Thread: LO to text conversion
Hi all. I need to store XML documents in database, so I use large objects for them. Now I have a problem: I want to search for patterns in documents, and it looks like there is no standard way to do it, so I write C function to do it, but it has limitation: it is not support regular expressions. Although it will be easy to use some library to handel them, but I know that PostgreSQL already has it. So, my question is: is there any function to search in lobjs? Another thing which would be enought is convertion of lo to text. BTW: Somebody can say, that I could use text field for documents as well, but it is inefficient to embed document in sql query and then extract it back -- Regards, ASK
On Wed, 24 Jan 2001, Alexander Klimov wrote:
> I need to store XML documents in database, so I use large objects for
> them. Now I have a problem: I want to search for patterns in documents,
> and it looks like there is no standard way to do it, so I write
> C function to do it, but it has limitation: it is not support regular
> expressions. Although it will be easy to use some library to handel
> them, but I know that PostgreSQL already has it.
>
> So, my question is: is there any function to search in lobjs? Another
> thing which would be enought is convertion of lo to text.
>
> BTW: Somebody can say, that I could use text field for documents as
> well, but it is inefficient to embed document in sql query and then
> extract it back
You might want to use the full-text search facility, whihc you can find
under src/contrib. With the new TOADT features of 7.1, storing large
chunks of text is now quite feasible.
-- Brett
http://www.chapelperilous.net/~bmccoy/
---------------------------------------------------------------------------
I would like to electrocute everyone who uses the word 'fair' in connection
with income tax policies.
-- William F. Buckley
----- Original Message ----- From: "Alexander Klimov" <ask@wisdom.weizmann.ac.il> > Hi all. > > I need to store XML documents in database, so I use large objects for > them. Now I have a problem: I want to search for patterns in documents, > and it looks like there is no standard way to do it, so I write > C function to do it, but it has limitation: it is not support regular > expressions. Although it will be easy to use some library to handel > them, but I know that PostgreSQL already has it. > > So, my question is: is there any function to search in lobjs? Another > thing which would be enought is convertion of lo to text. Don't think so - the objects are external to the DB. You could try 7.1 (in beta at the moment) - that has TOAST to provide v.large text fields (seems to work fine). You might want to look at the full-text-index stuff in /contrib - something based on that would probably be the way to go. Failing that, I know there are some XML DBs out there somewhere. Presumably they are designed for this sort of thing. - Richard Huxton