Merge pull request #8371 from drinkcat/df-osstr

df: Move to using `OsString`

Cargo.lock

@@ -3977,6 +3977,7 @@ dependencies = [
  "bigdecimal",
  "blake2b_simd",
  "blake3",
+ "bstr",
  "chrono",
  "clap",
  "crc32fast",

fuzz/Cargo.lock

@@ -1584,6 +1584,7 @@ dependencies = [
  "bigdecimal",
  "blake2b_simd",
  "blake3",
+ "bstr",
  "clap",
  "crc32fast",
  "data-encoding",

src/uu/comm/src/comm.rs

@@ -101,7 +101,7 @@ impl OrderChecker {
             return true;
         }
 
-        let is_ordered = current_line >= &self.last_line;
+        let is_ordered = *current_line >= *self.last_line;
         if !is_ordered && !self.has_error {
             eprintln!(
                 "{}",

src/uu/df/src/df.rs

@@ -20,6 +20,7 @@ use uucore::{format_usage, show};
 use clap::{Arg, ArgAction, ArgMatches, Command, parser::ValueSource};
 
 use std::ffi::OsString;
+use std::io::stdout;
 use std::path::Path;
 use thiserror::Error;
 
@@ -431,7 +432,7 @@ pub fn uumain(args: impl uucore::Args) -> UResult<()> {
 
     let opt = Options::from(&matches).map_err(DfError::OptionsError)?;
     // Get the list of filesystems to display in the output table.
-    let filesystems: Vec<Filesystem> = match matches.get_many::<String>(OPT_PATHS) {
+    let filesystems: Vec<Filesystem> = match matches.get_many::<OsString>(OPT_PATHS) {
         None => {
             let filesystems = get_all_filesystems(&opt).map_err(|e| {
                 let context = get_message("df-error-cannot-read-table-of-mounted-filesystems");
@@ -464,7 +465,7 @@ pub fn uumain(args: impl uucore::Args) -> UResult<()> {
         }
     };
 
-    println!("{}", Table::new(&opt, filesystems));
+    Table::new(&opt, filesystems).write_to(&mut stdout())?;
 
     Ok(())
 }
@@ -611,6 +612,7 @@ pub fn uu_app() -> Command {
         .arg(
             Arg::new(OPT_PATHS)
                 .action(ArgAction::Append)
+                .value_parser(ValueParser::os_string())
                 .value_hint(clap::ValueHint::AnyPath),
         )
 }
@@ -629,9 +631,9 @@ mod tests {
                 dev_id: String::new(),
                 dev_name: String::from(dev_name),
                 fs_type: String::new(),
-                mount_dir: String::from(mount_dir),
+                mount_dir: mount_dir.into(),
                 mount_option: String::new(),
-                mount_root: String::from(mount_root),
+                mount_root: mount_root.into(),
                 remote: false,
                 dummy: false,
             }
@@ -679,9 +681,9 @@ mod tests {
                 dev_id: String::from(dev_id),
                 dev_name: String::new(),
                 fs_type: String::new(),
-                mount_dir: String::from(mount_dir),
+                mount_dir: mount_dir.into(),
                 mount_option: String::new(),
-                mount_root: String::new(),
+                mount_root: "/".into(),
                 remote: false,
                 dummy: false,
             }
@@ -724,9 +726,9 @@ mod tests {
                 dev_id: String::new(),
                 dev_name: String::new(),
                 fs_type: String::from(fs_type),
-                mount_dir: String::from(mount_dir),
+                mount_dir: mount_dir.into(),
                 mount_option: String::new(),
-                mount_root: String::new(),
+                mount_root: "/".into(),
                 remote,
                 dummy,
             }

src/uu/df/src/filesystem.rs

@@ -8,7 +8,7 @@
 //! filesystem mounted at a particular directory. It also includes
 //! information on amount of space available and amount of space used.
 // spell-checker:ignore canonicalized
-use std::path::Path;
+use std::{ffi::OsString, path::Path};
 
 #[cfg(unix)]
 use uucore::fsext::statfs;
@@ -28,7 +28,7 @@ pub(crate) struct Filesystem {
     /// When invoking `df` with a positional argument, it displays
     /// usage information for the filesystem that contains the given
     /// file. If given, this field contains that filename.
-    pub file: Option<String>,
+    pub file: Option<OsString>,
 
