Oracle 10g 完全参考手册之DECODE实现交叉报表--数据行转换为报表的列
本文主要摘自并翻译于《Oracle Database 10g 完全参考手册》英文版，并根据情况补充11g的内容，和自己的例子
（注：这里使用查询出来的数据结果是把源码中的数据建到本地使用，因为直接把书本中的查询出来的结果复制出来后格式乱，对不齐，原来格式请看书本）
The DECODE function is without doubt one of themost powerful in Oracle’s SQL. It is one of several extensions Oracle added to thestandard SQL language. This chapter will explore a number of ways that DECODE can be used, including the generationof “crosstab” reports. You can also use the CASE function and the COALESCE function to execute complex logicaltests within your SQL statements.
11g中补充：DECODE 和 CASE 常常用来旋转数据----也就是说将数据行转换为报表的列。
Here is DECODE’s format:
DECODE(value,if1,then1,if2,then2,if3,then3,.. . ,else)
Here, value represents any column in a table (regardless of datatype) orany result of a computation,
such as one date minus another, a SUBSTR of a character column, one numbertimes another,
and so on. value is tested for each row. If value equals if1, then theresult of the DECODE is
then1; if value equals if2, then the result of the DECODE is then2. This continues forvirtually
as many if-then pairs as you can construct. If value equals none of theifs, then the result of the
DECODE is else. Each of the ifs and thensas well as the else also can be a column or the result
of a function or computation. You can have up to 255 elements within theparentheses.
Let’s consider the checkout history for the bookshelf, as recorded in the BOOKSHELF_CHECKOUTtable:
select *
from BOOKSHELF_CHECKOUT;
| NAME |
TITLE |
CHECKOUTDATE |
RETURNEDDATE |
| JED HOPKINS |
INNUMERACY |
2002-1-1 |
2002-1-22 |
| GERHARDT KENTGEN |
WONDERFUL LIFE |
2002-1-2 |
2002-2-2 |
| DORAH TALBOT |
EITHER/OR |
2002-1-2 |
2002-1-10 |
| EMILY TALBOT |
ANNE OF GREEN GABLES |
2002-1-2 |
2002-1-20 |
| PAT LAVAY |
THE SHIPPING NEWS |
2002-1-2 |
2002-1-12 |
| ROLAND BRANDT |
THE SHIPPING NEWS |
2002-1-12 |
2002-3-12 |
| ROLAND BRANDT |
THE DISCOVERERS |
2002-1-12 |
2002-3-1 |
| ROLAND BRANDT |
WEST WITH THE NIGHT |
2002-1-12 |
2002-3-1 |
| EMILY TALBOT |
MIDNIGHT MAGIC |
2002-1-20 |
2002-2-3 |
| EMILY TALBOT |
HARRY POTTER AND THE GOBLET OF FIRE |
2002-2-3 |
2002-2-14 |
| PAT LAVAY |
THE MISMEASURE OF MAN |
2002-1-12 |
2002-2-12 |
| DORAH TALBOT |
POLAR EXPRESS |
2002-2-1 |
2002-2-15 |
| DORAH TALBOT |
GOOD DOG, CARL |
2002-2-1 |
2002-2-15 |
| GERHARDT KENTGEN |
THE MISMEASURE OF MAN |
2002-2-13 |
2002-3-5 |
| FRED FULLER |
JOHN ADAMS |
2002-2-1 |
2002-3-1 |
| FRED FULLER |
TRUMAN |
2002-3-1 |
2002-3-20 |
| JED HOPKINS |
TO KILL A MOCKINGBIRD |
2002-2-15 |
2002-3-1 |
| DORAH TALBOT |
MY LEDGER |
2002-2-15 |
2002-3-3 |
| GERHARDT KENTGEN |
MIDNIGHT MAGIC |
2002-2-5 |
2002-2-10 |
NOTE
Your output may contain an extra blank line due to word wrapping.
Extra blank lines are not shown here for readability.
