Perl problem with substituting UTF-8 string on windows

I am trying to substitute substrings in a text file with Perl on Windows 10 using command line.

C:\Windows\System32\chcp 65001 & type test.txt | c:\Strawberry\perl\bin\perl -CSD -pe "use open ':std', ':encoding(UTF-8)'; binmode(STDOUT, ':utf8'); binmode(STDIN, ':encoding(utf8)'); s/__compare_loan__/So sánh sản phẩm cho vay/g" 

File test.txt (saved as UTF-8):

Our benefit: __compare_loan__

Output:

> Active code page: 65001
> Our benefit: So sánh s?n ph?m cho vay

If I add use utf8; at the beginning of the Perl script, I get:

> Active code page: 65001 Malformed UTF-8 character: \xe1\x6e\x68
> (unexpected non-continuation byte 0x6e, immediately after start byte
> 0xe1; need 3 bytes, got 1) at -e line 1. Malformed UTF-8 character
> (fatal) at -e line 1.

Please any idea how do I get rid of the question marks in the output?

That error when you add use utf8; to your one liner suggests that the arguments to perl are being given in CP-1252 or a similar code page, not in UTF-8 (0xE1 in CP-1252 is á, 0x6E is n and 0x68 is h).

One portable fix is to use character escapes instead of trying to include the non-ascii characters directly:

C:\Code\SO> chcp 65001 & type test.txt | perl -CSD -pe "s/__compare_loan__/So s\x{e1}nh s\x{1ea3}n ph\x{1ea9}m cho vay/g"
Active code page: 65001
Our benefit: So sánh sản phẩm cho vay

(Tested with Strawberry Perl 5.32.1 and the standard Windows 10 Command Prompt application)

Note that using -CSD means you don't need all that use open and binmode stuff; it's all implied by the arguments to -C.