Delicious Bookmark this on Delicious Share on Facebook SlashdotSlashdot It! Digg! Digg



PHP : Function Reference : PostgreSQL Functions : pg_last_oid

pg_last_oid

Returns the last row's OID (PHP 4 >= 4.2.0, PHP 5)
string pg_last_oid ( resource result )

Example 1942. pg_last_oid() example

<?php
 $pgsql_conn
= pg_connect("dbname=mark host=localhost");
 
 
$res1 = pg_query("CREATE TABLE test (a INTEGER) WITH OIDS");

 
$res2 = pg_query("INSERT INTO test VALUES (1)");
 
 
$oid = pg_last_oid($res2);
?>

Code Examples / Notes » pg_last_oid

bens

[Editor's Note: If another record is inserted after the nextval is obtained and before you the INSERT query this code will fail.  This should not be done on busy sites.]
Especially now with an optional oid field, getting an implicit serial key is harder than ever. The solution is to get the serial key first, and then use that value in an insert:
<?
pg_query($conn, 'CREATE TABLE foo (key SERIAL, foo TEXT)');
$res=pg_query("SELECT nextval('foo_key_seq') as key");
$key=pg_fetch_array($res, 0);
$key=$key[key];
// now we have the serial value in $key, let's do the insert
pg_query("INSERT INTO foo (key, foo) VALUES ($key, 'blah blah')");
?>
Hope this helps...


webmaster

You could use this to get the last insert id...
CREATE TABLE test (
 id serial,
 something int not null
);
This creates the sequence test_id_seq. Now do the following after inserting something into table test:
INSERT INTO test (something) VALUES (123);
SELECT currval('test_id_seq') AS lastinsertid;
lastinsertid should contain your last insert id.


dtutar

This is very useful function :)
function sql_last_inserted_id($connection, $result, $table_name, $column_name) {
  $oid = pg_last_oid ( $result);
  $query_for_id = "SELECT $column_name FROM $table_name WHERE oid=$oid";
  $result_for_id = pg_query($connection,$query_for_id);
  if(pg_num_rows($result_for_id))
     $id=pg_fetch_array($result_for_id,0,PGSQL_ASSOC);
  return $id[$column_name];
}
Call after insert, simply ;)


wes

There seems to be some confusion about the PostgreSQL currval() function in these notes.  currval() isn't a general-purpose access to sequences, it has a very specific function.  From the PostgreSQL User's Guide (v 7.3.2) section 6.11:
currval - Return the value most recently obtained by nextval in the current session.  (An error is reported if nextval has never been called for this sequence in this session.)  Notice that because this is returning a session-local value, it gives a predictable answer even if other sessions are executing nextval meanwhile.
So wrapping a nextval (or other sequence insert) followed by currval in a transaction is not necessary.


julian

The way I nuderstand it, each value is emitted by a sequence only ONCE. If you retrieve a number (say 12) from a sequence using nextval(), the sequence will advance and subsequent calls to nextval() will return further numbers (after 12) in the sequence.
This means that if you use nextval() to retrieve a value to use as a primary key, you can be guaranteed that no other calls to nextval() on that sequence will return the same value. No race conditions, no transactions required.
That's what sequences are *for* afaik :^)


talk_biz

The sequence functions for autonumbering  can be read like an ordinary table.
ex. "select * from request_id_seq";
which returns all the current values for the sequence.  To get the last id number entered use
"select last_value from request_id_seq";
To have the query pull the values of the record with the last id, use a query with a subselect statement.
ex. "SELECT id, date_requested, time_requested, requestor_name, requestor_email, requestor_phone
FROM request where id = (select last_value from request_id_seq)";


luke

responding to a previous  editor's note-
No, the user is correct. when you do a 'select nextval' from a sequence, that increments the sequence.  If someone else does a 'select nextval' or inserts into the table with a 'null' value for the field defaulted to a nextval, they will get the next value-  thus if another record is inserted after the nextval and before the insert, that next record will still get a different value from the sequence.
Am I wrong?
[Editor's Note: If another record is inserted after the nextval is obtained and before you the INSERT query this code will fail.  This should not be done on busy sites.]
Especially now with an optional oid field, getting an implicit serial key is harder than ever. The solution is to get the serial key first, and then use that value in an insert:
<?
pg_query($conn, 'CREATE TABLE foo (key SERIAL, foo TEXT)');
$res=pg_query("SELECT nextval('foo_key_seq') as key");
$key=pg_fetch_array($res, 0);
$key=$key[key];
// now we have the serial value in $key, let's do the insert
pg_query("INSERT INTO foo (key, foo) VALUES ($key, 'blah blah')");
?>
Hope this helps...


php

Remember:  Although OID is somewhat unique (as others have mentioned), the OID isn't entirely unique.  Once it reaches the extent of the datatype it will go back to 0 and start again.
I suggest either using transactions or have another set of identifiers which you can use together to identify the inserted row... such as user_id and timestamp (which you've manually created, not using NOW)... If it's for something human based, a user would not be able to realistically click two submit buttons in the same second... if you do get that happening, then maybe you could assume it was spam.
Just a thought


gaagaagui

note the following:
"The oid type is currently implemented as an unsigned four-byte integer. Therefore, it is not large enough to provide database-wide uniqueness in large databases, or even in large individual tables. So, using a user-created table's OID column as a primary key is discouraged. OIDs are best used only for references to system tables."
and
"OIDs are 32-bit quantities and are assigned from a single cluster-wide counter. In a large or long-lived database, it is possible for the counter to wrap around. Hence, it is bad practice to assume that OIDs are unique, unless you take steps to ensure that they are unique."
from: http://www.phphub.com/postgres_manual/index.php?p=datatype-oid.html


juancri

Note that:
- OID is a unique id. It will not work if the table was created with "No oid".
- MySql's "mysql_insert_id" receives the conection handler as argument but PostgreSQL's "pg_last_oid" uses the result handler.


