Re: String searching peformance

On Mon, 24 Feb 2003 22:35:35 +0100, Ernest E Vogelsinger wrote: >At 21:22 24.02.2003, {R}ichard Ashton spoke out and said: >--------------------[snip]-------------------- >>while ( $flag == true ) >>if (strpos($body, $word[]) > 0) {$flag=false} >> >>What I really need to know is which is the fastest loop? >>Which is the fastest match, strpos? >>Which is the fastest comparison? >>What other optimisations are possible to maximise search speed? >> >>I would expect to keep a frequency hit count and sort the $words[] so >>that the most frequent hits are found first, remembering that only >>$body with NO $words in then are automatically posted, so every word >>must be tested. >> >>I don't know sufficient internals to pick the fastest method. I will be >>running with 4.3.1 on FreeBSD 4.6 >--------------------[snip]-------------------- > >I'd suggest something like this: > >$buzzwords = array('idiot', 'fool', 'shit', 'FOAD'); > >$re = '/(' . implode('|',$buzzwords).')/is'; >if (preg_match($re, $posting)) > // bad word found >else > // cleared Thank you very much that is brilliant, I would never have thought of that. Do you think that: if (preg_match($re, $posting, $hits)) would slow it down at all. The $buzzwords will be kept in a file to be loaded before each run, every 5 minutes. I could therefore keep a count of which words hit most frequently and move them to the top of the list. {R}