Fix wasip2 build (#6935)

.github/workflows/ci.yaml

@@ -257,6 +257,13 @@ jobs:
       - name: Check compilation for freeBSD
         run: cargo check --target x86_64-unknown-freebsd ${{ env.CARGO_ARGS_NO_SSL }}
 
+      - uses: dtolnay/rust-toolchain@stable
+        with:
+          target: wasm32-wasip2
+
+      - name: Check compilation for wasip2
+        run: cargo check --target wasm32-wasip2 ${{ env.CARGO_ARGS_NO_SSL }}
+
       # - name: Prepare repository for redox compilation
       #   run: bash scripts/redox/uncomment-cargo.sh
       # - name: Check compilation for Redox

crates/common/src/lib.rs

@@ -1,7 +1,5 @@
 //! A crate to hold types and functions common to all rustpython components.
 
-#![cfg_attr(all(target_os = "wasi", target_env = "p2"), feature(wasip2))]
-
 extern crate alloc;
 
 #[macro_use]

crates/common/src/os.rs

@@ -94,9 +94,61 @@ pub fn bytes_as_os_str(b: &[u8]) -> Result<&std::ffi::OsStr, Utf8Error> {
 
 #[cfg(unix)]
 pub use std::os::unix::ffi;
-#[cfg(target_os = "wasi")]
+
+// WASIp1 uses stable std::os::wasi::ffi
+#[cfg(all(target_os = "wasi", not(target_env = "p2")))]
 pub use std::os::wasi::ffi;
 
+// WASIp2: std::os::wasip2::ffi is unstable, so we provide a stable implementation
+// leveraging WASI's UTF-8 string guarantee
+#[cfg(all(target_os = "wasi", target_env = "p2"))]
+pub mod ffi {
+    use std::ffi::{OsStr, OsString};
+
+    pub trait OsStrExt: sealed::Sealed {
+        fn as_bytes(&self) -> &[u8];
+        fn from_bytes(slice: &[u8]) -> &Self;
+    }
+
+    impl OsStrExt for OsStr {
+        fn as_bytes(&self) -> &[u8] {
+            // WASI strings are guaranteed to be UTF-8
+            self.to_str().expect("wasip2 strings are UTF-8").as_bytes()
+        }
+
+        fn from_bytes(slice: &[u8]) -> &OsStr {
+            // WASI strings are guaranteed to be UTF-8
+            OsStr::new(std::str::from_utf8(slice).expect("wasip2 strings are UTF-8"))
+        }
+    }
+
+    pub trait OsStringExt: sealed::Sealed {
+        fn from_vec(vec: Vec<u8>) -> Self;
+        fn into_vec(self) -> Vec<u8>;
+    }
+
+    impl OsStringExt for OsString {
+        fn from_vec(vec: Vec<u8>) -> OsString {
+            // WASI strings are guaranteed to be UTF-8
+            OsString::from(String::from_utf8(vec).expect("wasip2 strings are UTF-8"))
+        }
+
+        fn into_vec(self) -> Vec<u8> {
+            // WASI strings are guaranteed to be UTF-8
+            self.to_str()
+                .expect("wasip2 strings are UTF-8")
+                .as_bytes()
+                .to_vec()
+        }
+    }
+
+    mod sealed {
+        pub trait Sealed {}
+        impl Sealed for std::ffi::OsStr {}
+        impl Sealed for std::ffi::OsString {}
+    }
+}
+
 #[cfg(windows)]
 pub fn errno_to_winerror(errno: i32) -> i32 {
     use libc::*;

crates/stdlib/src/lib.rs

@@ -2,7 +2,6 @@
 // how `mod` works, but we want this sometimes for pymodule declarations
 
 #![allow(clippy::module_inception)]
-#![cfg_attr(all(target_os = "wasi", target_env = "p2"), feature(wasip2))]
 
 #[macro_use]
 extern crate rustpython_derive;

crates/stdlib/src/select.rs

@@ -61,7 +61,7 @@ mod platform {
 #[cfg(target_os = "wasi")]
 mod platform {
     pub use libc::{FD_SETSIZE, timeval};
-    pub use std::os::wasi::io::RawFd;
+    pub use std::os::fd::RawFd;
 
     pub fn check_err(x: i32) -> bool {
         x < 0

src/lib.rs

@@ -44,7 +44,6 @@
 //!
 //! See [`rustpython_derive`](../rustpython_derive/index.html) crate for documentation on macros used in the example above.
 
-#![cfg_attr(all(target_os = "wasi", target_env = "p2"), feature(wasip2))]
 #![allow(clippy::needless_doctest_main)]
 
 #[macro_use]

src/settings.rs

@@ -407,8 +407,10 @@ pub(crate) use env::split_paths;
 pub(crate) fn split_paths<T: AsRef<std::ffi::OsStr> + ?Sized>(
     s: &T,
 ) -> impl Iterator<Item = std::path::PathBuf> + '_ {
-    use std::os::wasi::ffi::OsStrExt;
-    let s = s.as_ref().as_bytes();
-    s.split(|b| *b == b':')
-        .map(|x| std::ffi::OsStr::from_bytes(x).to_owned().into())
+    let s = s.as_ref().as_encoded_bytes();
+    s.split(|b| *b == b':').map(|x| {
+        unsafe { std::ffi::OsStr::from_encoded_bytes_unchecked(x) }
+            .to_owned()
+            .into()
+    })
 }