Add treat_east_asian_ambiguous_width_as_wide option

I've bundled this into termwiz's UnicodeVersion type as that is
a similar concept that is already routed through to the appropriate
function.

refs: https://github.com/wez/wezterm/issues/1888

config/src/config.rs

@@ -595,6 +595,9 @@ pub struct Config {
     #[serde(default = "default_unicode_version")]
     pub unicode_version: u8,
 
+    #[serde(default)]
+    pub treat_east_asian_ambiguous_width_as_wide: bool,
+
     #[serde(default = "default_true")]
     pub allow_download_protocols: bool,
 

config/src/terminal.rs

@@ -2,6 +2,7 @@
 
 use crate::{configuration, ConfigHandle, NewlineCanon};
 use std::sync::Mutex;
+use termwiz::cell::UnicodeVersion;
 use wezterm_term::color::ColorPalette;
 use wezterm_term::config::BidiMode;
 
@@ -80,8 +81,12 @@ impl wezterm_term::TerminalConfiguration for TermConfig {
         }
     }
 
-    fn unicode_version(&self) -> u8 {
-        self.configuration().unicode_version
+    fn unicode_version(&self) -> UnicodeVersion {
+        let config = self.configuration();
+        UnicodeVersion {
+            version: config.unicode_version,
+            ambiguous_are_wide: config.treat_east_asian_ambiguous_width_as_wide,
+        }
     }
 
     fn debug_key_events(&self) -> bool {

docs/changelog.md

@@ -19,6 +19,7 @@ As features stabilize some brief notes about them will accumulate here.
 * [AttachDomain](config/lua/keyassignment/AttachDomain.md) and [DetachDomain](config/lua/keyassignment/DetachDomain.md) key assignments
 * Specifying a domain name in a [SpawnCommand](config/lua/SpawnCommand.md) will cause that domain to be attached if it is in the detached state. This is useful when combined with [SwitchToWorkspace](config/lua/keyassignment/SwitchToWorkspace.md).
 * X11: wezterm now sets `_NET_WM_NAME` in addition to `WM_NAME` for clients that don't know how to fallback
+* [treat_east_asian_ambiguous_width_as_wide](config/lua/config/treat_east_asian_ambiguous_width_as_wide.md) for control how ambiguous width characters are resolved. [#1888](https://github.com/wez/wezterm/issues/1888)
 
 #### Changed
 * Debian packages now register wezterm as an alternative for `x-terminal-emulator`. Thanks to [@xpufx](https://github.com/xpufx)! [#1883](https://github.com/wez/wezterm/pull/1883)

docs/config/lua/config/treat_east_asian_ambiguous_width_as_wide.md

@@ -0,0 +1,17 @@
+# `treat_east_asian_ambiguous_width_as_wide = fasel`
+
+*Since: nightly builds only*
+
+Unicode defines a number of codepoints as having [Ambiguous
+Width](http://www.unicode.org/reports/tr11/#Ambiguous). These are characters
+whose width resolves differently according to context that is typically absent
+from the monospaced world of the terminal.
+
+WezTerm will by default treat ambiguous width as occupying a single cell.
+
+When `treat_east_asian_ambiguous_width_as_wide = true` WezTerm will treat them
+as being two cells wide.
+
+Note that changing this setting may have consequences for layout in text UI
+applications if their expectation of width differs from your choice of
+configuration.

term/src/config.rs

@@ -1,4 +1,5 @@
 use crate::color::ColorPalette;
+use termwiz::cell::UnicodeVersion;
 use termwiz::surface::{Line, SequenceNo};
 use wezterm_bidi::ParagraphDirectionHint;
 
