Backtranslating capitals AA's

[jrbowden]

In UEB, any word written fully capitalized is terminated by a non-alphabetic sign, see Rules of Unified English Braille 8.4.2.

Thus:
AA's should be written in UEB as ⠠⠠⠁⠁⠄⠎
AA'S should be wrriten in UEB as ⠠⠠⠁⠁⠄⠠⠎

Currently, backtranslating, Liblouis outputs:
(incorrect:) ⠠⠠⠁⠁⠄⠎ AA'S

Test:
- [AA's, ⠠⠠⠁⠁⠄⠎, xfail: backtranslation fails]

There are many examples in en-ueb-g2-dictionary_harness.yaml

How can you specify that after a non-alphabetic, the capital word is terminated?

[tibbsa]

Not a solution yet, but some more info on this.

This appears to happen specifically for 's endings, because of this line in ueb-en-charsdefs.uti:

endword 's 3-234

Note that the same thing happens with endings 'd, 'll, 'm, 'n, 're, s', 't, and 've, for the same reason.

But this isn't an engine problem per se, because but for those table rules, it does what you would expect. For other endings, albeit gibberish in the English context, it does properly recognize the end-of-caps, but it appears the endword rule is not being recognized as something that should terminate caps mode first, probably because the punctuation is a part of that rule itself.

Consider that DOG'd fails the translation test, but DOG'y works fine (there being no endword rule for 'y):

$ tools/lou_trace -b en-ueb-g1.ctb

,,dog'd
DOG'D

,,dog'y
DOG'y