As you look through the checkout list, you realize that some of the bookswere checked out for a rather long time. You can order this list by the numberof days checked out, to highlight the readers who keep books for the longesttime:
selectname
,TITLE
,RETURNEDDATE - CHECKOUTDATE DAYSOUT
from BOOKSHELF_CHECKOUT
orderby DAYSOUT desc;
| NAME |
TITLE |
DAYSOUT |
| ROLAND BRANDT |
THE SHIPPING NEWS |
59 |
| ROLAND BRANDT |
WEST WITH THE NIGHT |
48 |
| ROLAND BRANDT |
THE DISCOVERERS |
48 |
| GERHARDT KENTGEN |
WONDERFUL LIFE |
31 |
| PAT LAVAY |
THE MISMEASURE OF MAN |
31 |
| FRED FULLER |
JOHN ADAMS |
28 |
| JED HOPKINS |
INNUMERACY |
21 |
| GERHARDT KENTGEN |
THE MISMEASURE OF MAN |
20 |
| FRED FULLER |
TRUMAN |
19 |
| EMILY TALBOT |
ANNE OF GREEN GABLES |
18 |
| DORAH TALBOT |
MY LEDGER |
16 |
| DORAH TALBOT |
POLAR EXPRESS |
14 |
| DORAH TALBOT |
GOOD DOG, CARL |
14 |
| EMILY TALBOT |
MIDNIGHT MAGIC |
14 |
| JED HOPKINS |
TO KILL A MOCKINGBIRD |
14 |
| EMILY TALBOT |
HARRY POTTER AND THE GOBLET OF FIRE |
11 |
| PAT LAVAY |
THE SHIPPING NEWS |
10 |
| DORAH TALBOT |
EITHER/OR |
8 |
| GERHARDT KENTGEN |
MIDNIGHT MAGIC |
5 |
that take longer to read? Multiple variables are involved, and looking atthem in isolation may lead
to incorrect decisions. What is the average number of days out for thebooks in each category?
select B.CATEGORYNAME
,min(BC.RETURNEDDATE - BC.CHECKOUTDATE) MINOUT
,max(BC.RETURNEDDATE - BC.CHECKOUTDATE) MAXOUT
,avg(BC.RETURNEDDATE - BC.CHECKOUTDATE) AVGOUT
from BOOKSHELF_CHECKOUT BC
,BOOKSHELF B
where BC.TITLE = B.TITLE
groupby B.CATEGORYNAME
orderby B.CATEGORYNAME
| CATEGORYNAME |
MINOUT |
MAXOUT |
AVGOUT |
| ADULTFIC |
10 |
59 |
27.66666667 |
| ADULTNF |
16 |
48 |
29.11111111 |
| ADULTREF |
8 |
8 |
8 |
| CHILDRENFIC |
5 |
18 |
12 |
| CHILDRENPIC |
14 |
14 |
14 |
This is more useful, but it doesn’t factor in the impact of the differentpeople checking out books in these categories. To accomplish that, you coulduse an additional level of grouping, but the results will be easier to use ifyou create a crosstab report. The following listing generates a report thatshows the minimum, maximum, and average days out by person, by category. This queryuses the DECODE function to perform thecalculations. Three borrowers are used for the purposes of this example.
