Adapt `fsyspath` for C++20 conventions.

In C++20, `std::filesystem::u8path()` (that accepts a UTF-8 encoded
`std::string` and returns a `std::filesystem::path`) is deprecated. Instead,
to engage UTF-8 coding conversions, we're supposed to pass the `path`
constructor a `std::u8string`, i.e. a `std::basic_string<char8_t>`. Since
`char8_t` is a type distinct from both `char` and `unsigned char`, we must Do
Something to pass a UTF-8 encoded `std::string` into `std::filesystem::path`.

To avoid copying characters from a `std::string` into a temporary
`std::u8string` and from there into the `std::filesystem::path`, make a
`boost::transform_iterator` that accepts a `std::string_view::iterator` and
adapts it to dereference `char8_t` characters. Make `fsyspath(std::string_view)`
engage the base-class constructor accepting (iterator, iterator), adapting
`string_view::begin()` and `end()` to deliver `char8_t` characters.

Use the same tactic for `fsyspath::operator=(std::string_view)`, explicitly
calling `std::filesystem::path::assign()` with the adapted iterators.

To resolve ambiguities, provide both constructors and assignment operators
accepting `(const std::string&)` and `(const char*)`, explicitly converting
each to `std::string_view`.

At the same time, `std::filesystem::path::u8string()` now returns
`std::u8string` rather than `std::string`. Since `std::filesystem::path`
delivers only that `std::u8string` rather than iterators into its internal
representation, we can't avoid capturing it and copying to the returned
`std::string`.

Remove explicit `.u8string()` calls from a few existing `fsyspath` instances,
now that `fsyspath` supports implicit conversion to `std::string`.

2 files changed, 40 insertions, 24 deletions

diff --git a/indra/llcommon/fsyspath.h b/indra/llcommon/fsyspath.h
index 1b4aec09b4..f66970ed8f 100644
--- a/indra/llcommon/fsyspath.h
+++ b/indra/llcommon/fsyspath.h
@@ -12,7 +12,10 @@
 #if ! defined(LL_FSYSPATH_H)
 #define LL_FSYSPATH_H
 
+#include <boost/iterator/transform_iterator.hpp>
 #include <filesystem>
+#include <string>
+#include <string_view>
 
 // While std::filesystem::path can be directly constructed from std::string on
 // both Posix and Windows, that's not what we want on Windows. Per
@@ -33,42 +36,55 @@
 // char"), the "native narrow encoding" isn't UTF-8, so file paths containing
 // non-ASCII characters get mangled.
 //
-// Once we're building with C++20, we could pass a UTF-8 std::string through a
-// vector<char8_t> to engage std::filesystem::path's own UTF-8 conversion. But
-// sigh, as of 2024-04-03 we're not yet there.
-//
-// Anyway, encapsulating the important UTF-8 conversions in our own subclass
-// allows us to migrate forward to C++20 conventions without changing
-// referencing code.
+// Encapsulating the important UTF-8 conversions in our own subclass allows us
+// to migrate forward to C++20 conventions without changing referencing code.
 
 class fsyspath: public std::filesystem::path
 {
     using super = std::filesystem::path;
 
+    // In C++20 (__cpp_lib_char8_t), std::filesystem::u8path() is deprecated.
+    // std::filesystem::path(iter, iter) performs UTF-8 conversions when the
+    // value_type of the iterators is char8_t. While we could copy into a
+    // temporary std::u8string and from there into std::filesystem::path, to
+    // minimize string copying we'll define a transform_iterator that accepts
+    // a std::string_view::iterator and dereferences to char8_t.
+    struct u8ify
+    {
+        char8_t operator()(char c) const { return char8_t(c); }
+    };
+    using u8iter = boost::transform_iterator<u8ify, std::string_view::iterator>;
+
 public:
     // default
     fsyspath() {}
-    // construct from UTF-8 encoded std::string
-    fsyspath(const std::string& path): super(std::filesystem::u8path(path)) {}
-    // construct from UTF-8 encoded const char*
-    fsyspath(const char* path): super(std::filesystem::u8path(path)) {}
+    // construct from UTF-8 encoded string
+    fsyspath(const std::string& path): fsyspath(std::string_view(path)) {}
+    fsyspath(const char* path):        fsyspath(std::string_view(path)) {}
+    fsyspath(std::string_view path):
+        super(u8iter(path.begin(), u8ify()), u8iter(path.end(), u8ify()))
+    {}
     // construct from existing path
     fsyspath(const super& path): super(path) {}
 
-    fsyspath& operator=(const super& p) { super::operator=(p); return *this; }
-    fsyspath& operator=(const std::string& p)
-    {
-        super::operator=(std::filesystem::u8path(p));
-        return *this;
-    }
-    fsyspath& operator=(const char* p)
+    fsyspath& operator=(const super& p)       { super::operator=(p); return *this; }
+    fsyspath& operator=(const std::string& p) { return (*this) = std::string_view(p); }
+    fsyspath& operator=(const char* p)        { return (*this) = std::string_view(p); }
+    fsyspath& operator=(std::string_view p)
     {
-        super::operator=(std::filesystem::u8path(p));
+        assign(u8iter(p.begin(), u8ify()), u8iter(p.end(), u8ify()));
         return *this;
     }
 
     // shadow base-class string() method with UTF-8 aware method
-    std::string string() const { return super::u8string(); }
+    std::string string() const
+    {
+        // Short of forbidden type punning, I see no way to avoid copying this
+        // std::u8string to a std::string.
+        auto u8str{ super::u8string() };
+        // from https://github.com/tahonermann/char8_t-remediation/blob/master/char8_t-remediation.h#L180-L182
+        return { u8str.begin(), u8str.end() };
+    }
     // On Posix systems, where value_type is already char, this operator
     // std::string() method shadows the base class operator string_type()
     // method. But on Windows, where value_type is wchar_t, the base class
diff --git a/indra/llcommon/lua_function.cpp b/indra/llcommon/lua_function.cpp
index a9f88f3170..e28c0caa27 100644
--- a/indra/llcommon/lua_function.cpp
+++ b/indra/llcommon/lua_function.cpp
@@ -170,7 +170,7 @@ fsyspath source_path(lua_State* L)
     {
         lua_getinfo(L, i, "s", &ar);
     }
-    return ar.source;
+    return { ar.source };
 }
 
 } // namespace lluau
@@ -1108,7 +1108,7 @@ lua_function(source_path, "source_path(): return the source path of the running
 {
     lua_checkdelta(L, 1);
     lluau_checkstack(L, 1);
-    lua_pushstdstring(L, lluau::source_path(L).u8string());
+    lua_pushstdstring(L, lluau::source_path(L));
     return 1;
 }
 
@@ -1119,7 +1119,7 @@ lua_function(source_dir, "source_dir(): return the source directory of the runni
 {
     lua_checkdelta(L, 1);
     lluau_checkstack(L, 1);
-    lua_pushstdstring(L, lluau::source_path(L).parent_path().u8string());
+    lua_pushstdstring(L, lluau::source_path(L).parent_path());
     return 1;
 }
 
@@ -1132,7 +1132,7 @@ lua_function(abspath, "abspath(path): "
     lua_checkdelta(L);
     auto path{ lua_tostdstring(L, 1) };
     lua_pop(L, 1);
-    lua_pushstdstring(L, (lluau::source_path(L).parent_path() / path).u8string());
+    lua_pushstdstring(L, (lluau::source_path(L).parent_path() / path));
     return 1;
 }