How to upload linux file to DB depending on Column name

Question

Hi ,
ii'm using Oracle Database 11g R2 need to upload Telecom CDRs to the database on daily basis , it's huge data and changeable , an example of my file in linux Redhat 5 server as below ,
INDRtotalduration = 00:00:00
origin_matrix = 4186603ec003ef01
triggering_key = 665000207
Start_Date_And_Time = 03/04/2013 09:24:10
IMSI = 418666651000207
INDRenddatetime = 03/04/2013 09:24:10
locind = 0
Correlation_Identifier = d50648d80000012400000187
3572640402767000
IN_service_indicator = 0
service_context_identifier = [email protected]'m already create my table which is the columns names match the file left right names ( like column name =INDRtotalduration ),
there is no broblem with this i think i can use SQLLDR to upload this file , but the problem here the positions of the columns in the file could change depending on user behavior it could be the first row comes in the thired row or any row and maybe more rows appears ,
locind = 0
origin_matrix = 4186603ec003ef01
Start_Date_And_Time = 03/04/2013 09:24:10
INDRtotalduration = 00:00:00
IMSI = 418666651000207
triggering_key = 665000207
Service_Retailer_Name = sdpprov
INDRenddatetime = 03/04/2013 09:24:10
Correlation_Identifier = d50648d80000012400000187
3572640402767000
IN_service_indicator = 0
service_context_identifier = [email protected]
Type_of_IN_Triggering_Key = 1
Start_Client_Info = 1
Ordinary_Client_Id = 665000207
Service_Name = LTE_SOreally this is sample of the file i could be more than 100 rows , and the position of the field and field names could be change every time depending on the Subscriper usage ,
is there any way to upload the file but after checking the field name in the file and matching to corresponding column name in the table .
Please help

Answer 1

Hi,
If you are only loading one file as a single logical CDR then you could possible just sqlload it into a 2 column table so you have the first column containing the variable name and the second column the value. You can then run a select statement using pivot to turn that into a single logical row - see the example below:
This example assumes you load the first 3 values from your test data into a table called PIVOT_TEST using sqlloader (i did it with an insert just to demostrate the pivot). You need to list every possible value that the column value could be in the pivot command. See what you think....
CREATE TABLE pivot_test (
varname varchar2(100),
var_value varchar2(100)
INSERT INTO pivot_test VALUES ('INDRtotalduration','00:00:00');
INSERT INTO pivot_test VALUES ('origin_matrix','4186603ec003ef01');
INSERT INTO pivot_test VALUES ('triggering_key','665000207');
COMMIT;
SELECT *
FROM (SELECT varname, var_value
FROM pivot_test)
PIVOT (max(var_value) FOR (varname) IN ('INDRtotalduration' AS INDRtotalduration, 'origin_matrix' AS origin_matrix, 'triggering_key' AS triggering_key,'some_other_column' as some_other_column));
Regards,
Harry