Saturday, January 28, 2017

SQL101: Using Oracle Analytical function - range between unbounded preceding and 1 preceding

In SQL101 series, I will post some simple SQL related techniques. Most of them are posted in OTN time to time by me.
This SQL101 is related to Using Oracle Analytical function with range between unbounded preceding and 1 preceding, the problem was posted in OTN.
Test table and data setup:
First, I have used the following to create the test table and insert the sample data.

--Test table
create table daily_tr_tab
  (
    invid    number unique, --- though this is also not used in my query. if you need any unique column to give me a better solution.
    dates    date,
    purchase number,
    sales    number
  );
  

--Test data setup 
REM INSERTING into daily_tr_tab
SET DEFINE OFF;
Insert into daily_tr_tab (INVID,DATES,PURCHASE,SALES) values (1,to_date('22-DEC-15 00:00:00','DD-MON-RR hh24:mi:ss'),50,20);
Insert into daily_tr_tab (INVID,DATES,PURCHASE,SALES) values (2,to_date('23-DEC-15 00:00:00','DD-MON-RR hh24:mi:ss'),20,15);
Insert into daily_tr_tab (INVID,DATES,PURCHASE,SALES) values (3,to_date('24-DEC-15 00:00:00','DD-MON-RR hh24:mi:ss'),10,10);
Insert into daily_tr_tab (INVID,DATES,PURCHASE,SALES) values (4,to_date('25-DEC-15 00:00:00','DD-MON-RR hh24:mi:ss'),60,45);
Insert into daily_tr_tab (INVID,DATES,PURCHASE,SALES) values (5,to_date('26-DEC-15 00:00:00','DD-MON-RR hh24:mi:ss'),40,25);
Insert into daily_tr_tab (INVID,DATES,PURCHASE,SALES) values (6,to_date('27-DEC-15 00:00:00','DD-MON-RR hh24:mi:ss'),15,5);
Insert into daily_tr_tab (INVID,DATES,PURCHASE,SALES) values (7,to_date('28-DEC-15 00:00:00','DD-MON-RR hh24:mi:ss'),25,20);
Insert into daily_tr_tab (INVID,DATES,PURCHASE,SALES) values (8,to_date('29-DEC-15 00:00:00','DD-MON-RR hh24:mi:ss'),15,15);
Insert into daily_tr_tab (INVID,DATES,PURCHASE,SALES) values (9,to_date('30-DEC-15 00:00:00','DD-MON-RR hh24:mi:ss'),30,20);
Insert into daily_tr_tab (INVID,DATES,PURCHASE,SALES) values (10,to_date('31-DEC-15 00:00:00','DD-MON-RR hh24:mi:ss'),20,10);
Insert into daily_tr_tab (INVID,DATES,PURCHASE,SALES) values (11,to_date('01-JAN-16 00:00:00','DD-MON-RR hh24:mi:ss'),20,10);
Insert into daily_tr_tab (INVID,DATES,PURCHASE,SALES) values (12,to_date('02-JAN-16 00:00:00','DD-MON-RR hh24:mi:ss'),30,25);
Insert into daily_tr_tab (INVID,DATES,PURCHASE,SALES) values (13,to_date('03-JAN-16 00:00:00','DD-MON-RR hh24:mi:ss'),25,0);
Insert into daily_tr_tab (INVID,DATES,PURCHASE,SALES) values (14,to_date('04-JAN-16 00:00:00','DD-MON-RR hh24:mi:ss'),35,25);
Insert into daily_tr_tab (INVID,DATES,PURCHASE,SALES) values (15,to_date('05-JAN-16 00:00:00','DD-MON-RR hh24:mi:ss'),15,10);
Insert into daily_tr_tab (INVID,DATES,PURCHASE,SALES) values (16,to_date('06-JAN-16 00:00:00','DD-MON-RR hh24:mi:ss'),20,15);
Insert into daily_tr_tab (INVID,DATES,PURCHASE,SALES) values (17,to_date('07-JAN-16 00:00:00','DD-MON-RR hh24:mi:ss'),25,10);
Insert into daily_tr_tab (INVID,DATES,PURCHASE,SALES) values (18,to_date('08-JAN-16 00:00:00','DD-MON-RR hh24:mi:ss'),30,20);
Insert into daily_tr_tab (INVID,DATES,PURCHASE,SALES) values (19,to_date('09-JAN-16 00:00:00','DD-MON-RR hh24:mi:ss'),35,30);
Insert into daily_tr_tab (INVID,DATES,PURCHASE,SALES) values (20,to_date('10-JAN-16 00:00:00','DD-MON-RR hh24:mi:ss'),20,20);
Insert into daily_tr_tab (INVID,DATES,PURCHASE,SALES) values (21,to_date('11-JAN-16 00:00:00','DD-MON-RR hh24:mi:ss'),30,5);
Insert into daily_tr_tab (INVID,DATES,PURCHASE,SALES) values (22,to_date('12-JAN-16 00:00:00','DD-MON-RR hh24:mi:ss'),40,0);
Insert into daily_tr_tab (INVID,DATES,PURCHASE,SALES) values (23,to_date('13-JAN-16 00:00:00','DD-MON-RR hh24:mi:ss'),50,20);
Insert into daily_tr_tab (INVID,DATES,PURCHASE,SALES) values (24,to_date('14-JAN-16 00:00:00','DD-MON-RR hh24:mi:ss'),12,10);
Insert into daily_tr_tab (INVID,DATES,PURCHASE,SALES) values (25,to_date('15-JAN-16 00:00:00','DD-MON-RR hh24:mi:ss'),13,10);
Insert into daily_tr_tab (INVID,DATES,PURCHASE,SALES) values (26,to_date('16-JAN-16 00:00:00','DD-MON-RR hh24:mi:ss'),14,10);
Insert into daily_tr_tab (INVID,DATES,PURCHASE,SALES) values (27,to_date('17-JAN-16 00:00:00','DD-MON-RR hh24:mi:ss'),24,5);
Insert into daily_tr_tab (INVID,DATES,PURCHASE,SALES) values (28,to_date('18-JAN-16 00:00:00','DD-MON-RR hh24:mi:ss'),34,0);
Insert into daily_tr_tab (INVID,DATES,PURCHASE,SALES) values (29,to_date('19-JAN-16 00:00:00','DD-MON-RR hh24:mi:ss'),35,25);
insert into daily_tr_tab (invid,dates,purchase,sales) values (30,to_date('20-JAN-16 00:00:00','DD-MON-RR hh24:mi:ss'),40,10);