     /// Information about the mounted device, mount directory, and related options.
     pub mount_info: MountInfo,
@@ -123,22 +123,22 @@ where
 
 impl Filesystem {
     // TODO: resolve uuid in `mount_info.dev_name` if exists
-    pub(crate) fn new(mount_info: MountInfo, file: Option<String>) -> Option<Self> {
+    pub(crate) fn new(mount_info: MountInfo, file: Option<OsString>) -> Option<Self> {
         let _stat_path = if mount_info.mount_dir.is_empty() {
             #[cfg(unix)]
             {
-                mount_info.dev_name.clone()
+                mount_info.dev_name.clone().into()
             }
             #[cfg(windows)]
             {
                 // On windows, we expect the volume id
-                mount_info.dev_id.clone()
+                mount_info.dev_id.clone().into()
             }
         } else {
             mount_info.mount_dir.clone()
         };
         #[cfg(unix)]
-        let usage = FsUsage::new(statfs(_stat_path).ok()?);
+        let usage = FsUsage::new(statfs(&_stat_path).ok()?);
         #[cfg(windows)]
         let usage = FsUsage::new(Path::new(&_stat_path)).ok()?;
         Some(Self {
@@ -154,7 +154,7 @@ impl Filesystem {
     pub(crate) fn from_mount(
         mounts: &[MountInfo],
         mount: &MountInfo,
-        file: Option<String>,
+        file: Option<OsString>,
     ) -> Result<Self, FsError> {
         if is_over_mounted(mounts, mount) {
             Err(FsError::OverMounted)
@@ -165,7 +165,7 @@ impl Filesystem {
 
     /// Find and create the filesystem from the given mount.
     #[cfg(windows)]
-    pub(crate) fn from_mount(mount: &MountInfo, file: Option<String>) -> Result<Self, FsError> {
+    pub(crate) fn from_mount(mount: &MountInfo, file: Option<OsString>) -> Result<Self, FsError> {
         Self::new(mount.clone(), file).ok_or(FsError::MountMissing)
     }
 
@@ -189,7 +189,7 @@ impl Filesystem {
     where
         P: AsRef<Path>,
     {
-        let file = path.as_ref().display().to_string();
+        let file = path.as_ref().as_os_str().to_owned();
         let canonicalize = true;
 
         let result = mount_info_from_path(mounts, path, canonicalize);
@@ -205,6 +205,8 @@ mod tests {
 
     mod mount_info_from_path {
 
+        use std::ffi::OsString;
+
         use uucore::fsext::MountInfo;
 
         use crate::filesystem::{FsError, mount_info_from_path};
@@ -215,9 +217,9 @@ mod tests {
                 dev_id: String::default(),
                 dev_name: String::default(),
                 fs_type: String::default(),
-                mount_dir: String::from(mount_dir),
+                mount_dir: OsString::from(mount_dir),
                 mount_option: String::default(),
-                mount_root: String::default(),
+                mount_root: OsString::default(),
                 remote: Default::default(),
                 dummy: Default::default(),
             }
@@ -312,6 +314,8 @@ mod tests {
 
     #[cfg(not(windows))]
     mod over_mount {
+        use std::ffi::OsString;
+
         use crate::filesystem::{Filesystem, FsError, is_over_mounted};
         use uucore::fsext::MountInfo;
 
@@ -320,9 +324,9 @@ mod tests {
                 dev_id: String::default(),
                 dev_name: dev_name.map(String::from).unwrap_or_default(),
                 fs_type: String::default(),
-                mount_dir: String::from(mount_dir),
+                mount_dir: OsString::from(mount_dir),
                 mount_option: String::default(),
-                mount_root: String::default(),
+                mount_root: OsString::default(),
                 remote: Default::default(),
                 dummy: Default::default(),
             }

src/uu/stat/src/stat.rs

@@ -22,7 +22,6 @@ use std::ffi::{OsStr, OsString};
 use std::fs::{FileType, Metadata};
 use std::io::Write;
 use std::os::unix::fs::{FileTypeExt, MetadataExt};
-use std::os::unix::prelude::OsStrExt;
 use std::path::Path;
 use std::{env, fs};
 
@@ -258,7 +257,7 @@ struct Stater {
     show_fs: bool,
     from_user: bool,
     files: Vec<OsString>,
-    mount_list: Option<Vec<String>>,
+    mount_list: Option<Vec<OsString>>,
     default_tokens: Vec<Token>,
     default_dev_tokens: Vec<Token>,
 }
@@ -876,7 +875,7 @@ impl Stater {
                 })?
                 .iter()
                 .map(|mi| mi.mount_dir.clone())
-                .collect::<Vec<String>>();
+                .collect::<Vec<_>>();
             // Reverse sort. The longer comes first.
             mount_list.sort();
             mount_list.reverse();
@@ -899,7 +898,8 @@ impl Stater {
 
