Add truncate_str_short() which always returns a prefix of the input

Document that the input of `truncate_str()` may be cut on a double width
char ("fullwidth") - then the output is no longer a prefix of the input.

`truncate_str_short()` could be faster by not constructing the result and
instead only returning an index, but it will be used in rare edge cases so
reusing existing logic is good enough.

Common code moved to `truncate_str_impl()`.

Assume (with fallback) that graphemes are at most 2 wide.
Fewer allocations.
Tests.

src/ansi/mod.rs

@@ -25,24 +25,19 @@ pub fn measure_text_width(s: &str) -> usize {
     })
 }
 
-/// Truncate string such that `tail` is present as a suffix, preceded by as much of `s` as can be
-/// displayed in the requested width.
-// Return string constructed as follows:
-// 1. `display_width` characters are available. If the string fits, return it.
-//
-// 2. Contribute graphemes and ANSI escape sequences from `tail` until either (1) `tail` is
-//    exhausted, or (2) the display width of the result would exceed `display_width`.
-//
-// 3. If tail was exhausted, then contribute graphemes and ANSI escape sequences from `s` until the
-//    display_width of the result would exceed `display_width`.
-pub fn truncate_str<'a>(s: &'a str, display_width: usize, tail: &str) -> Cow<'a, str> {
+fn truncate_str_impl<'a>(
+    s: &'a str,
+    display_width: usize,
+    tail: &str,
+    fill2w: Option<char>,
+) -> Cow<'a, str> {
     let items = ansi_strings_iterator(s).collect::<Vec<(&str, bool)>>();
     let width = strip_ansi_codes_from_strings_iterator(items.iter().copied()).width();
     if width <= display_width {
         return Cow::from(s);
     }
     let result_tail = if !tail.is_empty() {
-        truncate_str(tail, display_width, "").to_string()
+        truncate_str_impl(tail, display_width, "", fill2w).to_string()
     } else {
         String::new()
     };
@@ -51,20 +46,60 @@ pub fn truncate_str<'a>(s: &'a str, display_width: usize, tail: &str) -> Cow<'a,
     for (t, is_ansi) in items {
         if !is_ansi {
             for g in t.graphemes(true) {
-                let w = g.width();
-                if used + w > display_width {
-                    result.push_str(&" ".repeat(display_width.saturating_sub(used)));
+                let width_of_grapheme = g.width();
+                if used + width_of_grapheme > display_width {
+                    // Handle case "2." mentioned in `truncate_str` docs and fill the
+                    // hole left by double-width (2w) truncation.
+                    if let Some(fillchar) = fill2w {
+                        if width_of_grapheme == 2 && used < display_width {
+                            result.push(fillchar);
+                        } else if width_of_grapheme > 2 {
+                            // Should not happen, this means either unicode_segmentation
+                            // graphemes are too wide, or the unicode_width is calculated wrong.
+                            // Fallback:
+                            debug_assert!(width_of_grapheme <= 2, "strange grapheme width");
+                            for _ in 0..display_width.saturating_sub(used) {
+                                result.push(fillchar);
+                            }
+                        }
+                    }
                     break;
                 }
                 result.push_str(g);
-                used += w;
+                used += width_of_grapheme;
             }
         } else {
             result.push_str(t);
         }
     }
 
