Turkish Language

[anysoftkeyboard]

Turkish Language

[menny]

@Sixthtry you can use ask:shiftCodes to specify special uppercase codes, in case the automatic guessing fails

[menny]

Done: https://github.com/AnySoftKeyboard/LanguagePack/tree/Turkish
Please make sure your branch is rebased on top latest https://github.com/AnySoftKeyboard/LanguagePack/tree/Turkish, and that you point your PR to my Turkish branch.

Also, how are you creating the dictionary?

[menny]

I see. This is how to get the real gz file. First, clone the repo to your local machine:
git clone https://android.googlesource.com/platform/packages/inputmethods/LatinIME
then cd to the dictionaries folder:
cd LatinIME/dictionaries
find the Turkish dictionary tr_wordlist.combined.gz (it's 926338 bytes), and unzip it. Use the uncompressed file as the input for parseAospForEnglishDictionary

Hello @Sixthtry,
I think that you have to focus on the Unicode first :
U+0130 = LATIN CAPITAL LETTER I WITH DOT ABOVE = İ
Wich has as lower case U+0069 = LATIN SMALL LETTER I = i
You have now to convert both to the right decimal on the keyboard starting with the small letter i.
Wainting for an answer.
Regards,

[lem0n4de]

Hi @BoFFire,
In the layout, I have both Latin Small Letter I, which is [ i ] and unicode "105", and Latin Small Letter Dotless I, which is [ ı ] and unicode 305. Both works well when writing in lowercase.

The problem is, when I turn on the uppercase writing, dotted little i with unicode 105 should turn to "LATIN CAPITAL LETTER I WITH DOT", which is [ İ ] and unicode 304 but instead it becomes "LATIN CAPITAL LETTER I", which is [ I ] and unicode 73.

Edit: Appearently, I misexplained here. Corrected the letters.

(Sorry for being late. I wrote an answer on phone but git client got crashed.)

[lem0n4de]

That was just what I needed for. Thank you very much sir. And may I ask you to open up a Turkish branch? I am trying to build a nice dictionary and then I will try to upload files here somehow in the next days(I am new to the Git).

[lem0n4de]

As I saw from a thread in here, "New Sicilian language pack" thread to be exact, I am collecting wikipedia pages. I also did WordsfromAosp but this is all I know about dictionaries.

[menny]

These are both very good sources. I mostly wanted to make sure that you are aware of them, and that the sources you are using are not copyrighted in any restricted way.

…

On Sat, Dec 16, 2017, 12:00 AM Sixthtry ***@***.***> wrote: As I saw from a thread in here, "New Sicilian language pack" thread to be exact, I am collecting wikipedia pages. I also did WordsfromAosp but this is all I know about dictionaries. — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#112 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AAdX-OKnOR4MIiKYpuBp1K2g_DoPBw4Fks5tA05cgaJpZM4RDA3B> .

[lem0n4de]

Hey, umm,
I was trying to get as many pages as I can in Turkish manually yesterday. Today I realized that there is dump of wikipedia as said in build.gradle comments https://dumps.wikimedia.org/other/static_html_dumps/current downloaded the one for Turkish, it was around 300 MB but it extracted as 6.8 GB .tar file. Now there is .html's inside this but they are in files and I don't know how can I put them in build.gradle.

Also I found another file in wikimedia dumps. It is in https://dumps.wikimedia.org/trwikisource/20171201/ and there is an .xml (trwikisource-20171201-pages-articles.xml.bz2) file which contains "Articles, templates, media/file descriptions, and primary meta-pages" as title says in a file which is 70 MB after extract.

I am wondering, what I will get from these two files will be the same?

[I am sorry that I am asking a lot of questions, but I am doing this for the first time and I want it to be usable enough.]

[menny]

Wiki dumps are good, but there are two general problems with them:

1. They include words in other languages, which sometime get added to the generated dictionary
2. They are not "dialog“ source, so the word frequency might seem strange

[lem0n4de]

I gave up on aosp word list as it just does not generate any wordlist. I can see that it is a long file in editor but I can not read of course but the code can't either. The parseAospForEnglishDictionary task generates a 2 line file from that which contains only; </wordlist>

[menny]

@Sixthtry can you send me a link to that aosp file?

[menny]

Also, you'll need to fix/complete the issues reported here: https://271-4270172-gh.circle-artifacts.com/0/lint_reports/app/checkstyle/checkstyle.html

[lem0n4de]

https://android.googlesource.com/platform/packages/inputmethods/LatinIME/+/master/dictionaries/tr_wordlist.combined.gz Here is the aosp wordlist link. I downloaded it from the [txt] link in the right corner

[lem0n4de]

Thank you for being so helpful, that solved the problem. I want to ask just one last question; Is there a document where I can find all ask: and android: codes?

[menny]

Glad it helped! Can't wait to see the new language pack 👍 Regarding attributes, you can check the source code of AnysoftKeyboard API, and specifically, the styleables: https://github.com/AnySoftKeyboard/AnySoftKeyboard-API/blob/master/api/src/main/res/values/keyboard_layout_api.xml

…

On Wed, Dec 20, 2017, 3:05 PM mkdor ***@***.***> wrote: Thank you for being so helpful, that solved the problem. I want to ask just one last question; Is there a document where I can find all ask: and android:codes? — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#112 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AAdX-AyXvyL3RazLHqpbVZGcBao-iSzdks5tCWdTgaJpZM4RDA3B> .

[MK8T]

I dont see the ask:shiftCodes to specify special uppercase codes
in https://github.com/AnySoftKeyboard/LanguagePack/blob/Turkish/src/main/res/xml/qwerty.xml ... perhaps dont read the good file ?

[lem0n4de]

@MK8T
Okay, sorry for that but there seems to be a problem with file. I will reupload it to github manually then, it works in my keyboard.

[lem0n4de]

@MK8T
Oh, I understand now. You are looking at the 'Turkish' branch of original AnySoftKeyboard repo. My repo did not merged yet with it. You should look at here = https://github.com/Sixthtry/LanguagePack/blob/master/src/main/res/xml/qwerty.xml

[friesenkiwi]

Closing this in favor of #116