Extracting META information from a PostgreSQL database using the INFORMATION_SCHEMA views and the system catalogs (pg_class, pg_user, pg_view)
Following my tutorial on how to extract meta informations from Firebird SQL, I'm now going to show how to retrieve the same informations from PostgreSQL, using the INFORMATION_SCHEMA (available since PostgreSQL 7.4) and with system catalogs (pg_class, pg_user, pg_view, etc).
NB: as you probably know, you can list tables/indices/sequences/views from the command line with the \d{t|i|s|v} command, but here I want to show how to extract these informations using standard SQL queries.
We need a few sample tables, indices and views to test the following queries, so let's create them. We also create a sample TRIGGER and a function.
-- sample data to test PostgreSQL INFORMATION_SCHEMA
-- TABLE TEST
CREATE TABLE TEST (
TEST_NAME CHAR(30) NOT NULL,
TEST_ID INTEGER DEFAULT '0' NOT NULL,
TEST_DATE TIMESTAMP NOT NULL
);
ALTER TABLE TEST ADD CONSTRAINT PK_TEST PRIMARY KEY (TEST_ID);
-- TABLE TEST2 with some CONSTRAINTs and an INDEX
CREATE TABLE TEST2 (
ID INTEGER NOT NULL,
FIELD1 INTEGER,
FIELD2 CHAR(15),
FIELD3 VARCHAR(50),
FIELD4 INTEGER,
FIELD5 INTEGER,
ID2 INTEGER NOT NULL
);
ALTER TABLE TEST2 ADD CONSTRAINT PK_TEST2 PRIMARY KEY (ID2);
ALTER TABLE TEST2 ADD CONSTRAINT TEST2_FIELD1ID_IDX UNIQUE (ID, FIELD1);
ALTER TABLE TEST2 ADD CONSTRAINT TEST2_FIELD4_IDX UNIQUE (FIELD4);
CREATE INDEX TEST2_FIELD5_IDX ON TEST2(FIELD5);
-- TABLE NUMBERS
CREATE TABLE NUMBERS (
NUMBER INTEGER DEFAULT '0' NOT NULL,
EN CHAR(100) NOT NULL,
FR CHAR(100) NOT NULL
);
-- TABLE NEWTABLE
CREATE TABLE NEWTABLE (
ID INT DEFAULT 0 NOT NULL,
SOMENAME VARCHAR (12),
SOMEDATE TIMESTAMP NOT NULL
);
ALTER TABLE NEWTABLE ADD CONSTRAINT PKINDEX_IDX PRIMARY KEY (ID);
CREATE SEQUENCE NEWTABLE_SEQ INCREMENT 1 START 1;
-- VIEW on TEST
CREATE VIEW "testview"(
TEST_NAME,
TEST_ID,
TEST_DATE
) AS
SELECT *
FROM TEST
WHERE TEST_NAME LIKE 't%';
-- VIEW on NUMBERS
CREATE VIEW "numbersview"(
NUMBER,
TRANS_EN,
TRANS_FR
) AS
SELECT *
FROM NUMBERS
WHERE NUMBER > 100;
-- TRIGGER on NEWTABLE
CREATE FUNCTION add_stamp() RETURNS OPAQUE AS '
BEGIN
IF (NEW.somedate IS NULL OR NEW.somedate = 0) THEN
NEW.somedate := CURRENT_TIMESTAMP;
RETURN NEW;
END IF;
END;
' LANGUAGE 'plpgsql';
CREATE TRIGGER ADDCURRENTDATE
BEFORE INSERT OR UPDATE
ON newtable FOR EACH ROW
EXECUTE PROCEDURE add_stamp();
-- TABLEs for testing CONSTRAINTs
CREATE TABLE testconstraints (
someid integer NOT NULL,
somename character varying(10) NOT NULL,
CONSTRAINT testconstraints_id_pk PRIMARY KEY (someid)
);
CREATE TABLE testconstraints2 (
ext_id integer NOT NULL,
modified date,
uniquefield character varying(10) NOT NULL,
usraction integer NOT NULL,
CONSTRAINT testconstraints_id_fk FOREIGN KEY (ext_id)
REFERENCES testconstraints (someid) MATCH SIMPLE
ON UPDATE CASCADE ON DELETE CASCADE,
CONSTRAINT unique_2_fields_idx UNIQUE (modified, usraction),
CONSTRAINT uniquefld_idx UNIQUE (uniquefield)
);Here's the query that will return the names of the tables defined in the current database:
SELECT relname
FROM pg_class
WHERE relname !~ '^(pg_|sql_)'
AND relkind = 'r';
-- using INFORMATION_SCHEMA:
SELECT table_name
FROM information_schema.tables
WHERE table_type = 'BASE TABLE'
