numfmt: add unit tests for detailed_error_message, parse_number_part, and apply_grouping

sylvestre · sylvestre · commit 0c5be9fb85cc · 2026-04-01T09:53:49.000+02:00
diff --git a/src/uu/numfmt/src/format.rs b/src/uu/numfmt/src/format.rs
@@ -262,8 +262,10 @@ fn apply_grouping(s: &str) -> String {
     grouped.push_str(&integer[..first_group]);
     for chunk in integer.as_bytes()[first_group..].chunks(3) {
         grouped.push_str(grouping_separator);
-        // SAFETY: integer is known to be valid UTF-8 ASCII digits
-        grouped.push_str(std::str::from_utf8(chunk).unwrap());
+        // integer is valid UTF-8 ASCII digits, so this cannot fail
+        if let Ok(s) = std::str::from_utf8(chunk) {
+            grouped.push_str(s);
+        }
     }
 
     if !fraction.is_empty() {
@@ -995,4 +997,140 @@ mod tests {
         assert_eq!(raw_suffix as i32, RawSuffix::Q as i32);
         assert_eq!(value, 5.0);
     }
+
+    #[test]
+    fn test_detailed_error_message_empty() {
+        let result = detailed_error_message("", Unit::Auto, "");
+        assert!(result.is_some());
+    }
+
+    #[test]
+    fn test_detailed_error_message_valid_number() {
+        // A plain valid number should return None (no error)
+        assert!(detailed_error_message("123", Unit::Auto, "").is_none());
+        assert!(detailed_error_message("5K", Unit::Auto, "").is_none());
+        assert!(detailed_error_message("-3.5M", Unit::Auto, "").is_none());
+    }
+
+    #[test]
+    fn test_detailed_error_message_trailing_garbage() {
+        // Number with suffix followed by extra chars
+        let result = detailed_error_message("5Kx", Unit::Auto, "").unwrap();
+        assert!(
+            result.contains("numfmt-error-invalid-specific-suffix")
+                || result.contains("invalid suffix")
+        );
+    }
+
+    #[test]
+    fn test_detailed_error_message_dot_only() {
+        let result = detailed_error_message(".", Unit::Auto, "").unwrap();
+        assert!(
+            result.contains("numfmt-error-invalid-suffix") || result.contains("invalid suffix")
+        );
+    }
+
+    #[test]
+    fn test_detailed_error_message_trailing_dot() {
+        let result = detailed_error_message("5.", Unit::Auto, "").unwrap();
+        assert!(
+            result.contains("numfmt-error-invalid-number") || result.contains("invalid number")
+        );
+    }
+
+    #[test]
+    fn test_detailed_error_message_unit_separator() {
+        // With unit separator, "5 K" is valid
+        assert!(detailed_error_message("5 K", Unit::Auto, " ").is_none());
+
+        // "5 Kx" should report trailing garbage after the suffix
+        let result = detailed_error_message("5 Kx", Unit::Auto, " ");
+        assert!(result.is_some());
+    }
+
+    #[test]
+    fn test_parse_number_part_valid() {
+        assert_eq!(parse_number_part("42", "42").unwrap(), 42.0);
+        assert_eq!(parse_number_part("-3.5", "-3.5").unwrap(), -3.5);
+        assert_eq!(parse_number_part("0", "0").unwrap(), 0.0);
+    }
+
+    #[test]
+    fn test_parse_number_part_trailing_dot() {
+        assert!(parse_number_part("5.", "5.").is_err());
+    }
+
+    #[test]
+    fn test_parse_number_part_non_numeric() {
+        assert!(parse_number_part("abc", "abc").is_err());
+        assert!(parse_number_part("", "").is_err());
+    }
+
+    #[test]
+    fn test_apply_grouping_short_numbers() {
+        // Numbers with fewer than 4 digits should be unchanged
+        assert_eq!(apply_grouping("0"), "0");
+        assert_eq!(apply_grouping("999"), "999");
+        assert_eq!(apply_grouping("-99"), "-99");
+    }
+
+    #[test]
+    fn test_apply_grouping_with_fraction() {
+        // Fraction part should not be grouped
+        let result = apply_grouping("1234.567");
+        // Depending on locale, separator may or may not be present
+        assert!(result.contains("567"));
+        assert!(result.contains('.'));
+    }
+
+    #[test]
+    fn test_apply_grouping_negative() {
+        let result = apply_grouping("-1234");
+        assert!(result.starts_with('-'));
+    }
+
+    #[test]
+    fn test_apply_grouping_large_numbers() {
+        // These tests verify grouping structure; actual separator depends on locale
+        let result = apply_grouping("1000000");
+        // Should have separators inserted (length grows if separator is non-empty)
+        assert!(result.len() >= 7);
+
+        let result = apply_grouping("1234567890");
+        assert!(result.len() >= 10);
+
+        let result = apply_grouping("-9999999999999");
+        assert!(result.starts_with('-'));
+        assert!(result.len() >= 13);
+    }
+
+    #[test]
+    fn test_apply_grouping_tiny_fraction() {
+        // Small decimal: integer part < 4 digits, so no grouping
+        assert_eq!(apply_grouping("0.000001"), "0.000001");
+        assert_eq!(apply_grouping("1.23456789"), "1.23456789");
+    }
+
+    #[test]
+    fn test_apply_grouping_exactly_four_digits() {
+        let result = apply_grouping("1000");
+        // Should be grouped (4 digits)
+        assert!(result.len() >= 4);
+    }
+
+    #[test]
+    fn test_parse_number_part_large_and_tiny() {
+        assert_eq!(
+            parse_number_part("999999999999", "999999999999").unwrap(),
+            999_999_999_999.0
+        );
+        assert_eq!(
+            parse_number_part("0.000000001", "0.000000001").unwrap(),
+            0.000_000_001
+        );
+        assert_eq!(
+            parse_number_part("-99999999", "-99999999").unwrap(),
+            -99_999_999.0
+        );
+    }
 }
diff --git a/src/uu/numfmt/src/numfmt.rs b/src/uu/numfmt/src/numfmt.rs
@@ -32,13 +32,17 @@ mod units;
 