The output needed:
  1. DATES     OPENING_QTY   PURCHASE      SALES CLOSING_QTY  
  2. --------- ----------- ---------- ---------- -----------  
  3. 01-JAN-16         100         20         10         110  
  4. 02-JAN-16         110         30         25         115  
  5. 03-JAN-16         115         25          0         140  
  6. 04-JAN-16         140         35         25         150  
  7. 05-JAN-16         150         15         10         155  
  8. 06-JAN-16         155         20         15         160  
  9. 07-JAN-16         160         25         10         175  
  10. 08-JAN-16         175         30         20         185  
  11. 09-JAN-16         185         35         30         190  
  12. 10-JAN-16         190         20         20         190  
  13. 11-JAN-16         190         30          5         215  
 Opening quantity in a date is total of previous purchase minus total sales. The closing balance is opening quantity of that date minus the sale.

The solution is:
 
SCOTT@orclSB 28-JAN-17> ed
Wrote file afiedt.buf

  1  select *
  2  from
  3    (select dates,
  4        sum(purchase) over(order by dates range between unbounded preceding and 1 preceding )
  5      - sum(sales) over(order by dates range between unbounded preceding and 1 preceding) opening_qty ,
  6      purchase,
  7      sales ,
  8        sum(purchase) over(order by dates range between unbounded preceding and 1 preceding)
  9      - sum(sales) over(order by dates range between unbounded preceding and 1 preceding) + purchase - sales closing_qty
 10    from daily_tr_tab
 11    )
 12  where dates >= to_date('01/01/16', 'dd/mm/rr')
 13* order by dates
SCOTT@orclSB 28-JAN-17> /

DATES     OPENING_QTY   PURCHASE      SALES CLOSING_QTY
--------- ----------- ---------- ---------- -----------
01-JAN-16         100         20         10         110
02-JAN-16         110         30         25         115
03-JAN-16         115         25          0         140
04-JAN-16         140         35         25         150
05-JAN-16         150         15         10         155
06-JAN-16         155         20         15         160
07-JAN-16         160         25         10         175
08-JAN-16         175         30         20         185
09-JAN-16         185         35         30         190
10-JAN-16         190         20         20         190
11-JAN-16         190         30          5         215

DATES     OPENING_QTY   PURCHASE      SALES CLOSING_QTY
--------- ----------- ---------- ---------- -----------
12-JAN-16         215         40          0         255
13-JAN-16         255         50         20         285
14-JAN-16         285         12         10         287
15-JAN-16         287         13         10         290
16-JAN-16         290         14         10         294
17-JAN-16         294         24          5         313
18-JAN-16         313         34          0         347
19-JAN-16         347         35         25         357
20-JAN-16         357         40         10         387

20 rows selected.

