I've been searching everywhere in order to find a way to transliterate a search term before initiating the search in a different language - greek in particular. So I would expect a term written in greeklish (greek in english chracters) to be searched as:

  • the search term 'as is'
  • or search term transliterated in greek characters

I was thinking about taking the search term and use str_replace to replace characters, but there might be different variations of the same character. Any ideas on how to tackle this?


If you have PHP's Intl extension installed, you can use the wonderful Transliterator class.

I get some pretty good results, like this:

$t = Transliterator::create('Latin-Greek', Transliterator::FORWARD );

var_dump( [
    'alfa' => $t->transliterate('alfa'),
    'alpha' => $t->transliterate('alpha'),
] );

Both these greeklish forms give me "άλφα". I don't know how it knows where to put the accent, but it's pretty impressive.

Once you have the greek script, the remaining nuances should be taken care of by the database collation. "α" will match "ά" and "σ" will even match "ς" (This works on my system running MySQL with collation utf8mb4_unicode_520_ci)

If you don't have this extension, or the underlying ICU library isn't available, then you'd be looking at expanding the characters yourself, but transliteration is hard so I don't recommend it.

  • 1
    After playing with this code some more, I think your biggest problem is going to be spelling!
    – Tim
    Feb 27 '19 at 13:23
  • Hello and thanks for your comment. I have never heard of this class, nor I am so advanced in php, but this seems fairly straightforward. I'll give it a go tomorrow. I was too busy with other parts of the project that I left this part for later. I can only assume that if this works then I have to make a custom query that uses relation => 'OR'. Anyway I'll consult back! Feb 27 '19 at 20:40
  • Hello, using my local server I managed to install the library and tried the transliterator, however, when quering for alfa I get ἀλφα which is not the same as άλφα (different kind of accent). Will check the library documentation in case there's a solution to this. Feb 28 '19 at 9:33
  • Probably baked into the ICU library, but shouldn't matter as DB collation should handle it as already mentioned. I'd test that rather than chase library issues
    – Tim
    Feb 28 '19 at 12:47
  • Collation works for me. Blog post with title "άλφα" is found with queries for "αλφα" (no accent) and "ἀλφα" (wrong accent). I think you can ignore this.
    – Tim
    Feb 28 '19 at 12:54

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.