWORD stdoutMode;
+        GetConsoleMode(stdoutHandle, &stdoutMode);
+        SetConsoleMode(stdoutHandle, ENABLE_PROCESSED_OUTPUT | ENABLE_WRAP_AT_EOL_OUTPUT | ENABLE_VIRTUAL_TERMINAL_PROCESSING);
+        SetConsoleOutputCP(CP_UTF8);
+#endif
+    }
+
+    ~SetupTeardown() = default;
+};
+
+} // namespace
+
+int main(int argc, char* argv[])
 {
-    int const result = Catch::Session().run(argc, argv);
+    auto const _ = SetupTeardown {};
 
-    // avoid closing extern console to close on VScode/windows
-    // system("pause");
+    int result = Catch::Session().run(argc, argv);
 
     return result;
 }

src/libunicode/unicode_test.cpp

@@ -29,7 +29,7 @@ std::string escape(std::string const& s)
 {
     std::string t;
     for (char ch: s)
-        if (std::isprint(ch))
+        if (ch >= 0x20 && ch <= 0x7E) // printable ASCII (We don't use isprint() due to MSVC)
             t += ch;
         else
             t += fmt::format("\\x{:02X}", ((unsigned) ch) & 0xFF);