Merge pull request #2164 from GitoxideLabs/improvements

fix: make non-UTF8 unified diffs possible again.

Author: Byron

gix-diff/src/blob/unified_diff/impls.rs

@@ -200,7 +200,7 @@ where
         buf.clear();
         for &(line_type, content) in lines {
             buf.push(line_type.to_prefix() as u8);
-            buf.push_str(std::str::from_utf8(content).map_err(std::io::Error::other)?);
+            buf.extend_from_slice(content);
 
             if !content.ends_with_str(self.newline) {
                 buf.push_str(self.newline);

gix-diff/tests/diff/blob/unified_diff.rs

@@ -3,6 +3,7 @@ use gix_diff::blob::{
     unified_diff::{ConsumeHunk, ContextSize, DiffLineKind, HunkHeader},
     Algorithm, UnifiedDiff,
 };
+use gix_object::bstr::BString;
 
 #[test]
 fn removed_modified_added() -> crate::Result {
@@ -131,6 +132,45 @@ fn context_overlap_by_one_line_move_up() -> crate::Result {
     Ok(())
 }
 
+#[test]
+fn non_utf8() -> crate::Result {
+    let a = &b"\xC0\x80"[..];
+    let b = b"ascii";
+
+    let interner = gix_diff::blob::intern::InternedInput::new(a, b);
+    let err = gix_diff::blob::diff(
+        Algorithm::Myers,
+        &interner,
+        UnifiedDiff::new(
+            &interner,
+            ConsumeBinaryHunk::new(String::new(), "\n"),
+            ContextSize::symmetrical(3),
+        ),
+    )
+    .unwrap_err();
+    assert_eq!(
+        err.to_string(),
+        "invalid UTF-8 found at byte offset 1",
+        "strings enforce an encoding, which fails here"
+    );
+
+    let actual = gix_diff::blob::diff(
+        Algorithm::Myers,
+        &interner,
+        UnifiedDiff::new(
+            &interner,
+            ConsumeBinaryHunk::new(BString::default(), "\n"),
+            ContextSize::symmetrical(3),
+        ),
+    )?;
+    insta::assert_snapshot!(actual, @r"
+    @@ -1,1 +1,1 @@
+    -��
+    +ascii
+    ");
+    Ok(())
+}
+
 #[test]
 fn context_overlap_by_one_line_move_down() -> crate::Result {
     let a = "2\n3\n4\n5\n6\n7\n";