         for root in self.mount_list.as_ref()? {
             if path.starts_with(root) {
-                return Some(root.clone());
+                // TODO: This is probably wrong, we should pass the OsString
+                return Some(root.to_string_lossy().into_owned());
             }
         }
         None
@@ -992,7 +992,7 @@ impl Stater {
                     'h' => OutputType::Unsigned(meta.nlink()),
                     // inode number
                     'i' => OutputType::Unsigned(meta.ino()),
-                    // mount point
+                    // mount point: TODO: This should be an OsStr
                     'm' => OutputType::Str(self.find_mount_point(file).unwrap()),
                     // file name
                     'n' => OutputType::Str(display_name.to_string()),
@@ -1092,11 +1092,7 @@ impl Stater {
             OsString::from(file)
         };
         if self.show_fs {
-            #[cfg(unix)]
+            match statfs(&file) {
-            let p = file.as_bytes();
-            #[cfg(not(unix))]
-            let p = file.into_string().unwrap();
-            match statfs(p) {
                 Ok(meta) => {
                     let tokens = &self.default_tokens;
 

src/uucore/Cargo.toml

@@ -19,6 +19,7 @@ all-features = true
 path = "src/lib/lib.rs"
 
 [dependencies]
+bstr = { workspace = true }
 chrono = { workspace = true, optional = true }
 clap = { workspace = true }
 uucore_procs = { workspace = true }

src/uucore/src/lib/features/fsext.rs

@@ -19,17 +19,18 @@ const MAX_PATH: usize = 266;
 static EXIT_ERR: i32 = 1;
 
 #[cfg(any(
-    windows,
     target_os = "freebsd",
     target_vendor = "apple",
     target_os = "netbsd",
     target_os = "openbsd"
 ))]
+use crate::os_str_from_bytes;
 #[cfg(windows)]
 use crate::show_warning;
 
-#[cfg(windows)]
 use std::ffi::OsStr;
+#[cfg(unix)]
+use std::os::unix::ffi::OsStrExt;
 #[cfg(windows)]
 use std::os::windows::ffi::OsStrExt;
 #[cfg(windows)]
@@ -61,17 +62,15 @@ fn to_nul_terminated_wide_string(s: impl AsRef<OsStr>) -> Vec<u16> {
 use libc::{
     S_IFBLK, S_IFCHR, S_IFDIR, S_IFIFO, S_IFLNK, S_IFMT, S_IFREG, S_IFSOCK, mode_t, strerror,
 };
-use std::borrow::Cow;
 #[cfg(unix)]
-use std::ffi::CStr;
+use std::ffi::{CStr, CString};
-#[cfg(unix)]
-use std::ffi::CString;
 use std::io::Error as IOError;
 #[cfg(unix)]
 use std::mem;
 #[cfg(windows)]
 use std::path::Path;
 use std::time::UNIX_EPOCH;
+use std::{borrow::Cow, ffi::OsString};
 
 #[cfg(any(
     target_os = "linux",
@@ -123,14 +122,16 @@ impl BirthTime for Metadata {
     }
 }
 
+// TODO: Types for this struct are probably mostly wrong. Possibly, most of them
+// should be OsString.
 #[derive(Debug, Clone)]
 pub struct MountInfo {
     /// Stores `volume_name` in windows platform and `dev_id` in unix platform
     pub dev_id: String,
     pub dev_name: String,
     pub fs_type: String,
-    pub mount_root: String,
+    pub mount_root: OsString,
-    pub mount_dir: String,
+    pub mount_dir: OsString,
     /// We only care whether this field contains "bind"
     pub mount_option: String,
     pub remote: bool,
@@ -138,7 +139,9 @@ pub struct MountInfo {
 }
 
 #[cfg(any(target_os = "linux", target_os = "android"))]
-fn replace_special_chars(s: String) -> String {
+fn replace_special_chars(s: &[u8]) -> Vec<u8> {
+    use bstr::ByteSlice;
+
     // Replace
     //
     // * ASCII space with a regular space character,
@@ -152,7 +155,11 @@ fn replace_special_chars(s: String) -> String {
 
