1.语法基础
http://www.jianshu.com/p/d8cf42a96cd5
2.高级语法编程
http://www.imooc.com/article/1415
3.高级查询
http://www.cnblogs.com/hoojo/archive/2011/07/16/2108129.html
高级查询在数据库中用得是最频繁的,也是应用最广泛的。 Ø 基本常用查询 --select select * from student; --all 查询所有 select all sex from student; --distinct 过滤重复 select distinct sex from student; --count 统计 select count(*) from student; select count(sex) from student; select count(distinct sex) from student; --top 取前N条记录 select top 3 * from student; --alias column name 列重命名 select id as 编号, name '名称', sex 性别 from student; --alias table name 表重命名 select id, name, s.id, s.name from student s; --column 列运算 select (age + id) col from student; select s.name + '-' + c.name from classes c, student s where s.cid = c.id; --where 条件 select * from student where id = 2; select * from student where id > 7; select * from student where id < 3; select * from student where id <> 3; select * from student where id >= 3; select * from student where id <= 5; select * from student where id !> 3; select * from student where id !< 5; --and 并且 select * from student where id > 2 and sex = 1; --or 或者 select * from student where id = 2 or sex = 1; --between ... and ... 相当于并且 select * from student where id between 2 and 5; select * from student where id not between 2 and 5; --like 模糊查询 select * from student where name like '%a%'; select * from student where name like '%[a][o]%'; select * from student where name not like '%a%'; select * from student where name like 'ja%'; select * from student where name not like '%[j,n]%'; select * from student where name like '%[j,n,a]%'; select * from student where name like '%[^ja,as,on]%'; select * from student where name like '%[ja_on]%'; --in 子查询 select * from student where id in (1, 2); --not in 不在其中 select * from student where id not in (1, 2); --is null 是空 select * from student where age is null; --is not null 不为空 select * from student where age is not null; --order by 排序 select * from student order by name; select * from student order by name desc; select * from student order by name asc; --group by 分组 按照年龄进行分组统计 select count(age), age from student group by age; 按照性别进行分组统计 select count(*), sex from student group by sex; 按照年龄和性别组合分组统计,并排序 select count(*), sex from student group by sex, age order by age; 按照性别分组,并且是id大于2的记录最后按照性别排序 select count(*), sex from student where id > 2 group by sex order by sex; 查询id大于2的数据,并完成运算后的结果进行分组和排序 select count(*), (sex * id) new from student where id > 2 group by sex * id order by sex * id; --group by all 所有分组 按照年龄分组,是所有的年龄 select count(*), age from student group by all age; --having 分组过滤条件 按照年龄分组,过滤年龄为空的数据,并且统计分组的条数和现实年龄信息 select count(*), age from student group by age having age is not null; 按照年龄和cid组合分组,过滤条件是cid大于1的记录 select count(*), cid, sex from student group by cid, sex having cid > 1; 按照年龄分组,过滤条件是分组后的记录条数大于等于2 select count(*), age from student group by age having count(age) >= 2; 按照cid和性别组合分组,过滤条件是cid大于1,cid的最大值大于2 select count(*), cid, sex from student group by cid, sex having cid > 1 and max(cid) > 2; Ø 嵌套子查询 子查询是一个嵌套在select、insert、update或delete语句或其他子查询中的查询。任何允许使用表达式的地方都可以使用子查询。子查询也称为内部查询或内部选择,而包含子查询的语句也成为外部查询或外部选择。 # from (select … table)示例 将一个table的查询结果当做一个新表进行查询 select * from ( select id, name from student where sex = 1 ) t where t.id > 2; 上面括号中的语句,就是子查询语句(内部查询)。