fix(ui): UnifiedDiffParser mishandles paths with spaces and git-quoted paths

diff --git headers pack two paths on one space-delimited line, which broke for unquoted paths containing spaces and for git's C-style octal-quoted paths (non-ASCII filenames). Add quote-aware header splitting plus a git unquote helper, and prefer the unambiguous "--- a/"/"+++ b/" lines to correct the file's identity when present.
2026-07-23 18:11:52 +02:00
parent f18e03354a
commit 0f2d202b01
3 changed files with 206 additions and 10 deletions
@@ -106,4 +106,72 @@ public class UnifiedDiffParserTests
        Assert.False(file.HasLines);
        Assert.True(file.IsEmptyContent);
    }
+
+    [Fact]
+    public void Path_with_spaces_is_parsed_in_full()
+    {
+        const string raw =
+            "diff --git a/pfad mit space.txt b/pfad mit space.txt\n" +
+            "index 111..222 100644\n" +
+            "--- a/pfad mit space.txt\t\n" +
+            "+++ b/pfad mit space.txt\t\n" +
+            "@@ -1,1 +1,1 @@\n" +
+            "-old\n" +
+            "+new\n";
+
+        var file = Assert.Single(UnifiedDiffParser.Parse(raw));
+        Assert.Equal("pfad mit space.txt", file.Path);
+        Assert.Equal(DiffFileStatus.Modified, file.Status);
+    }
+
+    [Fact]
+    public void Rename_with_spaces_on_both_sides_captures_full_paths()
+    {
+        const string raw =
+            "diff --git a/old name with space.txt b/new name with space.txt\n" +
+            "similarity index 75%\n" +
+            "rename from old name with space.txt\n" +
+            "rename to new name with space.txt\n" +
+            "index 111..222 100644\n" +
+            "--- a/old name with space.txt\t\n" +
+            "+++ b/new name with space.txt\t\n";
+
+        var file = Assert.Single(UnifiedDiffParser.Parse(raw));
+        Assert.Equal(DiffFileStatus.Renamed, file.Status);
+        Assert.Equal("old name with space.txt", file.OldPath);
+        Assert.Equal("new name with space.txt", file.Path);
+    }
+
+    [Fact]
+    public void Git_quoted_unicode_path_is_unquoted()
+    {
+        // git quotes paths containing non-ASCII bytes as C-style octal escapes,
+        // e.g. "pfad_ä.txt" -> "pfad_\303\244.txt" (UTF-8 bytes for 'ä').
+        const string raw =
+            "diff --git \"a/pfad_\\303\\244.txt\" \"b/pfad_\\303\\244.txt\"\n" +
+            "index 111..222 100644\n" +
+            "--- \"a/pfad_\\303\\244.txt\"\n" +
+            "+++ \"b/pfad_\\303\\244.txt\"\n" +
+            "@@ -1,1 +1,1 @@\n" +
+            "-old\n" +
+            "+new\n";
+
+        var file = Assert.Single(UnifiedDiffParser.Parse(raw));
+        Assert.Equal("pfad_ä.txt", file.Path);
+    }
+
+    [Fact]
+    public void Binary_file_with_spaces_is_parsed_in_full()
+    {
+        const string raw =
+            "diff --git a/bin file.png b/bin file.png\n" +
+            "new file mode 100644\n" +
+            "index 000..abc\n" +
+            "Binary files /dev/null and b/bin file.png differ\n";
+
+        var file = Assert.Single(UnifiedDiffParser.Parse(raw));
+        Assert.Equal("bin file.png", file.Path);
+        Assert.True(file.IsBinary);
+        Assert.False(file.HasLines);
+    }
 }