zig fmt: fix non-UTF-8 encoding #2820

Fixes #2820

After reading the source code, the first two bytes are inspected, and
if they correspond to a UTF-16 BOM in little-endian order, the source
code is converted to UTF-8.

1 files changed, 9 insertions, 0 deletions

diff --git a/test/cli.zig b/test/cli.zig
index c0702fa54c..db58511d14 100644
--- a/test/cli.zig
+++ b/test/cli.zig
@@ -174,4 +174,13 @@ fn testZigFmt(zig_exe: []const u8, dir_path: []const u8) !void {
     const run_result3 = try exec(dir_path, true, &[_][]const u8{ zig_exe, "fmt", dir_path });
     // both files have been formatted, nothing should change now
     testing.expect(run_result3.stdout.len == 0);
+
+    // Check UTF-16 decoding
+    const fmt4_zig_path = try fs.path.join(a, &[_][]const u8{ dir_path, "fmt4.zig" });
+    var unformatted_code_utf16 = "\xff\xfe \x00 \x00 \x00 \x00/\x00/\x00 \x00n\x00o\x00 \x00r\x00e\x00a\x00s\x00o\x00n\x00";
+    try fs.cwd().writeFile(fmt4_zig_path, unformatted_code_utf16);
+
+    const run_result4 = try exec(dir_path, true, &[_][]const u8{ zig_exe, "fmt", dir_path });
+    testing.expect(std.mem.startsWith(u8, run_result4.stdout, fmt4_zig_path));
+    testing.expect(run_result4.stdout.len == fmt4_zig_path.len + 1 and run_result4.stdout[run_result4.stdout.len - 1] == '\n');
 }