Japanese Index Dropping Entries

[patrickq]

I'm building a bookmap that contains Japanese text. The PDF opens fine, and all looks well, but the output log has about 60 occurrences of the following error (with slight variations in the number of ?s):

[index-preprocess] [ERROR] Index entry '???????' is dropped, because corresponding group is not found

From what I've been able to dig up, this may be due to a character(s) that the index processing does not know how to sort. Does anyone know how to fix this issue?

Thanks,
patrickq

[Derek Read]

We are suspecting one of the following two issues might be part of the issue:

1. The index is not configured for the ja_JP locale.
2. The locale isn't set on the bookmap correctly. Though the RFC4646 spec says it should be "ja-JP" (note the hyphen), the DITA OT needs it to be ja_JP (an underscore, not a hyphen).

[patrickq]

Thanks Derek. Can you expand on option 1. ? The xml:lang is set correctly on the bookmap.

[Derek Read]

xml:lang should (ideally as far as I know) be set for the root element of the map file, that is assuming the entire document should be treated as Japanese (I am not sure how well the DITA OT deals with xml:lang otherwise in all cases where you might mark different portions of a document using different languages).

[Derek Read]

You may wish to broaden your audience to general DITA OT users as well in this case as they may have some advice. The best location for that would likely be here:

http://tech.groups.yahoo.com/group/dita-users/

Probably best to minimize mention of XMetaL on that discussion board if possible as many people there are pure DITA OT users (or users of other tools). In this case, if you can reproduce your issue with PDF2 (in our UI this is Book via RenderX) that would be best. PDF3 (XMetaL Enhanced PDF via RenderX XEP) is based on PDF2. We have not modified any files there, just added a few things to it.