SQLPlus 中的时间维度表

Question

I'm doing a datawarehouse project for college with Oracle Database (SQLPlus). I need to create the time dimension table and to populate it. The table needs to be like this:

It needs to go from 2004 to 2019.

I've tried different things and queries that I've found but they don't works and, sadly, I don't know enough about SQLPlus to create one on my own (or to successfully modify one). I'm completely lost.

Thank you very much for your help and patience.

Answer 1

Do not store all the columns; use virtual columns instead to calculate derived data otherwise you will find that your columns could be inconsistent:

CREATE TABLE table_name (
  id             NUMBER(10,0)
                 GENERATED ALWAYS AS IDENTITY
                 CONSTRAINT table_name__id__pk PRIMARY KEY,
  "DATE"         DATE
                 CONSTRAINT table_name__date__nn NOT NULL
                 CONSTRAINT table_name__date__u UNIQUE
                 CONSTRAINT table_name__date__chk CHECK ( "DATE" = TRUNC( "DATE" ) ),
  id_day_of_week NUMBER(1,0) 
                 GENERATED ALWAYS AS ( "DATE" - TRUNC( "DATE", 'IW' ) + 1 ),
  day_of_week    VARCHAR2(9)
                 GENERATED ALWAYS AS ( CAST( TRIM( TO_CHAR( "DATE", 'DAY', 'NLS_DATE_LANGUAGE = AMERICAN' ) ) AS VARCHAR2(9) ) ),
  is_holiday     NUMBER(1,0)
                 CONSTRAINT table_name__id_holiday__chk CHECK ( is_holiday IN ( 0, 1 ) ),
  id_month       NUMBER(2,0)
                 GENERATED ALWAYS AS ( EXTRACT( MONTH FROM "DATE" ) ),
  month          VARCHAR2(9)
                 GENERATED ALWAYS AS ( CAST( TRIM( TO_CHAR( "DATE", 'MONTH', 'NLS_DATE_LANGUAGE = AMERICAN' ) ) AS VARCHAR2(9) ) ),
  id_year        NUMBER(5,0)
                 GENERATED ALWAYS AS ( EXTRACT( YEAR FROM "DATE" ) ),
  id_total       NUMBER(1,0)
                 GENERATED ALWAYS AS ( 1 ),
  total          CHAR(5)
                 GENERATED ALWAYS AS ( 'Total' )
);

Note:

• You should not name the column DATE as its a keyword and you will need to surround it in double-quotes and use the same case every time you use it.
• The id_day_of_week is based on the day of the ISO8601 week because relying on TO_CHAR( "DATE", 'D' ) depends on the NLS_TERRITORY setting as to which day of the week is the first day; this way it is independent of any settings.
• The day_of_week and month columns have a fixed language.
• It is unclear what id_total and total should contain so these are generated as literal values; if you want to have non-static data in these columns then remove the GENERATED... part of the declaration.

Then you can populate it using:

INSERT INTO table_name ( "DATE", is_holiday )
SELECT DATE '2004-01-01' + LEVEL - 1, 0
FROM  DUAL
CONNECT BY DATE '2004-01-01' + LEVEL - 1 < DATE '2020-01-01';

And update the holiday dates using an UPDATE statement according to your territory.

Then if you do:

SELECT *
FROM   table_name
ORDER BY "DATE" ASC
FETCH FIRST 32 ROWS ONLY;

The output is:

