greenplum 和 posgresq 通用合并表函数

为了方便，创建以下合并函数

1、创建合并配置表 和 合并统计表

--创建合并配置表
drop table merge_config_table;
CREATE TABLE merge_config_table(tablename varchar,ispartitiontable varchar,to_merge_temp_table varchar,to_merge_table varchar,merge_field varchar,merge_flag int) distributed by (tablename);
alter table merge_config_table drop constraint pk_tablename;
alter table  merge_config_table add constraint pk_tablename primary key (tablename);
comment on column merge_config_table.tablename is '需要合并表名称：如需要合并 a 表，那么此处配置为 a 表';
comment on column merge_config_table.ispartitiontable is '是否是分区表： 0 否,1 是';
comment on column merge_config_table.to_merge_table is '待合并表名称:如将 b 表的数据合进 a 表，那么此配置为 b 表';
comment on column merge_config_table.to_merge_temp_table is '待合并临时表名称:如先将 c 表（临时表）的数据合并 b 表,再将 b 表的数据合进 a 表，此处配置为 c 表';
comment on column merge_config_table.merge_field is '合并字段';
comment on column merge_config_table.merge_flag is '是否需要合并：0 否,1 是';
--创建合并统计表
drop table merge_count;
create table merge_count(tablename varchar,merge_count int,total_count int,merge_time date) distributed randomly;
comment column on merge_count.tablename is '合并表名';
comment column on merge_count.merge_count is '合并数据统计';
comment column on merge_count.total_count is '表总数据';
comment column on merge_count.merge_time is '合并时间';

2、为合并配置表添加待合并表

--为配置表配置待合并的表
insert into merge_config_table values('tmp1',0,null,'tmp2','id,name',1)

3、创建合并函数

create or replace function proc_merge_table(tabname varchar,uname varchar)
returns void as
$FUNCTION$
declare
	isExist integer;
	sqltext text;
	merge_cnt int;
	orign_count int;
	increment_count int;
	merge_tabname varchar;
	middle_tabname varchar;
	totalcnt int;
	getresult int;
	merge_tablename cursor for select tablename from merge_config_table;
	all_field varchar;
	field cursor for select merge_field from merge_config_table where tablename = lower(tabname) and merge_flag = 1;
begin
	if lower(tabname) <> '' and uname <> '' then 
		--指定表合并
		SELECT count(*) INTO isExist
		FROM pg_tables pt,
 	    	merge_config_table mct
		WHERE pt.tablename = lower(mct.tablename)
		AND mct.tablename = lower(tabname)
  		AND pt.tableowner = lower(uname);
		if isExist = 0 then
			raise notice 'The query table % does not exists , please check tablename or username !',tabname;
		else
			raise notice '合并的表名为: % ',tabname;
			--获取合并字段
			open field;
			fetch field into all_field;
				--从配置表中将合并的表取出
				select to_merge_table into middle_tabname
				from merge_config_table 
				where tablename = lower(tabname) 
				and merge_flag = 1;
				raise notice '字段为： %' ,all_field;
				sqltext := 	'insert into ' 		|| lower(tabname) || 
							' select * from ' 	|| middle_tabname || 
							' where (' 	|| all_field || 
							') not in ' 	|| 
								' ( select ' || all_field || ' from ' || lower(tabname) || ' )';
				execute sqltext;
				GET DIAGNOSTICS getresult := ROW_COUNT;
				if getresult <> 0 then
					raise  notice '合并 % 行',getresult;
					sqltext := 'insert into merge_count(tablename,merge_count,merge_time)  
					values('''||tabname || ''','''|| getresult||''','''||now()::date||''')';
					execute sqltext;
				else 
					raise notice '未合并数据: %',getresult;
				end if;
				raise notice 'sqltext : %',sqltext;
				close field;
		end if;
	else
		--合并所有表
		open merge_tablename;
		fetch merge_tablename into merge_tabname;
	end if;
end;
$FUNCTION$
language plpgsql;

4、初始化表数据

truncate table tmp2;
truncate table tmp1;
truncate table merge_count;
insert into tmp1 values(1,'a');
insert into tmp1 values(1,'a');
insert into tmp1 values(2,'b');
insert into tmp1 values(3,'c');
insert into tmp1 values(4,'d');
insert into tmp1 values(1,'a');
insert into tmp2 values(2,'b');
insert into tmp2 values(3,'c');
insert into tmp2 values(4,'d');
insert into tmp2 values(5,'f');
insert into tmp2 values(6,'e');
insert into tmp2 values(7,'e');
insert into tmp2 values(8,'e');
insert into tmp2 values(9,'e');

5、执行合并函数

select proc_merge_table('tmp1','test');
NOTICE:  合并的表名为: tmp1
NOTICE:  字段为： id,name
NOTICE:  合并 5 行
NOTICE:  sqltext : insert into merge_count(tablename,merge_count,merge_time)
DETAIL:  values('tmp1','5','2019-06-30')
 proc_merge_table 
------------------
 
(1 row)

6、查看合并统计表记录

 tablename | merge_count | total_count | merge_time 
-----------+-------------+-------------+------------
 tmp1      |           5 |             | 2019-06-30
(1 row)

以上函数可以根据实际需求稍作修改即可满足