在外面的是外部查询,其中外部查询可以包含以下语句: 1、 包含常规选择列表组件的常规select查询 2、 包含一个或多个表或视图名称的常规from语句 3、 可选的where子句 4、 可选的group by子句 5、 可选的having子句 # 示例 查询班级信息,统计班级学生人生 select *, (select count(*) from student where cid = classes.id) as num from classes order by num; # in, not in子句查询示例 查询班级id大于小于的这些班级的学生信息 select * from student where cid in ( select id from classes where id > 2 and id < 4 ); 查询不是班的学生信息 select * from student where cid not in ( select id from classes where name = '2班' ) in、not in 后面的子句返回的结果必须是一列,这一列的结果将会作为查询条件对应前面的条件。如cid对应子句的id; # exists和not exists子句查询示例 查询存在班级id为的学生信息 select * from student where exists ( select * from classes where id = student.cid and id = 3 ); 查询没有分配班级的学生信息 select * from student where not exists ( select * from classes where id = student.cid ); exists和not exists查询需要内部查询和外部查询进行一个关联的条件,如果没有这个条件将是查询到的所有信息。如:id等于student.id; # some、any、all子句查询示例 查询班级的学生年龄大于班级的学生的年龄的信息 select * from student where cid = 5 and age > all ( select age from student where cid = 3 ); select * from student where cid = 5 and age > any ( select age from student where cid = 3 ); select * from student where cid = 5 and age > some ( select age from student where cid = 3 ); Ø 聚合查询 1、 distinct去掉重复数据 select distinct sex from student; select count(sex), count(distinct sex) from student; 2、 compute和compute by汇总查询 对年龄大于的进行汇总 select age from student where age > 20 order by age compute sum(age) by age; 对年龄大于的按照性别进行分组汇总年龄信息 select id, sex, age from student where age > 20 order by sex, age compute sum(age) by sex; 按照年龄分组汇总 select age from student where age > 20 order by age, id compute sum(age); 按照年龄分组,年龄汇总,id找最大值 select id, age from student where age > 20 order by age compute sum(age), max(id); compute进行汇总前面是查询的结果,后面一条结果集就是汇总的信息。compute子句中可以添加多个汇总表达式,可以添加的信息如下: a、 可选by关键字。它是每一列计算指定的行聚合 b、 行聚合函数名称。包括sum、avg、min、max、count等 c、 要对其执行聚合函数的列 compute by适合做先分组后汇总的业务。compute by后面的列一定要是order by中出现的列。 3、 cube汇总 cube汇总和compute效果类似,但语法较简洁,而且返回的是一个结果集。 select count(*), sex from student group by sex with cube; select count(*), age, sum(age) from student where age is not null group by age with cube; cube要结合group by语句完成分组汇总 Ø 排序函数 排序在很多地方需要用到,需要对查询结果进行排序并且给出序号。比如: 1、 对某张表进行排序,序号需要递增不重复的 2、 对学生的成绩进行排序,得出名次,名次可以并列,但名次的序号是连续递增的 3、 在某些排序的情况下,需要跳空序号,虽然是并列 基本语法 排序函数 over([分组语句] 排序子句[desc][asc]) 排序子句 order by 列名, 列名 分组子句 partition by 分组列, 分组列 # row_number函数 根据排序子句给出递增连续序号 按照名称排序的顺序递增 select s.id, s.name, cid, c.name, row_number() over(order by c.name) as number from student s, classes c where cid = c.id; # rank函数函数 根据排序子句给出递增的序号,但是存在并列并且跳空 顺序递增 select id, name, rank() over(order by cid) as rank from student; 跳过相同递增 select s.id, s.name, cid, c.name, rank() over(order by c.name) as rank from student s, classes c where cid = c.id; # dense_rank函数 根据排序子句给出递增的序号,但是存在并列不跳空 不跳过,直接递增 select s.id, s.name, cid, c.name, dense_rank() over(order by c.name) as dense from student s, classes c where cid = c.id; # partition by分组子句 可以完成对分组的数据进行增加排序,partition by可以与以上三个函数联合使用。 利用partition by按照班级名称分组,学生id排序 select s.id, s.name, cid, c.name, row_number() over(partition by c.name order by s.id) as rank from student s, classes c where cid = c.id; select s.id, s.name, cid, c.name, rank() over(partition by c.name order by s.id) as rank from student s, classes c where cid = c.id; select s.id, s.name, cid, c.name, dense_rank() over(partition by c.name order by s.id) as rank from student s, classes c where cid = c.id; # ntile平均排序函数 将要排序的数据进行平分,然后按照等分排序。