I don’t have a working example at hand right now.
But the point is that
the compiled source code in the binary file
in Hexrays pseudocode is shown in reverse order.
This doesn’t affect the result.
But the compiled binary file from this pseudocode will differ from the original.
Therefore, to get pseudocode closer to the original, you need to compile the pseudocode and output another pseudocode from this binary.
It turns out to be some kind of Russian nesting doll.
I remember a few years ago, Rolf Rolles wrote something about this on Twitter, saying that he seemed to have reversed the order of AST output and obtained a more acceptable result,
but he never shared his findings publicly.