Oracle hierarchy query - sql

Scoured the interwebs for a few days looking for a solution to this hierarchy riddle but no luck. All solutions I've found have either a row with parent ID null, have only one ultimate parent, or I'm just missing some piece of knowledge that's preventing me from adapting the solutions for the various needs out there to fit my need.
Data appears like so:
SELECT 3225 PARENT_ID,'TYPE2_A' PARENT,3227 CHILD_ID,'TYPE2_C' CHILD FROM DUAL UNION ALL
SELECT 148,'TYPE2_H',150,'TYPE2_G' FROM DUAL UNION ALL
SELECT 3225,'TYPE2_A',3226,'TYPE2_B' FROM DUAL UNION ALL
SELECT 3227,'TYPE2_C',3222,'TYPE3_E' FROM DUAL UNION ALL
SELECT 3220,'TYPE3_D',3221,'TYPE3_I' FROM DUAL UNION ALL
SELECT 3226,'TYPE2_B',3220,'TYPE3_D' FROM DUAL UNION ALL
SELECT 2379,'TYPE1_K',148,'TYPE2_H' FROM DUAL UNION ALL
SELECT 113,'TYPE1_L',91,'TYPE3_F' FROM DUAL UNION ALL
SELECT 148,'TYPE2_H',128,'TYPE3_N' FROM DUAL UNION ALL
SELECT 3223,'TYPE1_J',3226,'TYPE2_B' FROM DUAL UNION ALL
SELECT 2379,'TYPE1_K',150,'TYPE2_G' FROM DUAL UNION ALL
SELECT 150,'TYPE2_G',91,'TYPE3_F' FROM DUAL UNION ALL
SELECT 2487,'TYPE1_A',3225,'TYPE2_A' FROM DUAL UNION ALL
SELECT 98,'TYPE1_M',91,'TYPE3_F' FROM DUAL;
TYPE1 can never be a child, it's always a parent.
TYPE1 can be a parent of either TYPE2 or TYPE3.
TYPE2 can be a parent of either TYPE2 or TYPE3.
TYPE2 can never be a child of TYPE3.
TYPE3 can be a parent of TYPE3.
TYPE2 and TYPE3 always have at least one TYPE1 ultimate parent but can have more than one TYPE1 parent.
The goal of the query is to take a distinct list of all TYPE2 and TYPE3 children like so...
SELECT PARENT FROM
(query from above)
WHERE PARENT LIKE 'TYPE2%' OR PARENT LIKE 'TYPE3%'
UNION
SELECT CHILD FROM
(query from above)
WHERE CHILD LIKE 'TYPE2%' OR CHILD LIKE 'TYPE3%';
...and find who their ultimate TYPE1 parent(s) is.
For example, based on the data set above...
TYPE3_F has ultimate parents TYPE1_L and TYPE1_M
TYPE3_E has ultimate parent TYPE1_A
TYPE2_B has ultimate parents TYPE1_A and TYPE1_J
The result set for inputs TYPE3_F, TYPE3_E, and TYPE2_B would appear like...
Child Ultimate_Parent
TYPE3_F TYPE1_L
TYPE3_F TYPE1_M
TYPE3_E TYPE1_A
TYPE2_B TYPE1_A
TYPE2_B TYPE1_J