ntile中的参数代表分成多少等分。 select s.id, s.name, cid, c.name, ntile(5) over(order by c.name) as ntile from student s, classes c where cid = c.id; Ø 集合运算 操作两组查询结果,进行交集、并集、减集运算 1、 union和union all进行并集运算 --union 并集、不重复 select id, name from student where name like 'ja%' union select id, name from student where id = 4; --并集、重复 select * from student where name like 'ja%' union all select * from student; 2、 intersect进行交集运算 --交集(相同部分) select * from student where name like 'ja%' intersect select * from student; 3、 except进行减集运算 --减集(除相同部分) select * from student where name like 'ja%' except select * from student where name like 'jas%'; Ø 公式表表达式 查询表的时候,有时候中间表需要重复使用,这些子查询被重复查询调用,不但效率低,而且可读性低,不利于理解。那么公式表表达式可以解决这个问题。 我们可以将公式表表达式(CET)视为临时结果集,在select、insert、update、delete或是create view语句的执行范围内进行定义。 --表达式 with statNum(id, num) as ( select cid, count(*) from student where id > 0 group by cid ) select id, num from statNum order by id; with statNum(id, num) as ( select cid, count(*) from student where id > 0 group by cid ) select max(id), avg(num) from statNum; Ø 连接查询 1、 简化连接查询 --简化联接查询 select s.id, s.name, c.id, c.name from student s, classes c where s.cid = c.id; 2、 left join左连接 --左连接 select s.id, s.name, c.id, c.name from student s left join classes c on s.cid = c.id; 3、 right join右连接 --右连接 select s.id, s.name, c.id, c.name from student s right join classes c on s.cid = c.id; 4、 inner join内连接 --内连接 select s.id, s.name, c.id, c.name from student s inner join classes c on s.cid = c.id; --inner可以省略 select s.id, s.name, c.id, c.name from student s join classes c on s.cid = c.id; 5、 cross join交叉连接 --交叉联接查询,结果是一个笛卡儿乘积 select s.id, s.name, c.id, c.name from student s cross join classes c --where s.cid = c.id; 6、 自连接(同一张表进行连接查询) --自连接 select distinct s.* from student s, student s1 where s.id <> s1.id and s.sex = s1.sex; Ø 函数 1、 聚合函数 max最大值、min最小值、count统计、avg平均值、sum求和、var求方差 select max(age) max_age, min(age) min_age, count(age) count_age, avg(age) avg_age, sum(age) sum_age, var(age) var_age from student; 2、 日期时间函数 select dateAdd(day, 3, getDate());--加天 select dateAdd(year, 3, getDate());--加年 select dateAdd(hour, 3, getDate());--加小时 --返回跨两个指定日期的日期边界数和时间边界数 select dateDiff(day, '2011-06-20', getDate()); --相差秒数 select dateDiff(second, '2011-06-22 11:00:00', getDate()); --相差小时数 select dateDiff(hour, '2011-06-22 10:00:00', getDate()); select dateName(month, getDate());--当前月份 select dateName(minute, getDate());--当前分钟 select dateName(weekday, getDate());--当前星期 select datePart(month, getDate());--当前月份 select datePart(weekday, getDate());--当前星期 select datePart(second, getDate());--当前秒数 select day(getDate());--返回当前日期天数 select day('2011-06-30');--返回当前日期天数 select month(getDate());--返回当前日期月份 select month('2011-11-10'); select year(getDate());--返回当前日期年份 select year('2010-11-10'); select getDate();--当前系统日期 select getUTCDate();--utc日期 3、 数学函数 select pi();--PI函数 select rand(100), rand(50), rand(), rand();--随机数 select round(rand(), 3), round(rand(100), 5);--精确小数位 --精确位数,负数表示小数点前 select round(123.456, 2), round(254.124, -2); select round(123.4567, 1, 2); 4、 元数据 select col_name(object_id('student'), 1);--返回列名 select col_name(object_id('student'), 2); --该列数据类型长度 select col_length('student', col_name(object_id('student'), 2)); --该列数据类型长度 select col_length('student', col_name(object_id('student'), 