Command: handle exe and batch files separately

Author: ChrisDenton

library/std/src/sys/windows/args.rs

@@ -299,3 +299,89 @@ pub(crate) fn append_arg(cmd: &mut Vec<u16>, arg: &Arg, force_quotes: bool) -> i
     }
     Ok(())
 }
+
+pub(crate) fn make_bat_command_line(
+    script: &[u16],
+    args: &[Arg],
+    force_quotes: bool,
+) -> io::Result<Vec<u16>> {
+    // Set the start of the command line to `cmd.exe /c "`
+    // It is necessary to surround the command in an extra pair of quotes,
+    // hence The trailing quote here. It will be closed after all arguments
+    // have been added.
+    let mut cmd: Vec<u16> = "cmd.exe /c \"".encode_utf16().collect();
+
+    // Push the script name surrounded by its quote pair.
+    cmd.push(b'"' as u16);
+    cmd.extend_from_slice(script.strip_suffix(&[0]).unwrap_or(script));
+    cmd.push(b'"' as u16);
+
+    // Append the arguments.
+    // FIXME: This needs tests to ensure that the arguments are properly
+    // reconstructed by the batch script by default.
+    for arg in args {
+        cmd.push(' ' as u16);
+        append_arg(&mut cmd, arg, force_quotes)?;
+    }
+
+    // Close the quote we left opened earlier.
+    cmd.push(b'"' as u16);
+
+    Ok(cmd)
+}
+
+/// Takes a path and tries to return a non-verbatim path.
+///
+/// This is necessary because cmd.exe does not support verbatim paths.
+pub(crate) fn to_user_path(mut path: Vec<u16>) -> io::Result<Vec<u16>> {
+    use crate::ptr;
+    use crate::sys::windows::fill_utf16_buf;
+
+    // UTF-16 encoded code points, used in parsing and building UTF-16 paths.
+    // All of these are in the ASCII range so they can be cast directly to `u16`.
+    const SEP: u16 = b'\\' as _;
+    const QUERY: u16 = b'?' as _;
+    const COLON: u16 = b':' as _;
+    const U: u16 = b'U' as _;
+    const N: u16 = b'N' as _;
+    const C: u16 = b'C' as _;
+
+    // Early return if the path is too long to remove the verbatim prefix.
+    const LEGACY_MAX_PATH: usize = 260;
+    if path.len() > LEGACY_MAX_PATH {
+        return Ok(path);
+    }
+
+    match &path[..] {
+        // `\\?\C:\...` => `C:\...`
+        [SEP, SEP, QUERY, SEP, _, COLON, SEP, ..] => unsafe {
+            let lpfilename = path[4..].as_ptr();
+            fill_utf16_buf(
+                |buffer, size| c::GetFullPathNameW(lpfilename, size, buffer, ptr::null_mut()),
+                |full_path: &[u16]| {
+                    if full_path == &path[4..path.len() - 1] { full_path.into() } else { path }
+                },
+            )
+        },
+        // `\\?\UNC\...` => `\\...`
+        [SEP, SEP, QUERY, SEP, U, N, C, SEP, ..] => unsafe {
+            // Change the `C` in `UNC\` to `\` so we can get a slice that starts with `\\`.
+            path[6] = b'\\' as u16;
+            let lpfilename = path[6..].as_ptr();
+            fill_utf16_buf(
+                |buffer, size| c::GetFullPathNameW(lpfilename, size, buffer, ptr::null_mut()),
+                |full_path: &[u16]| {
+                    if full_path == &path[6..path.len() - 1] {
+                        full_path.into()
+                    } else {
+                        // Restore the 'C' in "UNC".
+                        path[6] = b'C' as u16;
+                        path
+                    }
+                },
+            )
+        },
+        // For everything else, leave the path unchanged.
+        _ => Ok(path),
+    }
+}

library/std/src/sys/windows/process.rs

@@ -267,8 +267,19 @@ impl Command {
             program.len().checked_sub(5).and_then(|i| program.get(i..)),
             Some([46, 98 | 66, 97 | 65, 116 | 84, 0] | [46, 99 | 67, 109 | 77, 100 | 68, 0])
         );
-        let mut cmd_str =
-            make_command_line(&program, &self.args, self.force_quotes_enabled, is_batch_file)?;
+        let (program, mut cmd_str) = if is_batch_file {
+            (
+                command_prompt()?,
+                args::make_bat_command_line(
+                    &args::to_user_path(program)?,
+                    &self.args,
+                    self.force_quotes_enabled,
+                )?,
+            )
+        } else {
+            let cmd_str = make_command_line(&self.program, &self.args, self.force_quotes_enabled)?;
+            (program, cmd_str)
+        };
         cmd_str.push(0); // add null terminator
 