@@ -181,8 +182,11 @@ pub trait TerminalConfiguration: std::fmt::Debug {
     /// that also alter the width of certain sequences, and that is too
     /// new for most deployed applications.
     // Coupled with config/src/lib.rs:default_unicode_version
-    fn unicode_version(&self) -> u8 {
-        9
+    fn unicode_version(&self) -> UnicodeVersion {
+        UnicodeVersion {
+            version: 9,
+            ambiguous_are_wide: false,
+        }
     }
 
     fn debug_key_events(&self) -> bool {

term/src/terminalstate/mod.rs

@@ -497,7 +497,7 @@ impl TerminalState {
 
         let color_map = default_color_map();
 
-        let unicode_version = UnicodeVersion(config.unicode_version());
+        let unicode_version = config.unicode_version();
 
         TerminalState {
             config,

term/src/terminalstate/performer.rs

@@ -7,7 +7,7 @@ use log::{debug, error};
 use num_traits::FromPrimitive;
 use std::fmt::Write;
 use std::ops::{Deref, DerefMut};
-use termwiz::cell::{grapheme_column_width, Cell, CellAttributes, SemanticType, UnicodeVersion};
+use termwiz::cell::{grapheme_column_width, Cell, CellAttributes, SemanticType};
 use termwiz::escape::csi::{CharacterPath, EraseInDisplay};
 use termwiz::escape::osc::{
     ChangeColorPair, ColorOrQuery, FinalTermSemanticPrompt, ITermProprietary,
@@ -559,7 +559,7 @@ impl<'a> Performer<'a> {
                 self.palette.take();
                 self.top_and_bottom_margins = 0..self.screen().physical_rows as VisibleRowIndex;
                 self.left_and_right_margins = 0..self.screen().physical_cols;
-                self.unicode_version = UnicodeVersion(self.config.unicode_version());
+                self.unicode_version = self.config.unicode_version();
                 self.unicode_version_stack.clear();
                 self.suppress_initial_title_change = false;
                 self.accumulating_title.take();
@@ -638,7 +638,7 @@ impl<'a> Performer<'a> {
                     }
                 }
                 ITermProprietary::UnicodeVersion(ITermUnicodeVersionOp::Set(n)) => {
-                    self.unicode_version = UnicodeVersion(n);
+                    self.unicode_version.version = n;
                 }
                 ITermProprietary::UnicodeVersion(ITermUnicodeVersionOp::Push(label)) => {
                     let vers = self.unicode_version;

termwiz/src/cell.rs

@@ -829,9 +829,15 @@ impl Cell {
 }
 
 #[derive(Clone, Copy, Debug, Eq, PartialEq)]
-pub struct UnicodeVersion(pub u8);
+pub struct UnicodeVersion {
+    pub version: u8,
+    pub ambiguous_are_wide: bool,
+}
 
-pub const LATEST_UNICODE_VERSION: UnicodeVersion = UnicodeVersion(14);
+pub const LATEST_UNICODE_VERSION: UnicodeVersion = UnicodeVersion {
+    version: 14,
+    ambiguous_are_wide: false,
+};
 
 /// Returns the number of cells visually occupied by a sequence
 /// of graphemes.
@@ -875,7 +881,9 @@ pub fn unicode_column_width(s: &str, version: Option<UnicodeVersion>) -> usize {
 /// the Cell that is used to hold a grapheme, and that per-Cell version
 /// can then be used to calculate width.
 pub fn grapheme_column_width(s: &str, version: Option<UnicodeVersion>) -> usize {
-    let version = version.unwrap_or(LATEST_UNICODE_VERSION).0;
+    let version = version.unwrap_or(LATEST_UNICODE_VERSION);
+    let ambiguous_are_wide = version.ambiguous_are_wide;
+    let version = version.version;
 
     let width: usize = s
         .chars()
@@ -887,6 +895,8 @@ pub fn grapheme_column_width(s: &str, version: Option<UnicodeVersion>) -> usize
             // <https://github.com/wez/wezterm/issues/1864>
             if c == WcWidth::Unassigned {
                 1
+            } else if c == WcWidth::Ambiguous && ambiguous_are_wide {
+                2
             } else if version >= 9 {
                 c.width_unicode_9_or_later()
             } else {