 impl MountInfo {
     #[cfg(any(target_os = "linux", target_os = "android"))]
-    fn new(file_name: &str, raw: &[&str]) -> Option<Self> {
+    fn new(file_name: &str, raw: &[&[u8]]) -> Option<Self> {
+        use std::ffi::OsStr;
+        use std::os::unix::ffi::OsStrExt;
+        use std::os::unix::ffi::OsStringExt;
+
         let dev_name;
         let fs_type;
         let mount_root;
@@ -165,21 +172,24 @@ impl MountInfo {
             // "man proc" for more details
             LINUX_MOUNTINFO => {
                 const FIELDS_OFFSET: usize = 6;
-                let after_fields = raw[FIELDS_OFFSET..].iter().position(|c| *c == "-").unwrap()
+                let after_fields = raw[FIELDS_OFFSET..]
+                    .iter()
+                    .position(|c| *c == b"-")
+                    .unwrap()
                     + FIELDS_OFFSET
                     + 1;
-                dev_name = raw[after_fields + 1].to_string();
+                dev_name = String::from_utf8_lossy(raw[after_fields + 1]).to_string();
-                fs_type = raw[after_fields].to_string();
+                fs_type = String::from_utf8_lossy(raw[after_fields]).to_string();
-                mount_root = raw[3].to_string();
+                mount_root = OsStr::from_bytes(raw[3]).to_owned();
-                mount_dir = replace_special_chars(raw[4].to_string());
+                mount_dir = OsString::from_vec(replace_special_chars(raw[4]));
-                mount_option = raw[5].to_string();
+                mount_option = String::from_utf8_lossy(raw[5]).to_string();
             }
             LINUX_MTAB => {
-                dev_name = raw[0].to_string();
+                dev_name = String::from_utf8_lossy(raw[0]).to_string();
-                fs_type = raw[2].to_string();
+                fs_type = String::from_utf8_lossy(raw[2]).to_string();
-                mount_root = String::new();
+                mount_root = OsString::new();
-                mount_dir = replace_special_chars(raw[1].to_string());
+                mount_dir = OsString::from_vec(replace_special_chars(raw[1]));
-                mount_option = raw[3].to_string();
+                mount_option = String::from_utf8_lossy(raw[3]).to_string();
             }
             _ => return None,
         };
@@ -233,6 +243,8 @@ impl MountInfo {
             // TODO: support the case when `GetLastError()` returns `ERROR_MORE_DATA`
             return None;
         }
+        // TODO: This should probably call `OsString::from_wide`, but unclear if
+        // terminating zeros need to be striped first.
         let mount_root = LPWSTR2String(&mount_root_buf);
 
         let mut fs_type_buf = [0u16; MAX_PATH];
@@ -263,8 +275,8 @@ impl MountInfo {
             dev_id: volume_name,
             dev_name,
             fs_type: fs_type.unwrap_or_default(),
-            mount_root,
+            mount_root: mount_root.into(), // TODO: We should figure out how to keep an OsString here.
-            mount_dir: String::new(),
+            mount_dir: OsString::new(),
             mount_option: String::new(),
             remote,
             dummy: false,
@@ -292,12 +304,11 @@ impl From<StatFs> for MountInfo {
                 .to_string_lossy()
                 .into_owned()
         };
-        let mount_dir = unsafe {
+        let mount_dir_bytes = unsafe {
             // spell-checker:disable-next-line
-            CStr::from_ptr(&statfs.f_mntonname[0])
+            CStr::from_ptr(&statfs.f_mntonname[0]).to_bytes()
-                .to_string_lossy()
-                .into_owned()
         };
+        let mount_dir = os_str_from_bytes(mount_dir_bytes).unwrap().into_owned();
 
         let dev_id = mount_dev_id(&mount_dir);
         let dummy = is_dummy_filesystem(&fs_type, "");
@@ -308,7 +319,7 @@ impl From<StatFs> for MountInfo {
             dev_name,
             fs_type,
             mount_dir,
-            mount_root: String::new(),
+            mount_root: OsString::new(),
             mount_option: String::new(),
             remote,
             dummy,
@@ -343,7 +354,7 @@ fn is_remote_filesystem(dev_name: &str, fs_type: &str) -> bool {
 }
 
 #[cfg(all(unix, not(any(target_os = "aix", target_os = "redox"))))]
-fn mount_dev_id(mount_dir: &str) -> String {
+fn mount_dev_id(mount_dir: &OsStr) -> String {
     use std::os::unix::fs::MetadataExt;
 
     if let Ok(stat) = std::fs::metadata(mount_dir) {
@@ -426,10 +437,10 @@ pub fn read_fs_list() -> UResult<Vec<MountInfo>> {
             .or_else(|_| File::open(LINUX_MTAB).map(|f| (LINUX_MTAB, f)))?;
         let reader = BufReader::new(f);
         Ok(reader
-            .lines()
+            .split(b'\n')
             .map_while(Result::ok)
             .filter_map(|line| {
-                let raw_data = line.split_whitespace().collect::<Vec<&str>>();
+                let raw_data = line.split(|c| *c == b' ').collect::<Vec<&[u8]>>();
                 MountInfo::new(file_name, &raw_data)
             })
             .collect::<Vec<_>>())
@@ -855,11 +866,13 @@ impl FsMeta for StatFs {
 }
 
