From b17c1b6bcc49fdfb530e3c25316bbfdfa4e67a11 Mon Sep 17 00:00:00 2001
From: Ed Page <eopage@gmail.com>
Date: Thu, 18 Apr 2024 16:38:51 -0500
Subject: [PATCH] fix(parser): Return correct span for multi-byte characters

This is a step towards addressing rust-lang/cargo#13772
---
 crates/toml_edit/src/error.rs               | 18 +++++++++++-------
 crates/toml_edit/tests/testsuite/invalid.rs |  4 ++--
 2 files changed, 13 insertions(+), 9 deletions(-)

diff --git a/crates/toml_edit/src/error.rs b/crates/toml_edit/src/error.rs
index d9630fca..6a278024 100644
--- a/crates/toml_edit/src/error.rs
+++ b/crates/toml_edit/src/error.rs
@@ -21,19 +21,23 @@ impl TomlError {
     ) -> Self {
         use winnow::stream::Stream;
 
+        let message = error.inner().to_string();
+        let raw = raw.finish();
+        let raw = String::from_utf8(raw.to_owned()).expect("original document was utf8");
+
         let offset = error.offset();
-        let span = if offset == raw.len() {
-            offset..offset
+        let mut indices = raw[offset..].char_indices();
+        indices.next();
+        let len = if let Some((index, _)) = indices.next() {
+            index
         } else {
-            offset..(offset + 1)
+            raw.len()
         };
-
-        let message = error.inner().to_string();
-        let raw = raw.finish();
+        let span = offset..(offset + len);
 
         Self {
             message,
-            raw: Some(String::from_utf8(raw.to_owned()).expect("original document was utf8")),
+            raw: Some(raw),
             keys: Vec::new(),
             span: Some(span),
         }
diff --git a/crates/toml_edit/tests/testsuite/invalid.rs b/crates/toml_edit/tests/testsuite/invalid.rs
index a156792e..79724163 100644
--- a/crates/toml_edit/tests/testsuite/invalid.rs
+++ b/crates/toml_edit/tests/testsuite/invalid.rs
@@ -211,10 +211,10 @@ dotted key `k1` attempted to extend non-table type (integer)
 }
 
 #[test]
-#[should_panic]
 fn emoji_error_span() {
     let input = "😀";
     let err = input.parse::<toml_edit::DocumentMut>().unwrap_err();
     dbg!(err.span());
-    let _ = &input[err.span().unwrap()];
+    let actual = &input[err.span().unwrap()];
+    assert_eq!(actual, input);
 }