Prefer resolved executable symlinks in POSIX virtual environments (#9058)

Resolve executable symlinks for virtual-environment paths while retaining macOS framework launcher behavior. Include directory-canonicalization and launcher regressions, using core for UTF-8 checks.

Assisted-by: OpenAI Codex:model identifier unavailable

crates/vm/src/getpath.rs

@@ -299,12 +299,69 @@ fn calculate_exec_prefix(exe_dir: Option<&PathBuf>, prefix: &str) -> String {
     }
 }
 
+/// Resolve final-component links without resolving directory aliases (ref: getpath.c).
+#[cfg(unix)]
+fn resolve_executable_symlinks(executable: &Path) -> Option<PathBuf> {
+    use std::os::unix::ffi::OsStrExt;
+
+    let mut path = executable.to_path_buf();
+    // CPython rejects 40 successful hops, matching the Linux 4.2 kernel limit.
+    for _ in 0..40 {
+        let Ok(target) = std::fs::read_link(&path) else {
+            return Some(path);
+        };
+        if target.is_absolute() {
+            path = target;
+            continue;
+        }
+
+        // Match getpath.c's raw dirname join, then normalize lexically without
+        // resolving directory symlinks or losing a double-leading-slash root.
+        let bytes = path.as_os_str().as_bytes();
+        let end = bytes.iter().rposition(|&b| b == b'/');
+        let dirname = &bytes[..end.unwrap_or(bytes.len())];
+        let mut joined = PathBuf::from(std::ffi::OsStr::from_bytes(dirname));
+        // Filesystem text uses UTF-8/surrogateescape; count characters, not bytes.
+        let multiple_chars =
+            core::str::from_utf8(dirname).map_or(dirname.len() > 1, |s| s.chars().nth(1).is_some());
+        if multiple_chars && !dirname.ends_with(b"/") {
+            joined.as_mut_os_string().push("/");
+        }
+        joined.as_mut_os_string().push(target.as_os_str());
+        let bytes = joined.as_os_str().as_bytes();
+        let double_root = bytes.starts_with(b"//") && !bytes.starts_with(b"///");
+        path.clear();
+        for component in joined.components() {
+            if component == std::path::Component::ParentDir
+                && !path.as_os_str().is_empty()
+                && !path.ends_with("..")
+            {
+                path.pop();
+            } else if component != std::path::Component::CurDir {
+                path.push(component.as_os_str());
+            }
+        }
+        if double_root {
+            path = Path::new("//").join(path.strip_prefix("/").ok()?);
+        }
+    }
+    None
+}
+
 /// Calculate base_executable
 fn calculate_base_executable(executable: Option<&PathBuf>, home_dir: Option<&PathBuf>) -> String {
-    // If in venv and we have home, construct base_executable from home
+    // If in venv and we have home, prefer the actual target of a symlink.
     if let (Some(exe), Some(home)) = (executable, home_dir)
         && let Some(exe_name) = exe.file_name()
     {
+        #[cfg(unix)]
+        if let Some(resolved) = resolve_executable_symlinks(exe)
+            && !resolved.as_os_str().is_empty()
+            && resolved.as_os_str() != exe.as_os_str()
+        {
+            return resolved.to_string_lossy().into_owned();
+        }
+
         let base = home.join(exe_name);
         return base.to_string_lossy().into_owned();
     }

extra_tests/snippets/stdlib_sys.py

@@ -1,7 +1,10 @@
 import os
+import shutil
 import subprocess
 import sys
+import tempfile
 import warnings
+from pathlib import Path
 
