Improve word wrapping

Now supports long trailing whitespace as well as tabs.

examples/text_wrapping.rs

@@ -14,7 +14,20 @@ fn draw(f: &mut Frame) {
         "This\u{00a0}sentence\u{00a0}is\u{00a0}separated\u{00a0}by\u{00a0}non-\u{2060}breaking\u{00a0}spaces.\n",
         "\n",
         "It can also properly handle wide graphemes (like emoji 🤔), ",
-        "including ones usually displayed incorrectly by terminal emulators, like 👩‍🔬 (a female scientist emoji).",
+        "including ones usually displayed incorrectly by terminal emulators, like 👩‍🔬 (a female scientist emoji).\n",
+        "\n",
+        "Finally, tabs are supported as well. ",
+        "The following text is rendered with a tab width of 4:\n",
+        "\tx\n",
+        "1\tx\n",
+        "12\tx\n",
+        "123\tx\n",
+        "1234\tx\n",
+        "12345\tx\n",
+        "123456\tx\n",
+        "1234567\tx\n",
+        "12345678\tx\n",
+        "123456789\tx\n",
     );
 
     let breaks = f.wrap(text, f.size().width.into());
@@ -46,6 +59,7 @@ fn main() {
     // Automatically enters alternate screen and enables raw mode
     let mut term = Terminal::new().unwrap();
     term.set_measuring(true);
+    term.set_tab_width(4);
 
     loop {
         // Render and display a frame. A full frame is displayed on the terminal

src/buffer.rs

@@ -1,8 +1,8 @@
 use crossterm::style::ContentStyle;
-use unicode_segmentation::UnicodeSegmentation;
 
 use crate::styled::Styled;
 use crate::widthdb::WidthDB;
+use crate::wrap;
 
 #[derive(Debug, Default, Clone, Copy, PartialEq, Eq)]
 pub struct Size {
@@ -128,25 +128,31 @@ impl Buffer {
         }
     }
 
-    pub fn write(&mut self, widthdb: &mut WidthDB, mut pos: Pos, styled: &Styled) {
+    pub fn write(&mut self, widthdb: &mut WidthDB, tab_width: u8, pos: Pos, styled: &Styled) {
         // If we're not even visible, there's nothing to do
         if pos.y < 0 || pos.y >= self.size.height as i32 {
             return;
         }
         let y = pos.y as u16;
 
+        let mut col: usize = 0;
         for styled_grapheme in styled.styled_graphemes() {
-            let width = widthdb.grapheme_width(styled_grapheme.content());
+            let x = pos.x + col as i32;
-            if width > 0 {
+            let g = *styled_grapheme.content();
-                self.write_grapheme(
+            let style = *styled_grapheme.style();
-                    pos.x,
+            if g == "\t" {
-                    y,
+                let width = wrap::tab_width_at_column(tab_width, col);
-                    width,
+                col += width as usize;
-                    styled_grapheme.content(),
+                for dx in 0..width {
-                    *styled_grapheme.style(),
+                    self.write_grapheme(x + dx as i32, y, width, " ", style);
-                );
+                }
+            } else {
+                let width = widthdb.grapheme_width(g);
+                col += width as usize;
+                if width > 0 {
+                    self.write_grapheme(x, y, width, g, style);
+                }
             }
-            pos.x += width as i32;
         }
     }
 

src/frame.rs

@@ -1,18 +1,28 @@
 //! Rendering the next frame.
 
-use crossterm::style::ContentStyle;
-
 use crate::buffer::Buffer;
 pub use crate::buffer::{Pos, Size};
 use crate::styled::Styled;
 use crate::widthdb::WidthDB;
 use crate::wrap;
 
-#[derive(Debug, Default)]
+#[derive(Debug)]
 pub struct Frame {
     pub(crate) widthdb: WidthDB,
     pub(crate) buffer: Buffer,
     cursor: Option<Pos>,
+    pub(crate) tab_width: u8,
+}
+
+impl Default for Frame {
+    fn default() -> Self {
+        Self {
+            widthdb: Default::default(),
+            buffer: Default::default(),
+            cursor: None,
+            tab_width: 8,
+        }
+    }
 }
 
 impl Frame {
@@ -58,10 +68,11 @@ impl Frame {
     }
 
     pub fn wrap(&mut self, text: &str, width: usize) -> Vec<usize> {
-        wrap::wrap(text, width, &mut self.widthdb)
+        wrap::wrap(&mut self.widthdb, self.tab_width, text, width)
     }
 
     pub fn write<S: Into<Styled>>(&mut self, pos: Pos, styled: S) {
-        self.buffer.write(&mut self.widthdb, pos, &styled.into());
+        self.buffer
+            .write(&mut self.widthdb, self.tab_width, pos, &styled.into());
     }
 }

