I wonder why you implemented Charset interface here?
I couldn't find implemented encode & decode functions.

Good job to make select function
It'll make a shortest encoded data from sort.

Seeing this result, I thought it would be good to implement the Charset interface called AutoCharset.

class AutoCharset : public Charset {
  // encode & decode implemenation
}

However, what I expected is that select is gonna tell us type of charset.
this is utility for charset selection. so, function definition might be like below.
your proposal is also this one #222

Chraset* ChrasetAutoSelector::select(std::string src);

or

shared_ptr<Charset> CharsetAutoSelector::select(std::string src);

what do you think about it?

I think so. It has been revised and I will commit soon.

@DahyeonWoo
Here is my thinking.
There are so many methods to implement CharsetAutoSelector :)

Firstly, analyze the charset's charecteristics.

• ASCII is excluded because UTF-8 is included
• UTF32 is unconditionally 4 bytes, except that it always returns the maximum length.
• All that is left is to compare UTF-8 and UTF-16
• If there are only Korean characters, UTF16 is short (2 bytes per Korean character)
• If there is a lot of English and sometimes a little bit of Korean, UTF8 is short (because English is 1 byte and Korean is 3 bytes)

My pseudo code :)

std::shared_ptr<Charset> CharsetAutoSelector::select(std::string src) {
    std::shared_ptr<Charset> utf8Charset = make_shared<UTF8Charset>();
    std::shared_ptr<Charset> utf16Charset = make_shared<UTF16Charset>();

    return (utf8Charset->encode(src) > utf16Charset->encode(src) ? utc16Charset : utf8Charset;
}

And.. if return type is CharsetType,
we can use like below.

CharsetType CharsetAutoSelector::select(std::string src) {
    return (UTF8Charset().encode(src) > UTF16Charset().encode(src)) ? CharsetType::UTF16 : CharsetType::UTF8;
}

CharsetType is defined in Definition.h

Which one do you like?
if you have some question, let me know :)