 #[cfg(unix)]
-pub fn statfs<P>(path: P) -> Result<StatFs, String>
+pub fn statfs(path: &OsStr) -> Result<StatFs, String> {
-where
+    #[cfg(unix)]
-    P: Into<Vec<u8>>,
+    let p = path.as_bytes();
-{
+    #[cfg(not(unix))]
-    match CString::new(path) {
+    let p = path.into_string().unwrap();
+
+    match CString::new(p) {
         Ok(p) => {
             let mut buffer: StatFs = unsafe { mem::zeroed() };
             unsafe {
@@ -1060,8 +1073,8 @@ mod tests {
         // spell-checker:ignore (word) relatime
         let info = MountInfo::new(
             LINUX_MOUNTINFO,
-            &"106 109 253:6 / /mnt rw,relatime - xfs /dev/fs0 rw"
+            &b"106 109 253:6 / /mnt rw,relatime - xfs /dev/fs0 rw"
-                .split_ascii_whitespace()
+                .split(|c| *c == b' ')
                 .collect::<Vec<_>>(),
         )
         .unwrap();
@@ -1075,8 +1088,8 @@ mod tests {
         // Test parsing with different amounts of optional fields.
         let info = MountInfo::new(
             LINUX_MOUNTINFO,
-            &"106 109 253:6 / /mnt rw,relatime master:1 - xfs /dev/fs0 rw"
+            &b"106 109 253:6 / /mnt rw,relatime master:1 - xfs /dev/fs0 rw"
-                .split_ascii_whitespace()
+                .split(|c| *c == b' ')
                 .collect::<Vec<_>>(),
         )
         .unwrap();
@@ -1086,8 +1099,8 @@ mod tests {
 
         let info = MountInfo::new(
             LINUX_MOUNTINFO,
-            &"106 109 253:6 / /mnt rw,relatime master:1 shared:2 - xfs /dev/fs0 rw"
+            &b"106 109 253:6 / /mnt rw,relatime master:1 shared:2 - xfs /dev/fs0 rw"
-                .split_ascii_whitespace()
+                .split(|c| *c == b' ')
                 .collect::<Vec<_>>(),
         )
         .unwrap();
@@ -1101,8 +1114,8 @@ mod tests {
     fn test_mountinfo_dir_special_chars() {
         let info = MountInfo::new(
             LINUX_MOUNTINFO,
-            &r#"317 61 7:0 / /mnt/f\134\040\011oo rw,relatime shared:641 - ext4 /dev/loop0 rw"#
+            &br#"317 61 7:0 / /mnt/f\134\040\011oo rw,relatime shared:641 - ext4 /dev/loop0 rw"#
-                .split_ascii_whitespace()
+                .split(|c| *c == b' ')
                 .collect::<Vec<_>>(),
         )
         .unwrap();
@@ -1111,12 +1124,43 @@ mod tests {
 
         let info = MountInfo::new(
             LINUX_MTAB,
-            &r#"/dev/loop0 /mnt/f\134\040\011oo ext4 rw,relatime 0 0"#
+            &br#"/dev/loop0 /mnt/f\134\040\011oo ext4 rw,relatime 0 0"#
-                .split_ascii_whitespace()
+                .split(|c| *c == b' ')
                 .collect::<Vec<_>>(),
         )
         .unwrap();
 
         assert_eq!(info.mount_dir, r#"/mnt/f\ 	oo"#);
     }
+
+    #[test]
+    #[cfg(any(target_os = "linux", target_os = "android"))]
+    fn test_mountinfo_dir_non_unicode() {
+        let info = MountInfo::new(
+            LINUX_MOUNTINFO,
+            &b"317 61 7:0 / /mnt/some-\xc0-dir-\xf3 rw,relatime shared:641 - ext4 /dev/loop0 rw"
+                .split(|c| *c == b' ')
+                .collect::<Vec<_>>(),
+        )
+        .unwrap();
+
+        assert_eq!(
+            info.mount_dir,
+            crate::os_str_from_bytes(b"/mnt/some-\xc0-dir-\xf3").unwrap()
+        );
+
+        let info = MountInfo::new(
+            LINUX_MOUNTINFO,
+            &b"317 61 7:0 / /mnt/some-\\040-dir-\xf3 rw,relatime shared:641 - ext4 /dev/loop0 rw"
+                .split(|c| *c == b' ')
+                .collect::<Vec<_>>(),
+        )
+        .unwrap();
+
+        // Note that the \040 above will have been substituted by a space.
+        assert_eq!(
+            info.mount_dir,
+            crate::os_str_from_bytes(b"/mnt/some- -dir-\xf3").unwrap()
+        );
+    }
 }