src/terminal.rs

@@ -59,6 +59,14 @@ impl Terminal {
         Ok(())
     }
 
+    pub fn set_tab_width(&mut self, tab_width: u8) {
+        self.frame.tab_width = tab_width;
+    }
+
+    pub fn tab_width(&self) -> u8 {
+        self.frame.tab_width
+    }
+
     pub fn set_measuring(&mut self, active: bool) {
         self.frame.widthdb.active = active;
     }

src/wrap.rs

@@ -5,9 +5,11 @@ use unicode_segmentation::UnicodeSegmentation;
 
 use crate::widthdb::WidthDB;
 
-// TODO Handle tabs separately?
+pub fn tab_width_at_column(tab_width: u8, col: usize) -> u8 {
-// TODO Convert into an iterator?
+    tab_width - (col % tab_width as usize) as u8
-pub fn wrap(text: &str, width: usize, widthdb: &mut WidthDB) -> Vec<usize> {
+}
+
+pub fn wrap(widthdb: &mut WidthDB, tab_width: u8, text: &str, width: usize) -> Vec<usize> {
     let mut breaks = vec![];
 
     let mut break_options = unicode_linebreak::linebreaks(text).peekable();
@@ -16,8 +18,10 @@ pub fn wrap(text: &str, width: usize, widthdb: &mut WidthDB) -> Vec<usize> {
     let mut valid_break = None;
     let mut valid_break_width = 0;
 
-    // Width of the line at the current grapheme
+    // Width of the line at the current grapheme (with and without trailing
+    // whitespace)
     let mut current_width = 0;
+    let mut current_width_trimmed = 0;
 
     for (gi, g) in text.grapheme_indices(true) {
         // Advance break options
@@ -38,6 +42,7 @@ pub fn wrap(text: &str, width: usize, widthdb: &mut WidthDB) -> Vec<usize> {
                     valid_break = None;
                     valid_break_width = 0;
                     current_width = 0;
+                    current_width_trimmed = 0;
                 }
                 BreakOpportunity::Allowed => {
                     valid_break = Some(bi);
@@ -46,31 +51,52 @@ pub fn wrap(text: &str, width: usize, widthdb: &mut WidthDB) -> Vec<usize> {
             }
         }
 
-        let grapheme_width: usize = widthdb.grapheme_width(g).into();
+        // Calculate widths after current grapheme
-        if current_width + grapheme_width > width {
+        let g_width = if g == "\t" {
-            if current_width == 0 {
+            tab_width_at_column(tab_width, current_width) as usize
-                // The grapheme is wider than the maximum width, so we'll allow
+        } else {
-                // it, thereby forcing the following grapheme to break no matter
+            widthdb.grapheme_width(g) as usize
-                // what (either because of a mandatory or allowed break, or via
+        };
-                // a forced break).
+        let mut new_width = current_width + g_width;
-            } else if let Some(bi) = valid_break {
+        let mut new_width_trimmed = if g.chars().all(|c| c.is_whitespace()) {
-                // We can't fit the grapheme onto the current line, so we'll
+            current_width_trimmed
-                // just break at the last valid break point.
+        } else {
+            new_width
+        };
+
+        // Wrap at last break point if necessary
+        if new_width_trimmed > width {
+            if let Some(bi) = valid_break {
                 breaks.push(bi);
-                current_width -= valid_break_width;
+                new_width -= valid_break_width;
+                new_width_trimmed = new_width_trimmed.saturating_sub(valid_break_width);
                 valid_break = None;
                 valid_break_width = 0;
-            } else {
-                // Forced break in the midde of a normally non-breakable chunk
-                // because there have been no valid break points yet.
-                breaks.push(gi);
-                valid_break = None;
-                valid_break_width = 0;
-                current_width = 0;
             }
         }
 
-        current_width += grapheme_width;
+        // Perform a forced break if still necessary
+        if new_width_trimmed > width {
+            if new_width == g_width {
+                // The grapheme is the only thing on the current line and it is
+                // wider than the maximum width, so we'll allow it, thereby
+                // forcing the following grapheme to break no matter what
+                // (either because of a mandatory or allowed break, or via a
+                // forced break).
+            } else {
+                // Forced break in the midde of a normally non-breakable chunk
+                // because there are no valid break points.
+                breaks.push(gi);
+                new_width = 0;
+                new_width_trimmed = 0;
+                valid_break = None;
+                valid_break_width = 0;
+            }
+        }
+
+        // Update current width
+        current_width = new_width;
+        current_width_trimmed = new_width_trimmed;
     }
 
     breaks