Refactor pre-getopts command line argument handling

compiler/rustc_driver_impl/src/args.rs

@@ -1,51 +1,85 @@
-use std::error;
-use std::fmt;
-use std::fs;
-use std::io;
+use rustc_span::ErrorGuaranteed;
+use std::{env, error, fmt, fs, io};
 
 use rustc_session::EarlyErrorHandler;
 
-fn arg_expand(arg: String) -> Result<Vec<String>, Error> {
+fn arg_expand(arg: &str) -> Result<Vec<String>, Error> {
     if let Some(path) = arg.strip_prefix('@') {
         let file = match fs::read_to_string(path) {
             Ok(file) => file,
             Err(ref err) if err.kind() == io::ErrorKind::InvalidData => {
-                return Err(Error::Utf8Error(Some(path.to_string())));
+                return Err(Error::Utf8Error(path.to_string()));
             }
             Err(err) => return Err(Error::IOError(path.to_string(), err)),
         };
         Ok(file.lines().map(ToString::to_string).collect())
     } else {
-        Ok(vec![arg])
+        Ok(vec![arg.to_string()])
     }
 }
 
+/// Replaces any `@file` arguments with the contents of `file`, with each line of `file` as a
+/// separate argument.
+///
 /// **Note:** This function doesn't interpret argument 0 in any special way.
 /// If this function is intended to be used with command line arguments,
 /// `argv[0]` must be removed prior to calling it manually.
-pub fn arg_expand_all(handler: &EarlyErrorHandler, at_args: &[String]) -> Vec<String> {
-    let mut args = Vec::new();
+pub fn arg_expand_all(
+    handler: &EarlyErrorHandler,
+    at_args: &[String],
+) -> Result<Vec<String>, ErrorGuaranteed> {
+    let mut res = Ok(Vec::new());
     for arg in at_args {
-        match arg_expand(arg.clone()) {
-            Ok(arg) => args.extend(arg),
-            Err(err) => handler.early_error(format!("Failed to load argument file: {err}")),
+        match arg_expand(arg) {
+            Ok(arg) => {
+                if let Ok(args) = &mut res {
+                    args.extend(arg)
+                }
+            }
+            Err(err) => {
+                res =
+                    Err(handler
+                        .early_error_no_abort(format!("failed to load argument file: {err}")))
+            }
+        }
+    }
+    res
+}
+
+/// Gets the raw unprocessed command-line arguments as Unicode strings, without doing any further
+/// processing (e.g., without `@file` expansion).
+///
+/// This function is identical to [`env::args()`] except that it emits an error when it encounters
+/// non-Unicode arguments instead of panicking.
+pub fn raw_args(handler: &EarlyErrorHandler) -> Result<Vec<String>, ErrorGuaranteed> {
+    let mut res = Ok(Vec::new());
+    for (i, arg) in env::args_os().enumerate() {
+        match arg.into_string() {
+            Ok(arg) => {
+                if let Ok(args) = &mut res {
+                    args.push(arg);
+                }
+            }
+            Err(arg) => {
+                res = Err(handler
+                    .early_error_no_abort(format!("argument {i} is not valid Unicode: {arg:?}")))
+            }
         }
     }
-    args
+    res
 }
 
 #[derive(Debug)]
-pub enum Error {
-    Utf8Error(Option<String>),
+enum Error {
+    Utf8Error(String),
     IOError(String, io::Error),
 }
 
 impl fmt::Display for Error {
     fn fmt(&self, fmt: &mut fmt::Formatter<'_>) -> fmt::Result {
         match self {
-            Error::Utf8Error(None) => write!(fmt, "Utf8 error"),
-            Error::Utf8Error(Some(path)) => write!(fmt, "Utf8 error in {path}"),
-            Error::IOError(path, err) => write!(fmt, "IO Error: {path}: {err}"),
+            Error::Utf8Error(path) => write!(fmt, "UTF-8 error in {path}"),
+            Error::IOError(path, err) => write!(fmt, "IO error: {path}: {err}"),
         }
     }
 }

compiler/rustc_driver_impl/src/lib.rs

@@ -280,7 +280,7 @@ fn run_compiler(
     // the compiler with @empty_file as argv[0] and no more arguments.
     let at_args = at_args.get(1..).unwrap_or_default();
 
-    let args = args::arg_expand_all(&early_error_handler, at_args);
+    let args = args::arg_expand_all(&early_error_handler, at_args)?;
 
     let Some(matches) = handle_options(&early_error_handler, &args) else { return Ok(()) };
 