SCOTT@orclSB 28-JAN-17> 

SQL101: Using sequence.nextval with group and analytical function.

In SQL101 series, I will post some simple SQL related techniques. Most of them are posted in OTN time to time by me.
This SQL101 is related to Using sequence.nextval with group and analytical function, the problem was posted in OTN.
The problem with sequence.nextval is it increments within SQL statement even within the group. Carefully watch this test case:
SCOTT@orclSB 28-JAN-17> --My database version.
SCOTT@orclSB 28-JAN-17> select * from v$version;

BANNER
--------------------------------------------------------------------------------
Oracle Database 11g Enterprise Edition Release 11.2.0.1.0 - 64bit Production
PL/SQL Release 11.2.0.1.0 - Production
CORE    11.2.0.1.0      Production
TNS for Linux: Version 11.2.0.1.0 - Production
NLSRTL Version 11.2.0.1.0 - Production

SCOTT@orclSB 28-JAN-17> ed
Wrote file afiedt.buf

  1  --Creating the sequence.
  2* create sequence file_seq minvalue 1 maxvalue 9999999999999999999999999999 increment by 1 start with 100 cache 20 noorder nocycle
SCOTT@orclSB 28-JAN-17> /

Sequence created.

SCOTT@orclSB 28-JAN-17> ed
Wrote file afiedt.buf

  1  --Creating the test table.
  2  create table test4
  3    ( file_number number, seq_no number
  4*   )
SCOTT@orclSB 28-JAN-17> /

Table created.

SCOTT@orclSB 28-JAN-17> ed
Wrote file afiedt.buf

  1  --Inserting my test data.
  2  insert into test4
  3  select 10,null from dual
  4  union all
  5  select 10,null from dual
  6  union all
  7  select 10,null from dual
  8  union all
  9  select 20,null from dual
 10  union all
 11  select 20,null from dual
 12  union all
 13* select 30,null from dual
SCOTT@orclSB 28-JAN-17> /

6 rows created.

SCOTT@orclSB 28-JAN-17> commit;

Commit complete.

SCOTT@orclSB 28-JAN-17> select file_number,dense_rank() over (order by file_number) dnk from test4;

FILE_NUMBER        DNK
----------- ----------
         10          1
         10          1
         10          1
         20          2
         20          2
         30          3

6 rows selected.

SCOTT@orclSB 28-JAN-17> 
 
 
Required Output
FILE_NUMBER, DNK
10                          100
10                          100
10                          100
20                          101
20                          101
30                          103

Rules:
If you noticed that I created sequence name as file_SEQ, system should  user this sequence to generate dnk column  and if file_number change then get next sequence. So far I created the below query but it is not working.
SCOTT@orclSB 28-JAN-17> ed
Wrote file afiedt.buf

  1  select file_number,
  2    dense_rank() over (order by file_number) + file_seq.nextval dnk
  3* from test4
SCOTT@orclSB 28-JAN-17> /

FILE_NUMBER        DNK
----------- ----------
         10        101
         10        102
         10        103
         20        105
         20        106
         30        108

6 rows selected.

SCOTT@orclSB 28-JAN-17> 
Now, you can see the problem, it is creating counting without grouping of file_number.
This is the actual solution of this problem:

SCOTT@orclSB 28-JAN-17> ed
Wrote file afiedt.buf

  1  create or replace function retfileseq(
  2      pi_indicator integer)
  3    return number
  4  as
  5  begin
  6    if pi_indicator = 1 then
  7      return (file_seq.nextval);
  8    else
  9      return (file_seq.currval);
 10    end if;
 11* end retfileseq;
SCOTT@orclSB 28-JAN-17> /

Function created.

SCOTT@orclSB 28-JAN-17> ed
Wrote file afiedt.buf

  1  select file_number ,
  2    case
  3      when lag(file_number) over(order by file_number) is null
  4      or file_number <> lag(file_number) over(order by file_number)
  5      then retfileseq(1)
  6      else retfileseq(0)
  7    end as file_seq
  8* from test4
SCOTT@orclSB 28-JAN-17> /

FILE_NUMBER   FILE_SEQ
----------- ----------
         10        106
         10        106
         10        106
         20        107
         20        107
         30        108

6 rows selected.

