Thread: Generic Join on Arrays

Generic Join on Arrays

From
KÖPFERL Robert
Date:
Hi,
I'm currently on retrieving meta infromation about db-schemas.

As I found out, pg_proc relation provides me with data about defined stored
procedures. Togehter with other relations as pg_type I can retrieve readable
information, like:
select proname, pd.description FROM  pg_proc pp left outer join
pg_description   pd on  pp.oid=pd.objoid

BUT, how about arrays of type "oidvector" (int4-array) containing foreign
keys (to be joined) as pg_proc's column  "proargtypes" 

Is there a way to make some kind of indetermine number of JOINs on the
single array items?
I want to retrieve a list of argument type names, not plain numbers.
I want to avoid doing several resolutions oid->name


Re: Generic Join on Arrays

From
Michael Fuhr
Date:
On Mon, May 30, 2005 at 03:42:49PM +0200, KÖPFERL Robert wrote:
> 
> As I found out, pg_proc relation provides me with data about defined stored
> procedures. Togehter with other relations as pg_type I can retrieve readable
> information, like:
> select proname, pd.description FROM  pg_proc pp left outer join
> pg_description   pd on  pp.oid=pd.objoid
> 
> BUT, how about arrays of type "oidvector" (int4-array) containing foreign
> keys (to be joined) as pg_proc's column  "proargtypes" 

Have you used ECHO_HIDDEN (or psql -E) to look at the queries that \df
generates?  Would oidvectortypes() be useful?

SELECT proname, oidvectortypes(proargtypes)
FROM pg_proc
WHERE proname = 'substr';proname |     oidvectortypes      
---------+-------------------------substr  | bytea, integersubstr  | text, integersubstr  | bytea, integer,
integersubstr | text, integer, integer
 
(4 rows)

Another possibility might be to cast the function's oid to regprocedure:

SELECT oid::regprocedure
FROM pg_proc
WHERE proname = 'substr';             oid              

-------------------------------substr(bytea,integer)substr(text,integer)substr(bytea,integer,integer)substr(text,integer,integer)
(4 rows)

Is either of those what you're looking for?  They don't address the
problem in the general case, but they might serve in this particular
case.

-- 
Michael Fuhr
http://www.fuhr.org/~mfuhr/


Re: Generic Join on Arrays

From
KÖPFERL Robert
Date:
Thanks.
Since I was just interested in this special case (while the general wasn't
interesting, either) this helped me. Also psql -E is a nice new feature to
me.

|-----Original Message-----
|From: Michael Fuhr [mailto:mike@fuhr.org]
|Sent: Montag, 30. Mai 2005 16:09
|To: KÖPFERL Robert
|Cc: pgsql-sql@postgresql.org
|Subject: Re: [SQL] Generic Join on Arrays
|
|
|On Mon, May 30, 2005 at 03:42:49PM +0200, KÖPFERL Robert wrote:
|>
|> As I found out, pg_proc relation provides me with data about
|defined stored
|> procedures. Togehter with other relations as pg_type I can
|retrieve readable
|> information, like:
|> select proname, pd.description FROM  pg_proc pp left outer join
|> pg_description   pd on  pp.oid=pd.objoid
|>
|> BUT, how about arrays of type "oidvector" (int4-array)
|containing foreign
|> keys (to be joined) as pg_proc's column  "proargtypes"
|
|Have you used ECHO_HIDDEN (or psql -E) to look at the queries that \df
|generates?  Would oidvectortypes() be useful?
|
|SELECT proname, oidvectortypes(proargtypes)
|FROM pg_proc
|WHERE proname = 'substr';
| proname |     oidvectortypes
|---------+-------------------------
| substr  | bytea, integer
| substr  | text, integer
| substr  | bytea, integer, integer
| substr  | text, integer, integer
|(4 rows)
|
|Another possibility might be to cast the function's oid to
|regprocedure:
|
|SELECT oid::regprocedure
|FROM pg_proc
|WHERE proname = 'substr';
|              oid
|-------------------------------
| substr(bytea,integer)
| substr(text,integer)
| substr(bytea,integer,integer)
| substr(text,integer,integer)
|(4 rows)
|
|Is either of those what you're looking for?  They don't address the
|problem in the general case, but they might serve in this particular
|case.
|
|--
|Michael Fuhr
|http://www.fuhr.org/~mfuhr/
|