cksum: correctly output non-utf8 filename

src/uu/cksum/src/cksum.rs

@@ -117,52 +117,82 @@ where
  };
  // The BSD checksum output is 5 digit integer
  let bsd_width = 5;
- match (options.algo_name, not_file) {
- (ALGORITHM_OPTIONS_SYSV, true) => println!(
- "{} {}",
- sum.parse::<u16>().unwrap(),
- div_ceil(sz, options.output_bits)
+ let (before_filename, print_filename, after_filename) = match (options.algo_name, not_file)
+ {
+ (ALGORITHM_OPTIONS_SYSV, true) => (
+ format!(
+ "{} {}",
+ sum.parse::<u16>().unwrap(),
+ div_ceil(sz, options.output_bits)
+ ),
+ false,
+ String::new(),
  ),
- (ALGORITHM_OPTIONS_SYSV, false) => println!(
- "{} {} {}",
- sum.parse::<u16>().unwrap(),
- div_ceil(sz, options.output_bits),
- filename.display()
+ (ALGORITHM_OPTIONS_SYSV, false) => (
+ format!(
+ "{} {} ",
+ sum.parse::<u16>().unwrap(),
+ div_ceil(sz, options.output_bits)
+ ),
+ true,
+ String::new(),
  ),
- (ALGORITHM_OPTIONS_BSD, true) => println!(
- "{:0bsd_width$} {:bsd_width$}",
- sum.parse::<u16>().unwrap(),
- div_ceil(sz, options.output_bits)
+ (ALGORITHM_OPTIONS_BSD, true) => (
+ format!(
+ "{:0bsd_width$} {:bsd_width$}",
+ sum.parse::<u16>().unwrap(),
+ div_ceil(sz, options.output_bits)
+ ),
+ false,
+ String::new(),
  ),
- (ALGORITHM_OPTIONS_BSD, false) => println!(
- "{:0bsd_width$} {:bsd_width$} {}",
- sum.parse::<u16>().unwrap(),
- div_ceil(sz, options.output_bits),
- filename.display()
+ (ALGORITHM_OPTIONS_BSD, false) => (
+ format!(
+ "{:0bsd_width$} {:bsd_width$} ",
+ sum.parse::<u16>().unwrap(),
+ div_ceil(sz, options.output_bits),
+ ),
+ true,
+ String::new(),
  ),
- (ALGORITHM_OPTIONS_CRC, true) => println!("{sum} {sz}"),
- (ALGORITHM_OPTIONS_CRC, false) => println!("{sum} {sz} {}", filename.display()),
+ (ALGORITHM_OPTIONS_CRC, true) => (format!("{sum} {sz}"), false, String::new()),
+ (ALGORITHM_OPTIONS_CRC, false) => (format!("{sum} {sz} "), true, String::new()),
  (ALGORITHM_OPTIONS_BLAKE2B, _) if options.tag => {
- if let Some(length) = options.length {
- // Multiply by 8 here, as we want to print the length in bits.
- println!("BLAKE2b-{} ({}) = {sum}", length * 8, filename.display());
- } else {
- println!("BLAKE2b ({}) = {sum}", filename.display());
- }
+ (
+ if let Some(length) = options.length {
+ // Multiply by 8 here, as we want to print the length in bits.
+ format!("BLAKE2b-{} (", length * 8)
+ } else {
+ format!("BLAKE2b (")
+ },
+ true,
+ format!(") = {sum}"),
+ )
  }
  _ => {
  if options.tag {
- println!(
- "{} ({}) = {sum}",
- options.algo_name.to_ascii_uppercase(),
- filename.display()
- );
+ (
+ format!("{} (", options.algo_name.to_ascii_uppercase()),
+ true,
+ format!(") = {sum}")
+ )
  } else {
  let prefix = if options.asterisk { "*" } else { " " };
- println!("{sum} {prefix}{}", filename.display());
+ (
+ format!("{sum} {prefix}"),
+ true,
+ String::new()
+ )
  }
  }
+ };
+ print!("{}", before_filename);
+ if print_filename {
+ // The filename might not be valid UTF-8, and filename.display() would mangle the names.
+ // Therefore, emit the bytes directly to stdout, without any attempt at encoding them.
+ stdout().write(filename.as_os_str().as_encoded_bytes()).expect("FIXME");
  }
+ println!("{}", after_filename);
  }
 
  Ok(())

tests/by-util/test_cksum.rs

@@ -6,6 +6,11 @@
 
 use crate::common::util::TestScenario;
 
+#[cfg(unix)]
+use std::ffi::OsString;
+#[cfg(unix)]
+use std::os::unix::ffi::OsStringExt;
+
 const ALGOS: [&str; 11] = [
  "sysv", "bsd", "crc", "md5", "sha1", "sha224", "sha256", "sha384", "sha512", "blake2b", "sm3",
 ];
@@ -1250,3 +1255,27 @@ fn test_several_files_error_mgmt() {
  .stderr_contains("empty: no properly ")
  .stderr_contains("incorrect: no properly ");
 }
+
+#[cfg(unix)]
+#[test]
+fn test_non_utf8_filename() {
+ let scene = TestScenario::new(util_name!());
+ let at = &scene.fixtures;
+ let filename: OsString = OsStringExt::from_vec(b"funky\xffname".to_vec());
+
+ at.touch(filename.clone());
+
+ scene
+ .ucmd()
+ .arg(filename.clone())
+ .succeeds()
+ .stdout_is_bytes(b"4294967295 0 funky\xffname\n")
+ .no_stderr();
+ scene
+ .ucmd()
+ .arg("-asha256")
+ .arg(filename)
+ .succeeds()
+ .stdout_is_bytes(b"SHA256 (funky\xffname) = e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855\n")
+ .no_stderr();
+}