280512 - speller accepts words made of capital letter followed by Hebrew characters

Status: NEW

(Core :: Spelling checker, defect, P3)

Description

[Dan Kenigsberg]

User-Agent:       Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.6) Gecko/20040124
Build Identifier: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.6) Gecko/20040124

In Thunderbird 1.0, with the English (US) spell checker, the spell checker
accepts  words begining with a capital English letter, followed by foreign
charecters.
For example, the string E人 (capital E followed by the Kanji for "person") is
accepted. And worse - it is even suggested as a correction if one type just 人.

It was first reported in http://ivrix.org.il/bugzilla/show_bug.cgi?id=76 but has
nothing Hebrew- or Windows- specific.

Reproducible: Always

Steps to Reproduce:
1. Type a capital English letter followed by Hebrew or Japanese charachters
2. Spell-check using the English dictionary


Actual Results:  
silly string is accepted.

Expected Results:  
Reject the string since it is non English, as Mozilla does.
It would be even better if we could spell-check multiligual texts, but that is
irrelevant to this bug.

Comment 1

[Prognathous]

(In reply to Dan Kenigsberg, comment #0)

> For example, the string E人 (capital E followed by the Kanji for "person") is
> accepted. 

This may be a dupe of other bugs.

> And worse - it is even suggested as a correction if one type just 人.

To the best of my b.m.o searching abilities, this behavior is a new bug. I'm
confirming this part of the bug, though the summary may need to change.
Screenshot of the same problem using Japanese characters is coming.

Prog.

Comment 2

[Prognathous]

Attachment: Same problem, Japanese instead of Hebrew

Comment 3

[Dan Kenigsberg]

Prog, would you please explain why do you consider my report as made of two
different issues?
According to my understanding, E+japanese is a suggested correction exactly
because it is accepted as valid string by the speller. In my opinion, the two
parts are a cause and an effect.

Comment 4

[Prognathous]

Generally speaking, words that are considered as *possibly* having valid
spelling (e.g. E+misspelling), should merely not be tagged as misspelt, they
should not be actively suggested as possible corrections. I know for sure that
E+misspelling can be valid, I don't know enough to rule out the possibility of
E+foreign_language being valid as well. That's why these two issues may be a
case of cause and effect, but not necessarily.

Prog.

Comment 5

[Mike Cowperthwaite]

Is this the same as bug 257073?

Comment 6

[Dan Kenigsberg]

(In reply to comment #5)
Would you please explain why bug 257073 has any relevance here? I fail to find a
connection.

Comment 7

[Matt Caywood]

'e' is accepted by the spell checker as a valid word, but characters outside the language are not checked so e人人人人 is accepted. ee人人 is not.

E人 will never be suggested by the spell checker.

This is probably related to the way the spell checker breaks words, which is why bug 257073 is relevant.

Comment 8

[Jens Stutte [:jstutte]]

So this seems to be still a thing, but I believe that bug 1402822 could help here to some extent?

Comment 9

[Dan Minor [:dminor]]

I did some quick testing, and E人 passes with the en-US spellchecker, but not with en-CA or es-ES. But unfortunately bug 1402822 won't help here since it checks if a word is valid in any installed dictionary. So as long as E人 passes en-US, it won't matter what other installed dictionaries are present.