Re: Fast prefix search?

On 10/14/07, Nathan Nobbe <quickshiftin@gmail.com> wrote: > how big is your dataset; have you tested against a potential data set and > gotten > long execution times? The dataset consists of about several million lines and I've tested script like above against the dataset. it took almost a hour. (perl script that uses set_range finished the job within 2 minutes) > notice how i used strpos rather than strstr, because i got this tip from the > docs > Note: If you only want to determine if a particular needle occurs within > haystack, > use the faster and less memory intensive function strpos() instead. On 10/14/07, Robert Cummings <robert@interjinn.com> wrote: > So don't use strstr() use strpos(). Specifically use it like follows: > > if( strpos( $haystack, $prefix ) === 0 ) > { > // it's a prefix. > } > Great tip. Thank you! > there are some optimization points as well. if you use a for loop, i think > it > will run a hair faster than the foreach, just make sure to store the length > of the > dictionary in an variable and use that as the sentinel control variable. > also, if > your search can be case insensitive you can use stripos instead of strpos, > that will > probly get you a little speed bump as well. I'll try. > if the string methods in a loop are too doggy due to the size of your > datasets, > you might write a program that uses berkdb and call it using the shell from > within php. that seems a shellscript... Threre seems no easy solution for this job. The best bet could be to write up do-it-your-self B-tree implementation in PHP... Anyway, thank you. your advice extended my knowledge a little :)