AND table_schema NOT IN
('pg_catalog', 'information_schema');Here's the query that will return the names of the VIEWs defined in the current database:
-- with postgresql 7.2:
SELECT viewname
FROM pg_views
WHERE viewname !~ '^pg_';
-- with postgresql 7.4 and later:
SELECT viewname
FROM pg_views
WHERE schemaname NOT IN
('pg_catalog', 'information_schema')
AND viewname !~ '^pg_';
-- using INFORMATION_SCHEMA:
SELECT table_name
FROM information_schema.tables
WHERE table_type = 'VIEW'
AND table_schema NOT IN
('pg_catalog', 'information_schema')
AND table_name !~ '^pg_';
-- or
SELECT table_name
FROM information_schema.views
WHERE table_schema NOT IN ('pg_catalog', 'information_schema')
AND table_name !~ '^pg_';
SELECT usename FROM pg_user;
Here's the query that will return the names of the fields of the TEST2 table:
SELECT a.attname FROM pg_class c, pg_attribute a, pg_type t WHERE c.relname = 'test2' AND a.attnum > 0 AND a.attrelid = c.oid AND a.atttypid = t.oid -- with INFORMATION_SCHEMA: SELECT column_name FROM information_schema.columns WHERE table_name = 'test2';
If you want some more info about the field definitions, you can retrieve a larger subset of the fields available in the schema:
SELECT a.attnum AS ordinal_position,
a.attname AS column_name,
t.typname AS data_type,
a.attlen AS character_maximum_length,
a.atttypmod AS modifier,
a.attnotnull AS notnull,
a.atthasdef AS hasdefault
FROM pg_class c,
pg_attribute a,
pg_type t
WHERE c.relname = 'test2'
AND a.attnum > 0
AND a.attrelid = c.oid
AND a.atttypid = t.oid
ORDER BY a.attnum;
-- with INFORMATION_SCHEMA:
SELECT ordinal_position,
column_name,
data_type,
column_default,
is_nullable,
character_maximum_length,
numeric_precision
FROM information_schema.columns
WHERE table_name = 'test2'
ORDER BY ordinal_position;Here's the query that will return the names of the INDICES defined in the TEST2 table. Unfortunately I have no idea how to extract them from the INFORMATION_SCHEMA. If you do, please let me know.
NB: the CONSTRAINTs are not listed
SELECT relname
FROM pg_class
WHERE oid IN (
SELECT indexrelid
FROM pg_index, pg_class
WHERE pg_class.relname='test2'
AND pg_class.oid=pg_index.indrelid
AND indisunique != 't'
AND indisprimary != 't';If you want to know which table columns are referenced by an index, you can do it in two steps: first you get the table name and field(s) position with this query:
SELECT relname, indkey
FROM pg_class, pg_index
WHERE pg_class.oid = pg_index.indexrelid
AND pg_class.oid IN (
SELECT indexrelid
FROM pg_index, pg_class
WHERE pg_class.relname='test2'
AND pg_class.oid=pg_index.indrelid
AND indisunique != 't'
AND indisprimary != 't'
);Then, using your favorite language, you explode the indkey (the key separator is a space), and for each key you run this query:
SELECT t.relname, a.attname, a.attnum
FROM pg_index c
LEFT JOIN pg_class t
ON c.indrelid = t.oid
LEFT JOIN pg_attribute a
ON a.attrelid = t.oid
AND a.attnum = ANY(indkey)
WHERE t.relname = 'test2'
AND a.attnum = 6; -- this is the index keyHere's the query that will return the names of the CONSTRAINTs defined in the TEST2 table:
SELECT relname
FROM pg_class
WHERE oid IN (
SELECT indexrelid
FROM pg_index, pg_class
WHERE pg_class.relname='test2'
AND pg_class.oid=pg_index.indrelid
AND ( indisunique = 't'