I'm using the connect_by_root function to obtain the parent at the highest level, no matter where in the tree you started.
with parent_child_data
as
(SELECT 2487 PARENT_ID,'TYPE1_A' PARENT, 3225 CHILD_ID,'TYPE2_A' CHILD FROM DUAL UNION ALL
SELECT 3225 ,'TYPE2_A', 3226 ,'TYPE2_B' FROM DUAL UNION ALL
SELECT 3225 ,'TYPE2_A', 3227 ,'TYPE2_C' FROM DUAL UNION ALL
SELECT 3226 ,'TYPE2_B', 3220 ,'TYPE3_D' FROM DUAL UNION ALL
SELECT 3227 ,'TYPE2_C', 3222 ,'TYPE3_E' FROM DUAL UNION ALL
SELECT 113 ,'TYPE1_L', 91 ,'TYPE3_F' FROM DUAL UNION ALL
SELECT 98 ,'TYPE1_M', 91 ,'TYPE3_F' FROM DUAL UNION ALL
SELECT 2379 ,'TYPE1_K', 150 ,'TYPE2_G' FROM DUAL UNION ALL
SELECT 2379 ,'TYPE1_K', 148 ,'TYPE2_H' FROM DUAL UNION ALL
SELECT 150 ,'TYPE2_G', 1291 ,'TYPE3_F' FROM DUAL UNION ALL
SELECT 148 ,'TYPE2_H', 150 ,'TYPE2_G' FROM DUAL UNION ALL
SELECT 148 ,'TYPE2_H', 128 ,'TYPE3_N' FROM DUAL UNION ALL
SELECT 150 ,'TYPE2_G', 1291 ,'TYPE3_F' FROM DUAL UNION ALL
SELECT 3220 ,'TYPE3_D', 3221 ,'TYPE3_I' FROM DUAL UNION ALL
SELECT 3226 ,'TYPE2_B', 3220 ,'TYPE3_D' FROM DUAL UNION ALL
SELECT 3223 ,'TYPE1_J', 3226 ,'TYPE2_B' FROM DUAL UNION ALL
SELECT 3225 ,'TYPE2_A', 3226 ,'TYPE2_B' FROM DUAL UNION ALL
SELECT 2487 ,'TYPE1_A', 3225 ,'TYPE2_A' FROM DUAL)
select * from (
select d.child_id, d.child, d.parent_id, d.parent, connect_by_root d.parent ultimate_parent
from parent_child_data d
where child like 'TYPE3%' or child like 'TYPE2%'
connect by prior child_id = parent_id)
where ultimate_parent like 'TYPE1%'
;

Related

Generate Result based on max count in secondary column after a join

I have two tables which have a common key between them, and quite a lot of other important infos ; for the sake of simplicity i will be using Combination A and Combination B. When a combination is met, whichever table has the maximum number of records should be the source where i collect the information ; in this case say IDs. The priority when counts are same is Table1.
COMMONKEY column is the combination/join condition in my tables.
(Table 1)
SELECT '123' table1_id,'Comb A' commonkey from dual UNION
SELECT '124' table1_id,'Comb A' commonkey from dual UNION
SELECT '125' table1_id,'Comb A' commonkey from dual UNION
SELECT '126' table1_id,'Comb A' commonkey from dual UNION
SELECT '215' table1_id,'Comb B' commonkey from dual UNION
SELECT '216' table1_id,'Comb B' commonkey from dual UNION
SELECT '559' table1_id,'Random Combination 1' commonkey from dual UNION
SELECT '560' table1_id,'Random Combination 2' commonkey from dual ;
( Table 2 )
SELECT 'abc1' table2_id,'Comb A' commonkey from dual UNION
SELECT 'abc2' table2_id,'Comb A' commonkey from dual UNION
SELECT 'abc3' table2_id,'Comb A' commonkey from dual UNION
SELECT 'abc4' table2_id,'Comb A' commonkey from dual UNION
SELECT 'xyz1' table2_id,'Comb B' commonkey from dual UNION
SELECT 'xyz2' table2_id,'Comb B' commonkey from dual UNION
SELECT 'xyz3' table2_id,'Comb B' commonkey from dual UNION
SELECT 'xyz2' table2_id,'Comb B' commonkey from dual UNION
SELECT '416abc1' table2_id,'Random Combination 91' commonkey from dual UNION
SELECT '416abc2' table2_id,'Random Combination 92' commonkey from dual;
Result Set Expected :
ID COMMONKEY
123 Comb A
124 Comb A
125 Comb A
126 Comb A
xyz1 Comb B
xyz2 Comb B
xyz3 Comb B
559 Random Combination 1
560 Random Combination 1
416abc1 Random Combination 91
416abc2 Random Combination 92
Updated Image :
( the image shows a screenshot of the trail data in an excel; The Requirement and Strategy are color mapped to make it quickly understandable )
I need to generate the result set using SQL as follows :
When table1.commonkey = table2.commonkey hits, I need to-
If table1 has 10 IDs, table2 has 5 IDs -> Pick 10 IDs from table1.
If table1 has 15 IDs, table2 has 30 IDs -> Pick 30 IDs from table2.
If table1 has 4 IDs, table2 has 4 IDs -> Pick 4 IDs from table1.
( when equal, choose table1 IDs )
When no matches occur with the common key, prevent a cross join and add in the rowsets linearly to the result table.
Edit : I've initially gone on routes with
a left join b where b.key IS null ;
a full outer join b where b.key IS NULL or a.key is NULL ;
to achieve workarounds with A-B or B-A result sets but both these approaches were quite wrong. Gathering Delta sets or Exclusion sets didnt go well.
Here's one option; see comments within code
SQL> with
2 -- sample data
3 a (id, ckey) as
4 (select '123', 'ca' from dual union all
5 select '124', 'ca' from dual union all
6 select '125', 'ca' from dual union all
7 select '126', 'ca' from dual union all
8 select '215', 'cb' from dual union all
9 select '216', 'cb' from dual union all
10 select '551', 'r1' from dual union all
11 select '552', 'r2' from dual
12 ),
13 b (id, ckey) as
14 (select 'abc1', 'ca' from dual union all
15 select 'abc2', 'ca' from dual union all
16 select 'abc3', 'ca' from dual union all
17 select 'abc4', 'ca' from dual union all
18 select 'xyz1', 'cb' from dual union all
19 select 'xyz2', 'cb' from dual union all
20 select 'xyz3', 'cb' from dual union all
21 select '9991', 'r3' from dual union all
22 select '9992', 'r4' from dual
23 ),
24 -- count rows per each CKEY (common key)
25 tempa as
26 (select id, ckey, count(*) over (partition by ckey) cnt
27 from a
28 ),
29 tempb as
30 (select id, ckey, count(*) over (partition by ckey) cnt
31 from b
32 )
33 -- final query
34 select distinct
35 case when a.cnt >= b.cnt then a.id
36 else b.id
37 end id,
38 a.ckey
39 from tempa a join tempb b on b.ckey = a.ckey
40 union all
41 select ckey, id from a
42 where not exists (select null from b where a.ckey = b.ckey)
43 union all
44 select ckey, id from b
45 where not exists (select null from a where a.ckey = b.ckey)
46 order by 1, 2;
which results in
ID CKEY
---- -----
r1 551
r2 552
r3 9991
r4 9992
xyz1 cb
xyz2 cb
xyz3 cb
123 ca
124 ca
125 ca
126 ca
11 rows selected.
SQL>