@@ -1565,17 +1565,8 @@ pub fn main() -> ! {
     signal_handler::install();
     let mut callbacks = TimePassesCallbacks::default();
     install_ice_hook(DEFAULT_BUG_REPORT_URL, |_| ());
-    let exit_code = catch_with_exit_code(|| {
-        let args = env::args_os()
-            .enumerate()
-            .map(|(i, arg)| {
-                arg.into_string().unwrap_or_else(|arg| {
-                    handler.early_error(format!("argument {i} is not valid Unicode: {arg:?}"))
-                })
-            })
-            .collect::<Vec<_>>();
-        RunCompiler::new(&args, &mut callbacks).run()
-    });
+    let exit_code =
+        catch_with_exit_code(|| RunCompiler::new(&args::raw_args(&handler)?, &mut callbacks).run());
 
     if let Some(format) = callbacks.time_passes {
         let end_rss = get_resident_set_size();

src/librustdoc/lib.rs

@@ -175,19 +175,15 @@ pub fn main() {
     init_logging(&handler);
     rustc_driver::init_env_logger(&handler, "RUSTDOC_LOG");
 
-    let exit_code = rustc_driver::catch_with_exit_code(|| match get_args(&handler) {
-        Some(args) => main_args(&mut handler, &args),
-        _ =>
-        {
-            #[allow(deprecated)]
-            Err(ErrorGuaranteed::unchecked_claim_error_was_emitted())
-        }
+    let exit_code = rustc_driver::catch_with_exit_code(|| {
+        let at_args = rustc_driver::args::raw_args(&handler)?;
+        main_args(&mut handler, &at_args)
     });
     process::exit(exit_code);
 }
 
 fn init_logging(handler: &EarlyErrorHandler) {
-    let color_logs = match std::env::var("RUSTDOC_LOG_COLOR").as_deref() {
+    let color_logs = match env::var("RUSTDOC_LOG_COLOR").as_deref() {
         Ok("always") => true,
         Ok("never") => false,
         Ok("auto") | Err(VarError::NotPresent) => io::stdout().is_terminal(),
@@ -217,19 +213,6 @@ fn init_logging(handler: &EarlyErrorHandler) {
     tracing::subscriber::set_global_default(subscriber).unwrap();
 }
 
-fn get_args(handler: &EarlyErrorHandler) -> Option<Vec<String>> {
-    env::args_os()
-        .enumerate()
-        .map(|(i, arg)| {
-            arg.into_string()
-                .map_err(|arg| {
-                    handler.early_warn(format!("Argument {i} is not valid Unicode: {arg:?}"));
-                })
-                .ok()
-        })
-        .collect()
-}
-
 fn opts() -> Vec<RustcOptGroup> {
     let stable: fn(_, fn(&mut getopts::Options) -> &mut _) -> _ = RustcOptGroup::stable;
     let unstable: fn(_, fn(&mut getopts::Options) -> &mut _) -> _ = RustcOptGroup::unstable;
@@ -715,7 +698,7 @@ fn main_args(handler: &mut EarlyErrorHandler, at_args: &[String]) -> MainResult
     // the compiler with @empty_file as argv[0] and no more arguments.
     let at_args = at_args.get(1..).unwrap_or_default();
 
-    let args = rustc_driver::args::arg_expand_all(handler, at_args);
+    let args = rustc_driver::args::arg_expand_all(handler, at_args)?;
 
     let mut options = getopts::Options::new();
     for option in opts() {

src/tools/clippy/src/driver.rs

@@ -209,7 +209,7 @@ pub fn main() {
     });
 
     exit(rustc_driver::catch_with_exit_code(move || {
-        let mut orig_args: Vec<String> = env::args().collect();
+        let mut orig_args: Vec<String> = rustc_driver::args::raw_args(&handler)?;
         let has_sysroot_arg = arg_value(&orig_args, "--sysroot", |_| true).is_some();
 
         let sys_root_env = std::env::var("SYSROOT").ok();

src/tools/miri/src/bin/miri.rs

@@ -291,6 +291,8 @@ fn main() {
     // (`install_ice_hook` might change `RUST_BACKTRACE`.)
     let env_snapshot = env::vars_os().collect::<Vec<_>>();
 
+    let args = rustc_driver::args::raw_args(&handler).unwrap_or_else(|_| std::process::exit(rustc_driver::EXIT_FAILURE));
+
     // If the environment asks us to actually be rustc, then do that.
     if let Some(crate_kind) = env::var_os("MIRI_BE_RUSTC") {
         // Earliest rustc setup.
@@ -307,7 +309,7 @@ fn main() {
 
