fix: prevent UTF-8 slicing panic in Differ::fancy_replace

src/differ.rs

@@ -252,6 +252,15 @@ impl Differ {
         res
     }
 
+    fn split_at_char_boundary<'a>(line: &'a str, char_idx: usize) -> (&'a str, &'a str) {
+        let byte_idx = line
+            .char_indices()
+            .nth(char_idx)
+            .map(|(i, _)| i)
+            .unwrap_or(line.len());
+        line.split_at(byte_idx)
+    }
+
     fn qformat(
         &self,
         first_line: &str,
@@ -267,9 +276,9 @@ impl Differ {
             count_leading(second_line, '\t'),
         );
         common = cmp::min(common, count_leading(first_tags.split_at(common).0, ' '));
-        common = cmp::min(common, count_leading(first_tags.split_at(common).0, ' '));
-        first_tags = first_tags.split_at(common).1.trim_right();
-        second_tags = second_tags.split_at(common).1.trim_right();
+        common = cmp::min(common, count_leading(second_tags.split_at(common).0, ' '));
+        first_tags = Self::split_at_char_boundary(first_tags, common).1.trim_right();
+        second_tags = Self::split_at_char_boundary(second_tags, common).1.trim_right();
         let mut s = format!("- {}", first_line);
         res.push(s);
         if first_tags != "" {
@@ -338,3 +347,18 @@ fn test_qformat() {
         ]
     );
 }
+
+#[test]
+fn test_qformat_uses_both_tag_lines_for_alignment() {
+    let differ = Differ::new();
+    let result = differ.qformat("\t\tabc\n", "\t\tadc\n", "  ^", " ^");
+    assert_eq!(
+        result,
+        vec![
+            "- \t\tabc\n",
+            "? \t ^\n",
+            "+ \t\tadc\n",
+            "? \t^\n",
+        ]
+    );
+}

src/sequencematcher.rs

@@ -114,7 +114,7 @@ impl<'a, T: Sequence> SequenceMatcher<'a, T> {
         let second_sequence = self.second_sequence;
         let mut second_sequence_elements = HashMap::new();
         for (i, item) in second_sequence.iter().enumerate() {
-            let mut counter = second_sequence_elements
+            let counter = second_sequence_elements
                 .entry(item)
                 .or_insert_with(Vec::new);
             counter.push(i);

tests/tests.rs

@@ -133,6 +133,16 @@ fn test_differ_restore() {
     assert_eq!(second_text, Differ::restore(&diff, 2));
 }
 
+#[test]
+fn test_utf8_intraline_diff_no_panic() {
+    let first_text = vec!["\tcrab 🦀\n"];
+    let second_text = vec!["\tcrab 🦞\n"];
+    let differ = Differ::new();
+    let result = differ.compare(&first_text, &second_text).join("");
+    assert!(result.contains("🦀"));
+    assert!(result.contains("🦞"));
+}
+
 #[test]
 fn test_unified_diff() {
     let first_text = "one two three four".split(" ").collect::<Vec<&str>>();