OR indisprimary = 't'
)
);
-- with INFORMATION_SCHEMA:
SELECT constraint_name, constraint_type
FROM information_schema.table_constraints
WHERE table_name = 'test2';
If you want to retrieve detailed info from any constraint (fields, type, rules, referenced table and fields for FOREIGN KEYs, etc.) given its name and table, here's the query to do so:
SELECT c.conname AS constraint_name,
CASE c.contype
WHEN 'c' THEN 'CHECK'
WHEN 'f' THEN 'FOREIGN KEY'
WHEN 'p' THEN 'PRIMARY KEY'
WHEN 'u' THEN 'UNIQUE'
END AS "constraint_type",
CASE WHEN c.condeferrable = 'f' THEN 0 ELSE 1 END AS is_deferrable,
CASE WHEN c.condeferred = 'f' THEN 0 ELSE 1 END AS is_deferred,
t.relname AS table_name,
array_to_string(c.conkey, ' ') AS constraint_key,
CASE confupdtype
WHEN 'a' THEN 'NO ACTION'
WHEN 'r' THEN 'RESTRICT'
WHEN 'c' THEN 'CASCADE'
WHEN 'n' THEN 'SET NULL'
WHEN 'd' THEN 'SET DEFAULT'
END AS on_update,
CASE confdeltype
WHEN 'a' THEN 'NO ACTION'
WHEN 'r' THEN 'RESTRICT'
WHEN 'c' THEN 'CASCADE'
WHEN 'n' THEN 'SET NULL'
WHEN 'd' THEN 'SET DEFAULT'
END AS on_delete,
CASE confmatchtype
WHEN 'u' THEN 'UNSPECIFIED'
WHEN 'f' THEN 'FULL'
WHEN 'p' THEN 'PARTIAL'
END AS match_type,
t2.relname AS references_table,
array_to_string(c.confkey, ' ') AS fk_constraint_key
FROM pg_constraint c
LEFT JOIN pg_class t ON c.conrelid = t.oid
LEFT JOIN pg_class t2 ON c.confrelid = t2.oid
WHERE t.relname = 'testconstraints2'
AND c.conname = 'testconstraints_id_fk';
-- with INFORMATION_SCHEMA:
SELECT tc.constraint_name,
tc.constraint_type,
tc.table_name,
kcu.column_name,
tc.is_deferrable,
tc.initially_deferred,
rc.match_option AS match_type,
rc.update_rule AS on_update,
rc.delete_rule AS on_delete,
ccu.table_name AS references_table,
ccu.column_name AS references_field
FROM information_schema.table_constraints tc
LEFT JOIN information_schema.key_column_usage kcu
ON tc.constraint_catalog = kcu.constraint_catalog
AND tc.constraint_schema = kcu.constraint_schema
AND tc.constraint_name = kcu.constraint_name
LEFT JOIN information_schema.referential_constraints rc
ON tc.constraint_catalog = rc.constraint_catalog
AND tc.constraint_schema = rc.constraint_schema
AND tc.constraint_name = rc.constraint_name
LEFT JOIN information_schema.constraint_column_usage ccu
ON rc.unique_constraint_catalog = ccu.constraint_catalog
AND rc.unique_constraint_schema = ccu.constraint_schema
AND rc.unique_constraint_name = ccu.constraint_name
WHERE tc.table_name = 'testconstraints2'
AND tc.constraint_name = 'testconstraints_id_fk';The "constraint_key" and "fk_constraint_key"
fields returned by the first query are space-separated strings containing the
position of the fields involved (in the FOREIGN KEY
constraint and those referenced by it), so you may need to retrieve them with
another query on the respective tables. Since the field positions are stored
as arrays, you can't (to the best of my knowledge) get all the field names
with an unique query (well, you could with a stored procedure).
The second query, the one using the INFORMATION_SCHEMA,
is certainly more straightforward, albeit slower.
A SEQUENCE is an object that automatically generate sequence numbers.
A SEQUENCE is often used to ensure a unique value in a PRIMARY KEY that must uniquely identify the associated row.
