Refactor utf8_to_windows_encoding

mamba-org · Oct 26, 2023 · 2007b5b · 2007b5b
1 parent 08cbcb7
commit 2007b5b
Show file tree

Hide file tree

Showing 8 changed files with 87 additions and 56 deletions.
diff --git a/libmamba/include/mamba/core/util_os.hpp b/libmamba/include/mamba/core/util_os.hpp
@@ -49,7 +49,6 @@ namespace mamba
     std::string to_utf8(const wchar_t* windows_unicode_text, size_t size);
     std::string to_utf8(const wchar_t* windows_unicode_text);
     std::string to_utf8(const std::wstring& windows_unicode_text);
-    std::wstring to_windows_unicode(const std::string_view utf8_text);
 #endif
 
     /* Test whether a given `std::ostream` object refers to a terminal. */

diff --git a/libmamba/include/mamba/util/os_win.hpp b/libmamba/include/mamba/util/os_win.hpp
@@ -7,8 +7,10 @@
 #ifndef MAMBA_UTIL_OS_WIN_HPP
 #define MAMBA_UTIL_OS_WIN_HPP
 
-#include "mamba/fs/filesystem.hpp"
+#include <string>
+#include <string_view>
 
+#include "mamba/fs/filesystem.hpp"
 
 namespace mamba::util
 {
@@ -23,5 +25,7 @@ namespace mamba::util
     };
 
     auto get_windows_known_user_folder(WindowsKnowUserFolder dir) -> fs::u8path;
+
+    auto utf8_to_windows_encoding(const std::string_view utf8_text) -> std::wstring;
 }
 #endif
diff --git a/libmamba/src/core/environment.cpp b/libmamba/src/core/environment.cpp
@@ -51,7 +51,7 @@ namespace mamba::env
             );
         };
 
-        const std::wstring unicode_key = to_windows_unicode(key);
+        const std::wstring unicode_key = util::utf8_to_windows_encoding(key);
         size_t required_size = 0;
         if (auto error_code = _wgetenv_s(&required_size, nullptr, 0, unicode_key.c_str());
             error_code == 0)
@@ -99,8 +99,8 @@ namespace mamba::env
                                                           // functions are not thread-safe, this
                                                           // is to prevent related issues.
 
-        const std::wstring unicode_key = to_windows_unicode(key);
-        const std::wstring unicode_value = to_windows_unicode(value);
+        const std::wstring unicode_key = util::utf8_to_windows_encoding(key);
+        const std::wstring unicode_value = util::utf8_to_windows_encoding(value);
         auto res = _wputenv_s(unicode_key.c_str(), unicode_value.c_str());
         if (res != 0)
         {

diff --git a/libmamba/src/core/util_os.cpp b/libmamba/src/core/util_os.cpp
@@ -538,43 +538,6 @@ namespace mamba
         return to_utf8(s.data(), s.size());
     }
 
-    std::wstring to_windows_unicode(const std::string_view utf8_text)
-    {
-        std::wstring output;
-        if (!utf8_text.empty())
-        {
-            assert(utf8_text.size() <= INT_MAX);
-            const int size = MultiByteToWideChar(
-                CP_UTF8,
-                0,
-                utf8_text.data(),
-                utf8_text.size(),
-                nullptr,
-                0
-            );
-            if (size <= 0)
-            {
-                unsigned long last_error = ::GetLastError();
-                LOG_ERROR << "Failed to convert UTF-8 string to Windows Unicode (UTF-16)"
-                          << std::system_category().message(static_cast<int>(last_error));
-                throw std::runtime_error("Failed to convert UTF-8 string to UTF-16");
-            }
-
-            output.resize(size);
-            int res_size = MultiByteToWideChar(
-                CP_UTF8,
-                0,
-                utf8_text.data(),
-                utf8_text.size(),
-                output.data(),
-                output.size()
-            );
-            assert(res_size == size);
-        }
-
-        return output;
-    }
-
 #endif
 
     /* From https://github.com/ikalnytskyi/termcolor

diff --git a/libmamba/src/util/os_win.cpp b/libmamba/src/util/os_win.cpp
@@ -6,8 +6,11 @@
 
