checksum: do not let improperly formatted lines select the line format

src/uucore/src/lib/features/checksum/validate.rs

@@ -862,10 +862,13 @@ fn process_checksum_line(
 
     // Use `LineInfo` to extract the data of a line.
     // Then, depending on its format, apply a different pre-treatment.
-    let line_info = LineInfo::parse(line, cached_line_format, opts.allow_empty_filename)
+    // The detected format is kept only if the line turns out to be properly
+    // formatted: an improper line must not decide how the next ones are read.
+    let mut line_format = *cached_line_format;
+    let line_info = LineInfo::parse(line, &mut line_format, opts.allow_empty_filename)
         .ok_or(LineCheckError::ImproperlyFormatted)?;
 
-    if line_info.format == LineFormat::AlgoBased {
+    let result = if line_info.format == LineFormat::AlgoBased {
         process_algo_based_line(&line_info, cli_algo_name, opts, last_algo)
     } else if let Some(cli_algo) = cli_algo_name {
         // If we match a non-algo based parser, we expect a cli argument
@@ -874,7 +877,12 @@ fn process_checksum_line(
     } else {
         // We have no clue of what algorithm to use
         Err(LineCheckError::ImproperlyFormatted)
+    };
+
+    if !matches!(result, Err(LineCheckError::ImproperlyFormatted)) {
+        *cached_line_format = line_format;
     }
+    result
 }
 
 fn process_checksum_file(
@@ -1334,6 +1342,82 @@ mod tests {
         assert!(cached_line_format.is_none());
     }
 
+    fn check_md5_line(
+        line: &str,
+        cached_line_format: &mut Option<LineFormat>,
+    ) -> Result<(), LineCheckError> {
+        let opts = ChecksumValidateOptions {
+            ignore_missing: true,
+            ..Default::default()
+        };
+        process_checksum_line(
+            OsStr::new(line),
+            Some(AlgoKind::Md5),
+            None,
+            opts,
+            cached_line_format,
+            &mut None,
+        )
+    }
+
+    #[test]
+    fn test_improperly_formatted_line_does_not_select_single_space_format() {
+        let mut cached_line_format = None;
+
+        let res = check_md5_line("2024 release", &mut cached_line_format);
+        assert!(matches!(res, Err(LineCheckError::ImproperlyFormatted)));
+        assert_eq!(cached_line_format, None);
+
+        let res = check_md5_line(
+            "d41d8cd98f00b204e9800998ecf8427e  missing",
+            &mut cached_line_format,
+        );
+        assert!(matches!(res, Err(LineCheckError::FileNotFound)));
+        assert_eq!(cached_line_format, Some(LineFormat::Untagged));
+    }
+
+    #[test]
+    fn test_improperly_formatted_line_does_not_select_untagged_format() {
+        let mut cached_line_format = None;
+
+        let res = check_md5_line("Checksums  for release", &mut cached_line_format);
+        assert!(matches!(res, Err(LineCheckError::ImproperlyFormatted)));
+        assert_eq!(cached_line_format, None);
+
+        let res = check_md5_line(
+            "d41d8cd98f00b204e9800998ecf8427e missing",
+            &mut cached_line_format,
+        );
+        assert!(matches!(res, Err(LineCheckError::FileNotFound)));
+        assert_eq!(cached_line_format, Some(LineFormat::SingleSpace));
+    }
+
+    #[test]
+    fn test_properly_formatted_line_selects_format_whatever_the_digest_check_outcome() {
+        let mut cached_line_format = None;
+        let res = check_md5_line(
+            "d41d8cd98f00b204e9800998ecf8427e  missing",
+            &mut cached_line_format,
+        );
+        assert!(matches!(res, Err(LineCheckError::FileNotFound)));
+        assert_eq!(cached_line_format, Some(LineFormat::Untagged));
+
+        let mut cached_line_format = None;
+        let res = check_md5_line(
+            "d41d8cd98f00b204e9800998ecf8427e Cargo.toml",
+            &mut cached_line_format,
+        );
+        assert!(matches!(res, Err(LineCheckError::DigestMismatch)));
+        assert_eq!(cached_line_format, Some(LineFormat::SingleSpace));
+
+        let res = check_md5_line(
+            "d41d8cd98f00b204e9800998ecf8427e  missing",
+            &mut cached_line_format,
+        );
+        assert!(matches!(res, Err(LineCheckError::FileNotFound)));
+        assert_eq!(cached_line_format, Some(LineFormat::SingleSpace));
+    }
+
     #[test]
     fn test_get_expected_digest() {
         let ck = "47DEQpj8HBSa+/TImW+5JCeuQeRkm5NMpJWZG3hSuFU=".to_owned();

tests/by-util/test_cksum.rs

@@ -3920,3 +3920,46 @@ fn test_numeric_looking_directory_names_are_kept_verbatim() {
         .fails_with_code(1)
         .stderr_is("cksum: 007: Is a directory\ncksum: 1e3: Is a directory\n");
 }
+
+#[test]
+fn test_check_improperly_formatted_line_does_not_select_format() {
+    let scene = TestScenario::new(util_name!());
+    let at = &scene.fixtures;
+
+    // The listed files were replaced, their genuine content is kept under the
+    // same names with a leading space.
+    for f in ["a", "c"] {
+        at.write(f, "tampered\n");
+        at.write(&format!(" {f}"), &format!("{f}\n"));
+    }
+    at.write(
+        "check.md5sum",
+        "2024 release\n\
+        60b725f10c9c85c70d97880dfe8191b3  a\n\
+        2cd6ee2c70b0bde53fbe6cac3c8b8bb1  c\n",
+    );
+    scene
+        .ucmd()
+        .arg("--check")
+        .arg("--algorithm=md5")
+        .arg("check.md5sum")
+        .fails_with_code(1)
+        .stdout_is("a: FAILED\nc: FAILED\n")
+        .stderr_contains("1 line is improperly formatted")
+        .stderr_contains("2 computed checksums did NOT match");
+
+    at.write("a", "a\n");
+    at.write(
+        "check_single_space.md5sum",
+        "Checksums  for release\n\
+        60b725f10c9c85c70d97880dfe8191b3 a\n",
+    );
+    scene
+        .ucmd()
+        .arg("--check")
+        .arg("--algorithm=md5")
+        .arg("check_single_space.md5sum")
+        .succeeds()
+        .stdout_is("a: OK\n")
+        .stderr_contains("1 line is improperly formatted");
+}