 ID | DATE | ID_DAY_OF_WEEK |  DAY_OF_WEEK |  IS_HOLIDAY |  ID_MONTH |  MONTH | ID_YEAR |  ID_TOTAL |  TOTAL -: |:-------- | -------------: |:---------- |  ---------: |  -------: |:------- |  ------: |  -------: |:---- 1 |  01-JAN-04 |  4 |  THURSDAY | 0 |  1 |  JANUARY | 2004 |  1 |  Total 2 | 02-JAN-04 |  5 |  FRIDAY | 0 |  1 |  JANUARY | 2004 |  1 |  Total 3 | 03-JAN-04 |  6 |  SATURDAY | 0 |  1 |  JANUARY | 2004 |  1 |  Total 4 | 04-JAN-04 |  7 |  SUNDAY | 0 |  1 |  JANUARY | 2004 |  1 |  Total 5 | 05-JAN-04 |  1 |  MONDAY | 0 |  1 |  JANUARY | 2004 |  1 |  Total 6 | 06-JAN-04 |  2 |  TUESDAY | 0 |  1 |  JANUARY | 2004 |  1 |  Total 7 | 07-JAN-04 |  3 |  WEDNESDAY | 0 |  1 |  JANUARY | 2004 |  1 |  Total 8 | 08-JAN-04 |  4 |  THURSDAY | 0 |  1 |  JANUARY | 2004 |  1 |  Total 9 | 09-JAN-04 |  5 |  FRIDAY | 0 |  1 |  JANUARY | 2004 |  1 |  Total 10 | 10-JAN-04 |  6 |  SATURDAY | 0 |  1 |  JANUARY | 2004 |  1 |  Total 11 | 11-JAN-04 |  7 |  SUNDAY | 0 |  1 |  JANUARY | 2004 |  1 |  Total 12 | 12-JAN-04 |  1 |  MONDAY | 0 |  1 |  JANUARY | 2004 |  1 |  Total 13 | 13-JAN-04 |  2 |  TUESDAY | 0 |  1 |  JANUARY | 2004 |  1 |  Total 14 | 14-JAN-04 |  3 |  WEDNESDAY | 0 |  1 |  JANUARY | 2004 |  1 |  Total 15 | 15-JAN-04 |  4 |  THURSDAY | 0 |  1 |  JANUARY | 2004 |  1 |  Total 16 | 16-JAN-04 |  5 |  FRIDAY | 0 |  1 |  JANUARY | 2004 |  1 |  Total 17 | 17-JAN-04 |  6 |  SATURDAY | 0 |  1 |  JANUARY | 2004 |  1 |  Total 18 | 18-JAN-04 |  7 |  SUNDAY | 0 |  1 |  JANUARY | 2004 |  1 |  Total 19 | 19-JAN-04 |  1 |  MONDAY | 0 |  1 |  JANUARY | 2004 |  1 |  Total 20 | 20-JAN-04 |  2 |  TUESDAY | 0 |  1 |  JANUARY | 2004 |  1 |  Total 21 | 21-JAN-04 |  3 |  WEDNESDAY | 0 |  1 |  JANUARY | 2004 |  1 |  Total 22 | 22-JAN-04 |  4 |  THURSDAY | 0 |  1 |  JANUARY | 2004 |  1 |  Total 23 | 23-JAN-04 |  5 |  FRIDAY | 0 |  1 |  JANUARY | 2004 |  1 |  Total 24 | 24-JAN-04 |  6 |  SATURDAY | 0 |  1 |  JANUARY | 2004 |  1 |  Total 25 | 25-JAN-04 |  7 |  SUNDAY | 0 |  1 |  JANUARY | 2004 |  1 |  Total 26 | 26-JAN-04 |  1 |  MONDAY | 0 |  1 |  JANUARY | 2004 |  1 |  Total 27 | 27-JAN-04 |  2 |  TUESDAY | 0 |  1 |  JANUARY | 2004 |  1 |  Total 28 | 28-JAN-04 |  3 |  WEDNESDAY | 0 |  1 |  JANUARY | 2004 |  1 |  Total 29 | 29-JAN-04 |  4 |  THURSDAY | 0 |  1 |  JANUARY | 2004 |  1 |  Total 30 | 30-JAN-04 |  5 |  FRIDAY | 0 |  1 |  JANUARY | 2004 |  1 |  Total 31 | 31-JAN-04 |  6 |  SATURDAY | 0 |  1 |  JANUARY | 2004 |  1 |  Total 32 | 01-FEB-04 |  7 |  SUNDAY | 0 |  2 |  FEBRUARY | 2004 |  1 |  Total

db<>fiddle here

Answer 2

create table date_dim
(id number(38),
date date,
id_dayofweek number(38),
dayofweek varchar(100),
id_holiday number(38),
 id_month number(38),
 month varchar(100),  
 id_year number(38),
 id_total number(38),
 Total varchar(100));

Use above query to create the table.

Regarding the data, you can generate it through connect by clause.

insert into date_dim
(select level as id, to_date('31-DEC-2003', 'DD-MON-YYYY') + level as date1,   
case when ltrim(rtrim(to_char(to_date('31-DEC-2003', 'DD-MON-YYYY') + level, 'Day'))) = 'Monday' then 2
when ltrim(rtrim(to_char(to_date('31-DEC-2003', 'DD-MON-YYYY') + level, 'Day'))) = 'Tuesday' then 3
when ltrim(rtrim(to_char(to_date('31-DEC-2003', 'DD-MON-YYYY') + level, 'Day'))) = 'Wednesday' then 4
when ltrim(rtrim(to_char(to_date('31-DEC-2003', 'DD-MON-YYYY') + level, 'Day'))) = 'Thursday' then 5
when ltrim(rtrim(to_char(to_date('31-DEC-2003', 'DD-MON-YYYY') + level, 'Day'))) = 'Friday' then 6
when ltrim(rtrim(to_char(to_date('31-DEC-2003', 'DD-MON-YYYY') + level, 'Day'))) = 'Saturday' then 7
when ltrim(rtrim(to_char(to_date('31-DEC-2003', 'DD-MON-YYYY') + level, 'Day'))) = 'Sunday' then 1 end as id_dayofweek,
to_char(to_date('31-DEC-2003', 'DD-MON-YYYY') + level, 'Day') as dayofweek,
0 as id_holiday,
to_char(to_date('31-DEC-2003', 'DD-MON-YYYY') + level, 'MM') as id_month,
to_char(to_date('31-DEC-2003', 'DD-MON-YYYY') + level, 'Month') as month,
to_char(to_date('31-DEC-2003', 'DD-MON-YYYY') + level, 'YYYY') as year,
1 as id_total,
'Total' as Total
from dual
connect by level < = 5844);