1)); --返回类型名称、类型id select type_name(type_id('varchar')), type_id('varchar'); --返回列类型长度 select columnProperty(object_id('student'), 'name', 'PRECISION'); --返回列所在索引位置 select columnProperty(object_id('student'), 'sex', 'ColumnId'); 5、 字符串函数 select ascii('a');--字符转换ascii值 select ascii('A'); select char(97);--ascii值转换字符 select char(65); select nchar(65); select nchar(45231); select nchar(32993);--unicode转换字符 select unicode('A'), unicode('中');--返回unicode编码值 select soundex('hello'), soundex('world'), soundex('word'); select patindex('%a', 'ta'), patindex('%ac%', 'jack'), patindex('dex%', 'dexjack');--匹配字符索引 select 'a' + space(2) + 'b', 'c' + space(5) + 'd';--输出空格 select charIndex('o', 'hello world');--查找索引 select charIndex('o', 'hello world', 6);--查找索引 select quoteName('abc[]def'), quoteName('123]45'); --精确数字 select str(123.456, 2), str(123.456, 3), str(123.456, 4); select str(123.456, 9, 2), str(123.456, 9, 3), str(123.456, 6, 1), str(123.456, 9, 6); select difference('hello', 'helloWorld');--比较字符串相同 select difference('hello', 'world'); select difference('hello', 'llo'); select difference('hello', 'hel'); select difference('hello', 'hello'); select replace('abcedef', 'e', 'E');--替换字符串 select stuff('hello world', 3, 4, 'ABC');--指定位置替换字符串 select replicate('abc#', 3);--重复字符串 select subString('abc', 1, 1), subString('abc', 1, 2), subString('hello Wrold', 7, 5);--截取字符串 select len('abc');--返回长度 select reverse('sqlServer');--反转字符串 select left('leftString', 4);--取左边字符串 select left('leftString', 7); select right('leftString', 6);--取右边字符串 select right('leftString', 3); select lower('aBc'), lower('ABC');--小写 select upper('aBc'), upper('abc');--大写 --去掉左边空格 select ltrim(' abc'), ltrim('# abc#'), ltrim(' abc'); --去掉右边空格 select rtrim(' abc '), rtrim('# abc# '), rtrim('abc'); 6、 安全函数 select current_user; select user; select user_id(), user_id('dbo'), user_id('public'), user_id('guest'); select user_name(), user_name(1), user_name(0), user_name(2); select session_user; select suser_id('sa'); select suser_sid(), suser_sid('sa'), suser_sid('sysadmin'), suser_sid('serveradmin'); select is_member('dbo'), is_member('public'); select suser_name(), suser_name(1), suser_name(2), suser_name(3); select suser_sname(), suser_sname(0x01), suser_sname(0x02), suser_sname(0x03); select is_srvRoleMember('sysadmin'), is_srvRoleMember('serveradmin'); select permissions(object_id('student')); select system_user; select schema_id(), schema_id('dbo'), schema_id('guest'); select schema_name(), schema_name(1), schema_name(2), schema_name(3); 7、 系统函数 select app_name();--当前会话的应用程序名称 select cast(2011 as datetime), cast('10' as money), cast('0' as varbinary);--类型转换 select convert(datetime, '2011');--类型转换 select coalesce(null, 'a'), coalesce('123', 'a');--返回其参数中第一个非空表达式 select collationProperty('Traditional_Spanish_CS_AS_KS_WS', 'CodePage'); select current_timestamp;--当前时间戳 select current_user; select isDate(getDate()), isDate('abc'), isNumeric(1), isNumeric('a'); select dataLength('abc'); select host_id(); select host_name(); select db_name(); select ident_current('student'), ident_current('classes');--返回主键id的最大值 select ident_incr('student'), ident_incr('classes');--id的增量值 select ident_seed('student'), ident_seed('classes'); select @@identity;--最后一次自增的值 select identity(int, 1, 1) as id into tab from student;--将studeng表的烈属,以/1自增形式创建一个tab