You'd better avoid the use of Preview.app for searching rather rare kanji characters or copying and pasting text containing them to other applications. I just noticed that Preview (5.0.3 [504.1], running in OS X 10.6.6) substitutes U+798F and U+76CA for U+FA1B and U+FA17. I mean, U+FA1B and U+FA17 copied in Preview will be pasted as U+798F and U+76CA in NWP, TextEdit, etc. And you have to enter U+798F or U+76CA in Preview's search field to find U+FA1B or U+FA17. Presumably other characters belonging to CJK Compatibility Ideographs are affected by the same bug.

Admittedly U+FA1B and U+FA17 are rather rare character shapes at least in  modern Japanese texts. So it would be a nice *addition* if Preview enabled you to find U+FA1B or U+FA17 by searching more common U+798F or U+76CA *too*. But searching for U+FA1B or U+FA17 should find U+FA1B or U+FA17 for it is these characters that you *see* in Preview.

Also, U+FA1B, U+FA17, and other Kanji forms in the same range should be copied as U+FA1B, U+FA17, etc., without substitution. Although they are rather rare Kanji forms, if they are used in a document, the author has good reasons to do so. This is a serious problem, especially when you use text copied and pasted from PDF to proof read an article or a book.

A longer version of what I wrote above has been submitted to Apple's bug reporter (ID 8942869).

Adobe Reader 9 is not affected by such an oddity.


Kino