On 12/16/2011 1:08 PM, Roedy Green wrote:
> Let's say I had a million records each with a text field. I wanted to
> find all records that contained a given substring. Are there fast
> algorithms to do that or do you have to scan the whole thing linearly?
>
> I imagine you could build an index of words. Then maybe you could
> figure out which words contain the string, and narrow your search to
> records with those words.
>
> Similarly, I was wondering how you might build an index to all the
> files on a computer so that you could find one just by giving a wild
> card or partial filename.

Use a database that support full text search.

Arne