How to count distinct flags in the sql

BELOW IS SNIPPET OF MY DATA
Here is the sample creation code of testing.
CREATE TABLE MYGROUP ( Category,PERSON,Flag ) AS
SELECT 'Cat1','A','1' FROM DUAL
UNION ALL SELECT 'Cat1','A','0' FROM DUAL
UNION ALL SELECT 'Cat1','A','1' FROM DUAL
UNION ALL SELECT 'Cat1','B','1' FROM DUAL
UNION ALL SELECT 'Cat1','B','0' FROM DUAL
UNION ALL SELECT 'Cat2','A','0' FROM DUAL
UNION ALL SELECT 'Cat2','A','0' FROM DUAL
UNION ALL SELECT 'Cat2','A','0' FROM DUAL
UNION ALL SELECT 'Cat2','B','1' FROM DUAL
UNION ALL SELECT 'Cat2','B','1' FROM DUAL
UNION ALL SELECT 'Cat2','B','0' FROM DUAL
UNION ALL SELECT 'Cat3','X','0' FROM DUAL
UNION ALL SELECT 'Cat3','Y','0' FROM DUAL;
Desired Output:
Category - Count of Distinct Persons with Flag = 1
Cat1 - 2
Cat2 - 1
Cat3 - 0
I need to get my code in Big query to get distinct counts of persons. It shouldnt double count.
You can use conditional aggregation
SELECT
Category,
COUNT(DISTINCT CASE WHEN Flag = 1 THEN PERSON END)
FROM MYGROUP
GROUP BY Category;

How to make dual table with many rows in one col?

I tried to do so, but it returns me in columns not rows
SELECT 386 ,417,420,421,422,423 ... from dual
Try this:
select column_value col1 from
table(sys.odcinumberlist((386) ,(417),(420),(421),(422),(423)))
use union all
SELECT 386 from dual
union all
SELECT 417 from dual
union all
SELECT 420 from dual
union all
SELECT 386 from dual
You can use UNION or UNION ALL as a work around to get the desired output.
SELECT 386 from dual
UNION ALL
SELECT 417 from dual
UNION ALL
SELECT 420 from dual
...