SELECT relname
FROM pg_class
WHERE relkind = 'S'
AND relnamespace IN (
SELECT oid
FROM pg_namespace
WHERE nspname NOT LIKE 'pg_%'
AND nspname != 'information_schema'
);SELECT trg.tgname AS trigger_name
FROM pg_trigger trg, pg_class tbl
WHERE trg.tgrelid = tbl.oid
AND tbl.relname !~ '^pg_';
-- or
SELECT tgname AS trigger_name
FROM pg_trigger
WHERE tgname !~ '^pg_';
-- with INFORMATION_SCHEMA:
SELECT DISTINCT trigger_name
FROM information_schema.triggers
WHERE trigger_schema NOT IN
('pg_catalog', 'information_schema');List only the triggers for a given table:
SELECT trg.tgname AS trigger_name
FROM pg_trigger trg, pg_class tbl
WHERE trg.tgrelid = tbl.oid
AND tbl.relname = 'newtable';
-- with INFORMATION_SCHEMA:
SELECT DISTINCT trigger_name
FROM information_schema.triggers
WHERE event_object_table = 'newtable'
AND trigger_schema NOT IN
('pg_catalog', 'information_schema');Show more informations about the trigger definitions:
SELECT trg.tgname AS trigger_name,
tbl.relname AS table_name,
p.proname AS function_name,
CASE trg.tgtype & cast(2 as int2)
WHEN 0 THEN 'AFTER'
ELSE 'BEFORE'
END AS trigger_type,
CASE trg.tgtype & cast(28 as int2)
WHEN 16 THEN 'UPDATE'
WHEN 8 THEN 'DELETE'
WHEN 4 THEN 'INSERT'
WHEN 20 THEN 'INSERT, UPDATE'
WHEN 28 THEN 'INSERT, UPDATE, DELETE'
WHEN 24 THEN 'UPDATE, DELETE'
WHEN 12 THEN 'INSERT, DELETE'
END AS trigger_event,
CASE trg.tgtype & cast(1 as int2)
WHEN 0 THEN 'STATEMENT'
ELSE 'ROW'
END AS action_orientation
FROM pg_trigger trg,
pg_class tbl,
pg_proc p
WHERE trg.tgrelid = tbl.oid
AND trg.tgfoid = p.oid
AND tbl.relname !~ '^pg_';
-- with INFORMATION_SCHEMA:
SELECT *
FROM information_schema.triggers
WHERE trigger_schema NOT IN
('pg_catalog', 'information_schema');SELECT proname
FROM pg_proc pr,
pg_type tp
WHERE tp.oid = pr.prorettype
AND pr.proisagg = FALSE
AND tp.typname <> 'trigger'
AND pr.pronamespace IN (
SELECT oid
FROM pg_namespace
WHERE nspname NOT LIKE 'pg_%'
AND nspname != 'information_schema'
);
-- with INFORMATION_SCHEMA:
SELECT routine_name
FROM information_schema.routines
WHERE specific_schema NOT IN
('pg_catalog', 'information_schema')
AND type_udt_name != 'trigger';Albe Laurenz sent me the following function that is even more informative: for a function name and schema, it selects the position in the argument list, the direction, the name and the data-type of each argument. This procedure requires PostgreSQL 8.1 or later.
CREATE OR REPLACE FUNCTION public.function_args(
IN funcname character varying,
IN schema character varying,
OUT pos integer,
OUT direction character,
OUT argname character varying,
OUT datatype character varying)
RETURNS SETOF RECORD AS $$DECLARE
rettype character varying;
argtypes oidvector;
allargtypes oid[];
argmodes "char"[];
argnames text[];
mini integer;
maxi integer;
BEGIN
/* get object ID of function */
SELECT INTO rettype, argtypes, allargtypes, argmodes, argnames
CASE
WHEN pg_proc.proretset
THEN 'setof ' || pg_catalog.format_type(pg_proc.prorettype, NULL)
ELSE pg_catalog.format_type(pg_proc.prorettype, NULL) END,
pg_proc.proargtypes,
pg_proc.proallargtypes,
pg_proc.proargmodes,
pg_proc.proargnames
FROM pg_catalog.pg_proc
JOIN pg_catalog.pg_namespace
ON (pg_proc.pronamespace = pg_namespace.oid)
WHERE pg_proc.prorettype <> 'pg_catalog.cstring'::pg_catalog.regtype
AND (pg_proc.proargtypes[0] IS NULL
OR pg_proc.proargtypes[0] <> 'pg_catalog.cstring'::pg_catalog.regtype)
AND NOT pg_proc.proisagg
AND pg_proc.proname = funcname
AND pg_namespace.nspname = schema
AND pg_catalog.pg_function_is_visible(pg_proc.oid);
/* bail out if not found */
IF NOT FOUND THEN
RETURN;
END IF;
/* return a row for the return value */
pos = 0;
direction = 'o'::char;
argname = 'RETURN VALUE';
datatype = rettype;
RETURN NEXT;
/* unfortunately allargtypes is NULL if there are no OUT parameters */
IF allargtypes IS NULL THEN
mini = array_lower(argtypes, 1); maxi = array_upper(argtypes, 1);
ELSE
mini = array_lower(allargtypes, 1); maxi = array_upper(allargtypes, 1);
END IF;
IF maxi < mini THEN RETURN; END IF;
/* loop all the arguments */
FOR i IN mini .. maxi LOOP
pos = i - mini + 1;
IF argnames IS NULL THEN
argname = NULL;
ELSE
argname = argnames[pos];
END IF;
IF allargtypes IS NULL THEN
direction = 'i'::char;
datatype = pg_catalog.format_type(argtypes[i], NULL);