select * from tab; select @@rowcount;--影响行数 select @@cursor_rows;--返回连接上打开的游标的当前限定行的数目 select @@error;--T-SQL的错误号 select @@procid; 8、 配置函数 set datefirst 7;--设置每周的第一天,表示周日 select @@datefirst as '星期的第一天', datepart(dw, getDate()) AS '今天是星期'; select @@dbts;--返回当前数据库唯一时间戳 set language 'Italian'; select @@langId as 'Language ID';--返回语言id select @@language as 'Language Name';--返回当前语言名称 select @@lock_timeout;--返回当前会话的当前锁定超时设置(毫秒) select @@max_connections;--返回SQL Server 实例允许同时进行的最大用户连接数 select @@MAX_PRECISION AS 'Max Precision';--返回decimal 和numeric 数据类型所用的精度级别 select @@SERVERNAME;--SQL Server 的本地服务器的名称 select @@SERVICENAME;--服务名 select @@SPID;--当前会话进程id select @@textSize; select @@version;--当前数据库版本信息 9、 系统统计函数 select @@CONNECTIONS;--连接数 select @@PACK_RECEIVED; select @@CPU_BUSY; select @@PACK_SENT; select @@TIMETICKS; select @@IDLE; select @@TOTAL_ERRORS; select @@IO_BUSY; select @@TOTAL_READ;--读取磁盘次数 select @@PACKET_ERRORS;--发生的网络数据包错误数 select @@TOTAL_WRITE;--sqlserver执行的磁盘写入次数 select patIndex('%soft%', 'microsoft SqlServer'); select patIndex('soft%', 'software SqlServer'); select patIndex('%soft', 'SqlServer microsoft'); select patIndex('%so_gr%', 'Jsonisprogram'); 10、 用户自定义函数 # 查看当前数据库所有函数 --查询所有已创建函数 select definition,* from sys.sql_modules m join sys.objects o on m.object_id = o.object_id and type in('fn', 'if', 'tf'); # 创建函数 if (object_id('fun_add', 'fn') is not null) drop function fun_add go create function fun_add(@num1 int, @num2 int) returns int with execute as caller as begin declare @result int; if (@num1 is null) set @num1 = 0; if (@num2 is null) set @num2 = 0; set @result = @num1 + @num2; return @result; end go 调用函数 select dbo.fun_add(id, age) from student; --自定义函数,字符串连接 if (object_id('fun_append', 'fn') is not null) drop function fun_append go create function fun_append(@args nvarchar(1024), @args2 nvarchar(1024)) returns nvarchar(2048) as begin return @args + @args2; end go select dbo.fun_append(name, 'abc') from student; # 修改函数 alter function fun_append(@args nvarchar(1024), @args2 nvarchar(1024)) returns nvarchar(1024) as begin declare @result varchar(1024); --coalesce返回第一个不为null的值 set @args = coalesce(@args, ''); set @args2 = coalesce(@args2, '');; set @result = @args + @args2; return @result; end go select dbo.fun_append(name, '#abc') from student; # 返回table类型函数 --返回table对象函数 select name, object_id, type from sys.objects where type in ('fn', 'if', 'tf') or type like '%f%'; if (exists (select * from sys.objects where type in ('fn', 'if', 'tf') and name = 'fun_find_stuRecord')) drop function fun_find_stuRecord go create function fun_find_stuRecord(@id int) returns table as return (select * from student where id = @id); go select * from dbo.fun_find_stuRecord(2);
4.时间转换
1 SELECT CONVERT(varchar(100), GETDATE(), 0) 05 9 2011 9:12AM 2 SELECT CONVERT(varchar(100), GETDATE(), 1) 05/09/11 3 SELECT CONVERT(varchar(100), GETDATE(), 2) 11.05.09 4 SELECT CONVERT(varchar(100), GETDATE(), 3) 09/05/11 5 SELECT CONVERT(varchar(100), GETDATE(), 4) 09.05.11 6 SELECT CONVERT(varchar(100), GETDATE(), 5) 09-05-11 7 SELECT CONVERT(varchar(100), GETDATE(), 6) 09 05 11 8 SELECT CONVERT(varchar(100), GETDATE(), 7) 05 09, 11 9 SELECT CONVERT(varchar(100), GETDATE(), 8) 09:13:14 10 SELECT CONVERT(varchar(100), GETDATE(), 9) 05 9 2011 9:13:14:670AM 11 SELECT CONVERT(varchar(100), GETDATE(), 10) 