         // We cannot use `rustc_driver::main` as we need to adjust the CLI arguments.
         run_compiler(
-            env::args().collect(),
+            args,
             target_crate,
             &mut MiriBeRustCompilerCalls { target_crate },
         )
@@ -328,7 +330,7 @@ fn main() {
 
     // If user has explicitly enabled/disabled isolation
     let mut isolation_enabled: Option<bool> = None;
-    for arg in env::args() {
+    for arg in args {
         if rustc_args.is_empty() {
             // Very first arg: binary name.
             rustc_args.push(arg);

src/tools/tidy/src/ui_tests.rs

@@ -11,7 +11,7 @@ use std::path::{Path, PathBuf};
 const ENTRY_LIMIT: usize = 900;
 // FIXME: The following limits should be reduced eventually.
 const ISSUES_ENTRY_LIMIT: usize = 1891;
-const ROOT_ENTRY_LIMIT: usize = 866;
+const ROOT_ENTRY_LIMIT: usize = 860;
 
 const EXPECTED_TEST_FILE_EXTENSIONS: &[&str] = &[
     "rs",     // test source files
@@ -25,8 +25,8 @@ const EXPECTED_TEST_FILE_EXTENSIONS: &[&str] = &[
 const EXTENSION_EXCEPTION_PATHS: &[&str] = &[
     "tests/ui/asm/named-asm-labels.s", // loading an external asm file to test named labels lint
     "tests/ui/check-cfg/my-awesome-platform.json", // testing custom targets with cfgs
-    "tests/ui/commandline-argfile-badutf8.args", // passing args via a file
-    "tests/ui/commandline-argfile.args", // passing args via a file
+    "tests/ui/argfile/commandline-argfile-badutf8.args", // passing args via a file
+    "tests/ui/argfile/commandline-argfile.args", // passing args via a file
     "tests/ui/crate-loading/auxiliary/libfoo.rlib", // testing loading a manually created rlib
     "tests/ui/include-macros/data.bin", // testing including data with the include macros
     "tests/ui/include-macros/file.txt", // testing including data with the include macros

tests/rustdoc-ui/commandline-argfile-badutf8.rs → tests/rustdoc-ui/argfile/commandline-argfile-badutf8.rs

@@ -1,6 +1,6 @@
 // Check to see if we can get parameters from an @argsfile file
 //
-// compile-flags: --cfg cmdline_set @{{src-base}}/commandline-argfile-badutf8.args
+// compile-flags: --cfg cmdline_set @{{src-base}}/argfile/commandline-argfile-badutf8.args
 
 #[cfg(not(cmdline_set))]
 compile_error!("cmdline_set not set");

tests/rustdoc-ui/argfile/commandline-argfile-badutf8.stderr

@@ -0,0 +1,2 @@
+error: failed to load argument file: UTF-8 error in $DIR/commandline-argfile-badutf8.args
+

tests/ui/commandline-argfile-missing.rs → tests/rustdoc-ui/argfile/commandline-argfile-missing.rs

@@ -2,7 +2,7 @@
 //
 // normalize-stderr-test: "os error \d+" -> "os error $$ERR"
 // normalize-stderr-test: "commandline-argfile-missing.args:[^(]*" -> "commandline-argfile-missing.args: $$FILE_MISSING "
-// compile-flags: --cfg cmdline_set @{{src-base}}/commandline-argfile-missing.args
+// compile-flags: --cfg cmdline_set @{{src-base}}/argfile/commandline-argfile-missing.args
 
 #[cfg(not(cmdline_set))]
 compile_error!("cmdline_set not set");

tests/rustdoc-ui/argfile/commandline-argfile-missing.stderr

@@ -0,0 +1,2 @@
+error: failed to load argument file: IO error: $DIR/commandline-argfile-missing.args: $FILE_MISSING (os error $ERR)
+

tests/rustdoc-ui/argfile/commandline-argfile-multiple.rs

@@ -0,0 +1,14 @@
+// Check to see if we can get parameters from an @argsfile file
+//
+// normalize-stderr-test: "os error \d+" -> "os error $$ERR"
+// normalize-stderr-test: "commandline-argfile-missing.args:[^(]*" -> "commandline-argfile-missing.args: $$FILE_MISSING "
+// compile-flags: --cfg cmdline_set @{{src-base}}/argfile/commandline-argfile-missing.args @{{src-base}}/argfile/commandline-argfile-badutf8.args @{{src-base}}/argfile/commandline-argfile-missing2.args
+
+#[cfg(not(cmdline_set))]
+compile_error!("cmdline_set not set");
+
+#[cfg(not(unbroken))]
+compile_error!("unbroken not set");
+
+fn main() {
+}

tests/rustdoc-ui/argfile/commandline-argfile-multiple.stderr

@@ -0,0 +1,6 @@
+error: failed to load argument file: IO error: $DIR/commandline-argfile-missing.args: $FILE_MISSING (os error $ERR)
+
+error: failed to load argument file: UTF-8 error in $DIR/commandline-argfile-badutf8.args
+
+error: failed to load argument file: IO error: $DIR/commandline-argfile-missing2.args: No such file or directory (os error $ERR)
+

tests/rustdoc-ui/commandline-argfile.rs → tests/rustdoc-ui/argfile/commandline-argfile.rs