 from testutils import assert_raises
 
@@ -192,3 +195,132 @@ finally:
         del os.environ["PYTHONBREAKPOINT"]
     else:
         os.environ["PYTHONBREAKPOINT"] = saved_breakpoint_env
+
+
+# A venv's symlink target takes precedence over home/<invoked name>.
+if os.name == "posix" and sys.platform not in ("wasi", "emscripten"):
+    with tempfile.TemporaryDirectory() as directory:
+        root = Path(directory).resolve()
+        interpreter = Path(sys.executable).resolve()
+        # A framework build can also run its embedded app directly. Match the
+        # installed launcher's file identity before applying launcher semantics.
+        framework_base = (
+            Path(sys.base_prefix)
+            / "bin"
+            / f"python{sys.version_info.major}.{sys.version_info.minor}"
+        )
+        framework_launcher = (
+            sys.platform == "darwin"
+            and bool(getattr(sys, "_framework", ""))
+            and framework_base.is_file()
+            and interpreter.samefile(framework_base)
+        )
+        home = root / "home"
+        home.mkdir()
+        venv_env = dict(os.environ)
+        for name in ("PYTHONHOME", "PYTHONEXECUTABLE", "__PYVENV_LAUNCHER__"):
+            venv_env.pop(name, None)
+        # Keep startup imports available even with the synthetic home directory.
+        venv_env["PYTHONPATH"] = os.pathsep.join(sys.path)
+        path_code = (
+            "import sys; print(sys.executable); "
+            "print(sys._base_executable); print(sys.prefix)"
+        )
+
+        def executable_paths(executable, *, cwd=None, env=venv_env):
+            result = subprocess.run(
+                [str(executable), "-B", "-S", "-c", path_code],
+                cwd=cwd,
+                env=env,
+                capture_output=True,
+                text=True,
+                timeout=10,
+            )
+            assert result.returncode == 0, result.stderr
+            return result.stdout.splitlines()
+
+        for kind in ("absolute", "relative", "chain"):
+            bin_dir = root / kind / "bin"
+            bin_dir.mkdir(parents=True)
+            # Exercise both parent and adjacent pyvenv.cfg discovery.
+            prefix = bin_dir if kind == "chain" else bin_dir.parent
+            (prefix / "pyvenv.cfg").write_text(f"home = {home}\n")
+            executable = bin_dir / kind
+            # This regular file must never be preferred to a symlink's target.
+            (home / kind).touch()
+            if kind == "absolute":
+                executable.symlink_to(interpreter)
+            elif kind == "relative":
+                executable.symlink_to(os.path.relpath(interpreter, bin_dir))
+            else:
+                (bin_dir / "target").symlink_to(interpreter)
+                executable.symlink_to("target")
+            invocation = f"./bin/{kind}" if kind == "relative" else executable
+            assert executable_paths(invocation, cwd=bin_dir.parent) == [
+                str(executable),
+                str(framework_base if framework_launcher else interpreter),
+                str(prefix),
+            ]
+
+        # Framework launchers preset their base; ordinary binaries keep argv[0].
+        plain = root / "plain" / "python"
+        plain.parent.mkdir()
+        plain.symlink_to(interpreter)
+        assert executable_paths(plain)[:2] == [
+            str(plain),
+            str(framework_base if framework_launcher else plain),
+        ]
+
+        # Without a launcher preset, copies fall back to home/<invoked name>.
+        copied = root / "copied" / "python"
+        copied.parent.mkdir()
+        shutil.copy2(interpreter, copied)
+        (home / copied.name).touch()
+        (copied.parent / "pyvenv.cfg").write_text(f"home = {home}\n")
+        assert executable_paths(copied) == [
+            str(copied),
+            str(framework_base if framework_launcher else home / copied.name),
+            str(copied.parent),
+        ]
+
+        # A directory alias does not turn a copied executable into a symlink.
+        directory_alias = root / "directory-alias"
+        directory_alias.symlink_to(copied.parent, target_is_directory=True)
+        aliased_copy = directory_alias / copied.name
+        # The framework launcher canonicalizes its directory before dispatch.
+        assert executable_paths(aliased_copy) == [
+            str(copied if framework_launcher else aliased_copy),
+            str(framework_base if framework_launcher else home / copied.name),
+            str(copied.parent if framework_launcher else directory_alias),
+        ]
+
+        # Without a launcher preset, final-link resolution preserves target aliases.
+        target_link = root / "linked" / "python"
+        target_link.parent.mkdir()
+        (target_link.parent / "pyvenv.cfg").write_text(f"home = {home}\n")
+        target_link.symlink_to("../directory-alias/python")
+        assert executable_paths(target_link) == [
+            str(target_link),
+            str(framework_base if framework_launcher else aliased_copy),
+            str(target_link.parent),
+        ]
+
+        # macOS framework builds use a framework-derived base for overrides.
+        if sys.platform == "linux":
+            # Relative targets also preserve a double-leading-slash directory.
+            double_target = root / "double-target"
+            double_target.symlink_to("directory-alias/python")
+            target_link.unlink()
+            target_link.symlink_to("/" + str(double_target))
+            assert executable_paths(target_link) == [
+                str(target_link),
+                "/" + str(aliased_copy),
+                str(target_link.parent),
+            ]
+
+            override_env = dict(venv_env, __PYVENV_LAUNCHER__=str(executable))
+            assert executable_paths(plain, env=override_env) == [
+                str(executable),
+                str(plain),
+                str(prefix),
+            ]