         // stolen from the libuv code.
@@ -719,30 +730,17 @@ fn zeroed_process_information() -> c::PROCESS_INFORMATION {
 
 // Produces a wide string *without terminating null*; returns an error if
 // `prog` or any of the `args` contain a nul.
-fn make_command_line(
-    prog: &[u16],
-    args: &[Arg],
-    force_quotes: bool,
-    is_batch_file: bool,
-) -> io::Result<Vec<u16>> {
+fn make_command_line(argv0: &OsStr, args: &[Arg], force_quotes: bool) -> io::Result<Vec<u16>> {
     // Encode the command and arguments in a command line string such
     // that the spawned process may recover them using CommandLineToArgvW.
     let mut cmd: Vec<u16> = Vec::new();
 
-    // CreateFileW has special handling for .bat and .cmd files, which means we
-    // need to add an extra pair of quotes surrounding the whole command line
-    // so they are properly passed on to the script.
-    // See issue #91991.
-    if is_batch_file {
-        cmd.push(b'"' as u16);
-    }
-
     // Always quote the program name so CreateProcess to avoid ambiguity when
     // the child process parses its arguments.
     // Note that quotes aren't escaped here because they can't be used in arg0.
     // But that's ok because file paths can't contain quotes.
     cmd.push(b'"' as u16);
-    cmd.extend_from_slice(prog.strip_suffix(&[0]).unwrap_or(prog));
+    cmd.extend(argv0.encode_wide());
     cmd.push(b'"' as u16);
 
     for arg in args {
@@ -752,6 +750,16 @@ fn make_command_line(
     Ok(cmd)
 }
 
+// Get `cmd.exe` for use with bat scripts, encoded as a UTF-16 string.
+fn command_prompt() -> io::Result<Vec<u16>> {
+    let mut system: Vec<u16> = super::fill_utf16_buf(
+        |buf, size| unsafe { c::GetSystemDirectoryW(buf, size) },
+        |buf| buf.into(),
+    )?;
+    system.extend("\\cmd.exe".encode_utf16().chain([0]));
+    Ok(system)
+}
+
 fn make_envp(maybe_env: Option<BTreeMap<EnvKey, OsString>>) -> io::Result<(*mut c_void, Vec<u16>)> {
     // On Windows we pass an "environment block" which is not a char**, but
     // rather a concatenation of null-terminated k=v\0 sequences, with a final

library/std/src/sys/windows/process/tests.rs

@@ -3,12 +3,11 @@ use super::Arg;
 use crate::env;
 use crate::ffi::{OsStr, OsString};
 use crate::process::Command;
-use crate::sys::to_u16s;
 
 #[test]
 fn test_raw_args() {
     let command_line = &make_command_line(
-        &to_u16s("quoted exe").unwrap(),
+        OsStr::new("quoted exe"),
         &[
             Arg::Regular(OsString::from("quote me")),
             Arg::Raw(OsString::from("quote me *not*")),
@@ -17,7 +16,6 @@ fn test_raw_args() {
             Arg::Regular(OsString::from("optional-quotes")),
         ],
         false,
-        false,
     )
     .unwrap();
     assert_eq!(
@@ -30,10 +28,9 @@ fn test_raw_args() {
 fn test_make_command_line() {
     fn test_wrapper(prog: &str, args: &[&str], force_quotes: bool) -> String {
         let command_line = &make_command_line(
-            &to_u16s(prog).unwrap(),
+            OsStr::new(prog),
             &args.iter().map(|a| Arg::Regular(OsString::from(a))).collect::<Vec<_>>(),
             force_quotes,
-            false,
         )
         .unwrap();
         String::from_utf16(command_line).unwrap()