scim-tomoe does not recognize any characters I draw

[gundelgauk]

Hello,

I've used scim-anthy for some time and it works perfectly for inputting Japanese characters. However I wish to look up Kanji I don't know and have been trying to get scim-tomoe (handwriting recognition) to work for this purpose. I can launch the scim-tomoe GUI from the scim panel but no matter what I draw, it never offers anything in the text field below. It's as if it's unable to find the dictionary files. However if I go to the "search by reading" tab, I can search for characters with a specific number of strokes and it turns up results. Then, if I go back and try to draw one of them, nothing. No suggestions at all. Another strange thing is, I've seen a website that has screenshots of a dictionary-tab in the scim-tomoe GUI where one can add user dictionaries. However I do not have such a tab.

The packages tomoe-0.6.0, scim-tomoe-0.6.0 and libtomoe-gtk-0.6.0 are all installed and otherwise scim is working perfectly, as far as I can tell.

The file ~/.tomoe/dict.xml file contains the following:

Code: Select all

<?xml version ="1.0" encoding="UTF-8"?>
<!DOCTYPE dictionary SYSTEM "/usr/share/tomoe/dict.dtd">
<dictionary name="User dictionary">
</dictionary>

The file /usr/share/tomoe/dict.dtd does exist and contains the following:

Code: Select all

<!ELEMENT dictionary (character*)>
<!ATTLIST dictionary
          name CDATA #IMPLIED
 >
<!ELEMENT character (utf8, variant?, number-of-strokes?, strokes?,
                     readings?, radicals?, meta?)*>
<!ELEMENT utf8 (#PCDATA)>
<!ELEMENT variant (#PCDATA)>
<!ELEMENT number-of-strokes (#PCDATA)>
<!ELEMENT strokes (stroke*)>
<!ELEMENT stroke (point*)>
<!ELEMENT point EMPTY>
<!ATTLIST point
          x CDATA #REQUIRED
          y CDATA #REQUIRED
 >
<!ELEMENT readings (reading*)>
<!ELEMENT reading (#PCDATA)>
<!ATTLIST reading
          type CDATA #IMPLIED
 >
<!ELEMENT radicals (radical*)>
<!ELEMENT radical (#PCDATA)>
<!ELEMENT meta ANY>

However there is also a directory /usr/share/tomoe/dict which is empty. Is it supposed to be that way or is that what I'm missing? I have tried reinstalling the packages multiple times but this directory never contained any files.

Also, there is a directory /usr/share/tomoe/recognizer which contains the following files:

Code: Select all

-rw-r--r-- 1 root root 8919277 23. Okt 23:00 handwriting-ja.xml
-rw-r--r-- 1 root root 8751834 23. Okt 23:00 handwriting-zh_CN.xml

So the character dictionary file seems to be there, a wild guess of mine is that scim-tomoe somehow fails to find or read it. Does anyone have an idea what could be wrong or suggestions what I might try?

[codergeek42]

gundelgauk, I recently acquired a Wacom tablet and was also trying out Tomoe.

For me, however, it only works if I set the language to Japanese before running it (e.g., "LANG=ja scim-tomoe"). Maybe this will correct the issue for you, as well?

Hope that helps.

[gundelgauk]

Thank you for your response, I just tried running it that way but unfortunately it doesn't seem to help.

However I did notice an error message that I didn't notice before, because I never ran it from a shell so far:

Code: Select all

(tomoe:17629): Pango-CRITICAL **: pango_font_description_from_string: assertion `str != NULL' failed

I don't know if that's what's causing my problem, but there seems to be some kind of problem with fonts. I will try to do more research about that.

[thesun]

I'm not a Gentoo user, but I had exactly the same problem in Fedora Core 8 and found a post that was very helpful. Unfortunately, I can't find the post, but the gist was that by editing the "/usr/bin/scim-tomoe" file and adding 2 language-related export lines (not 1 as mentioned above), Tomoe will work by running "scim-tomoe" from the command line. For some reason, I still can't get any love if I use it from within the SCIM toolbar, probably because the SCIM toolbar uses a different default language environment and thus borks the dictionary loading.

But it's relatively easy to just type "scim-tomoe" at the command line and have it run independently from the toolbar. If that sounds useful, add:

export LANG=ja
export LC_MESSAGES=ja

to your /usr/bin/scim-tomoe file, just before the final line. Like so:

###more above this line
sysconfdir=/etc

###new lines start
export LANG=ja
export LC_MESSAGES=ja
###new lines end
/usr/lib/scim-1.0/scim-helper-launcher tomoe b1bfe2b4-6930-41b0-8c07-d05bce8c92e2 $*

If you happen to figure out how to get the toolbar method working, I'd love to know. Apologies for not being a Gentoo user (yet?) but this might be a usable workaround.

[DarkAn9el]

Same problem here... tried

Code: Select all

export LANG=ja
export LC_MESSAGES=ja

and

Code: Select all

export LANG=ja_JP.UTF-8
export LC_MESSAGES=ja_JP.UTF-8

The search with reading function works, but the handwritting recognition not...

What can we do?