元宵佳节:看Oracle技术粉们用SQL画团圆

话团圆,画团圆,元宵佳节倍思亲,可是大家知道吗,万能的SQL可以帮助大家绘制团圆。

在ITPUB论坛里,一群SQL爱好者们会用SQL来描摹一切可能。请看如下这段SQL,为大家绘制了团团圆圆的五连环:

with a as (select distinct round(a.x + b.x) x,round(a.y + b.y) y from
(select (sum(x) over(order by n)) x,
                            round(sum(y) over(order by n)) y
              from (select n, cos(n/30 * 3.1415926)*2  x,
                           sin(n/30 * 3.1415926) y
                           from (select rownum - 1 n from all_objects where rownum <= 30 +30))) a,
            (select n, (sum(x) over(order by n)) x,
                            round(sum(y) over(order by n)) y
              from (select n,
                           cos( m /3 * 3.1415926) * 2 * 15 x,
                           sin( m /3 * 3.1415926)* 15 y
                      from (select case when rownum <= 2 then 3 
                      when rownum = 3 then -2 else -6 end m, rownum - 1 n
                              from all_objects where rownum <= 5))) b
          )
select replace(sys_connect_by_path(point, '/'), '/', null) star
  from (select b.y, b.x, decode(a.x, null, ' ', '*') point
          from a,
               (select *
                  from (select rownum - 1 + (select min(x) from a) x
                          from all_objects
                         where rownum <= (select max(x) - min(x) + 1 from a)),
                       (select rownum - 1 + (select min(y) from a) y
                          from all_objects
                         where rownum <= (select max(y) - min(y) + 1 from a))) b
         where a.x(+) = b.x
           and a.y(+) = b.y)
where x = (select max(x) from a)
start with x = (select min(x) from a)
connect by y = prior y
       and x = prior x + 1;

这段SQL在Oracle中输出了下图,请用SQL执行:

好吧,这是五个连环,事实上是奥运会的五环旗,在庆祝奥运期间,网友 nyfor 的随手创作。

再看如下一段SQL,则是输出了一个五角星:

with a as (
            select distinct round(sum(x) over(order by n)) x,
                            round(sum(y) over(order by n)) y
              from (select n,
                           cos(trunc(n / 20) * (1-1/5) * 3.1415926) * 2 x,
                           sin(trunc(n / 20) * (1-1/5) * 3.1415926) y
                      from (select rownum - 1 n from all_objects where rownum <= 20 * 5))
          )
select replace(sys_connect_by_path(point, '/'), '/', null) star
  from (select b.y, b.x, decode(a.x, null, ' ', '*') point
          from a,
               (select *
                  from (select rownum - 1 + (select min(x) from a) x
                          from all_objects
                         where rownum <= (select max(x) - min(x) + 1 from a)),
                       (select rownum - 1 + (select min(y) from a) y
                          from all_objects
                         where rownum <= (select max(y) - min(y) + 1 from a))) b
         where a.x(+) = b.x
           and a.y(+) = b.y)
where x = (select max(x) from a)
start with x = (select min(x) from a)
connect by y = prior y
       and x = prior x + 1;

这个SQL的解释如下

其中数字20表示五角星每一条边上的点的个数(你也可以设置的大一些或小一些), 其中的数字5表示五角星的边数, 其中的数字2是为了调整横向字符间距与纵向行距之间的差异而设置的, 你也可以不乘以这个2, 这里只是为了输出稍微好看一些.

调整期中数字5, 你还可以输出7角星, 9角星.... 注意我的SQL不能输出6角星,8角星,因为我的SQL算法中是以一笔画能够画成的星为基础设计的算法的.

比如,以下是7角形输出:

在一轮讨论之后,newkid 大神给出了一个系列的SQL改写,小编就列举如下。

SQL一:

with a as ( select distinct round(sum(x) over(order by n)) x, round(sum(y) over(order by n)) y from (select n, cos(trunc(n / 20) * (1-1/5) * 3.1415926) * 2 x, sin(trunc(n / 20) * (1-1/5) * 3.1415926) y from (select rownum - 1 n from DUAL CONNECT BY rownum <= 20 * 5)) ) SELECT LPAD(REPLACE(SUM(POWER(10,x-1)),'0',' '),(SELECT MAX(x) FROM a)) AS star FROM a GROUP BY y ORDER BY y;

