Improve handling of Unicode in paths

This commit is contained in:
Victor Zverovich 2023-04-30 09:09:35 -07:00
parent 53162142b2
commit 02cae7e48a
2 changed files with 27 additions and 32 deletions

View File

@ -56,12 +56,23 @@ void write_escaped_path(basic_memory_buffer<Char>& quoted,
} }
# ifdef _WIN32 # ifdef _WIN32
template <> template <>
inline void write_escaped_path<char>(basic_memory_buffer<char>& quoted, inline void write_escaped_path<char>(memory_buffer& quoted,
const std::filesystem::path& p) { const std::filesystem::path& p) {
auto s = p.u8string(); auto buf = basic_memory_buffer<wchar_t>();
write_escaped_string<char>( write_escaped_string<wchar_t>(std::back_inserter(buf), p.native());
std::back_inserter(quoted), for (unsigned c : buf) {
string_view(reinterpret_cast<const char*>(s.c_str()), s.size())); // Convert UTF-16 to UTF-8.
if (c < 0x80) {
quoted.push_back(static_cast<unsigned char>(c));
} else if (c < 0x800) {
quoted.push_back(0b1100'0000 | ((c >> 6) & 0b01'1111));
quoted.push_back(0b1000'0000 | (c & 0b11'1111));
} else {
quoted.push_back(0b1110'0000 | ((c >> 12) & 0b01'1111));
quoted.push_back(0b1000'0000 | ((c >> 6) & 0b11'1111));
quoted.push_back(0b1000'0000 | (c & 0b11'1111));
}
}
} }
# endif # endif
template <> template <>
@ -86,7 +97,7 @@ struct formatter<std::filesystem::path, Char>
template <typename FormatContext> template <typename FormatContext>
auto format(const std::filesystem::path& p, FormatContext& ctx) const -> auto format(const std::filesystem::path& p, FormatContext& ctx) const ->
typename FormatContext::iterator { typename FormatContext::iterator {
basic_memory_buffer<Char> quoted; auto quoted = basic_memory_buffer<Char>();
detail::write_escaped_path(quoted, p); detail::write_escaped_path(quoted, p);
return formatter<basic_string_view<Char>>::format( return formatter<basic_string_view<Char>>::format(
basic_string_view<Char>(quoted.data(), quoted.size()), ctx); basic_string_view<Char>(quoted.data(), quoted.size()), ctx);

View File

@ -17,8 +17,8 @@
using testing::StartsWith; using testing::StartsWith;
TEST(std_test, path) {
#ifdef __cpp_lib_filesystem #ifdef __cpp_lib_filesystem
TEST(std_test, path) {
EXPECT_EQ(fmt::format("{:8}", std::filesystem::path("foo")), "\"foo\" "); EXPECT_EQ(fmt::format("{:8}", std::filesystem::path("foo")), "\"foo\" ");
EXPECT_EQ(fmt::format("{}", std::filesystem::path("foo\"bar.txt")), EXPECT_EQ(fmt::format("{}", std::filesystem::path("foo\"bar.txt")),
"\"foo\\\"bar.txt\""); "\"foo\\\"bar.txt\"");
@ -26,32 +26,25 @@ TEST(std_test, path) {
"\"foo\\\"bar.txt\""); "\"foo\\\"bar.txt\"");
# ifdef _WIN32 # ifdef _WIN32
// File.txt in Russian. EXPECT_EQ(fmt::format("{}", std::filesystem::path(
const wchar_t unicode_path[] = {0x424, 0x430, 0x439, 0x43b, 0x2e, L"\x0428\x0447\x0443\x0447\x044B\x043D\x0448"
0x74, 0x78, 0x74, 0}; L"\x0447\x044B\x043D\x0430")),
const char unicode_u8path[] = {'"', char(0xd0), char(0xa4), char(0xd0), "\"Шчучыншчына\"");
char(0xb0), char(0xd0), char(0xb9), char(0xd0), // EXPECT_EQ(fmt::format("{}", std::filesystem::path(L"\xd800")),
char(0xbb), '.', 't', 'x', // "\\x{d800}");
't', '"', '\0'};
EXPECT_EQ(fmt::format("{}", std::filesystem::path(unicode_path)),
unicode_u8path);
# endif # endif
#endif
} }
TEST(ranges_std_test, format_vector_path) {
// Test ambiguity problem described in #2954. // Test ambiguity problem described in #2954.
#ifdef __cpp_lib_filesystem TEST(ranges_std_test, format_vector_path) {
auto p = std::filesystem::path("foo/bar.txt"); auto p = std::filesystem::path("foo/bar.txt");
auto c = std::vector<std::string>{"abc", "def"}; auto c = std::vector<std::string>{"abc", "def"};
EXPECT_EQ(fmt::format("path={}, range={}", p, c), EXPECT_EQ(fmt::format("path={}, range={}", p, c),
"path=\"foo/bar.txt\", range=[\"abc\", \"def\"]"); "path=\"foo/bar.txt\", range=[\"abc\", \"def\"]");
#endif
} }
// Test that path is not escaped twice in the debug mode.
TEST(ranges_std_test, format_quote_path) { TEST(ranges_std_test, format_quote_path) {
// Test that path is not escaped twice in the debug mode.
#ifdef __cpp_lib_filesystem
auto vec = auto vec =
std::vector<std::filesystem::path>{"path1/file1.txt", "path2/file2.txt"}; std::vector<std::filesystem::path>{"path1/file1.txt", "path2/file2.txt"};
EXPECT_EQ(fmt::format("{}", vec), EXPECT_EQ(fmt::format("{}", vec),
@ -61,8 +54,8 @@ TEST(ranges_std_test, format_quote_path) {
EXPECT_EQ(fmt::format("{}", o), "optional(\"path/file.txt\")"); EXPECT_EQ(fmt::format("{}", o), "optional(\"path/file.txt\")");
EXPECT_EQ(fmt::format("{:?}", o), "optional(\"path/file.txt\")"); EXPECT_EQ(fmt::format("{:?}", o), "optional(\"path/file.txt\")");
# endif # endif
#endif
} }
#endif
TEST(std_test, thread_id) { TEST(std_test, thread_id) {
EXPECT_FALSE(fmt::format("{}", std::this_thread::get_id()).empty()); EXPECT_FALSE(fmt::format("{}", std::this_thread::get_id()).empty());
@ -215,13 +208,4 @@ TEST(std_test, exception) {
} catch (const std::system_error& ex) { } catch (const std::system_error& ex) {
EXPECT_THAT(fmt::format("{:t}", ex), StartsWith("std::system_error: ")); EXPECT_THAT(fmt::format("{:t}", ex), StartsWith("std::system_error: "));
} }
#ifdef __cpp_lib_filesystem
try {
throw std::filesystem::filesystem_error("message", std::error_code());
} catch (const std::filesystem::filesystem_error& ex) {
EXPECT_THAT(fmt::format("{:t}", ex),
StartsWith("std::filesystem::filesystem_error: "));
}
#endif
} }