locate, updatedb, xargs: avoid panic on a non-UTF-8 argument

src/locate/main.rs

@@ -3,7 +3,21 @@
 // https://opensource.org/licenses/MIT.
 
 fn main() {
-    let args = std::env::args().collect::<Vec<String>>();
+    // `std::env::args` panics on a non-UTF-8 argument, so read the arguments
+    // with `args_os` and report the first invalid one instead of aborting.
+    let args = match std::env::args_os()
+        .map(std::ffi::OsString::into_string)
+        .collect::<Result<Vec<String>, _>>()
+    {
+        Ok(args) => args,
+        Err(invalid) => {
+            eprintln!(
+                "locate: invalid (non-UTF-8) argument: {}",
+                invalid.to_string_lossy()
+            );
+            std::process::exit(1);
+        }
+    };
     let strs: Vec<&str> = args.iter().map(std::convert::AsRef::as_ref).collect();
     std::process::exit(findutils::locate::locate_main(strs.as_slice()));
 }

src/updatedb/main.rs

@@ -3,7 +3,21 @@
 // https://opensource.org/licenses/MIT.
 
 fn main() {
-    let args = std::env::args().collect::<Vec<String>>();
+    // `std::env::args` panics on a non-UTF-8 argument, so read the arguments
+    // with `args_os` and report the first invalid one instead of aborting.
+    let args = match std::env::args_os()
+        .map(std::ffi::OsString::into_string)
+        .collect::<Result<Vec<String>, _>>()
+    {
+        Ok(args) => args,
+        Err(invalid) => {
+            eprintln!(
+                "updatedb: invalid (non-UTF-8) argument: {}",
+                invalid.to_string_lossy()
+            );
+            std::process::exit(1);
+        }
+    };
     let strs: Vec<&str> = args.iter().map(std::convert::AsRef::as_ref).collect();
     std::process::exit(findutils::updatedb::updatedb_main(strs.as_slice()));
 }

src/xargs/main.rs

@@ -5,7 +5,21 @@
 // https://opensource.org/licenses/MIT.
 
 fn main() {
-    let args = std::env::args().collect::<Vec<String>>();
+    // `std::env::args` panics on a non-UTF-8 argument, so read the arguments
+    // with `args_os` and report the first invalid one instead of aborting.
+    let args = match std::env::args_os()
+        .map(std::ffi::OsString::into_string)
+        .collect::<Result<Vec<String>, _>>()
+    {
+        Ok(args) => args,
+        Err(invalid) => {
+            eprintln!(
+                "xargs: invalid (non-UTF-8) argument: {}",
+                invalid.to_string_lossy()
+            );
+            std::process::exit(1);
+        }
+    };
     std::process::exit(findutils::xargs::xargs_main(
         &args
             .iter()

tests/db_tests.rs

@@ -323,3 +323,41 @@ fn test_locate_one_byte_db() {
         .assert()
         .code(1);
 }
+
+#[cfg(unix)]
+#[test]
+fn test_locate_non_utf8_argument() {
+    // A non-UTF-8 argument must produce a clean error, not a panic in
+    // std::env::args.
+    use std::ffi::OsString;
+    use std::os::unix::ffi::OsStringExt;
+    let assert = Command::cargo_bin("locate")
+        .expect("couldn't find locate binary")
+        .arg(OsString::from_vec(vec![0xff]))
+        .assert()
+        .code(1);
+    let stderr = String::from_utf8_lossy(&assert.get_output().stderr);
+    assert!(
+        stderr.contains("non-UTF-8"),
+        "expected an error naming the invalid argument, got: {stderr:?}"
+    );
+}
+
+#[cfg(unix)]
+#[test]
+fn test_updatedb_non_utf8_argument() {
+    // A non-UTF-8 argument must produce a clean error, not a panic in
+    // std::env::args.
+    use std::ffi::OsString;
+    use std::os::unix::ffi::OsStringExt;
+    let assert = Command::cargo_bin("updatedb")
+        .expect("couldn't find updatedb binary")
+        .arg(OsString::from_vec(vec![0xff]))
+        .assert()
+        .code(1);
+    let stderr = String::from_utf8_lossy(&assert.get_output().stderr);
+    assert!(
+        stderr.contains("non-UTF-8"),
+        "expected an error naming the invalid argument, got: {stderr:?}"
+    );
+}

tests/test_xargs.rs

@@ -601,3 +601,17 @@ fn xargs_arg_file_missing_strips_errno() {
         .stderr_contains(MISSING)
         .stderr_str_check(|s| !s.contains("(os error"));
 }
+
+#[cfg(unix)]
+#[test]
+fn xargs_non_utf8_argument_is_rejected_gracefully() {
+    // A non-UTF-8 argument must produce a clean error, not a panic in
+    // std::env::args.
+    use std::ffi::OsString;
+    use std::os::unix::ffi::OsStringExt;
+    ucmd()
+        .arg(OsString::from_vec(vec![b'-', b'I', 0xff]))
+        .fails_with_code(1)
+        .stderr_contains("non-UTF-8")
+        .no_stdout();
+}