NormalizerNFKC: add unify_katakana_z_sounds option

[HashidaTKS]

When unify_katakana_z_sounds is specified, NormalizerNFKC* normalize characters as below.

ズァ -> ザ
ズィ -> ジ
ズェ -> ゼ
ズォ -> ゾ

Usage:

normalize \
  'NormalizerNFKC130("unify_katakana_z_sounds", true, \
                     "report_source_offset", true)' \
  "ズァズィズェズォ" \
  WITH_CHECKS|WITH_TYPES

[HashidaTKS]

Or directly specify 0xe3.
I have used current[0] because current implementation for other normalizer options use the current variable as possible as we can.

[kou]

I think that current[0] is better because existing implementation already uses the style.

[HashidaTKS]

Thanks.
I see.

[kou]

If you want to introduce a descriptive variable for this condition, could you use more meaningful name like is_du_or_zu or something?

In this case, I don't thing that we need a descriptive variable because the following code is well descriptive but it's not a strong opinion:

if ((char_length == 3) &&
    /* U+30C5 KATAKANA LETTER DU */
    (current[0] == 0xe3 && current[1] == 0x83 && current[2] == 0x85) ||
    /* U+30BA KATAKANA LETTER ZU */
    (current[0] == 0xe3 && current[1] == 0x82 && current[2] == 0xba)) {

[HashidaTKS]

We need to add extra parentheses before and after character checks because && has higher priority than ||, and I feel it somehow undermines readability.
That is the reason why I introduced a descriptive variable.

if ((char_length == 3) &&
    /* U+30C5 KATAKANA LETTER DU */
    ((current[0] == 0xe3 && current[1] == 0x83 && current[2] == 0x85) ||
    /* U+30BA KATAKANA LETTER ZU */
    (current[0] == 0xe3 && current[1] == 0x82 && current[2] == 0xba))) 

But this is also not a strong opinion...
As a conclusion, I will remove the descriptive variable and fix it like above.

[HashidaTKS]

Fixed

[kou]

I think that current[0] is better because existing implementation already uses the style.

[kou]

Suggested change

	(current[0] == 0xe3 && current[1] == 0x82 && current[2] == 0xba))) {
	(current[0] == 0xe3 && current[1] == 0x82 && current[2] == 0xba))) {

[HashidaTKS]

Fixed.

[HashidaTKS]

@kou

Thanks, I have addressed your comments.

[kou]

Could you split ヅァヅィヅヅェヅォ to unify_katakana_d_sounds?

[HashidaTKS]

Could you split ヅァヅィヅヅェヅォ to unify_katakana_d_sounds?

Let me confirm your suggestion.
Do you mean unify_katakana_d_sounds converts ヅァヅィヅヅェヅォ to ザジヅゼゾ ?

[kou]

Yes. They aren't started with z.

[HashidaTKS]

Yes. They aren't started with z.

Thanks, I will create unify_katakana_d_sounds as such behaviour.

[HashidaTKS]

Could you split ヅァヅィヅヅェヅォ to unify_katakana_d_sounds?

I have removed ヅァヅィヅヅェヅォ from the unify_katakana_z_sounds target.
I will create unify_katakana_d_sounds in another pull request.