SQL二:

with a as ( select distinct round(sum(x) over(order by n)) x, round(sum(y) over(order by n)) y from (select n, cos(trunc(n / 20) * (1-1/5) * 3.1415926) x, sin(trunc(n / 20) * (1-1/5) * 3.1415926) y from (select rownum - 1 n from DUAL CONNECT BY rownum <= 20 * 5)) ) SELECT LPAD(REPLACE(SUM(POWER(10,x)),'0',' '),(SELECT MAX(x)+1 FROM a)) AS star FROM a GROUP BY y ORDER BY y;

SQL三:

with a as ( select distinct round(sum(x) over(order by n)) x, round(sum(y) over(order by n)) y from (select n, cos(trunc(n / 20) * (1-1/5) * 3.1415926) * 2 x, sin(trunc(n / 20) * (1-1/5) * 3.1415926) y from (select rownum - 1 n from DUAL CONNECT BY rownum <= 20 * 5)) ) SELECT TRANSLATE(LPAD(NVL(SUM(POWER(10,CASE WHEN x>=40 THEN x-40 END)),0),(SELECT MAX(x)-39 FROM a WHERE x>=40)) ||LPAD(SUM(POWER(10,CASE WHEN x<40 THEN x END)),40) ,'01',' *' ) AS star FROM a GROUP BY y ORDER BY y;

SQL四:

with a as (SELECT x,y ,ROW_NUMBER() OVER(PARTITION BY y ORDER BY x) rn ,MAX(x) OVER(PARTITION BY y) maxx FROM (select distinct round(sum(x) over(order by n)) x, round(sum(y) over(order by n)) y from (select n, cos(trunc(n / 20) * (1-1/5) * 3.1415926) * 2 x, sin(trunc(n / 20) * (1-1/5) * 3.1415926) y from (select rownum - 1 n from DUAL CONNECT BY rownum <= 20 * 5) ) ) ) ,t(rn,x,y,str,maxx) AS ( SELECT 1,x,y,LPAD('*',x+1),maxx FROM a WHERE rn=1 UNION ALL SELECT a.rn,a.x,t.y,str||RPAD(' ',a.x-t.x-1)||'*',t.maxx FROM t,a WHERE t.rn=a.rn-1 AND t.y=a.y ) CYCLE x,y SET cycle_flag TO 'Y' DEFAULT 'N' SELECT str FROM t WHERE x=maxx ORDER BY y;

SQL五:

VAR SCALE NUMBER; EXEC :SCALE :=3; with a as (SELECT x,y ,ROW_NUMBER() OVER(PARTITION BY y ORDER BY x) rn ,MAX(x) OVER(PARTITION BY y) maxx FROM (select distinct round(sum(x) over(order by n)) x, round(sum(y) over(order by n)) y from (select n, cos(trunc(n / (10*:SCALE)) * (1-1/5) * 3.1415926) * 2 x, sin(trunc(n / (10*:SCALE)) * (1-1/5) * 3.1415926) y from (select rownum - 1 n from DUAL CONNECT BY rownum <= 10*:SCALE * 5) ) ) ) ,t(rn,x,y,str,maxx) AS ( SELECT 1,x,y,LPAD('*',x+1),maxx FROM a WHERE rn=1 UNION ALL SELECT a.rn,a.x,t.y,str||RPAD(' ',a.x-t.x-1)||'*',t.maxx FROM t,a WHERE t.rn=a.rn-1 AND t.y=a.y ) CYCLE x,y SET cycle_flag TO 'Y' DEFAULT 'N' SELECT str FROM t WHERE x=maxx ORDER BY y;

SQL六 - 利用wmsys.wm_concat的写法其实更简单

with a as (SELECT x,y ,LAG(x,1,0) OVER(PARTITION BY y ORDER BY x) last_x FROM (select distinct round(sum(x) over(order by n)) x, round(sum(y) over(order by n)) y from (select n, cos(trunc(n / (10*:SCALE)) * (1-1/5) * 3.1415926) * 2 x, sin(trunc(n / (10*:SCALE)) * (1-1/5) * 3.1415926) y from (select rownum - 1 n from DUAL CONNECT BY rownum <= 10*:SCALE * 5) ) ) ) SELECT REPLACE(MAX(str),',') STR FROM (SELECT y,wmsys.wm_concat(LPAD('*',x-last_x)) OVER(PARTITION BY y ORDER BY x) str FROM a ) GROUP BY y ORDER BY y;

SQL之七 - wmsys.wm_concat的connect by替代写法:

with a as (SELECT x,y ,LAG(x,1,0) OVER(PARTITION BY y ORDER BY x) last_x ,ROW_NUMBER() OVER(PARTITION BY y ORDER BY x) rn FROM (select distinct round(sum(x) over(order by n)) x, round(sum(y) over(order by n)) y from (select n, cos(trunc(n / (10*:SCALE)) * (1-1/5) * 3.1415926) * 2 x, sin(trunc(n / (10*:SCALE)) * (1-1/5) * 3.1415926) y from (select rownum - 1 n from DUAL CONNECT BY rownum <= 10*:SCALE * 5) ) ) ) SELECT REPLACE(MAX(str),',') STR FROM (SELECT y,SYS_CONNECT_BY_PATH(LPAD('*',x-last_x),',') str FROM a START WITH rn=1 CONNECT BY y=PRIOR y AND rn=PRIOR rn+1 ) GROUP BY y ORDER BY y;

SQL如神,学习入化,动手为王,祝愿大家元宵节快乐!

