Merge pull request #4260 from sbentmar/expand-fix-perl-failures

expand: improve plus specifier handling

src/uu/expand/src/expand.rs

@@ -126,12 +126,8 @@ fn tabstops_parse(s: &str) -> Result<(RemainingMode, Vec<usize>), ParseError> {
         let bytes = word.as_bytes();
         for i in 0..bytes.len() {
             match bytes[i] {
-                b'+' => {
+                b'+' => remaining_mode = RemainingMode::Plus,
-                    remaining_mode = RemainingMode::Plus;
+                b'/' => remaining_mode = RemainingMode::Slash,
-                }
-                b'/' => {
-                    remaining_mode = RemainingMode::Slash;
-                }
                 _ => {
                     // Parse a number from the byte sequence.
                     let s = from_utf8(&bytes[i..]).unwrap();
@@ -191,6 +187,10 @@ fn tabstops_parse(s: &str) -> Result<(RemainingMode, Vec<usize>), ParseError> {
     if nums.is_empty() {
         nums = vec![DEFAULT_TABSTOP];
     }
+
+    if nums.len() < 2 {
+        remaining_mode = RemainingMode::None;
+    }
     Ok((remaining_mode, nums))
 }
 
@@ -336,17 +336,19 @@ fn open(path: &str) -> BufReader<Box<dyn Read + 'static>> {
 /// in the `tabstops` slice is interpreted as a relative number of
 /// spaces, which this function will return for every input value of
 /// `col` beyond the end of the second-to-last element of `tabstops`.
-///
-/// If `remaining_mode` is [`RemainingMode::Plus`], then the last entry
-/// in the `tabstops` slice is interpreted as a relative number of
-/// spaces, which this function will return for every input value of
-/// `col` beyond the end of the second-to-last element of `tabstops`.
 fn next_tabstop(tabstops: &[usize], col: usize, remaining_mode: &RemainingMode) -> usize {
     let num_tabstops = tabstops.len();
     match remaining_mode {
         RemainingMode::Plus => match tabstops[0..num_tabstops - 1].iter().find(|&&t| t > col) {
             Some(t) => t - col,
-            None => tabstops[num_tabstops - 1] - 1,
+            None => {
+                let step_size = tabstops[num_tabstops - 1];
+                let last_fixed_tabstop = tabstops[num_tabstops - 2];
+                let characters_since_last_tabstop = col - last_fixed_tabstop;
+
+                let steps_required = 1 + characters_since_last_tabstop / step_size;
+                steps_required * step_size - characters_since_last_tabstop
+            }
         },
         RemainingMode::Slash => match tabstops[0..num_tabstops - 1].iter().find(|&&t| t > col) {
             Some(t) => t - col,
@@ -487,8 +489,8 @@ mod tests {
     #[test]
     fn test_next_tabstop_remaining_mode_plus() {
         assert_eq!(next_tabstop(&[1, 5], 0, &RemainingMode::Plus), 1);
-        assert_eq!(next_tabstop(&[1, 5], 3, &RemainingMode::Plus), 4);
+        assert_eq!(next_tabstop(&[1, 5], 3, &RemainingMode::Plus), 3);
-        assert_eq!(next_tabstop(&[1, 5], 6, &RemainingMode::Plus), 4);
+        assert_eq!(next_tabstop(&[1, 5], 6, &RemainingMode::Plus), 5);
     }
 
     #[test]

tests/by-util/test_expand.rs

@@ -298,3 +298,93 @@ fn test_tabs_and_tabs_shortcut_mixed() {
         //          01234567890
         .stdout_is("  a  b c");
 }
+
+#[test]
+fn test_ignore_initial_plus() {
+    new_ucmd!()
+        .args(&["--tabs=+3"])
+        .pipe_in("\ta\tb\tc")
+        .succeeds()
+        //          01234567890
+        .stdout_is("   a  b  c");
+}
+
+#[test]
+fn test_ignore_initial_pluses() {
+    new_ucmd!()
+        .args(&["--tabs=++3"])
+        .pipe_in("\ta\tb\tc")
+        .succeeds()
+        //          01234567890
+        .stdout_is("   a  b  c");
+}
+
+#[test]
+fn test_ignore_initial_slash() {
+    new_ucmd!()
+        .args(&["--tabs=/3"])
+        .pipe_in("\ta\tb\tc")
+        .succeeds()
+        //          01234567890
+        .stdout_is("   a  b  c");
+}
+
+#[test]
+fn test_ignore_initial_slashes() {
+    new_ucmd!()
+        .args(&["--tabs=//3"])
+        .pipe_in("\ta\tb\tc")
+        .succeeds()
+        //          01234567890
+        .stdout_is("   a  b  c");
+}
+
+#[test]
+fn test_ignore_initial_plus_slash_combination() {
+    new_ucmd!()
+        .args(&["--tabs=+/3"])
+        .pipe_in("\ta\tb\tc")
+        .succeeds()
+        //          01234567890
+        .stdout_is("   a  b  c");
+}
+
+#[test]
+fn test_comma_with_plus_1() {
+    new_ucmd!()
+        .args(&["--tabs=3,+6"])
+        .pipe_in("\t111\t222\t333")
+        .succeeds()
+        //          01234567890
+        .stdout_is("   111   222   333");
+}
+
+#[test]
+fn test_comma_with_plus_2() {
+    new_ucmd!()
+        .args(&["--tabs=1,+5"])
+        .pipe_in("\ta\tb\tc")
+        .succeeds()
+        //          01234567890
+        .stdout_is(" a    b    c");
+}
+
+#[test]
+fn test_comma_with_plus_3() {
+    new_ucmd!()
+        .args(&["--tabs=2,+5"])
+        .pipe_in("a\tb\tc")
+        .succeeds()
+        //          01234567890
+        .stdout_is("a b    c");
+}
+
+#[test]
+fn test_comma_with_plus_4() {
+    new_ucmd!()
+        .args(&["--tabs=1,3,+5"])
+        .pipe_in("a\tb\tc")
+        .succeeds()
+        //          01234567890
+        .stdout_is("a  b    c");
+}