Merge pull request #465 from foriequal0/fix/unicode-width

Fix misalignment in report highlight and indent when there are double width characters

Cargo.lock

@@ -1576,6 +1576,7 @@ dependencies = [
  "typos-vars",
  "unicase",
  "unicode-segmentation",
+ "unicode-width",
  "varcon-core",
  "yansi",
 ]

Cargo.toml

@@ -91,6 +91,7 @@ encoding = "0.2"
 kstring = { version = "2.0.0", features = ["serde"] }
 typed-arena = "2.0.1"
 maplit = "1.0"
+unicode-width = "0.1.9"
 
 [dev-dependencies]
 assert_fs = "1.0"

src/bin/typos-cli/report.rs

@@ -3,6 +3,8 @@
 use std::io::{self, Write};
 use std::sync::atomic;
 
+use unicode_width::UnicodeWidthStr;
+
 use typos_cli::report::{Context, Message, Report, Typo};
 
 #[derive(Copy, Clone, Debug)]
@@ -218,8 +220,11 @@ fn print_long_correction(msg: &Typo, palette: Palette) -> Result<(), std::io::Er
         let line_num = context.line_num.to_string();
         let line_indent: String = itertools::repeat_n(" ", line_num.len()).collect();
 
-        let hl_indent: String = itertools::repeat_n(" ", column).collect();
-        let hl: String = itertools::repeat_n("^", msg.typo.len()).collect();
+        let visible_column = UnicodeWidthStr::width(start.as_ref());
+        let visible_len = UnicodeWidthStr::width(msg.typo);
+
+        let hl_indent: String = itertools::repeat_n(" ", visible_column).collect();
+        let hl: String = itertools::repeat_n("^", visible_len).collect();
 
         writeln!(handle, "{} |", line_indent)?;
         writeln!(

tests/cmd/stdin-failure-multiwidth.stdin

@@ -0,0 +1,34 @@
+---
+Korean character (NFC)
+Grapheme clusters: 1, codepoints: 1, UnicodeWidthStr::width() == 2
+
+한 Apropriate world
+   ^^^^^^^^^^ highlight here
+
+---
+Korean character (NFD).
+Grapheme clusters: 1, codepoints: 3, UnicodeWidthStr::width() == 2
+
+한 Apropriate world
+   ^^^^^^^^^^ highlight here
+
+---
+Eye in Speech Bubble Emoji (U+1F441 U+FE0F U+200D U+1F5E8 U+FE0F, Recommended Emoji ZWJ Sequences, v2.0)
+Grapheme clusters: 1, codepoints: 5, UnicodeWidthStr::width() == 2 (Read NOTE: https://github.com/unicode-rs/unicode-width)
+
+👁️‍🗨️ Apropriate world
+   ^^^^^^^^^^ highlight here
+
+---
+Face with spiral eyes (U+1F635 U+200D U+1F4AB, Recommended Emoji ZWJ Sequence, v13.1)
+Grapheme clusters: 1, codepoints: 3, UnicodeWidthStr::width() == 4 (Read NOTE: https://github.com/unicode-rs/unicode-width)
+
+😵‍💫 Apropriate world
+     ^^^^^^^^^^ highlight here
+
+---
+Horizontal tab (\t, U+09)
+Grapheme clusters: 1, codepoints: 1, UnicodeWidthStr::width() == 0
+
+	Apropriate world
+^^^^^^^^^^ highlight here

tests/cmd/stdin-failure-multiwidth.stdout

@@ -0,0 +1,30 @@
+error: `Apropriate` should be `Appropriate`
+  --> -:5:2
+  |
+5 | 한 Apropriate world
+  |    ^^^^^^^^^^
+  |
+error: `Apropriate` should be `Appropriate`
+  --> -:12:2
+   |
+12 | 한 Apropriate world
+   |    ^^^^^^^^^^
+   |
+error: `Apropriate` should be `Appropriate`
+  --> -:19:2
+   |
+19 | 👁️‍🗨️ Apropriate world
+   |    ^^^^^^^^^^
+   |
+error: `Apropriate` should be `Appropriate`
+  --> -:26:2
+   |
+26 | 😵‍💫 Apropriate world
+   |      ^^^^^^^^^^
+   |
+error: `Apropriate` should be `Appropriate`
+  --> -:33:1
+   |
+33 |  Apropriate world
+   | ^^^^^^^^^^
+   |

tests/cmd/stdin-failure-multiwidth.toml

@@ -0,0 +1,3 @@
+bin.name = "typos"
+args = "-"
+status.code = 2