tests/by-util/test_df.rs

@@ -117,7 +117,7 @@ fn test_df_output() {
         .arg("-H")
         .arg("--total")
         .succeeds()
-        .stdout_move_str();
+        .stdout_str_lossy();
     let actual = output.lines().take(1).collect::<Vec<&str>>()[0];
     let actual = actual.split_whitespace().collect::<Vec<_>>();
     assert_eq!(actual, expected);
@@ -151,7 +151,7 @@ fn test_df_output_overridden() {
         .arg("-hH")
         .arg("--total")
         .succeeds()
-        .stdout_move_str();
+        .stdout_str_lossy();
     let actual = output.lines().take(1).collect::<Vec<&str>>()[0];
     let actual = actual.split_whitespace().collect::<Vec<_>>();
     assert_eq!(actual, expected);
@@ -181,7 +181,7 @@ fn test_default_headers() {
             "on",
         ]
     };
-    let output = new_ucmd!().succeeds().stdout_move_str();
+    let output = new_ucmd!().succeeds().stdout_str_lossy();
     let actual = output.lines().take(1).collect::<Vec<&str>>()[0];
     let actual = actual.split_whitespace().collect::<Vec<_>>();
     assert_eq!(actual, expected);
@@ -195,7 +195,7 @@ fn test_precedence_of_human_readable_and_si_header_over_output_header() {
         let output = new_ucmd!()
             .args(&[arg, "--output=size"])
             .succeeds()
-            .stdout_move_str();
+            .stdout_str_lossy();
         let header = output.lines().next().unwrap();
         assert_eq!(header, " Size");
     }
@@ -207,7 +207,7 @@ fn test_used_header_starts_with_space() {
         // using -h here to ensure the width of the column's content is <= 4
         .args(&["-h", "--output=used"])
         .succeeds()
-        .stdout_move_str();
+        .stdout_str_lossy();
     let header = output.lines().next().unwrap();
     assert_eq!(header, " Used");
 }
@@ -226,11 +226,11 @@ fn test_order_same() {
     let output1 = new_ucmd!()
         .arg("--output=source")
         .succeeds()
-        .stdout_move_str();
+        .stdout_str_lossy();
     let output2 = new_ucmd!()
         .arg("--output=source")
         .succeeds()
-        .stdout_move_str();
+        .stdout_str_lossy();
     assert_eq!(output1, output2);
 }
 
@@ -238,7 +238,7 @@ fn test_order_same() {
 #[cfg(all(unix, not(target_os = "freebsd")))] // FIXME: fix this test for FreeBSD
 #[test]
 fn test_output_mp_repeat() {
-    let output1 = new_ucmd!().arg("/").arg("/").succeeds().stdout_move_str();
+    let output1 = new_ucmd!().arg("/").arg("/").succeeds().stdout_str_lossy();
     let output1: Vec<String> = output1
         .lines()
         .map(|l| String::from(l.split_once(' ').unwrap().0))
@@ -272,7 +272,7 @@ fn test_type_option() {
     let fs_types = new_ucmd!()
         .arg("--output=fstype")
         .succeeds()
-        .stdout_move_str();
+        .stdout_str_lossy();
     let fs_type = fs_types.lines().nth(1).unwrap().trim();
 
     new_ucmd!().args(&["-t", fs_type]).succeeds();
@@ -292,7 +292,7 @@ fn test_type_option_with_file() {
     let fs_type = new_ucmd!()
         .args(&["--output=fstype", "."])
         .succeeds()
-        .stdout_move_str();
+        .stdout_str_lossy();
     let fs_type = fs_type.lines().nth(1).unwrap().trim();
 
     new_ucmd!().args(&["-t", fs_type, "."]).succeeds();
@@ -310,7 +310,7 @@ fn test_type_option_with_file() {
     let fs_types = new_ucmd!()
         .arg("--output=fstype")
         .succeeds()
-        .stdout_move_str();
+        .stdout_str_lossy();
     let fs_types: Vec<_> = fs_types
         .lines()
         .skip(1)
@@ -335,7 +335,7 @@ fn test_exclude_all_types() {
     let fs_types = new_ucmd!()
         .arg("--output=fstype")
         .succeeds()
-        .stdout_move_str();
+        .stdout_str_lossy();
     let fs_types: HashSet<_> = fs_types.lines().skip(1).collect();
 
     let mut args = Vec::new();
@@ -379,7 +379,7 @@ fn test_total() {
     //     ...
     //     /dev/loop14               63488    63488         0 100% /snap/core20/1361
     //     total                 258775268 98099712 148220200  40% -
-    let output = new_ucmd!().arg("--total").succeeds().stdout_move_str();
+    let output = new_ucmd!().arg("--total").succeeds().stdout_str_lossy();
 
