cksum: correctly output non-utf8 filename

2 files changed, 76 insertions, 37 deletions

diff --git a/src/uu/cksum/src/cksum.rs b/src/uu/cksum/src/cksum.rs
index ca0badae5c..1a3c51b3e7 100644
--- a/src/uu/cksum/src/cksum.rs
+++ b/src/uu/cksum/src/cksum.rs
@@ -19,7 +19,7 @@ use uucore::checksum::{
 use uucore::{
     encoding,
     error::{FromIo, UResult, USimpleError},
-    format_usage, help_about, help_section, help_usage, show,
+    format_usage, help_about, help_section, help_usage, os_str_as_bytes, show,
     sum::{div_ceil, Digest},
 };
 
@@ -117,52 +117,64 @@ 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, should_print_filename, after_filename) = match options.algo_name {
+            ALGORITHM_OPTIONS_SYSV => (
+                format!(
+                    "{} {}{}",
+                    sum.parse::<u16>().unwrap(),
+                    div_ceil(sz, options.output_bits),
+                    if not_file { "" } else { " " }
+                ),
+                !not_file,
+                String::new(),
             ),
-            (ALGORITHM_OPTIONS_SYSV, false) => println!(
-                "{} {} {}",
-                sum.parse::<u16>().unwrap(),
-                div_ceil(sz, options.output_bits),
-                filename.display()
+            ALGORITHM_OPTIONS_BSD => (
+                format!(
+                    "{:0bsd_width$} {:bsd_width$}{}",
+                    sum.parse::<u16>().unwrap(),
+                    div_ceil(sz, options.output_bits),
+                    if not_file { "" } else { " " }
+                ),
+                !not_file,
+                String::new(),
             ),
-            (ALGORITHM_OPTIONS_BSD, true) => println!(
-                "{:0bsd_width$} {:bsd_width$}",
-                sum.parse::<u16>().unwrap(),
-                div_ceil(sz, options.output_bits)
+            ALGORITHM_OPTIONS_CRC => (
+                format!("{sum} {sz}{}", if not_file { "" } else { " " }),
+                !not_file,
+                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_CRC, true) => println!("{sum} {sz}"),
-            (ALGORITHM_OPTIONS_CRC, false) => println!("{sum} {sz} {}", filename.display()),
-            (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());
-                }
+            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.
+                        format!("BLAKE2b-{} (", length * 8)
+                    } else {
+                        "BLAKE2b (".to_owned()
+                    },
+                    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 should_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.
+            let _dropped_result = stdout().write_all(os_str_as_bytes(filename.as_os_str())?);
         }
+        println!("{}", after_filename);
     }
 
     Ok(())
diff --git a/tests/by-util/test_cksum.rs b/tests/by-util/test_cksum.rs
index b2aafc2cbd..117e54e1ef 100644
--- a/tests/by-util/test_cksum.rs
+++ b/tests/by-util/test_cksum.rs
@@ -2,7 +2,7 @@
 //
 // For the full copyright and license information, please view the LICENSE
 // file that was distributed with this source code.
-// spell-checker:ignore (words) asdf algo algos mgmt
+// spell-checker:ignore (words) asdf algo algos asha mgmt xffname
 
 use crate::common::util::TestScenario;
 
@@ -1250,3 +1250,30 @@ fn test_several_files_error_mgmt() {
         .stderr_contains("empty: no properly ")
         .stderr_contains("incorrect: no properly ");
 }
+
+#[cfg(target_os = "linux")]
+#[test]
+fn test_non_utf8_filename() {
+    use std::ffi::OsString;
+    use std::os::unix::ffi::OsStringExt;
+
+    let scene = TestScenario::new(util_name!());
+    let at = &scene.fixtures;
+    let filename: OsString = OsStringExt::from_vec(b"funky\xffname".to_vec());
+
+    at.touch(&filename);
+
+    scene
+        .ucmd()
+        .arg(&filename)
+        .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();
+}