[Java|Go] Check string is ascii before using meta string encoding

[chaokunyang]

Is your feature request related to a problem? Please describe.

In #1514 and #1566 , we compress every char using 5/6 bytes. But we didn't check the encoding is ascii. In UTF-8, some byte may be in range of ascii char, but some not. We may take a utf-8 string as a meta string by accident?

Describe the solution you'd like

Check string is ascii before using meta string encoding

Additional context

[jasonmokk]

Hi @chaokunyang I'd like to contribute to Apache Fury, can you please assign this issue to me?

[LiangliangSui]

Great! thanks for the willingness to contribute to Fury.

[LiangliangSui]

// org/apache/fury/meta/MetaStringEncoder.java
public MetaString encode(String input) {
  if (input.isEmpty()) {
    return new MetaString(input, Encoding.UTF_8, specialChar1, specialChar2, new byte[0]);
  }
  Encoding encoding = computeEncoding(input);
  return encode(input, encoding);
}

Could we judge here whether the input is all ASCII encoded? If not, just return UTF-8 encoded MetaString directly, so that we can save time on computeEncoding and encode.

WDYT @chaokunyang

[LiangliangSui]

In addition, we also need to add unit tests to cover this issue

[jasonmokk]

@LiangliangSui Thanks for bringing that up, can I submit another PR to correct that/add unit tests? I also think it would be optimal to have the ASCII check early and just directly return a UTF-8 encoded MetaString.

[LiangliangSui]

can I submit another PR to correct that/add unit tests?

Sure, that is great!