numfmt: implement --delimiter

closes #1454

src/uu/numfmt/src/format.rs

@@ -226,12 +226,31 @@ fn format_string(
     })
 }
 
-/// Format a line of text according to the selected options.
-///
-/// Given a line of text `s`, split the line into fields, transform and format
-/// any selected numeric fields, and print the result to stdout. Fields not
-/// selected for conversion are passed through unmodified.
-pub fn format_and_print(s: &str, options: &NumfmtOptions) -> Result<()> {
+fn format_and_print_delimited(s: &str, options: &NumfmtOptions) -> Result<()> {
+    let delimiter = options.delimiter.as_ref().unwrap();
+
+    for (n, field) in (1..).zip(s.split(delimiter)) {
+        let field_selected = uucore::ranges::contain(&options.fields, n);
+
+        // print delimiter before second and subsequent fields
+        if n > 1 {
+            print!("{}", delimiter);
+        }
+
+        if field_selected {
+            print!("{}", format_string(&field.trim_start(), options, None)?);
+        } else {
+            // print unselected field without conversion
+            print!("{}", field);
+        }
+    }
+
+    println!();
+
+    Ok(())
+}
+
+fn format_and_print_whitespace(s: &str, options: &NumfmtOptions) -> Result<()> {
     for (n, (prefix, field)) in (1..).zip(WhitespaceSplitter { s: Some(s) }) {
         let field_selected = uucore::ranges::contain(&options.fields, n);
 
@@ -263,3 +282,15 @@ pub fn format_and_print(s: &str, options: &NumfmtOptions) -> Result<()> {
 
     Ok(())
 }
+
+/// Format a line of text according to the selected options.
+///
+/// Given a line of text `s`, split the line into fields, transform and format
+/// any selected numeric fields, and print the result to stdout. Fields not
+/// selected for conversion are passed through unmodified.
+pub fn format_and_print(s: &str, options: &NumfmtOptions) -> Result<()> {
+    match &options.delimiter {
+        Some(_) => format_and_print_delimited(s, options),
+        None => format_and_print_whitespace(s, options),
+    }
+}

src/uu/numfmt/src/numfmt.rs

@@ -128,11 +128,20 @@ fn parse_options(args: &ArgMatches) -> Result<NumfmtOptions> {
         None => unreachable!(),
     };
 
+    let delimiter = args.value_of(options::DELIMITER).map_or(Ok(None), |arg| {
+        if arg.len() == 1 {
+            Ok(Some(arg.to_string()))
+        } else {
+            Err("the delimiter must be a single character".to_string())
+        }
+    })?;
+
     Ok(NumfmtOptions {
         transform,
         padding,
         header,
         fields,
+        delimiter,
     })
 }
 
@@ -145,6 +154,13 @@ pub fn uumain(args: impl uucore::Args) -> i32 {
         .usage(&usage[..])
         .after_help(LONG_HELP)
         .setting(AppSettings::AllowNegativeNumbers)
+        .arg(
+            Arg::with_name(options::DELIMITER)
+                .short("d")
+                .long(options::DELIMITER)
+                .value_name("X")
+                .help("use X instead of whitespace for field delimiter"),
+        )
         .arg(
             Arg::with_name(options::FIELD)
                 .long(options::FIELD)

src/uu/numfmt/src/options.rs

@@ -1,6 +1,7 @@
 use crate::units::Transform;
 use uucore::ranges::Range;
 
+pub const DELIMITER: &str = "delimiter";
 pub const FIELD: &str = "field";
 pub const FIELD_DEFAULT: &str = "1";
 pub const FROM: &str = "from";
@@ -22,4 +23,5 @@ pub struct NumfmtOptions {
     pub padding: isize,
     pub header: usize,
     pub fields: Vec<Range>,
+    pub delimiter: Option<String>,
 }

tests/by-util/test_numfmt.rs

@@ -383,3 +383,98 @@ fn test_field_df_example() {
         .succeeds()
         .stdout_is_fixture("df_expected.txt");
 }
+
+#[test]
+fn test_delimiter_must_not_be_empty() {
+    new_ucmd!().args(&["-d"]).fails();
+}
+
+#[test]
+fn test_delimiter_must_not_be_more_than_one_character() {
+    new_ucmd!()
+        .args(&["--delimiter", "sad"])
+        .fails()
+        .stderr_is("numfmt: the delimiter must be a single character");
+}
+
+#[test]
+fn test_delimiter_only() {
+    new_ucmd!()
+        .args(&["-d", ","])
+        .pipe_in("1234,56")
+        .succeeds()
+        .stdout_only("1234,56\n");
+}
+
+#[test]
+fn test_line_is_field_with_no_delimiter() {
+    new_ucmd!()
+        .args(&["-d,", "--to=iec"])
+        .pipe_in("123456")
+        .succeeds()
+        .stdout_only("121K\n");
+}
+
+#[test]
+fn test_delimiter_to_si() {
+    new_ucmd!()
+        .args(&["-d=,", "--to=si"])
+        .pipe_in("1234,56")
+        .succeeds()
+        .stdout_only("1.3K,56\n");
+}
+
+#[test]
+fn test_delimiter_skips_leading_whitespace() {
+    new_ucmd!()
+        .args(&["-d=,", "--to=si"])
+        .pipe_in("     \t               1234,56")
+        .succeeds()
+        .stdout_only("1.3K,56\n");
+}
+
+#[test]
+fn test_delimiter_preserves_leading_whitespace_in_unselected_fields() {
+    new_ucmd!()
+        .args(&["-d=|", "--to=si"])
+        .pipe_in("             1000|   2000")
+        .succeeds()
+        .stdout_only("1.0K|   2000\n");
+}
+
+#[test]
+fn test_delimiter_from_si() {
+    new_ucmd!()
+        .args(&["-d=,", "--from=si"])
+        .pipe_in("1.2K,56")
+        .succeeds()
+        .stdout_only("1200,56\n");
+}
+
+#[test]
+fn test_delimiter_overrides_whitespace_separator() {
+    // GNU numfmt reports this as “invalid suffix”
+    new_ucmd!()
+        .args(&["-d,"])
+        .pipe_in("1 234,56")
+        .fails()
+        .stderr_is("numfmt: invalid number: ‘1 234’\n");
+}
+
+#[test]
+fn test_delimiter_with_padding() {
+    new_ucmd!()
+        .args(&["-d=|", "--to=si", "--padding=5"])
+        .pipe_in("1000|2000")
+        .succeeds()
+        .stdout_only(" 1.0K|2000\n");
+}
+
+#[test]
+fn test_delimiter_with_padding_and_fields() {
+    new_ucmd!()
+        .args(&["-d=|", "--to=si", "--padding=5", "--field=-"])
+        .pipe_in("1000|2000")
+        .succeeds()
+        .stdout_only(" 1.0K| 2.0K\n");
+}