How to upside down the result rows of a select statment in oracle?

I can select a list of rows from a table. but I want to show them by swapping upside down.
Explaination:
with table1 as
(
select 1 ID, 'txt1' value from dual
union all
select 2, 'txt2' from dual
union all
select 7, 'txt7' from dual
union all
select 5, 'txt5' from dual
union all
select 3, 'txt3' from dual
)
select * from table1;
in above query I can obtain following result
ID | VALUE
------------------
1 txt1
2 txt2
7 txt7
5 txt5
3 txt3
but I want to show them as follows
ID | VALUE
------------------
3 txt3
5 txt5
7 txt7
2 txt2
1 txt1
How to do that?
One approach would be to add a computed column to your set of union queries, then order by that column:
WITH table1 AS (
SELECT 1 ID, 'txt1' value, 1 AS position FROM dual
UNION ALL
SELECT 2, 'txt2', 2 FROM dual
UNION ALL
SELECT 7, 'txt7', 3 FROM dual
UNION ALL
SELECT 5, 'txt5', 4 FROM dual
UNION ALL
SELECT 3, 'txt3', 5 FROM dual
)
SELECT *
FROM table1
ORDER BY pos DESC;
Note that there is no internal order to a SQL table in general. Actually, even the current ordering you are observing is not necessarily guaranteed by Oracle. If you expect a certain order in a result set, you need to impose it via a ORDER BY clause.
How's this?
with table1 as
(
select 1 ID, 'txt1' value from dual
union all
select 2, 'txt2' from dual
union all
select 7, 'txt7' from dual
union all
select 5, 'txt5' from dual
union all
select 3, 'txt3' from dual
)
select * from table1 order by rownum desc;
Actually this is not working for this perticular example. but it is working for normal table.

How to insert multimple records of numbers with one statement in ORACLE and Sql-server using the same sentence