php

If you want to get the value of a sequence out of pgsql, similar to mysql_insert_id(), try this:
<?
pg_query( $connection, "BEGIN TRANSACTION" );
pg_query( $connection, $your_insert_query );
$result = pg_query( $connection, "SELECT CURRVAL('$seq_name') AS seq" );
$data = pg_fetch_assoc( $result );
pg_free_result( $result );
pg_query( $connection, "COMMIT TRANSACTION" );
echo "Insert sequence value: ", $data[ 'seq' ], "
\n";
?>
Note the transaction that groups the INSERT query and the query that obtains the sequence value.  If you are already in a transaction, don't start another.  Also be aware that CURRVAL only works after an INSERT query, and always returns the value for the last INSERT for your connection, even if someone else has done an INSERT after you.
See:
http://www.postgresql.org/docs/7.3/interactive/functions-sequence.html


jonathan bond-caron

I'm sharing an elegant solution I found on the web (Vadim Passynkov):
CREATE RULE get_pkey_on_insert AS ON INSERT TO Customers DO SELECT currval('customers_customers_id_seq') AS id;
Every time you insert to the Customers table, postgreSQL will return a table with the id you just inserted. No need to worry about concurrency, the ressource is locked when the rule gets executed.
Note that in cases of multiple inserts:
INSERT INTO C1 ( ... ) ( SELECT * FROM C2);
we would return the id of the last inserted row.
For more info about PostgreSQL rules:
http://www.postgresql.org/docs/7.4/interactive/sql-createrule.html


paul

I do not understand the editors notes in this section.
They seem to suggest that you can't safely use nextval to get an identifier from a sequence that will be unique accross all users/sessions. That is categorically false.
nextval will pull a number from a sequence and that number is guarunteed to be distinct accross all sessions.
To quote the Postgres Docs:
"nextval
Advance the sequence object to its next value and return that value. This is done atomically: even if multiple sessions execute nextval concurrently, each will safely receive a distinct sequence value."


kevin

I am afraid that the editor is misleading people here.
QUOTE "Editor's Note: If another record is inserted after the nextval is obtained and before you [sic: execute] the INSERT query this code will fail.  This should not be done on busy sites."
This is not correct. A sequence is a multi-session safe table-like structure in postgresql.
From the postgresql manual for nextval()
"Advance the sequence object to its next value and return that value. This is done atomically: even if multiple sessions execute nextval concurrently, each will safely receive a distinct sequence value. "
Since the default for a serial column during an insert is to call nextval(), it will also get a unique identifier.
Sequences are _not_ defined as being linearly ordered however and may contain holes and return values out of order (due to cache settings). They will, however, be unique 100% of the time.
The following code is CORRECT and thread safe WITHOUT a transaction on even the most loaded server.
QUOTE (with corrections) "
$res=pg_query("SELECT nextval('foo_key_seq') as key");
$row=pg_fetch_array($res, 0);
$key=$row['key'];
// now we have the serial value in $key, let's do the insert
pg_query("INSERT INTO foo (key, foo) VALUES ($key, 'blah blah')");
"
In this case the value retrieved for $key will never again be retrieved by the database under any circumstance. Period. Ever.
Hope that clears this up.


a dot bardsley

As pointed out on a busy site some of the above methods might actually give you an incorrect answer as another record is inserted inbetween your insert  and the select. To combat this put it into a transaction and dont commit till you have done all the work

Change Language


Follow Navioo On Twitter
pg_affected_rows
pg_cancel_query
pg_client_encoding
pg_close
pg_connect
pg_connection_busy
pg_connection_reset
pg_connection_status
pg_convert
pg_copy_from
pg_copy_to
pg_dbname
pg_delete
pg_end_copy
pg_escape_bytea
pg_escape_string
pg_execute
pg_fetch_all_columns
pg_fetch_all
pg_fetch_array
pg_fetch_assoc
pg_fetch_object
pg_fetch_result
pg_fetch_row
pg_field_is_null
pg_field_name
pg_field_num
pg_field_prtlen
pg_field_size
pg_field_table
pg_field_type_oid
pg_field_type
pg_free_result
pg_get_notify
pg_get_pid
pg_get_result
pg_host
pg_insert
pg_last_error
pg_last_notice
pg_last_oid
pg_lo_close
pg_lo_create
pg_lo_export
pg_lo_import
pg_lo_open
pg_lo_read_all
pg_lo_read
pg_lo_seek
pg_lo_tell
pg_lo_unlink
pg_lo_write
pg_meta_data
pg_num_fields
pg_num_rows
pg_options
pg_parameter_status
pg_pconnect
pg_ping
pg_port
pg_prepare
pg_put_line
pg_query_params
pg_query
pg_result_error_field
pg_result_error
pg_result_seek
pg_result_status
pg_select
pg_send_execute
pg_send_prepare
pg_send_query_params
pg_send_query
pg_set_client_encoding
pg_set_error_verbosity
pg_trace
pg_transaction_status
pg_tty
pg_unescape_bytea
pg_untrace
pg_update
pg_version
eXTReMe Tracker