Apache OpenOffice (AOO) Bugzilla – Full Text Issue Listing
|Summary:||SpellCheck problem with Chinese|
|Component:||editing||Assignee:||AOO issues mailing list <issues>|
|Status:||CONFIRMED ---||QA Contact:|
|Issue Type:||DEFECT||Latest Confirmation in:||---|
Description jlcheng 2006-09-06 02:54:13 UTC
In SRC680_ROOT/i18npool/source/breakiterator/data/zh.dic(line 6,7), there is "ä¸€ä¸‡" and "ä¸€ä¸‹" (I think these are words, and OK?). Open bugdoc, press key "F7" and spellcheck the doc, press "Ignore Once" button, 1st word "ä¸€ä¸‡" is marked as a word. But next "ä¸€ä¸‹", only 2nd CJK char "ä¸‹" is marked. If you press "Ignore Once" button, the whole string become to "ä¸€ä¸‡ä¸€ä¸€ ä¸‹". Thus the "ä¸‹" is marked and changed to "ä¸€ä¸‹". I think this is problem and some English string has similar problem.
Comment 2 jlcheng 2006-09-06 03:09:47 UTC
Created attachment 38986 [details] Add: Spell Check option "Check in all languages" is checked
Comment 3 christian.guenther 2006-09-06 14:50:38 UTC
One for me.
Comment 4 jlcheng 2006-09-07 03:32:31 UTC
Created attachment 39010 [details] another problem: bugdoc "Ignore Once" always until auto close dialog of SpellCheck, then close simpress, OOo crash
Comment 5 christian.guenther 2006-09-08 12:08:22 UTC
Set to new and change the target.
Comment 6 christian.guenther 2006-09-08 12:09:59 UTC
I can reproduce the bug. With each ignore once the 'line' between the both words become longer. Please have a look.
Comment 7 thomas.lange 2006-09-11 11:29:12 UTC
I wrote issue 69451 because of the different behavior when doing the same in thing in the Writer application. Thus this issue now is only about the wrong slection of the next word to be checked (which should have been the third and fourth character) and the insertion of the third character if one keeps pressing the "ignore once" button.
Comment 8 thomas.lange 2006-09-11 11:47:33 UTC
TL->OS: The very first problem found is in SpellDialog.cxx in function SentenceEditWindow_Impl::MarkNextError(). There the line TextPaM aCursor(0, m_nErrorEnd ? m_nErrorEnd + 1 : 0); tries to set up a cursor to find the next error but since it is Chinese there is usually no space between the words an thus m_nErrorEnd + 1 will already start one character to far. A following problem will be that subsequent one error range will immediately follow another one and thus the current algorithm can not distinguish them properly anymore. Please take over. Thanks!
Comment 9 zhongqiyao 2007-07-24 07:49:06 UTC
In "Writer", red wavy lines sometimes appear under Chinese text. Is this a problem with the spellchecker thinking that Chinese text is misspelled English? Thanks. Qiyao
Comment 10 Mathias_Bauer 2008-01-11 12:54:41 UTC
Comment 11 Oliver Specht 2008-07-07 08:28:38 UTC
Retarget to 3.1
Comment 12 Oliver Specht 2008-12-10 14:07:51 UTC
Check in all languages has been removed in cws tl46 for OOo 3.0.1 Retarget to 3.x, Prio changed to P2 (no crash)