 /// Format a single line and write it, handling `--invalid` error modes.
 ///
+/// `fmt_buf` is a caller-owned scratch buffer reused across calls to avoid
+/// allocating a new `Vec` on every line in non-abort invalid modes.
+///
 /// Returns `true` if the line contained invalid input (only possible in
 /// non-abort modes).
 fn format_and_write<W: std::io::Write>(
     writer: &mut W,
     input_line: &[u8],
     options: &NumfmtOptions,
     eol: Option<u8>,
+    fmt_buf: &mut Vec<u8>,
 ) -> UResult<bool> {
     // GNU truncates at the first embedded null byte.
     let line = match memchr::memchr(b'\0', input_line) {
@@ -49,8 +53,8 @@ fn format_and_write<W: std::io::Write>(
     // In non-abort modes we buffer the formatted output so that on error we
     // can emit the original line instead.
     let buffer_output = !matches!(options.invalid, InvalidModes::Abort);
-    let mut buf = Vec::new();
-    let dest: &mut dyn std::io::Write = if buffer_output { &mut buf } else { writer };
+    fmt_buf.clear();
+    let dest: &mut dyn std::io::Write = if buffer_output { fmt_buf } else { writer };
 
     let result = if options.delimiter.is_some() {
         write_formatted_with_delimiter(dest, line, options, eol)
@@ -86,7 +90,7 @@ fn format_and_write<W: std::io::Write>(
     }
 
     if buffer_output {
-        writer.write_all(&buf)?;
+        writer.write_all(fmt_buf)?;
     }
     Ok(false)
 }
diff --git a/util/gnu-patches/tests_numfmt.patch b/util/gnu-patches/tests_numfmt.patch
@@ -29,28 +29,28 @@ Index: gnu/tests/numfmt/numfmt.pl
       ['field-range-err-6', '--field - --field 1- --to=si 10',
 -             {EXIT=>1}, {ERR=>"$prog: multiple field specifications\n"}],
 +             {EXIT=>1},
-+             {ERR_SUBST=>"s/.*//msg"},
-+             {ERR=>""}],
++             {ERR_SUBST=>"s/\n.*//s"},
++             {ERR=>"error: the argument '--field <FIELDS>' cannot be used multiple times"}],
       ['field-range-err-7', '--field -1 --field 1- --to=si 10',
 -             {EXIT=>1}, {ERR=>"$prog: multiple field specifications\n"}],
 +             {EXIT=>1},
-+             {ERR_SUBST=>"s/.*//msg"},
-+             {ERR=>""}],
++             {ERR_SUBST=>"s/\n.*//s"},
++             {ERR=>"error: the argument '--field <FIELDS>' cannot be used multiple times"}],
       ['field-range-err-8', '--field -1 --field 1,2,3 --to=si 10',
 -             {EXIT=>1}, {ERR=>"$prog: multiple field specifications\n"}],
 +             {EXIT=>1},
-+             {ERR_SUBST=>"s/.*//msg"},
-+             {ERR=>""}],
++             {ERR_SUBST=>"s/\n.*//s"},
++             {ERR=>"error: the argument '--field <FIELDS>' cannot be used multiple times"}],
       ['field-range-err-9', '--field 1- --field 1,2,3 --to=si 10',
 -             {EXIT=>1}, {ERR=>"$prog: multiple field specifications\n"}],
 +             {EXIT=>1},
-+             {ERR_SUBST=>"s/.*//msg"},
-+             {ERR=>""}],
++             {ERR_SUBST=>"s/\n.*//s"},
++             {ERR=>"error: the argument '--field <FIELDS>' cannot be used multiple times"}],
       ['field-range-err-10','--field 1,2,3 --field 1- --to=si 10',
 -             {EXIT=>1}, {ERR=>"$prog: multiple field specifications\n"}],
 +             {EXIT=>1},
-+             {ERR_SUBST=>"s/.*//msg"},
-+             {ERR=>""}],
++             {ERR_SUBST=>"s/\n.*//s"},
++             {ERR=>"error: the argument '--field <FIELDS>' cannot be used multiple times"}],
       ['field-range-err-11','--field 1-2-3 --to=si 10',
 -             {EXIT=>1}, {ERR=>"$prog: invalid field range\n$try"}],
 +             {EXIT=>1}, {ERR=>"$prog: range '1-2-3' was invalid: failed to parse range\n"}],
@@ -113,8 +113,8 @@ Index: gnu/tests/numfmt/numfmt.pl
       ['help-1', '--foobar',
 -             {ERR=>"$prog: unrecognized option\n$try"},
 -             {ERR_SUBST=>"s/option.*/option/; s/unknown/unrecognized/"},
-+             {ERR=>""},
-+             {ERR_SUBST=>"s/.*//msg"},
++             {ERR=>"error: unexpected argument '--foobar' found"},
++             {ERR_SUBST=>"s/\n.*//s"},
               {EXIT=>1}],
 
       ## Format string - check error detection
