Apache OpenOffice (AOO) Bugzilla – Issue 77846
Japanese input not displayed correctly
Last modified: 2007-07-05 08:41:51 UTC
Mac allows the user to select an input type such as japanese Hiragana through the System Preferences - International - Input Menu. Then the input method can be changed from the icon in the top right of the main menu bar (next to the sound icon on my menu bar). Now the text input will be interpreted as "romaji" which is romanised japanese phonetics. For example "nihongo" represents three japanese kanji characters corresponding to the sounds "ni hon go" as in the attached screenshot. As each romaji sound is entered, the corresponding hiragana character is displayed. So "ni" will cause a character to display, then another etc. These characters are underlined. Press the space bar toggles to a katakana character, press space bar again toggles to a pick list of various kanji. Press return and the chosen characters are now entered in the document. In OOo aqua this works fine except that nothing is displayed until the picklist is activated by pressing spacebar two times. Please see attached screenshot. Notice the OOo text input is absent above the picklist. The cursor did not move. In both cases the text imput from the keyboard is "nihongo"
Created attachment 45424 [details] input not displayed
Here is log extract - result of simply typing two characters "ni". This shows that OOo has reveived the correct character (from the TSM i guess) but it is not displayed inline. >*>_> SetItemImage >*>_> SetItemImage >*>_> SetItemImage >>>> HandleTSMEvent >>>> HandleOffsetToPos <<<< HandleTSMEvent -->RefreshRect refresh 148 - 212 - 2 - 23 >>>> HandleTSMEvent >>>> HandleOffsetToPos <<<< HandleTSMEvent >>>> HandleTSMEvent >>>> HandleUpdateActiveInputArea GetInputText len: 2 (1) text: に fixLen: 0 (0) <<<< HandleTSMEvent -->RefreshRect refresh 148 - 212 - 2 - 23 -->RefreshRect refresh 148 - 212 - 2 - 23 -->RefreshRect refresh 148 - 212 - 2 - 23 -->RefreshRect refresh 148 - 212 - 2 - 23 >*>_> SetPointer
btw, in case your browser displays the IZ text in the log extract above like "text: ?" - in the actual console log output and in browser on my Mac the question mark "?" is not seen, instead it displays as the correct hiragana character for "ni".
OK. Now preliminary implementation to draw preedit text is committed in aquavcl01 (1.46.112.93 of salframe.cxx). So it should display with current code. However, some work is still needed to draw proper text attributes of text segments.
This is fixed in vcl/aqua/source/window/salframe.cxx 1.46.112.95 thanks to ekato ;) see next screenshot. There is a different problem, that was already existing before this fix but it was not so easy to confirm until this fix is done. If I select a japanase font on a new document, when i enter some text the font reverts to tahoma. See second screenshot. This might be my bad usage, so I need to check if it is a real issue?
Created attachment 45460 [details] fixed
Created attachment 45461 [details] font reverts to tahoma
A workaround for the second font problem is: enter some text, select a range, apply a japanese font, delete, then enter again and it works correctly.
I meant the version tested was 1.46.112.94 (not 95)
Now text attributes is set according to the information from input method in salframe.cxx 1.46.112.106. @sparcmoz: I think "the font reverts to tahoma" problem is an another issue. Please fill new one.
@ekato: I created issue 78818 for font reverts to tahoma
sparcmoz: can you thus mark this issue as verified?
Set target to 2.3.
------- Additional comments from ekato Mon May 28 12:10:32 +0000 2007 ------- verified--> OK. Now preliminary implementation to draw preedit text is committed in aquavcl01 (1.46.112.93 of salframe.cxx). So it should display with current code. I don't know what this means :( --> However, some work is still needed to draw proper text attributes of text segments.
@spacmoz: > I don't know what this means :( --> >> However, some work is still needed to draw proper text attributes of text >> segments. It means underline types changes with the state of Japanese input. For example, in the state of inputting Hiragana, it shows dotted underline. And then in the state of converting the Hiragana to Kanji, it shows bold underline for a selected segment and normal underline for unselected segments, and so on. I've implemented this in salframe.cxx 1.46.112.106.
verified in m217 - ja and en-US builds: the underline behaves as described in the previous comment.
closed