     // Skip the header line.
     let lines: Vec<&str> = output.lines().skip(1).collect();
@@ -422,21 +422,21 @@ fn test_total_label_in_correct_column() {
     let output = new_ucmd!()
         .args(&["--output=source", "--total", "."])
         .succeeds()
-        .stdout_move_str();
+        .stdout_str_lossy();
     let last_line = output.lines().last().unwrap();
     assert_eq!(last_line.trim(), "total");
 
     let output = new_ucmd!()
         .args(&["--output=target", "--total", "."])
         .succeeds()
-        .stdout_move_str();
+        .stdout_str_lossy();
     let last_line = output.lines().last().unwrap();
     assert_eq!(last_line.trim(), "total");
 
     let output = new_ucmd!()
         .args(&["--output=source,target", "--total", "."])
         .succeeds()
-        .stdout_move_str();
+        .stdout_str_lossy();
     let last_line = output.lines().last().unwrap();
     assert_eq!(
         last_line.split_whitespace().collect::<Vec<&str>>(),
@@ -446,7 +446,7 @@ fn test_total_label_in_correct_column() {
     let output = new_ucmd!()
         .args(&["--output=target,source", "--total", "."])
         .succeeds()
-        .stdout_move_str();
+        .stdout_str_lossy();
     let last_line = output.lines().last().unwrap();
     assert_eq!(
         last_line.split_whitespace().collect::<Vec<&str>>(),
@@ -463,7 +463,7 @@ fn test_use_percentage() {
         // "percentage" values.
         .args(&["--total", "--output=used,avail,pcent", "--block-size=1"])
         .succeeds()
-        .stdout_move_str();
+        .stdout_str_lossy();
 
     // Skip the header line.
     let lines: Vec<&str> = output.lines().skip(1).collect();
@@ -488,7 +488,7 @@ fn test_iuse_percentage() {
     let output = new_ucmd!()
         .args(&["--total", "--output=itotal,iused,ipcent"])
         .succeeds()
-        .stdout_move_str();
+        .stdout_str_lossy();
 
     // Skip the header line.
     let lines: Vec<&str> = output.lines().skip(1).collect();
@@ -518,7 +518,7 @@ fn test_default_block_size() {
     let output = new_ucmd!()
         .arg("--output=size")
         .succeeds()
-        .stdout_move_str();
+        .stdout_str_lossy();
     let header = output.lines().next().unwrap().trim().to_string();
 
     assert_eq!(header, "1K-blocks");
@@ -527,7 +527,7 @@ fn test_default_block_size() {
         .arg("--output=size")
         .env("POSIXLY_CORRECT", "1")
         .succeeds()
-        .stdout_move_str();
+        .stdout_str_lossy();
     let header = output.lines().next().unwrap().trim().to_string();
 
     assert_eq!(header, "512B-blocks");
@@ -547,14 +547,14 @@ fn test_default_block_size_in_posix_portability_mode() {
             .to_string()
     }
 
-    let output = new_ucmd!().arg("-P").succeeds().stdout_move_str();
+    let output = new_ucmd!().arg("-P").succeeds().stdout_str_lossy();
     assert_eq!(get_header(&output), "1024-blocks");
 
     let output = new_ucmd!()
         .arg("-P")
         .env("POSIXLY_CORRECT", "1")
         .succeeds()
-        .stdout_move_str();
+        .stdout_str_lossy();
     assert_eq!(get_header(&output), "512-blocks");
 }
 
@@ -564,7 +564,7 @@ fn test_block_size_1024() {
         let output = new_ucmd!()
             .args(&["-B", &format!("{block_size}"), "--output=size"])
             .succeeds()
-            .stdout_move_str();
+            .stdout_str_lossy();
         output.lines().next().unwrap().trim().to_string()
     }
 
@@ -588,7 +588,7 @@ fn test_block_size_with_suffix() {
         let output = new_ucmd!()
             .args(&["-B", block_size, "--output=size"])
             .succeeds()
-            .stdout_move_str();
+            .stdout_str_lossy();
         output.lines().next().unwrap().trim().to_string()
     }
 
@@ -612,7 +612,7 @@ fn test_block_size_in_posix_portability_mode() {
         let output = new_ucmd!()
             .args(&["-P", "-B", block_size])
             .succeeds()
-            .stdout_move_str();
+            .stdout_str_lossy();
         output
             .lines()
             .next()
@@ -639,7 +639,7 @@ fn test_block_size_from_env() {
             .arg("--output=size")
             .env(env_var, env_value)
             .succeeds()
-            .stdout_move_str();
+            .stdout_str_lossy();
         output.lines().next().unwrap().trim().to_string()
     }
 
