e5342afe2a
* Add PostgreSQL DBI extension Bug #3849 This adds a postgresql database driver called "pgsql". The ambuild script changes could be very wrong ;) * Add pgsql support to clientprefs This was originally made by Lyfe in bug 3849! Thanks for that. PostgreSQL supports the 'IF NOT EXISTS' clause when creating tables since version 9.1, so i've switched to use that. * Added pgsql support to sql-admin-manager * Add --pgsql-path to ambuild configure * More ambuild fixes * Compile libpq with ambuild * Try to generate postgres configs * Add dummy config file * More ambuild fixes * Get errorcode and readable error in preparequery No need to stop getting the errorCode, if the error string should be retrieved. Just return both! * Fix indentation in AMBuilderPGSQL * Try to patch and configure postgres through ambuild * Revert "Try to patch and configure postgres through ambuild" This reverts commit 68dfc82b8eb0ce11f621af67ec247b5f47e4a189. * Update to use PostgreSQL 9.4 * Move postgres preparation into seperate script Fetching, preparing and configuring of the postgresql sourcecode is done in a seperate prepare_postgresql.sh script now. People can use this script for their manual builds, if they don't want to use checkout-deps.sh. * Add patch to configure.in again Remove the version check for autoconf in postgres 9.4 too. * Nit: Support older objdirs for hasPgSql ambuild option * Update to use PostgreSQL 9.4.6 * Use newer AMTL and PGSQL 9.6 * Fix threaded queries When running the thread part of a query, it already gets the atomic lock. Don't try to get it twice in that case. Use a seperate lock to protect access to the lastInsertID and lastAffectedRows variables. * Fix SetCharacterSet racing with threaded queries Same fix like in the mysql driver. * Use ke::AutoLock for lastInsertID mutex * Fix fetching one more row than available Don't try to access a row that doesn't exist when iterating rows. * Improve sanity checks on column access Don't call libpq functions with out of bounds column indices. * Let QuoteString return false if quoting failed * Fix build for x64 support changes and update to PGSQL 9.6.9 * Fix linux build * Fix building of connection options string snprintf doesn't support overlapping buffers. * Update to PostgreSQL 9.6.15 * Fix crash after freeing first IQuery object * Fix code crunch * Fix memory leak, cleanup * Nuke MSVC project and Makefile * Remove unsupported numeric error codes * Use STL threads * Add prebuilt libpq for Linux and Windows * Restore and fix version.rc file * Add PostgreSQL build instructions * Add prebuilt libpq for Mac * Add win64 libpq build This is version 9.6.15 since that's what I still had laying around. * Fix buildscript * Fix x64 build on linux and mac Co-authored-by: Erik Minekus <tsunami@tsunami-productions.nl>
41 lines
1.1 KiB
PL/PgSQL
41 lines
1.1 KiB
PL/PgSQL
CREATE TABLE IF NOT EXISTS sm_cookies
|
|
(
|
|
id serial,
|
|
name varchar(30) NOT NULL UNIQUE,
|
|
description varchar(255),
|
|
access INTEGER,
|
|
PRIMARY KEY (id)
|
|
);
|
|
|
|
CREATE TABLE IF NOT EXISTS sm_cookie_cache
|
|
(
|
|
player varchar(65) NOT NULL,
|
|
cookie_id int NOT NULL,
|
|
value varchar(100),
|
|
timestamp int NOT NULL,
|
|
PRIMARY KEY (player, cookie_id)
|
|
);
|
|
|
|
CREATE LANGUAGE plpgsql;
|
|
|
|
CREATE OR REPLACE FUNCTION add_or_update_cookie(in_player VARCHAR(65), in_cookie INT, in_value VARCHAR(100), in_time INT) RETURNS VOID AS
|
|
$$
|
|
BEGIN
|
|
LOOP
|
|
-- first try to update the it.
|
|
UPDATE sm_cookie_cache SET value = in_value, timestamp = in_time WHERE player = in_player AND cookie_id = in_cookie;
|
|
IF found THEN
|
|
RETURN;
|
|
END IF;
|
|
-- not there, so try to insert.
|
|
-- if someone else inserts the same key concurrently, we could get a unique-key failure.
|
|
BEGIN
|
|
INSERT INTO sm_cookie_cache (player, cookie_id, value, timestamp) VALUES (in_player, in_cookie, in_value, in_time);
|
|
RETURN;
|
|
EXCEPTION WHEN unique_violation THEN
|
|
-- do nothing... loop again, and we'll update.
|
|
END;
|
|
END LOOP;
|
|
END;
|
|
$$
|
|
LANGUAGE plpgsql; |