The mapping in the ICUFoldingFilter has been pretty static, as far as I know, and I'm wondering whether there's any willingness to apply adjustments.
At the Cornell Library, we're finding ourselves having to make separate PatternReplace mapping rules because of specific characters that should be handled by the filter's punctuation folding being handled by the diacritic removal instead. The result is extremely troublesome searching behavior for Romanized Arabic text, at least, and probably creates issues with other languages. I have conferred with a college at the Columbia University Library, who has had to make similar adjustments to the mapping. I would ideally like to see these changes made at the source, which would simplify my configuration, and improve searching for everyone depending on this filter to take care of the messy international characters out of the box. Would such an update be likely to be accepted, or is the filter wed to its current configuration? Thanks, Frances Frances Webb Developer Cornell University Library
