Localisation: Chinese

[kevinlin18]

Hi @JuliusSweetland ! Is that possible to separate 'Language Enum' current used by UI language and keyboard language? or a PR about it is welcome?
There are 3 English options: Canada, UK, US actually shared same UI, the difference is dictionary. I think we cam merge them in UI list, keep separate in keyboard list.

I translated UI into Chinese Traditional and tried to make bopomofo IME work in Optikey.
Localisation is 100% worked, but IME is not in ideal condition.
So I think we can make a PR about UI language first.
But it will crash if someone choose Chinese on keyboard list, so this the second reason I think we should have a new Enum.

Sorry for post again, but I didn't get e-mail reply or response in #148 ,maybe GitHub didn't notify you

[JuliusSweetland]

Hi @kevinlin18. Apologies for the lack of reply - I wasn't notified for some reason.

1.It would be possible to reduce the language list for the UI, but it's only English which is repeated in the drop down, so the impact would be pretty minimal. I'm happy to leave it.

2.I am happy that you are working on Chinese. This is a hard problem with the IME, but if it can be solved Optikey could help a lot of new people. In terms of the UI translation - that is great, but as you suggest it will not work until the IME problem can be solved. Once you have a way forward with that we can solve the crashing and integrate everything into a full Chinese solution.

Thanks for your work.

[kevinlin18]

Thanks, we can solve that after IME is done.

Please have a try with 'zh-tw' branch, I'll explain later where I need your help
https://github.com/kevinlin18/OptiKey/tree/zh-tw

https://youtu.be/Ax_l18Hz7QU

[JuliusSweetland]

@kevinlin18 Hi Kevin. I have checked out the branch and tested, looked over the commits, and watched the video. All very impressive! I'm really excited about the prospect of Optikey having support for Chinese, so thank you! I have a few questions at this point:

1.It looks like this new language is oriented around Taiwan. My (extremely basic!) understanding is that Taiwan speaks Mandarin and writes in "Traditional" Chinese. This lines up with what I see in the code; the Language enum entry is Taiwan specific and the dictionary commit is commented "Bopomofo" which I understand to be Mandarin specific.

a)Is the UI translation specific to Taiwan, or could it be considered and labelled as Chinese (Mandarin) in order to make it more inclusive of people from China, Taiwan, and Singapore? This would open up the path to adding Chinese (Cantonese) to cover Hong Kong. (please correct anything that is incorrect - I know very little about this stuff, but my motivation is for as many people as possible to be able to use Optikey).

b)Is the Keyboard specific to Taiwan, or could it be considered and labelled as Chinese (Traditional), which would give us some coverage of Hong Kong as well as Taiwan?

c)Am I on the right track to suggest that if we could support Mandarin and Cantonese UI languages, and Traditional and Simplified Chinese Keyboard languages (dictionary and word completion/prediction) then Optikey could support:

(from https://www.vengaglobal.com/blog/simplified-traditional-chinese-mandarin-cantonese/)

2.I know this is currently a work in progress, but how will the IME prediction work? When tested locally there was no prediction so I presume there's a 3rd party app/service I will need to run and configure Optikey to use?

3.I noticed the change to DictionaryEntry, with the addition of the 'EntryValue' - what is the extra string in the Chinese dictionary for? For example I see some entries like cj04|𪍺.

FYI in the dictionary I also see a lot of rows with no | (pipe separator). Are these working as the code looks to split on the pipe and then count the items on the row and only handle if it is 2 or 3? For example "憤而行兇" doesn't have a pipe so will split to 1 entry and then be ignored, I think?

[JuliusSweetland]

FYI I have closed the old Chinese localisation issue #515

[kevinlin18]

@JuliusSweetland Hi Julius.
Thanks for reply, you typed alot word🙄

For UI:

Area	S/T	Written	in menu
China	Simplified	Mandarin	Chinese Simplified, Chinese(China)
Taiwan	Traditional	Mandarin	Chinese Traditional, Chinese(Taiwan)
Hong Kong	Traditional	Mandarin	Chinese Traditional, Chinese(Hong Kong)

In my opinion, people in Hong Kong speak Cantonese and write in Mandarin with Trad. Chinese.
We could have only 2 UI: Chinese Traditional and Chinese Simplified.
If we want a UI for Hong Kong, we can copy from Chinese Traditional and edit some term.(The difference will like chips and french fries between UK and US English)
FYI: Firefox only have 2 options: Trad and Simp. Youtube have Chinese(Trad),Chinese(Hong Kong), Chinese(Simp), but it's hard for me to find the difference between Trad and Hong Kong.

For Keyboard:
We should consider more about IME like Pinyin, Bopomofo, Cangjie.
(I only list most popular and I familiar with)

• Pinyin (Phonetic-based) Mandarin
  Popular in China
  can typing Trad and Simp with different dic
• Bopomofo (aka Zhuyin) (Phonetic-based) Mandarin
  Popular in Taiwan
  can typing Trad and Simp with different dic, but it only used in Taiwan so we can skip Simp dic
• Cangjie (Shape-based)
  Popular in Hong Kong and Taiwan

I think the basic idea for these IME are same: Table lookup
the table will look like the dictionary current using by Optikey
(I'm not sure how other call it, let's temporary call it 'IME table')
for example: https://github.com/chewing/libchewing/blob/452f6221fbad90c0706a3963b17e226216e40dd7/data/phone.cin

As mentioned above we chould known the options in UI Language will be:

• Chinese Trad.
• Chinese Simp.