ELSE
direction = argmodes[i];
datatype = pg_catalog.format_type(allargtypes[i], NULL);
END IF;
RETURN NEXT;
END LOOP;
RETURN;
END;$$ LANGUAGE plpgsql STABLE STRICT SECURITY INVOKER;
COMMENT ON FUNCTION public.function_args(character varying, character
varying)
IS $$For a function name and schema, this procedure selects for each
argument the following data:
- position in the argument list (0 for the return value)
- direction 'i', 'o', or 'b'
- name (NULL if not defined)
- data type$$;SELECT p.proname AS procedure_name,
p.pronargs AS num_args,
t1.typname AS return_type,
a.rolname AS procedure_owner,
l.lanname AS language_type,
p.proargtypes AS argument_types_oids,
prosrc AS body
FROM pg_proc p
LEFT JOIN pg_type t1 ON p.prorettype=t1.oid
LEFT JOIN pg_authid a ON p.proowner=a.oid
LEFT JOIN pg_language l ON p.prolang=l.oid
WHERE proname = :PROCEDURE_NAME;pg_proc.proargtypes contains an array of oids pointing to pg_type.oid. You can use unnest(), generate_procedure() or the function in the previous paragraph to retrieve the data type of each parameter.
If you'd like to see some other example queries, or have some comments and/or suggestions, just drop me a mail (you can find my address in the footer of this page) and I'll add them to this list.
HTH.
Lorenzo
21 responses to "Extracting META information from PostgreSQL (INFORMATION_SCHEMA)"
hi,
reading through your queries, really good stuff. One slight but critical part is missing and I am wondering if you could add it.
Under the section 'Detailed CONSTRAINT info' you have 2 queries that retrieve index information, I am interested in getting 4 columns for foreign keys, src_table, src_column, dst_table, dst_column, your 2nd information_schema sql can do this fine but it is slowww, your longer query above that one is much faster but is missing just one column, the dst_column (foreign table column name) and I have tried but cant find how to reliable retrieve it as I dont have an ERD for the pg_* tables.
Any help is much appreciated, thanks!
Hi, what you want is a not easy with one single query, because to get the name of the fields you have to do a JOIN on pg_attribute, but the pg_constraint conkey and confkey fields are arrays (listing the column numbers within the tables in the constraint). If you know your FOREIGN KEY constraint is on a single field, you can get what you want with this query:
SELECT c.conname AS constraint_name, t.relname AS table_name, a.attname AS field_name, t2.relname AS references_table, a2.attname AS references_field FROM pg_constraint c LEFT JOIN pg_class t ON c.conrelid = t.oid LEFT JOIN pg_class t2 ON c.confrelid = t2.oid LEFT JOIN pg_attribute a ON t.oid = a.attrelid AND a.attnum = c.conkey[1] LEFT JOIN pg_attribute a2 ON t2.oid = a2.attrelid AND a2.attnum = c.confkey[1] WHERE t.relname = 'testconstraints2' AND c.conname = 'testconstraints_id_fk' AND c.contype = 'f';If the FK is on multiple columns, then either you parse the conkey/confkeys arrays and do another query to select the names of the matching columns, or you try something like generate_subscripts(): http://www.postgresql.org/docs/8.4/static/arrays.html
Thank you so much for this information. I am building a postgresql module for my open source statistics package SOFA Statistics and I found it very hard to get the required information in a useful form.
Lorenzo hi! Thank you for sharing your knowledge with us. I found your notes very useful. Have you got anything similar concerning how to retrieve users queries on a postgres database? I 've tried pg_stat_activity but it only shows current queries and only select queries (and not update, insert, delete). I 've tried pg_log directory by changing in the conf file the log_line_prefix and log_statement= 'all' but it has implications on the load (creates extreme big log file). Is there any easy way to track users queries? Thank you Pepy
Hi, I am new to PostgreSQL. I have started learning and written triggers and they are working fine. Currently I am using spi_exe_query to insert/update my log table when primary table updated. Could you please give me an example of spi_prepare and spi_execute_plan usage in PL/Perl that I can use in my trigger. I couldn\'t find any example in documentation.