@@ -658,7 +658,7 @@ fn test_block_size_from_env_precedences() {
             .env(k1, v1)
             .env(k2, v2)
             .succeeds()
-            .stdout_move_str();
+            .stdout_str_lossy();
         output.lines().next().unwrap().trim().to_string()
     }
 
@@ -677,7 +677,7 @@ fn test_precedence_of_block_size_arg_over_env() {
         .args(&["-B", "999", "--output=size"])
         .env("DF_BLOCK_SIZE", "111")
         .succeeds()
-        .stdout_move_str();
+        .stdout_str_lossy();
     let header = output.lines().next().unwrap().trim().to_string();
 
     assert_eq!(header, "999B-blocks");
@@ -691,7 +691,7 @@ fn test_invalid_block_size_from_env() {
         .arg("--output=size")
         .env("DF_BLOCK_SIZE", "invalid")
         .succeeds()
-        .stdout_move_str();
+        .stdout_str_lossy();
     let header = output.lines().next().unwrap().trim().to_string();
 
     assert_eq!(header, default_block_size_header);
@@ -701,7 +701,7 @@ fn test_invalid_block_size_from_env() {
         .env("DF_BLOCK_SIZE", "invalid")
         .env("BLOCK_SIZE", "222")
         .succeeds()
-        .stdout_move_str();
+        .stdout_str_lossy();
     let header = output.lines().next().unwrap().trim().to_string();
 
     assert_eq!(header, default_block_size_header);
@@ -717,7 +717,7 @@ fn test_ignore_block_size_from_env_in_posix_portability_mode() {
         .env("BLOCK_SIZE", "222")
         .env("BLOCKSIZE", "333")
         .succeeds()
-        .stdout_move_str();
+        .stdout_str_lossy();
     let header = output
         .lines()
         .next()
@@ -784,13 +784,13 @@ fn test_output_selects_columns() {
     let output = new_ucmd!()
         .args(&["--output=source"])
         .succeeds()
-        .stdout_move_str();
+        .stdout_str_lossy();
     assert_eq!(output.lines().next().unwrap(), "Filesystem");
 
     let output = new_ucmd!()
         .args(&["--output=source,target"])
         .succeeds()
-        .stdout_move_str();
+        .stdout_str_lossy();
     assert_eq!(
         output
             .lines()
@@ -804,7 +804,7 @@ fn test_output_selects_columns() {
     let output = new_ucmd!()
         .args(&["--output=source,target,used"])
         .succeeds()
-        .stdout_move_str();
+        .stdout_str_lossy();
     assert_eq!(
         output
             .lines()
@@ -821,7 +821,7 @@ fn test_output_multiple_occurrences() {
     let output = new_ucmd!()
         .args(&["--output=source", "--output=target"])
         .succeeds()
-        .stdout_move_str();
+        .stdout_str_lossy();
     assert_eq!(
         output
             .lines()
@@ -840,7 +840,7 @@ fn test_output_file_all_filesystems() {
     let output = new_ucmd!()
         .arg("--output=file")
         .succeeds()
-        .stdout_move_str();
+        .stdout_str_lossy();
     let mut lines = output.lines();
     assert_eq!(lines.next().unwrap(), "File");
     for line in lines {
@@ -862,7 +862,7 @@ fn test_output_file_specific_files() {
     let output = ucmd
         .args(&["--output=file", "a", "b", "c"])
         .succeeds()
-        .stdout_move_str();
+        .stdout_str_lossy();
     let actual: Vec<&str> = output.lines().collect();
     assert_eq!(actual, vec!["File", "a", "b", "c"]);
 }
@@ -876,7 +876,7 @@ fn test_file_column_width_if_filename_contains_unicode_chars() {
     let output = ucmd
         .args(&["--output=file,target", "äöü.txt"])
         .succeeds()
-        .stdout_move_str();
+        .stdout_str_lossy();
     let actual = output.lines().next().unwrap();
     // expected width: 7 chars (length of äöü.txt) + 1 char (column separator)
     assert_eq!(actual, "File    Mounted on");

tests/uutests/src/lib/util.rs

@@ -357,6 +357,11 @@ impl CmdResult {
         std::str::from_utf8(&self.stdout).unwrap()
     }
 
+    /// Returns the program's standard output as a string, automatically handling invalid utf8
+    pub fn stdout_str_lossy(self) -> String {
+        String::from_utf8_lossy(&self.stdout).to_string()
+    }
+
     /// Returns the program's standard output as a string
     /// consumes self
     pub fn stdout_move_str(self) -> String {