@@ -1,7 +1,7 @@
 // Check to see if we can get parameters from an @argsfile file
 //
 // check-pass
-// compile-flags: --cfg cmdline_set @{{src-base}}/commandline-argfile.args
+// compile-flags: --cfg cmdline_set @{{src-base}}/argfile/commandline-argfile.args
 
 #[cfg(not(cmdline_set))]
 compile_error!("cmdline_set not set");

tests/ui/commandline-argfile-badutf8.rs → tests/ui/argfile/commandline-argfile-badutf8.rs

@@ -1,6 +1,6 @@
 // Check to see if we can get parameters from an @argsfile file
 //
-// compile-flags: --cfg cmdline_set @{{src-base}}/commandline-argfile-badutf8.args
+// compile-flags: --cfg cmdline_set @{{src-base}}/argfile/commandline-argfile-badutf8.args
 
 #[cfg(not(cmdline_set))]
 compile_error!("cmdline_set not set");

tests/ui/argfile/commandline-argfile-badutf8.stderr

@@ -0,0 +1,2 @@
+error: failed to load argument file: UTF-8 error in $DIR/commandline-argfile-badutf8.args
+

tests/rustdoc-ui/commandline-argfile-missing.rs → tests/ui/argfile/commandline-argfile-missing.rs

@@ -2,7 +2,7 @@
 //
 // normalize-stderr-test: "os error \d+" -> "os error $$ERR"
 // normalize-stderr-test: "commandline-argfile-missing.args:[^(]*" -> "commandline-argfile-missing.args: $$FILE_MISSING "
-// compile-flags: --cfg cmdline_set @{{src-base}}/commandline-argfile-missing.args
+// compile-flags: --cfg cmdline_set @{{src-base}}/argfile/commandline-argfile-missing.args
 
 #[cfg(not(cmdline_set))]
 compile_error!("cmdline_set not set");

tests/ui/argfile/commandline-argfile-missing.stderr

@@ -0,0 +1,2 @@
+error: failed to load argument file: IO error: $DIR/commandline-argfile-missing.args: $FILE_MISSING (os error $ERR)
+

tests/ui/argfile/commandline-argfile-multiple.rs

@@ -0,0 +1,14 @@
+// Check to see if we can get parameters from an @argsfile file
+//
+// normalize-stderr-test: "os error \d+" -> "os error $$ERR"
+// normalize-stderr-test: "commandline-argfile-missing.args:[^(]*" -> "commandline-argfile-missing.args: $$FILE_MISSING "
+// compile-flags: --cfg cmdline_set @{{src-base}}/argfile/commandline-argfile-missing.args @{{src-base}}/argfile/commandline-argfile-badutf8.args @{{src-base}}/argfile/commandline-argfile-missing2.args
+
+#[cfg(not(cmdline_set))]
+compile_error!("cmdline_set not set");
+
+#[cfg(not(unbroken))]
+compile_error!("unbroken not set");
+
+fn main() {
+}

tests/ui/argfile/commandline-argfile-multiple.stderr

@@ -0,0 +1,6 @@
+error: failed to load argument file: IO error: $DIR/commandline-argfile-missing.args: $FILE_MISSING (os error $ERR)
+
+error: failed to load argument file: UTF-8 error in $DIR/commandline-argfile-badutf8.args
+
+error: failed to load argument file: IO error: $DIR/commandline-argfile-missing2.args: No such file or directory (os error $ERR)
+

tests/ui/commandline-argfile.rs → tests/ui/argfile/commandline-argfile.rs

@@ -1,7 +1,7 @@
 // Check to see if we can get parameters from an @argsfile file
 //
 // build-pass
-// compile-flags: --cfg cmdline_set @{{src-base}}/commandline-argfile.args
+// compile-flags: --cfg cmdline_set @{{src-base}}/argfile/commandline-argfile.args
 
 #[cfg(not(cmdline_set))]
 compile_error!("cmdline_set not set");