Re: full text indexing

On Thu, 28 Sep 2000, Gilles DAROLD wrote:
> Hi,
>
> After taking a look to many way from fast search into full text (flat
> field with regrep, UDMSearch,
> and other stuff like contrib FTI) i have developped my own indexer and
> fast search.
> It's still specific to my database but can be easily ported to any other
> database structure.
> It also use stop word
>
> The first load is very slow for big database but I have had incremental
> indexing and then
> following load are speeder.
>
> Here is the usage of the indexer :
>
> Usage: ./lincat-indexer.pl [-c -i file -h host -p port] -t table -f field
> -l language -u user -s file -d dbname
> -c => drop search tables and indexes, reset all indexing
> -i file => do incremental index with file
> -v => print a dot for each word processed
> -h host => database hostname [default: localhost]
> -p port => database service port [default: 5432]
> -t table => table to index
> -f field => field to index
> -l fr => language to use (can be fr, en, ...)
> -u user => postgres user who need read grant on search tables
> -s file => file with a list of all words not to be indexed
> -d dbname => name of the database
> -q => quiet mode, do not print anything else than error
> message
>
> I also have a search script that can be use to test the index and from you
> can cut and past
> some code to add to your programme.
>
> usage is ./lincat-search.pl "essai test".
>
> I don't remember exactly what it does with search (AND/OR) but this depend
> on what you want
> as result, so hack it.
>
> Of course it's Perl... Search is very fast !
> And in any way you can easily hack it to match your needs.
>
> Let me know if your are interested I will post it to the list.

I'm interested. I was starting to write a script that would store in another
table the information of certain words with the location, so when I look, I
do it on the table indexed and not with regex in a blob of TEXT.

Anyway I would like to see what you have.

--
"And I'm happy, because you make me feel good, about me." - Melvin Udall
-----------------------------------------------------------------
Martn Marqus email: martin(at)math(dot)unl(dot)edu(dot)ar
Santa Fe - Argentina http://math.unl.edu.ar/~martin/
Administrador de sistemas en math.unl.edu.ar
-----------------------------------------------------------------