select B.CATEGORYNAME
,max(DECODE(BC.NAME, 'FRED FULLER', BC.RETURNEDDATE - BC.CHECKOUTDATE, null)) MAXFF
,avg(DECODE(BC.NAME, 'FRED FULLER', BC.RETURNEDDATE - BC.CHECKOUTDATE, null)) AVGFF
,max(DECODE(BC.NAME, 'DORAH TALBOT', BC.RETURNEDDATE - BC.CHECKOUTDATE, null)) MAXDT
,avg(DECODE(BC.NAME, 'DORAH TALBOT', BC.RETURNEDDATE - BC.CHECKOUTDATE, null)) AVGDT
,max(DECODE(BC.NAME, 'GERHARDT KENTGEN', BC.RETURNEDDATE - BC.CHECKOUTDATE, null)) MAXGK
,avg(DECODE(BC.NAME, 'GERHARDT KENTGEN', BC.RETURNEDDATE - BC.CHECKOUTDATE, null)) AVGGK
from BOOKSHELF_CHECKOUT BC
,BOOKSHELF B
where BC.TITLE = B.TITLE
groupby B.CATEGORYNAME
orderby B.CATEGORYNAME;
| CATEGORYNAME |
MAXFF |
AVGFF |
MAXDT |
AVGDT |
MAXGK |
AVGGK |
| ADULTFIC |
||||||
| ADULTNF |
28 |
23.5 |
16 |
16 |
31 |
25.5 |
| ADULTREF |
8 |
8 |
||||
| CHILDRENFIC |
5 |
5 |
||||
| CHILDRENPIC |
14 |
14 |
The output now shows the borrowers across the top—Fred Fuller’s maximumcheckout time is the MaxFF column, Dorah Talbot’s is the MaxDT column, andGerhardt Kentgen’s is the MaxGK column. The output shows that the AdultNFcategory has the longest average checkout time for
each of the borrowers shown, and that in Gerhardt Kentgen’s case it issignificantly longer than the average checkout time in the other category hechecked out.
How was this report generated? In the query, the grouping is byCategoryName. The MaxFF column query is shown here:
（这个报表是如何生成的？在该查询中是通过CategoryName进行分组的。MaxFF列的查询如下所示：）
selectB.CategoryName,
MAX(DECODE(BC.Name, 'FRED FULLER',
BC.ReturnedDate-BC.CheckOutDate,NULL))MaxFF,
DECODE is performing an if-then check onthe data: If the BC.Name column value in a row is ‘FRED FULLER’, then calculatethe difference between the ReturnedDate and CheckOutDate;
else, return a NULL. That list of values is thenevaluated and the maximum value is returned.
A similar set of operations returns the average checkout time for FredFuller:
（DECODE 正对数据执行if-then 测试。如果某行的BC.NAME列的值为 FRED FULLER ，则计算ReturnedDate and CheckOutDate 之差；否则返回一个NULL值。然后计算该列表的值并返回最大值。返回Fred Fuller的平均借出时间的操作方法也类似）
AVG(DECODE(BC.Name,'FRED FULLER',
BC.ReturnedDate-BC.CheckOutDate,NULL))AvgFF,
自己补充：
一开始不是太明白为什么decode能够实现行转列，最后想起来sql的执行顺序以及中间虚拟的结果表，对理解这个还是有很大的帮助的。
这里加了一个中间过渡的结果集:
select BC.NAME
,B.CATEGORYNAME
,max(DECODE(BC.NAME, 'FRED FULLER', BC.RETURNEDDATE - BC.CHECKOUTDATE, null)) MAXFF