Options in Keyboard(IME) will be:

• Chinese Trad. Bopomofo
• Chinese Trad. Pinyin
• Chinese Simp. Bopomofo(Same as Trad but using different dic)
• Chinese Simp. Pinyin(Same...)
• Chinese Cangjie(as I known, IME table for Cangjie can have Trad and Simp in same time)

---

Now I can answer you questions:

1. Yes, there are for Taiwan and in Chinese Traditional since Bopomofo IME mainly used in Taiwan.

a) As above, the UI I translated can mark as Chinese Traditional.
for Chinese Simplified we still need help by people from mainland China, they can convert my file and edit some terms to they usually used, this can save their time.(This is important, please don't convert by machine 😁.)
FYI: I never seen the trem 'Mandarin' appear in any software lang options.

b) Keyboard layout specific to Bopomofo IME, Dictionary for Bopomofo IME in Trad Chinese.(The file 'ChineseTraditionalTaiwan.dic' should be separated, I'll explain below)

c) I think I have fully explained above😂.(There do have Contonese IME, if we can get the IME table we can add it into Optikey)

2. I'm not sure what you are asking for🤔.

• If you can't type in any single character, you might forgot to turn 'Suggestion method' from 'NGram' to 'Basic'
• After you typed few characters and suggestion stop showing up, that's cause by there is no white space between Chinese characters.
  I didn't use any 3rd party app. If these didn't answer your question, maybe a video can help

3. Those are related to 'IME table'
   The upper part of ChineseTraditionalTaiwan.dic are IME table this is for every single Chinese character, they looks like:

2k7|的
2u4|的
2u6|的

the code 2k7 can found in 'CharExtensions.cs', so they actually means

ㄉㄜ˙|的
ㄉㄧˋ|的
ㄉㄧˊ|的

These entries shows a Chinese char have 3 different phonetic

The lower part of ChineseTraditionalTaiwan.dic are for words, they works as Dictionary system as usual.
'ChineseTraditionalTaiwan.dic' should separate to 2 files, IME table is for Chinese Trad Bopomofo IME, 2nd part can be a shared dictionary between any Chinese Trad IME.
So there should be a new class to process IME table instead of modify DictionaryEntry

I didn't explain where I need your help, I'll explain ASAP😉

[JuliusSweetland]

@kevinlin18 Thank you for your reply - that's a really useful explanation of everything.

Ok, I understand how the dictionary works now - it IS an IME database. That's brilliant.

One thing we will want to do is default the prediction method to Basic when any Chinese language is selected as otherwise you get odd suggestions from NGram, e.g. "2k7" instead of "的"

One error I encountered was when trying to type "ㄉㄜ˙" to test your example:

I then tried "ㄉㄧˋ" and Optikey crashed. Are you aware of this?

No rush in terms of what help you need. I am here whenever you have time. I will be fairly busy tomorrow, but I will reply promptly and find some time in the coming week to do whatever work is needed.

[JuliusSweetland]

Also @kevinlin18 I am interested in your motivation for doing this work. If it's not too person would you email me directly julius@optikey.org ?

[kevinlin18]

Actually, I'm wondering if we can integration opensource third party IME engine, so we don't need to worry about the detail of how IME work.
But I'm not sure if they have enough English documents can let us known how it working. Maybe we can learn something from its front-end Weasel
RIME https://github.com/rime/librime
Weasel https://github.com/rime/weasel

One error I encountered was when trying to type "ㄉㄜ˙" to test your example:

I found it cause by the char "˙" \u02d9.
I didn't figure out what's going wrong. I notice the char didn't appear in blue textbox, but I can still copy it.

I then tried "ㄉㄧˋ" and Optikey crashed. Are you aware of this?

I can't reproduce this problem, maybe something goes wrong with char "ˋ" \u02cb ?

[JuliusSweetland]

@kevinlin18 A 3rd party IME could work well. I can lend some help, but I would need you to drive this forward as I don't understand how they work or how we could integrate.

[kevinlin18]

I built librime successfully and tested with rime_api_console.exe
But I don't know how to call the dll from C# application since it's a C++ dll, do you have any experience?

There is no API document, but 'tools/rime_api_console.cc' is an example shows how to call APIs.
I'll keep looking for any document can help

librime build instruction
https://github.com/rime/librime/blob/master/README-windows.md

[JuliusSweetland]

Hi @kevinlin18 Optikey already has a number of calls to C++ libraries, which you will find by looking for the PInvoke.cs files (e.g. https://github.com/OptiKey/OptiKey/blob/437435e89baa7b709a7cb5805fe0a66120a812e2/src/JuliusSweetland.OptiKey.Core/Native/PInvoke.cs). The way they work is to include the dll you want to call in the Optikey Core solution (as Content with copy always) and then create a PInvoke file that exposes the methods in that dll. From there you should just be able to call them from C# code.

Does that help?

[theoractice]

FYI: Tobii Dynavox partners with Tencent’s Sogou to empower people with disabilities in China

A Simplified Chinese keyboard example:
https://www.bilibili.com/video/BV1Qq4y1k7Xi