-    Cow::from(format!("{result}{result_tail}"))
+    result.push_str(&result_tail);
+    Cow::from(result)
+}
+
+/// Truncate string such that `tail` is present as a suffix, preceded by as much of `s` as can be
+/// displayed in the requested width. Even with `tail` empty the result may not be a prefix of `s`.
+// Return string constructed as follows:
+// 1. `display_width` characters are available. If the string fits, return it.
+//
+// 2. If a double-width (fullwidth) grapheme has to be cut in the following steps, replace the first
+//    half with a space (' '). If this happens the result is no longer a prefix of the input.
+//
+// 3. Contribute graphemes and ANSI escape sequences from `tail` until either (1) `tail` is
+//    exhausted, or (2) the display width of the result would exceed `display_width`.
+//
+// 4. If tail was exhausted, then contribute graphemes and ANSI escape sequences from `s` until the
+//    display_width of the result would exceed `display_width`.
+pub fn truncate_str<'a>(s: &'a str, display_width: usize, tail: &str) -> Cow<'a, str> {
+    truncate_str_impl(s, display_width, tail, Some(' '))
+}
+
+/// Truncate string `s` so it fits into `display_width`, ignoring any ANSI escape sequences when
+/// calculating the width. If a double-width ("fullwidth") grapheme has to be cut, it is omitted and
+/// the resulting string is *shorter* than `display_width`. But this way the result is always a
+/// prefix of the input `s`.
+pub fn truncate_str_short(s: &str, display_width: usize) -> Cow<str> {
+    truncate_str_impl(s, display_width, "", None)
 }
 
 pub fn parse_style_sections(s: &str) -> Vec<(ansi_term::Style, &str)> {
@@ -180,12 +215,12 @@ pub fn explain_ansi(line: &str, colorful: bool) -> String {
 
 #[cfg(test)]
 mod tests {
-    use crate::ansi::ansi_preserving_index;
+    use unicode_width::UnicodeWidthStr;
 
     // Note that src/ansi/console_tests.rs contains additional test coverage for this module.
     use super::{
-        ansi_preserving_slice, measure_text_width, parse_first_style,
-        string_starts_with_ansi_style_sequence, strip_ansi_codes, truncate_str,
+        ansi_preserving_index, ansi_preserving_slice, measure_text_width, parse_first_style,
+        string_starts_with_ansi_style_sequence, strip_ansi_codes, truncate_str, truncate_str_short,
     };
 
     #[test]
@@ -292,4 +327,39 @@ mod tests {
         assert_eq!(truncate_str("123", 2, "→"), "1→");
         assert_eq!(truncate_str("12ݶ", 1, "ݶ"), "ݶ");
     }
+
+    #[test]
+    fn test_truncate_str_at_double_width_grapheme() {
+        let one_double_four = "1＃4";
+        let double = "／";
+        assert_eq!(one_double_four.width(), 4);
+        assert_eq!(double.width(), 2);
+
+        assert_eq!(truncate_str(one_double_four, 1, ""), "1");
+        assert_eq!(truncate_str(one_double_four, 2, ""), "1 ");
+        assert_eq!(truncate_str(one_double_four, 3, ""), "1＃");
+        assert_eq!(truncate_str(one_double_four, 4, ""), "1＃4");
+
+        assert_eq!(truncate_str_short(one_double_four, 1), "1");
+        assert_eq!(truncate_str_short(one_double_four, 2), "1"); // !!
+        assert_eq!(truncate_str_short(one_double_four, 3), "1＃");
+        assert_eq!(truncate_str_short(one_double_four, 4), "1＃4");
+
+        assert_eq!(truncate_str(one_double_four, 1, double), " ");
+        assert_eq!(truncate_str(one_double_four, 2, double), "／");
+        assert_eq!(truncate_str(one_double_four, 3, double), "1／");
+        assert_eq!(truncate_str(one_double_four, 4, double), "1＃4");
+
+        assert_eq!(truncate_str(one_double_four, 0, ""), "");
+        assert_eq!(truncate_str(one_double_four, 0, double), "");
+        assert_eq!(truncate_str_short(one_double_four, 0), "");
+
+        assert_eq!(truncate_str(double, 0, double), "");
+        assert_eq!(truncate_str(double, 1, double), " ");
+        assert_eq!(truncate_str(double, 2, double), double);
+
+        assert_eq!(truncate_str_short(double, 0), "");
+        assert_eq!(truncate_str_short(double, 1), "");
+        assert_eq!(truncate_str_short(double, 2), double);
+    }
 }