SCOTT@orclSB 28-JAN-17> 

SQL101 : Oracle Check constraint to prevent alphanumeric value in column

In SQL101 series, I will post some simple SQL related techniques. Most of them are posted in OTN time to time by me.
This SQL101 is related to check constraint. We can implement check constraint very efficiently for many of our problems, sometimes people tends to use triggers for those, but remember trigger is not a good option in most of the time. You may want to read this article from Tom Kyte regarding trigger : Ask Tom: The Trouble with Triggers..
I will describe two scenarios, first we have a table with whole_name column, So we want to restrict bad entries in that column and we want to allow only alphabet and space as name can not be alphanumeric or purely numeric!
Second, we have a table with first_name and last_name column, So we want to allow only alphabet here, no numeric and no space.
 
SCOTT@orclSB 28-JAN-17> --My database version.
SCOTT@orclSB 28-JAN-17> select * from v$version;

BANNER
--------------------------------------------------------------------------------
Oracle Database 11g Enterprise Edition Release 11.2.0.1.0 - 64bit Production
PL/SQL Release 11.2.0.1.0 - Production
CORE    11.2.0.1.0      Production
TNS for Linux: Version 11.2.0.1.0 - Production
NLSRTL Version 11.2.0.1.0 - Production

SCOTT@orclSB 28-JAN-17> ed
Wrote file afiedt.buf

  1  -- In the whole_name column, we will only allow alphabet and space.
  2  CREATE TABLE contacts
  3      (
  4      whole_name varchar2(30) constraint chk_whole_name check (regexp_like (whole_name,'^[[:alpha:][:space:]]+$'))
  5*     )
SCOTT@orclSB 28-JAN-17> /

Table created.

SCOTT@orclSB 28-JAN-17> --This should work.
SCOTT@orclSB 28-JAN-17> insert into contacts values ('Saubhik');

1 row created.

SCOTT@orclSB 28-JAN-17> --This should not work as 123 is numeric.
SCOTT@orclSB 28-JAN-17> insert into contacts values ('Saubhik123');
insert into contacts values ('Saubhik123')
*
ERROR at line 1:
ORA-02290: check constraint (SCOTT.CHK_WHOLE_NAME) violated


SCOTT@orclSB 28-JAN-17> --This should not work as 1234 is numeric.
SCOTT@orclSB 28-JAN-17> insert into contacts values ('1234');
insert into contacts values ('1234')
*
ERROR at line 1:
ORA-02290: check constraint (SCOTT.CHK_WHOLE_NAME) violated


SCOTT@orclSB 28-JAN-17> --This should work as only alphabet and space.
SCOTT@orclSB 28-JAN-17> insert into contacts values ('Saubhik Banerjee');

1 row created.

SCOTT@orclSB 28-JAN-17> drop table contacts purge;

Table dropped.

SCOTT@orclSB 28-JAN-17> ed
Wrote file afiedt.buf

  1  -- Now, there is no whole_name column, instead we have broken down as first_name and last_name.
  2  -- So, Only alphabet should be allowed, no numeric character and no space.
  3  CREATE TABLE contacts
  4      (
  5      first_name varchar2(30) constraint chk_first_name check (regexp_like (first_name,'^[[:alpha:]]+$'))
  6*     )
SCOTT@orclSB 28-JAN-17> /

Table created.

SCOTT@orclSB 28-JAN-17> --This should work.
SCOTT@orclSB 28-JAN-17> insert into contacts values ('Saubhik');

1 row created.

SCOTT@orclSB 28-JAN-17> --This should not work as 123 is numeric.
SCOTT@orclSB 28-JAN-17> insert into contacts values ('Saubhik123');
insert into contacts values ('Saubhik123')
*
ERROR at line 1:
ORA-02290: check constraint (SCOTT.CHK_FIRST_NAME) violated


SCOTT@orclSB 28-JAN-17> --This should not work as 1234 is numeric.
SCOTT@orclSB 28-JAN-17> insert into contacts values ('1234');
insert into contacts values ('1234')
*
ERROR at line 1:
ORA-02290: check constraint (SCOTT.CHK_FIRST_NAME) violated


SCOTT@orclSB 28-JAN-17> --This should not work as space is not allowed within first_name.
SCOTT@orclSB 28-JAN-17> insert into contacts values ('Saubhik Banerjee');
insert into contacts values ('Saubhik Banerjee')
*
ERROR at line 1:
ORA-02290: check constraint (SCOTT.CHK_FIRST_NAME) violated


SCOTT@orclSB 28-JAN-17> drop table contacts purge;

Table dropped.

SCOTT@orclSB 28-JAN-17>