Improve name-based search handling no-ascii characters

[bjendres]

The getquick-lookup we use handles it perfectly, but the suspicion is that the later similar_text function doesn't.

[bjendres]

@vangelisp will take a look

[bjendres]

As an alternative, we should also look at http://www.artfulsoftware.com/infotree/qrytip.php?id=552

[VangelisP]

Since most (if not all) the techniques used (similar_text/levenshtein/soundex/metaphone) are used with English text, one acceptable solution would be to convert all strings for comparison to English (normalize), by using the Transliterator class, as seen here:

• http://stackoverflow.com/questions/26185163/mysql-matching-latin-english-form-input-to-utf8-non-english-data#answer-26186456

Then after normalizing, use either the Levenshtein function through PHP (as described on my above link's top part) OR through a custom function in MySQL that can do that (as the link you've posted in your comment). Using the MySQL method for string comparison would decrease the load of PHP.

PS1. Levenshtein method needs some tweaking to get the results adapted to CiviBanking's needs PS2. This php class is only available with PHP 5.4 and afterwards

Let me know @systopia what you think on this approach.

[bjendres]

I think we should tackle this in 0.8 along with #140