05-09-11 12 SELECT CONVERT(varchar(100), GETDATE(), 11) 11/05/09 13 SELECT CONVERT(varchar(100), GETDATE(), 12) 110509 14 SELECT CONVERT(varchar(100), GETDATE(), 13) 09 05 2011 09:13:14:670 15 SELECT CONVERT(varchar(100), GETDATE(), 14) 09:13:14:670 16 SELECT CONVERT(varchar(100), GETDATE(), 20) 2011-05-09 09:13:14 17 SELECT CONVERT(varchar(100), GETDATE(), 21) 2011-05-09 09:13:14.670 18 SELECT CONVERT(varchar(100), GETDATE(), 22) 05/09/11 9:15:33 AM 19 SELECT CONVERT(varchar(100), GETDATE(), 23) 2011-05-09 20 SELECT CONVERT(varchar(100), GETDATE(), 24) 09:15:33 21 SELECT CONVERT(varchar(100), GETDATE(), 25) 2011-05-09 09:15:33.140 22 SELECT CONVERT(varchar(100), GETDATE(), 100) 05 9 2011 9:15AM 23 SELECT CONVERT(varchar(100), GETDATE(), 101) 05/09/2011 24 SELECT CONVERT(varchar(100), GETDATE(), 102) 2011.05.09 25 SELECT CONVERT(varchar(100), GETDATE(), 103) 09/05/2011 26 SELECT CONVERT(varchar(100), GETDATE(), 104) 09.05.2011 27 SELECT CONVERT(varchar(100), GETDATE(), 105) 09-05-2011 28 SELECT CONVERT(varchar(100), GETDATE(), 106) 09 05 2011 29 SELECT CONVERT(varchar(100), GETDATE(), 107) 05 09, 2011 30 SELECT CONVERT(varchar(100), GETDATE(), 108) 09:16:38 31 SELECT CONVERT(varchar(100), GETDATE(), 109) 05 9 2011 9:16:38:543AM 32 SELECT CONVERT(varchar(100), GETDATE(), 110) 05-09-2011 33 SELECT CONVERT(varchar(100), GETDATE(), 111) 2011/05/09 34 SELECT CONVERT(varchar(100), GETDATE(), 112) 20110509 35 SELECT CONVERT(varchar(100), GETDATE(), 113) 09 05 2011 09:17:19:857 36 SELECT CONVERT(varchar(100), GETDATE(), 114) 09:17:19:857 37 SELECT CONVERT(varchar(100), GETDATE(), 120) 2011-05-09 09:17:19 38 SELECT CONVERT(varchar(100), GETDATE(), 121) 2011-05-09 09:17:19.857 39 SELECT CONVERT(varchar(100), GETDATE(), 126) 2011-05-09T09:17:19.857 40 SELECT CONVERT(varchar(100), GETDATE(), 130) 6 ????? ??????? 1432 9:17:19:857AM 41 SELECT CONVERT(varchar(100), GETDATE(), 131) 6/06/1432 9:17:19:857AM
5.数据分页
1 select * from 2 3 (select ROW_NUMBER() over (order by name ) as num, 4 Name, 5 Age, 6 Sex 7 from dt_People) as t 8 9 where t.num>=2 10 and t.num<4
6.高级语法
1.按class分组,分组后的结果按照name,age排序,如此根据rn可以进行数据筛选(例如取出年纪最大的)
1 row_number() over(partition by Class order by name,age desc) rn
2.生成连续的日期
注:生成的时间,可以通过convert方法转换为标准的日期格式
1 --方法1:从当前日期往回显示70天 2 select dateadd(d, number, GETDATE() + 1 - 70) as every_time 3 FROM master..spt_values n 4 WHERE n.type = 'p' 5 AND n.number <= 70; 6 7 --方法2:设定起止时间,显示时间段内的日期 8 declare @StartDate DATETIME = '2015/05/01' 9 declare @EndDate DATETIME ='2015/06/03' 10 select dateadd(day,number,@StartDate) as every_time 11 from master.dbo.spt_values 12 where type ='P' 13 and number <=DATEDIFF(day, @StartDate, @EndDate) 14 15 --方法3:设定开始时间到当天的所有时间 16 SELECT DateAdd(day,number,'2018-01-01') as time 17 FROM master..spt_values 18 WHERE type = 'p' 19 AND number <= DateDiff(day,'2018-01-01',GetDate()+1)
7.创建视图
1 if exists (select * from sysobjects where name = 'View_EdsProd') 2 drop view View_EdsProd 3 go 4 --创建视图 5 create view View_EdsProd 6 as 7 select * from Tab_EdsProd where Mid>1 8 go 9 --使用视图-- 10 select *from View_EdsProd
8.行转列
将:1 --> 1 转换为1 --> 1,2
1 --> 2
Oracle实现:
1 select id,wm_concat(name) names from table1 group by id
Sql Server实现:
1 select id, [names]=stuff((select ','+[name] from table1 A t where A.id=B.id for xml path('')), 1, 1, '') 2 from table1 B 3 group by id