[FR] Add support for the 'ar-cf' template

[BoboTiG]

• Wiktionary page: https://fr.wiktionary.org/wiki/Mourad
• Model link: https://fr.wiktionary.org/wiki/Mod%C3%A8le:ar-cf

[lasconic]

this is a hard one... We need to replicate https://fr.wiktionary.org/wiki/Module:arabe#L-474 arabe.appliquer like we did for arabiser.

[BoboTiG]

Yes. We can just postpone its handling until more words will be using it.

[BoboTiG]

Another one: https://fr.wiktionary.org/wiki/alm%C3%A9e

[lasconic]

ar-terme: https://fr.wiktionary.org/wiki/qasida

[BoboTiG]

ar-terme: https://fr.wiktionary.org/wiki/Muad-Dib

[BoboTiG]

Here are all words using that template:

 !! Missing 'ar-cf' template support for word 'cafir'
 !! Missing 'ar-cf' template support for word 'Coran'
 !! Missing 'ar-cf' template support for word 'casbah'
 !! Missing 'ar-cf' template support for word 'azulejo'
 !! Missing 'ar-cf' template support for word 'Ali'
 !! Missing 'ar-cf' template support for word 'Omar'
 !! Missing 'ar-cf' template support for word 'assorath'
 !! Missing 'ar-cf' template support for word 'alcade'
 !! Missing 'ar-cf' template support for word 'Imran'
 !! Missing 'ar-cf' template support for word 'Zahra'
 !! Missing 'ar-cf' template support for word 'adhan'
 !! Missing 'ar-cf' template support for word 'Ryad'
 !! Missing 'ar-cf' template support for word 'Mourad'
 !! Missing 'ar-cf' template support for word 'cabas'
 !! Missing 'ar-cf' template support for word 'Majlis'
 !! Missing 'ar-cf' template support for word 'djihad'
 !! Missing 'ar-cf' template support for word 'nadir'
 !! Missing 'ar-cf' template support for word 'génie'
 !! Missing 'ar-cf' template support for word 'ihram'
 !! Missing 'ar-cf' template support for word 'sura'
 !! Missing 'ar-cf' template support for word 'schème'

[BoboTiG]

So, I converted appliquer() to Python. And finally, I just saw we also need to fetch https://fr.wiktionary.org/wiki/Cat%C3%A9gorie:Mod%C3%A8les_de_racine_en_arabe_du_Wiktionnaire.

Work is still ongoing ⏲️

[BoboTiG]

#1414 needs way more work than I expected.

As stated in the documentation:

Le modèle :
 1. transcrit le paramètre en arabe et fait un lien sur la page correspondante ;
 2. donne la translittération, avec lien sur la convention de translittération ;
 3. donne la définition courte du terme (modèle de schème passé en paramètre du modèle de racine).

For now, the PR only tackles the last point (3).

For (1) and (2), we also need to support https://fr.wiktionary.org/wiki/Mod%C3%A8le:ar-abiser I guess? It seems we cannot rely on the current arabiser() fonction: results are not good (maybe it is my implementation that is incorrect too, though).

It will be for later, I'll pause the work for a moment. @lasconic if you want to move on, you can use the PR as a team effort. In any cases, I'll find some time to make progress the next week, hopefully.

[lasconic]

ar-ab https://fr.wiktionary.org/wiki/Loubna

[lasconic]

I took a deep look into the issue and #1414

1/ Lua a string are 1-indexed ... while python script are 0-indexed.
2/ The script is not enough, we need all the senses for the couple racine/scheme and probably the "forme".

I have 3 working correctly now. It's not given by appliquer.
Appliquer is supposed to give 2 and a call to arabiser with the result of appliquer should give 1..

Quite a bit of work and testing. I already spent some time on it. I will look further this week.

[lasconic]

With this PR, all the words from #643 (comment) are OK with check-word except for "schème" which uses other non supported templates (ar-racine/nom and ar-sch)

[lasconic]

The two big learning points about porting lua to python

• 1-indexed vs 0-indexed strings
• End of slices : sub(scheme, 1, position -1) --> scheme[:position]

[lasconic]

With this PR, all the words from #643 (comment) are OK with check-word except for "schème" which uses other non supported templates (ar-racine/nom and ar-sch)

schème is now OK with #1742 and the support of ar-racine/nom and ar-sch