On Thu, 30 Sep 1999, Michel Pelletier wrote:
Some people thing, 'why not use re (the Python regex module)?', because searching like '*ing' would require iterating over all the keys, a linear search like this could take multiple order of maginitude more time than a non-regex search.
But this is not a problem for many of us who arn't trying to index libraries. Perhaps a way of specifying 'extended' searches to ZCatalog (but still allowing normal, quick searches), implemented as a linear regex search through the index. And if this was too slow for some people, it would provide the hook they need to replace Catalog with something that met their requirements: GlimpseCatalog - indexes are dumped to text files that are indexed using Glimpse. Extended syntax would be the fuzzy regexp based matches used by Glimpse (as implemented by agrep) for those sites whose visitors can't spell :-) RDBCatalog - indexes are stored in a backend RDBMS, and substring matches are done using SQL. ConTextCatalog - interfaces to Oracle and the ConText option for people who require funky soundex matches and the various wierd options it provides. And Ultraseek uses python doesn't it? Might be a trivial mating. ___ // Zen (alias Stuart Bishop) Work: zen@cs.rmit.edu.au // E N Senior Systems Alchemist Play: zen@shangri-la.dropbear.id.au //__ Computer Science, RMIT WWW: http://www.cs.rmit.edu.au/~zen