415254 - Spellchecker doesn't handle well word preceded by an apostrophe (french)

Status: RESOLVED FIXED

(Core :: Spelling checker, defect)

Description

[jlm]

User-Agent:       Mozilla/5.0 (Windows; U; Windows NT 5.1; fr; rv:1.8.1.11) Gecko/20071127 Firefox/2.0.0.11
Build Identifier: Mozilla/5.0 (Windows; U; Windows NT 5.1; fr; rv:1.8.1.11) Gecko/20071127 Firefox/2.0.0.11

le correcteur orthographique ne semble pas connaitre des mots commençant par l'

Reproducible: Always

Steps to Reproduce:
1.
2.
3.

Comment 1

[Robert Strong (they/them - no direct email)]

The Extension Manager is not responsible for spellchecking or the words in any of the dictionaries. Over to General

Comment 2

[Régis Caspar]

WFM on Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.9b3pre) Gecko/2008020104 Minefield/3.0b3pre ID:2008020104

(with My Spell french dictionary 1.01, test: "l'allemand")

Comment 3

[jlm]

Dans wikipedia.fr, les mots précédés d'une apostrophe et qui commencent par une majuscule sont considérés comme erronés exemple : article Amazone de Tucuman  (cliquer sur modifier pour activer le correcteur)
jl maréchal

Comment 4

[Régis Caspar]

(In reply to comment #3)
> Dans wikipedia.fr, les mots précédés d'une apostrophe et qui commencent par
> une majuscule sont considérés comme erronés exemple : article Amazone de
> Tucuman  (cliquer sur modifier pour activer le correcteur)
Simplement parce que les noms propres ne sont pas dans le dictionnaire précédés d'un "l'". Note : si vous désirez que ce bug soit pris en compte il vaudrait mieux commenter en anglais.

Proper nouns starting with a vowel are not in the french ditionnary so "l'amazone" is marked as correct and "l'Amazone" is marked as not. 

Comment 5

[jlm]

Ok pour les noms propres, mais que pensez-vous de : en tant qu'instance supérieure
jl maréchal

Comment 6

[Brett Wilson]

This sounds more like grammar checker to me, which we do not pretend to have. But I can not read all the comments to understand what is desired. If true, this is WONTFIX or a dupe of a grammar checking bug which we may have.

Comment 7

[Régis Caspar]

Brett, it's not really grammar, in french when a noun begins with a vowel we remove the ending vowel of the preceding article:
for example "l'arbre" is decomposed in "l'" and the noun "arbre" and the problem reported is that the spellchecker read "l'arbre" as one word and spot it as misspelled if "l'arbre" is not in the dictionary.
Perhaps the spellchecker should detect "'" like space (but AFAIK, this is not the current idea of the dictionary, i.e. "l'arbre" is in the french dictionnary as a word)

Comment 8

[jlm]

comparez : l'instance (pas de problème) et : en tant qu'instance (problème)

           l'esprit   (pas de problème) et : il n'est qu'esprit  (problème)

jl m

Comment 9

[Brett Wilson]

I read this wrong. I though it was talking about the literal << >> characters for quotes, when it's about the l' part.

Is this bug with Firefox 2 or 3? We changed spellchecking libraries.

If it's in 3, I assume Hunspell doesn't support this? I suspect we're sending them the right word in this case and aren't breaking on the apostrophe (but I could be wrong on that). If I'm correct, the right approach would be to file a bug on them, since we have not modified the library.

Comment 10

[Régis Caspar]

Reporter see the problem with FF 2.0 and I reproduced it with "l'Amazone" and FF3.0. I think that FF doesn't break on the apostrophe and should not according to what is marked as correct and incorrect to the spellchecker.

So perhaps the reporter should fill a bug on hunspell/myspell (if FF 2.0 was using myspell)   

=> Updating summary

jlm> Je pense que nous avons bien compris le problème mais comme le dit Brett, ce problème se situe peut-être du côté des développeurs des librairies de corrections (ou des auteurs de dctionnaire) et non du côté des développeurs Firefox [meaning: I think we understood the problem and as Brett said, perhaps the problem should be reported to the spelling libraries developer or dictionary author and not here]

Comment 11

[jlm]

oui je suis assez d'accord. merci pour le temps passé
jlm

Comment 12

[Olivier R.]

Apostrophes are not considered as a word separator by Hunspell, so words with elisions have to be considered as a whole.

During one year, I have recreated the French dictionary from scratch. Now there is new affixation rules which handle elisions properly.
Words with an uppercase preceded with elisions are now recognized (unless there are verbs).

The result of this work is here:
http://dicollecte.free.fr/download.php?prj=fr

Benoit Leseul told me he had integrated this dictionary in the last nightly build. I was looking for the issue which is supposed to report the integration of the new dictionary when I found this one.

Comment 13

[Matt Caywood]

If this new dictionary is currently in the fr-FR nightly builds, then this bug is FIXED.

Although, this work could be improved once tokenization is improved (bug 355178).