Merge pull request #5158 from cakebaker/nl_section_delimiter

nl: implement -d/--section-delimiter

src/uu/nl/src/helper.rs

@@ -13,6 +13,15 @@ pub fn parse_options(settings: &mut crate::Settings, opts: &clap::ArgMatches) ->
     // This vector holds error messages encountered.
     let mut errs: Vec<String> = vec![];
     settings.renumber = opts.get_flag(options::NO_RENUMBER);
+    if let Some(delimiter) = opts.get_one::<String>(options::SECTION_DELIMITER) {
+        // check whether the delimiter is a single ASCII char (1 byte)
+        // because GNU nl doesn't add a ':' to single non-ASCII chars
+        settings.section_delimiter = if delimiter.len() == 1 {
+            format!("{delimiter}:")
+        } else {
+            delimiter.to_owned()
+        };
+    }
     if let Some(val) = opts.get_one::<String>(options::NUMBER_SEPARATOR) {
         settings.number_separator = val.to_owned();
     }

src/uu/nl/src/nl.rs

@@ -23,7 +23,7 @@ pub struct Settings {
     body_numbering: NumberingStyle,
     footer_numbering: NumberingStyle,
     // The variable corresponding to -d
-    section_delimiter: [char; 2],
+    section_delimiter: String,
     // The variables corresponding to the options -v, -i, -l, -w.
     starting_line_number: i64,
     line_increment: i64,
@@ -43,7 +43,7 @@ impl Default for Settings {
             header_numbering: NumberingStyle::None,
             body_numbering: NumberingStyle::NonEmpty,
             footer_numbering: NumberingStyle::None,
-            section_delimiter: ['\\', ':'],
+            section_delimiter: String::from("\\:"),
             starting_line_number: 1,
             line_increment: 1,
             join_blank_lines: 1,
@@ -134,6 +134,32 @@ impl NumberFormat {
     }
 }
 
+enum SectionDelimiter {
+    Header,
+    Body,
+    Footer,
+}
+
+impl SectionDelimiter {
+    // A valid section delimiter contains the pattern one to three times,
+    // and nothing else.
+    fn parse(s: &str, pattern: &str) -> Option<Self> {
+        if s.is_empty() || pattern.is_empty() {
+            return None;
+        }
+
+        let pattern_count = s.matches(pattern).count();
+        let is_length_ok = pattern_count * pattern.len() == s.len();
+
+        match (pattern_count, is_length_ok) {
+            (3, true) => Some(Self::Header),
+            (2, true) => Some(Self::Body),
+            (1, true) => Some(Self::Footer),
+            _ => None,
+        }
+    }
+}
+
 pub mod options {
     pub const HELP: &str = "help";
     pub const FILE: &str = "file";
@@ -307,14 +333,12 @@ fn nl<T: Read>(reader: &mut BufReader<T>, stats: &mut Stats, settings: &Settings
             stats.consecutive_empty_lines = 0;
         };
 
-        // FIXME section delimiters are hardcoded and settings.section_delimiter is ignored
-        // because --section-delimiter is not correctly implemented yet
-        let _ = settings.section_delimiter; // XXX suppress "field never read" warning
-        let new_numbering_style = match line.as_str() {
-            "\\:\\:\\:" => Some(&settings.header_numbering),
-            "\\:\\:" => Some(&settings.body_numbering),
-            "\\:" => Some(&settings.footer_numbering),
-            _ => None,
+        let new_numbering_style = match SectionDelimiter::parse(&line, &settings.section_delimiter)
+        {
+            Some(SectionDelimiter::Header) => Some(&settings.header_numbering),
+            Some(SectionDelimiter::Body) => Some(&settings.body_numbering),
+            Some(SectionDelimiter::Footer) => Some(&settings.footer_numbering),
+            None => None,
         };
 
         if let Some(new_style) = new_numbering_style {

tests/by-util/test_nl.rs

@@ -2,7 +2,8 @@
 //
 // For the full copyright and license information, please view the LICENSE
 // file that was distributed with this source code.
-// spell-checker:ignore binvalid finvalid hinvalid iinvalid linvalid ninvalid vinvalid winvalid
+//
+// spell-checker:ignore binvalid finvalid hinvalid iinvalid linvalid nabcabc nabcabcabc ninvalid vinvalid winvalid
 use crate::common::util::TestScenario;
 
 #[test]
@@ -537,3 +538,83 @@ fn test_line_number_overflow() {
         .stdout_is(format!("{}\ta\n", i64::MIN))
         .stderr_is("nl: line number overflow\n");
 }
+
+#[test]
+fn test_section_delimiter() {
+    for arg in ["-dabc", "--section-delimiter=abc"] {
+        new_ucmd!()
+            .arg(arg)
+            .pipe_in("a\nabcabcabc\nb") // header section
+            .succeeds()
+            .stdout_is("     1\ta\n\n       b\n");
+
+        new_ucmd!()
+            .arg(arg)
+            .pipe_in("a\nabcabc\nb") // body section
+            .succeeds()
+            .stdout_is("     1\ta\n\n     1\tb\n");
+
+        new_ucmd!()
+            .arg(arg)
+            .pipe_in("a\nabc\nb") // footer section
+            .succeeds()
+            .stdout_is("     1\ta\n\n       b\n");
+    }
+}
+
+#[test]
+fn test_one_char_section_delimiter_expansion() {
+    for arg in ["-da", "--section-delimiter=a"] {
+        new_ucmd!()
+            .arg(arg)
+            .pipe_in("a\na:a:a:\nb") // header section
+            .succeeds()
+            .stdout_is("     1\ta\n\n       b\n");
+
+        new_ucmd!()
+            .arg(arg)
+            .pipe_in("a\na:a:\nb") // body section
+            .succeeds()
+            .stdout_is("     1\ta\n\n     1\tb\n");
+
+        new_ucmd!()
+            .arg(arg)
+            .pipe_in("a\na:\nb") // footer section
+            .succeeds()
+            .stdout_is("     1\ta\n\n       b\n");
+    }
+}
+
+#[test]
+fn test_non_ascii_one_char_section_delimiter() {
+    for arg in ["-dä", "--section-delimiter=ä"] {
+        new_ucmd!()
+            .arg(arg)
+            .pipe_in("a\näää\nb") // header section
+            .succeeds()
+            .stdout_is("     1\ta\n\n       b\n");
+
+        new_ucmd!()
+            .arg(arg)
+            .pipe_in("a\nää\nb") // body section
+            .succeeds()
+            .stdout_is("     1\ta\n\n     1\tb\n");
+
+        new_ucmd!()
+            .arg(arg)
+            .pipe_in("a\nä\nb") // footer section
+            .succeeds()
+            .stdout_is("     1\ta\n\n       b\n");
+    }
+}
+
+#[test]
+fn test_empty_section_delimiter() {
+    for arg in ["-d ''", "--section-delimiter=''"] {
+        new_ucmd!()
+            .arg(arg)
+            .pipe_in("a\n\nb")
+            .succeeds()
+            .stdout_is("     1\ta\n       \n     2\tb\n");
+    }
+}