tests(cli): add test for improper unicode encoding

[kitsonk]

Fixes #8161

(Or actually will fix when swc-project/swc#1188 is fixed, this is a test case to help ensure it is fixed at the moment)

[bartlomieju]

@kitsonk please rebase, #8197 should include the fix

[kitsonk]

@bartlomieju @kdy1 the test case is still failing because the output is not as expected/valid. swc is still modifying the input string.

[kdy1]

@kitsonk I'm considering an option in swc_ecma_codegen to preserve input or emit ascii only. How do you think about it?

[kitsonk]

For us, we would always choose to preserve input, so yes. I am not sure what the advantages of re-encoding the input would be.

[kdy1]

This will be fixed by swc-project/swc#1221

[kdy1]

Can you try this again with the new version?

[kitsonk]

@kdy1 it still "modifies" the strings, but the way it re-encodes them exhibits the same behaviour, so I have updated the expected output to match the behaviour of swc.

[kdy1]

It should modify string anymore.

There's a test for the exact same input.

deno/Cargo.lock

Lines 2380 to 2382 in 686a17f

	[[package]]
	name = "swc_ecma_codegen"
	version = "0.41.0"

It seems like deno is using old version of codegen. It's fixed by v0.41.1. Can you try cargo update -p swc_ecma_codegen?

[bartlomieju]

LGTM