近期文章

新年贺礼:云和恩墨大讲堂期刊第二期

删繁就简-云和恩墨的一道面试题解析

用SQL解一道数学题:Gauss和Poincare

新年贺礼:云和恩墨大讲堂期刊发行

2015 Oracle 十大热门文章精选

Oracle 12c ASM 防火防盗新特性揭秘

DBA入门之路:学习与进阶之经验谈

DBA入门之路:关于日常工作的建议

电子渠道(网络销售)分析系统、数据治理

IT基础架构

分布式存储解决方案 | zData一体机 | 容灾环境建设

数据架构

Oracle DB2 MySQL NoSQL

专项服务:架构/安全/容灾/优化/整合/升级/迁移

运维服务:运维服务 代维服务

人才培养:个人认证 企业内训

软件产品:SQL审核、监控、数据恢复

应用架构

应用软件和中间件:数据建模 | SQL审核和优化 | 中间件服务

原文发布于微信公众号 - 数据和云(OraNews)

原文发表时间:2016-02-22

本文参与腾讯云自媒体分享计划,欢迎正在阅读的你也加入,一起分享。

发表于

我来说两句

0 条评论
登录 后参与评论

相关文章

来自专栏跟着阿笨一起玩NET

(1)显示每个类别最新更新的数据

在项目中经常遇到求每个类别最新显示的数据,比如显示某某某类别最新更新的5条数据。特写下这个sql记录于此:

821
来自专栏乐沙弥的世界

Oracle 索引质量分析

      索引质量的高低对数据库整体性能有着直接的影响。良好高质量的索引使得数据库性能得以数量级别的提升,而低效冗余的索引则使得数据库性能缓慢如牛,即便是使用...

841
来自专栏芋道源码1024

电商系统设计之订单

1. 前言2. 付款2.1 成功2.2 人祸2.4 天灾2.4 注释2.5 表结构2.5.1 交易表2.5.2 支付记录表2.5.3 订单表3. 运输4. 收货...

1733
来自专栏杨建荣的学习笔记

关于索引扫描的极速调优实战(第二篇)(r3笔记第82天)

在上一篇http://blog.itpub.net/23718752/viewspace-1364914/ 中我们大体介绍了下问题的情况,已经初步根据awr能...

3557
来自专栏Grace development

电商系统设计之商品 (下)

完成上述流程则是完成了一笔交易,经常网上购物的童鞋都懂这个。今天我们讲下从商品系统到交易系统和订单系统的存储过程及其设计上的应该注意的“坑”。

9922
来自专栏杨建荣的学习笔记

联系生活来简化sql(r3笔记第43天)

目前生产环境中有一条sql语句的CPU消耗很高。执行时间比较长。从awr中抓到的sql语句如下: SELECT run_request.run_mode, ...

2863
来自专栏沃趣科技

SQL优化案例-从执行计划定位SQL问题(三)

当SQL出现问题,能从执行计划中快速的定位哪部分出现问题很重要,SQL文本如下(为保证客户隐私,已经将注释和文字部分去掉):

1666
来自专栏Java帮帮-微信公众号-技术文章全总结

Oracle应用实战七——多表查询+PL/SQL

1 多表查询 内连接 使用一张以上的表做查询就是多表查询 语法: SELECT {DISTINCT} *|列名.. FROM 表名 别名,表名1 别名 {WH...

7574
来自专栏杨建荣的学习笔记

关于索引扫描的极速调优实战(第一篇) (r3笔记第81天)

一般在生产环境中,如果某个查询中涉及一个大表,走索引扫描是显然是最值得推荐的方式,但是索引扫描有unique index scan, range scan,sk...

2464
来自专栏沃趣科技

SQL优化案例-改变那些CBO无能为力的执行计划

用户写的sql,Oracle会进行等价改写,即使是RBO优化模式,Oracle也会给你做一些转换,这些转化都是基于一种固定的算法,oracle称这种转换是“启发...

4477

扫码关注云+社区

领取腾讯云代金券