In my C++ program I tried something like this:
INSERT INTO TEMP_TABELA (OSE_ID) values (7,12,16,17,19,21,24,26,30,33,35,38,42,46,53,58,59,72,73,74,77,78,82,86,87,88,89,91,92,93,100,101,102,104,106,109,113,115,127,133,139,140,142,143,144,148,149,150,151,153,155,160,164,166,167,170,172,178,188,189,191,192,198,199,200,201,202,203,205,207,208,219,220,223,225,231,233,236,240,241,242,244,245,253);
But all I got was:
Description: There are fewer columns in the INSERT statement than values specified in the VALUES clause. The number of values in the VALUES clause must match the number of columns specified in the INSERT statement.
Source: Microsoft OLE DB Provider for SQL Server
The same sentence must work also on Oracle, because my program works for both.
Can anyone help me?
Oracle's and SQL Server's bulk insertion syntax is different from one another, so the safest means is to use individual INSERT statements:
INSERT INTO TEMP_TABELA (OSE_ID) values (7);
INSERT INTO TEMP_TABELA (OSE_ID) values (12);
INSERT INTO TEMP_TABELA (OSE_ID) values (16);
INSERT INTO TEMP_TABELA (OSE_ID) values (17);
...
INSERT INTO TEMP_TABELA (OSE_ID) values (253);
This works in Oracle, it split's on the string (assuming 10g+)
CREATE TABLE TESTX(NUM NUMBER);
INSERT INTO TESTX(NUM)
select regexp_substr (x, '[^,]+', 1, level) as token
from (SELECT '7,12,16,17,19,21,24,26,30,33,35,38,42,46,53,58,59,72,73,74,77,78,82,86,87,88,89,91,92,93,100,101,102,104,106,109,113,115,127,133,139,140,142,143,144,148,149,150,151,153,155,160,164,166,167,170,172,178,188,189,191,192,198,199,200,201,202,203,205,207,208,219,220,223,225,231,233,236,240,241,242,244,245,253' X FROM DUAL) t
connect by regexp_instr (x, '[^,]+', 1, level) > 0
84 rows inserted
Do this for Oracle:
INSERT INTO TEMP_TABELA (OSE_ID)
select 7 from dual union all
select 12 from dual union all
select 16 from dual union all
select 17 from dual union all
select 19 from dual union all
select 21 from dual union all
select 24 from dual union all
select 26 from dual union all
select 30 from dual union all
select 33 from dual union all
select 35 from dual union all
select 38 from dual union all
select 42 from dual union all
select 46 from dual union all
select 53 from dual union all
select 58 from dual union all
select 59 from dual union all
select 72 from dual union all
select 73 from dual union all
select 74 from dual union all
select 77 from dual union all
select 78 from dual union all
select 82 from dual union all
select 86 from dual union all
select 87 from dual union all
select 88 from dual union all
select 89 from dual union all
select 91 from dual union all
select 92 from dual union all
select 93 from dual union all
select 100 from dual union all
select 101 from dual union all
select 102 from dual union all
select 104 from dual union all
select 106 from dual union all
select 109 from dual union all
select 113 from dual union all
select 115 from dual union all
select 127 from dual union all
select 133 from dual union all
select 139 from dual union all
select 140 from dual union all
select 142 from dual union all
select 143 from dual union all
select 144 from dual union all
select 148 from dual union all
select 149 from dual union all
select 150 from dual union all
select 151 from dual union all
select 153 from dual union all
select 155 from dual union all
select 160 from dual union all
select 164 from dual union all
select 166 from dual union all
select 167 from dual union all
select 170 from dual union all
select 172 from dual union all
select 178 from dual union all
select 188 from dual union all
select 189 from dual union all
select 191 from dual union all
select 192 from dual union all
select 198 from dual union all
select 199 from dual union all
select 200 from dual union all
select 201 from dual union all
select 202 from dual union all
select 203 from dual union all
select 205 from dual union all
select 207 from dual union all
select 208 from dual union all
select 219 from dual union all
select 220 from dual union all
select 223 from dual union all
select 225 from dual union all
select 231 from dual union all
select 233 from dual union all
select 236 from dual union all
select 240 from dual union all
select 241 from dual union all
select 242 from dual union all
select 244 from dual union all
select 245 from dual union all
select 253 from dual
And this for SQL Server:
INSERT INTO TEMP_TABELA (OSE_ID)
select 7 union all
select 12 union all
select 16 union all
select 17 union all
select 19 union all
select 21 union all
select 24 union all
select 26 union all
select 30 union all
select 33 union all
select 35 union all
select 38 union all
select 42 union all
select 46 union all
select 53 union all
select 58 union all
select 59 union all
select 72 union all
select 73 union all
select 74 union all
select 77 union all
select 78 union all
select 82 union all
select 86 union all
select 87 union all
select 88 union all
select 89 union all
select 91 union all
select 92 union all
select 93 union all
select 100 union all
select 101 union all
select 102 union all
select 104 union all
select 106 union all
select 109 union all
select 113 union all
select 115 union all
select 127 union all
select 133 union all
select 139 union all
select 140 union all
select 142 union all
select 143 union all
select 144 union all
select 148 union all
select 149 union all
select 150 union all
select 151 union all
select 153 union all
select 155 union all
select 160 union all
select 164 union all
select 166 union all
select 167 union all
select 170 union all
select 172 union all
select 178 union all
select 188 union all
select 189 union all
select 191 union all
select 192 union all
select 198 union all
select 199 union all
select 200 union all
select 201 union all
select 202 union all
select 203 union all
select 205 union all
select 207 union all
select 208 union all
select 219 union all
select 220 union all
select 223 union all
select 225 union all
select 231 union all
select 233 union all
select 236 union all
select 240 union all
select 241 union all
select 242 union all
select 244 union all
select 245 union all
select 253
The parentheses in your statement in the example indicate to SQL Server that you want all that to be on one row. You need to enclose each row in parentheses and separate those by commas, i.e:
INSERT INTO TEMP_TABELA (OSE_ID)
values (7),
(12),
(16),
(17),
(19),
(21),
(24),...
I'm not sure if this will work in Oracle or not, but if not you may need to create a condition in your calling application to determine which DB it's working with and modify the query appropriately.
Similar to #OMGPonies can also do something like:
INSERT INTO TEMP_TABLEA(OSE_ID)
SELECT 7 union all
SELECT 12 ...
Not sure which is faster or even if you care about speed given the number of items you're inserting.
Not tested on oracle but I'm about 90% sure it will work.