 #ifdef _WIN32
 
+#include <cassert>
+#include <limits>
 #include <stdexcept>
 
+#include <fmt/format.h>
 #include <Shlobj.h>
 
 #include "mamba/util/os_win.hpp"
@@ -22,17 +25,17 @@ namespace mamba::util
             switch (dir)
             {
                 case (WindowsKnowUserFolder::Documents):
-                    return FOLDERID_Documents;
+                    return ::FOLDERID_Documents;
                 case (WindowsKnowUserFolder::Profile):
-                    return FOLDERID_Profile;
+                    return ::FOLDERID_Profile;
                 case (WindowsKnowUserFolder::Programs):
-                    return FOLDERID_Programs;
+                    return ::FOLDERID_Programs;
                 case (WindowsKnowUserFolder::ProgramData):
-                    return FOLDERID_ProgramData;
+                    return ::FOLDERID_ProgramData;
                 case (WindowsKnowUserFolder::LocalAppData):
-                    return FOLDERID_LocalAppData;
+                    return ::FOLDERID_LocalAppData;
                 case (WindowsKnowUserFolder::RoamingAppData):
-                    return FOLDERID_RoamingAppData;
+                    return ::FOLDERID_RoamingAppData;
             }
             throw std::invalid_argument("Invalid enum");
         }
@@ -41,7 +44,7 @@ namespace mamba::util
         {
             ~COMwstr()
             {
-                CoTaskMemFree(str);
+                ::CoTaskMemFree(str);
             }
 
             wchar_t* str = nullptr;
@@ -52,7 +55,7 @@ namespace mamba::util
     {
         auto localAppData = COMwstr{ nullptr };
 
-        HRESULT const hres = SHGetKnownFolderPath(
+        const auto hres = SHGetKnownFolderPath(
             get_windows_known_user_folder_raw(dir),
             KF_FLAG_DONT_VERIFY,
             nullptr,
@@ -66,6 +69,37 @@ namespace mamba::util
 
         return fs::u8path(localAppData.str);
     }
+
+    auto utf8_to_windows_encoding(const std::string_view utf8_text) -> std::wstring
+    {
+        if (utf8_text.empty())
+        {
+            return {};
+        }
+
+        assert(utf8_text.size() <= std::numeric_limits<int>::max());
+        const int size = ::MultiByteToWideChar(CP_UTF8, 0, utf8_text.data(), utf8_text.size(), nullptr, 0);
+        if (size <= 0)
+        {
+            throw std::runtime_error(fmt::format(
+                R"(Failed to convert UTF-8 string "{}" to UTF-16: )",
+                utf8_text,
+                std::system_category().message(static_cast<int>(::GetLastError()))
+            ));
+        }
+
+        auto output = std::wstring(static_cast<std::size_t>(size), char(0));
+        [[maybe_unused]] const int res_size = ::MultiByteToWideChar(
+            CP_UTF8,
+            0,
+            utf8_text.data(),
+            utf8_text.size(),
+            output.data(),
+            output.size()
+        );
+        assert(res_size == size);
+        return output;
+    }
 }
 
 #else  // #ifdef _WIN32
@@ -91,6 +125,11 @@ namespace mamba::util
     {
         throw_not_implemented("get_windows_known_user_folder");
     }
+
+    auto utf8_to_windows_encoding(const std::string_view) -> std::wstring
+    {
+        throw_not_implemented("utf8_to_windows_unicode");
+    }
 }
 
 #endif  // #ifdef _WIN32
diff --git a/libmamba/tests/CMakeLists.txt b/libmamba/tests/CMakeLists.txt
@@ -36,6 +36,7 @@ set(LIBMAMBA_TEST_SRCS
     src/util/test_flat_bool_expr_tree.cpp
     src/util/test_graph.cpp
     src/util/test_iterator.cpp
+    src/util/test_os_win.cpp
     src/util/test_path_manip.cpp
     src/util/test_tuple_hash.cpp
     src/util/test_url_manip.cpp

diff --git a/libmamba/tests/src/core/test_util_os.cpp b/libmamba/tests/src/core/test_util_os.cpp
@@ -25,12 +25,6 @@ TEST_SUITE("basic_unicode_conversion")
         auto result = mamba::to_utf8(text_utf16);
         CHECK_EQ(text_utf8, result);
     }
-
-    TEST_CASE("to_windows_unicode")
-    {
-        auto result = mamba::to_windows_unicode(text_utf8);
-        CHECK_EQ(text_utf16, result);
-    }
 }
 
 TEST_SUITE("windows_path")

diff --git a/libmamba/tests/src/util/test_os_win.cpp b/libmamba/tests/src/util/test_os_win.cpp
@@ -0,0 +1,31 @@
+// Copyright (c) 2023, QuantStack and Mamba Contributors
+//
+// Distributed under the terms of the BSD 3-Clause License.
+//
+// The full license is in the file LICENSE, distributed with this software.
+
+#ifdef _WIN32
+
+#include <string>
+
+#include <doctest/doctest.h>
+
+#include "mamba/util/os_win.hpp"
+
+using namespace mamba::util;
+
+TEST_SUITE("uti::os_win")
+{
+    TEST_CASE("")
+    {
+        const std::wstring text_utf16 = L"Hello, I am Joël. 私のにほんごわへたです";
+        const std::string text_utf8 = u8"Hello, I am Joël. 私のにほんごわへたです";
+
+        SUBCASE("utf8_to_windows_encoding")
+        {
+            CHECK_EQ(to_windows_encoding(text_utf8), text_utf16);
+        }
+    }
+}
+
+#endif