capitalize_ascii don’t work for accentuated letters

[a2line]

The last modification from the 12 february 2018 in the master to be able to compile Gw with Ocaml > 4.05 changed the function capitalize to capitalize_ascii. Since that change, we have for eg. [*event/events]1 (that should write Évènements in french) left in lowercase and written évènements. What can be done to have these characters accentuated again?

[hgouraud]

It looks like proper handling of Unicode stuff requires the use of a separate library (Camomile, Batteries, Uchar, ...)
See https://discuss.ocaml.org/t/whats-the-function-of-uchar/939 and ocaml/ocaml#80
I have not yet figured out exactly what needs to be done, but Camomile seems to be the front runner!!
Possibly, we just have to replace our capitale_utf_8 function by its equivalent in Camomile.

[fablhx]

Yes this is indeed something that should be done, migrating to an external lib (thus creating a dependency). Long time ago, I was not sure which one should be used, but maybe camomile is the one to go with.

[fablhx]

And of course, we could get ride of this (horrible) hack in ged2gwd.

[hgouraud]

I tried this in util.ml (line 114):

replace:
let c1 = Char.uppercase_ascii (Char.chr (Char.code s.[1] + 0x40)) in
sprintf "%c%c%s" c (Char.chr (Char.code c1 - 0x40))
by
sprintf "%c%c%s" c (Char.chr (Char.code s.[1] - 0x20))

and it seems to work, at least for the é and â I tried!!
Switching to Camomile is a bit beyond my reach!!

[hgouraud]

By the way, what was happening before remains quite mystérious for me!!
If Char.uppercase was doing -20, then the whole thing was +40-20-40 = -20
If Char.uppercase_ascii is doing nothing for characters outside the ascii range, then we have +40-40=0
The reason why we were goint through this +40-40 step escapes me!

[hgouraud]

Independently of this fix, it appears that the capitalize (and possibly translate) function is broken for russian!
Two examples :
В браке с в 1995, Ailleurs, Aline Test
instead of
Браке с в 1995, Ailleurs, Aline Test

and

Свидетель:g:--я:a:--я (1985): похороны, Christian Vasseur 1925-1985
(french)
Témoin (1985) : inhumation, Christian Vasseur 1925-1985

[a2line]

This is fixed but dunno when!

[hgouraud]

Dunno if it is fully fixed, but the small change I made on Feb 26 does the job!