,avg(DECODE(BC.NAME, 'FRED FULLER', BC.RETURNEDDATE - BC.CHECKOUTDATE, null)) AVGFF
,max(DECODE(BC.NAME, 'DORAH TALBOT', BC.RETURNEDDATE -BC.CHECKOUTDATE, null)) MAXDT
,avg(DECODE(BC.NAME, 'DORAH TALBOT', BC.RETURNEDDATE -BC.CHECKOUTDATE, null)) AVGDT
,max(DECODE(BC.NAME, 'GERHARDT KENTGEN', BC.RETURNEDDATE -BC.CHECKOUTDATE, null)) MAXGK
,avg(DECODE(BC.NAME, 'GERHARDT KENTGEN', BC.RETURNEDDATE -BC.CHECKOUTDATE, null)) AVGGK
from BOOKSHELF_CHECKOUT BC
,BOOKSHELF B
where BC.TITLE = B.TITLE
groupby B.CATEGORYNAME
,BC.NAME
orderby B.CATEGORYNAME;
| NAME |
CATEGORYNAME |
MAXFF |
AVGFF |
MAXDT |
AVGDT |
MAXGK |
AVGGK |
| JED HOPKINS |
ADULTFIC |
||||||
| PAT LAVAY |
ADULTFIC |
||||||
| ROLAND BRANDT |
ADULTFIC |
||||||
| DORAH TALBOT |
ADULTNF |
16 |
16 |
||||
| FRED FULLER |
ADULTNF |
28 |
23.5 |
||||
| GERHARDT KENTGEN |
ADULTNF |
31 |
25.5 |
||||
| JED HOPKINS |
ADULTNF |
||||||
| PAT LAVAY |
ADULTNF |
||||||
| ROLAND BRANDT |
ADULTNF |
||||||
| DORAH TALBOT |
ADULTREF |
8 |
8 |
||||
| EMILY TALBOT |
CHILDRENFIC |
||||||
| GERHARDT KENTGEN |
CHILDRENFIC |
5 |
5 |
||||
| DORAH TALBOT |
CHILDRENPIC |
14 |
14 |
注：用黄色标记的是在书中没有用decode进行转换显示的
附：
SQL Select语句完整的执行顺序：
1、from子句组装来自不同数据源的数据；
2、where子句基于指定的条件对记录行进行筛选；
3、group by子句将数据划分为多个分组；
4、使用聚集函数进行计算，这里执行max()，avg()聚集函数；
5、使用having子句筛选分组；
6、计算所有的表达式；
7、select 的字段；
8、使用order by对结果集进行排序。
SQL语言不同于其他编程语言的最明显特征是处理代码的顺序。在大多数据库语言中，代码按编码顺序被处理。但在SQL语句中，第一个被处理的子句式FROM，而不是第一出现的SELECT。SQL查询处理的步骤序号：
(8)
1 SELECT (9) DISTINCT (11)<TOP_specification> <select_list>
2 (1) FROM <left_table>
3 (3) <join_type> JOIN<right_table>
4 (2) ON <join_condition>
5 (4) WHERE <where_condition>
6 (5) GROUP BY <group_by_list>
7 (6) WITH {CUBE | ROLLUP}
8 (7) HAVING <having_condition>
9 (10) ORDER BY <order_by_list>
以上每个步骤都会产生一个虚拟表，该虚拟表被用作下一个步骤的输入。这些虚拟表对调用者(客户端应用程序或者外部查询)不可用。只有最后一步生成的表才会会给调用者。如果没有在查询中指定某一个子句，将跳过相应的步骤。
逻辑查询处理阶段简介：
1、 FROM：对FROM子句中的前两个表执行笛卡尔积(交叉联接)，生成虚拟表VT1。
2、 ON：对VT1应用ON筛选器，只有那些使为真才被插入到TV2。
3、 OUTER (JOIN):如果指定了OUTER JOIN(相对于CROSS JOIN或INNER JOIN)，保留表中未找到匹配的行将作为外部行添加到VT2，生成TV3。如果FROM子句包含两个以上的表，则对上一个联接生成的结果表和下一个表重复执行步骤1到步骤3，直到处理完所有的表位置。
4、 WHERE：对TV3应用WHERE筛选器，只有使为true的行才插入TV4。
5、 GROUP BY：按GROUPBY子句中的列列表对TV4中的行进行分组，生成TV5。
6、 CUTE|ROLLUP：把超组插入VT5，生成VT6。
7、 HAVING：对VT6应用HAVING筛选器，只有使为true的组插入到VT7。
8、 SELECT：处理SELECT列表，产生VT8。
9、 DISTINCT：将重复的行从VT8中删除，产品VT9。
10、ORDER BY：将VT9中的行按ORDER BY子句中的列列表顺序，生成一个游标(VC10)。
11、TOP：从VC10的开始处选择指定数量或比例的行，生成表TV11，并返回给调用者。