Best way to remove characters?

[Kudose]

Hello all.

Basically, I am wondering what the best way to remove all non-alpha characters from a string without comprimising the content (whole words) is.

Preferably without having to use a hundred str_replace's after a single stripslashes and strip_tags.

I am building a meta keyword generator if your wondering why.

I would just build a kick arse regexp...but my book is still in the mail. :bemused:

Any suggestions would be appreciated.

[Drakla]

What's your source, is it a html page or something? First of all you could leave just the word characters with a preg replace '/\W/' with a space, then explode everything into an array on space... Then get all unique values in the array. You could also have a list of 'noise' words to dump - there are loads of premade ones out there if you google for it, dump all them then see what you're left with.