Basically, I want to prepare a plan and want to store it in %_SHARED hash so that I can use whenever I want. Moreover, if table definition changed then I should purge the cache.
Any suggestion would be much appreciated.
Thanks, Jignesh
Thanks Lorenzo.
Could you provied an example for conkey/confkeys arrays parsing in the case of multiple columns in fk ?
When you talk about to List Indices you said you don't have no idea to find using information schema. I found this way:
In this table you have 'constraint_type' to know the index type.
And, in addition, I found this select if someone are interested to know primary key(s) of one table:
Jordi, the first query you mention only lists constraints, not indices. The second one is not entirely correct either, since it will list all the columns having some constraint. A better way of listing the PRIMARY KEYs is:
Or, if you are interested in the full details for the PK, this one:
SELECT tc.constraint_name, tc.constraint_type, tc.table_name, kcu.column_name, tc.is_deferrable, tc.initially_deferred, rc.match_option AS match_type, rc.update_rule AS on_update, rc.delete_rule AS on_delete, ccu.table_name AS references_table, ccu.column_name AS references_field FROM information_schema.table_constraints tc LEFT JOIN information_schema.key_column_usage kcu ON tc.constraint_catalog = kcu.constraint_catalog AND tc.constraint_schema = kcu.constraint_schema AND tc.constraint_name = kcu.constraint_name LEFT JOIN information_schema.referential_constraints rc ON tc.constraint_catalog = rc.constraint_catalog AND tc.constraint_schema = rc.constraint_schema AND tc.constraint_name = rc.constraint_name LEFT JOIN information_schema.constraint_column_usage ccu ON rc.unique_constraint_catalog = ccu.constraint_catalog AND rc.unique_constraint_schema = ccu.constraint_schema AND rc.unique_constraint_name = ccu.constraint_name WHERE tc.table_name = 'test' AND tc.constraint_type = 'PRIMARY KEY';Jignesh, have a look at my article about table auditing: http://www.alberton.info/postgresql_table_audit.html. It's not something I would recommend on a live system, but it's OK if you only want to play around.
i'm getting some strange results for both information schema queries for detailed pk's and detailed constraint's. looks almost like a cartesian product.
how portable is information_schema, i hear its a standard but is this like browser standards or real standards?
re: my last comment, i neglected to notice or mention that for detailed constraint info i left off the constraint name in where clause.
if i modify the above PK query to look for tc.constraint_type = 'FOREIGN KEY' i get odd results as well.
any idea how to query the information schema to get full FK details on a given table without specifying the constraint name?
Thanks Lorenzo. You have an excellent resource for us here. In spite of all this I'm unable to solve my problem. Can you please help with this. I need to query column statistics to get tablename, rowcount, columnname, no. of nulls, null percentage and no. of distinct values. Thanks for your help.
Thanks for sharing it, a great post !
Hi,I am new to PostgreSql.I found your material very helpful.I am trying to write a generic function for auditing databases using pl/pgsql.I tried to use the same logic as you have used in your function using pl/tcl (http://www.alberton.info/postgresql_table_audit.html) but the use of new and old is giving errors.Is it not possible to convert new and old into arrays and then use them in pl/pgsql?If yes,then can you please help me out in this? Thank You
@KJ: unfortunately not, the reason why I used tcl instead of pl/pgsql is exactly that one: you can't use NEW and OLD with variable identifiers for the field names.
I am really Thankful to you for such a quick response.....I guess using another language would be a better option....
Hi! I need get the function body. How do I can get it? Thks
@Bruno: I updated the article with the query you need, to fetch the stored procedure body. HTH.
Hi Alberton, Thk to you for a quick solution. it works!
Do you know how to extract more information about the sequences in Postgres ??
Basically I need the sequence name (which I got ), its start value, increment value and max value (I tried the information_schema.sequences but dont see any inofrmation except names)
@Sameer Unfortunately I don't know. The INFORMATION_SCHEMA.SEQUENCE view is incomplete (as you can see from the definition [1], it's returning NULL for those values), probably waiting for the relevant info to be exposed. Looking at the sequence.c file [2], it looks like it shouldn't be that hard to add a function returning the sequence info. [1] http://git.postgresql.org/gitweb?p=postgresql.git;a=blob_plain;f=src/backend/catalog/information_schema.sql;hb=HEAD [2] http://git.postgresql.org/gitweb?p=postgresql.git;a=blob_plain;f=src/backend/commands/sequence.c;hb=HEAD