SQL LOADER is a very powerful tool that lets you load data from a delimited or position based data file into Oracle tables. We have received many questions regarding SQL LOADER features from many users. Here is the brief explanation on the same.
Please note that the basic knowledge of SQL LOADER is required to understand this article.
This article covers the below topics:
1.Load multiple data files into a single table
2.Load a single data file into multiple tables
3.Skip a column while loading using “FILLER” and Load field in the delimited data file into two different columns in a table using “POSITION”
4.Usage of BOUNDFILLER
5.Load the same record twice into a single table
6.Using WHEN to selectively load the records into the table
7.Run SQLLDR from SQL PLUS
8.Default path for Discard, bad and log files
1)Load multiple files into a single table:
SQL LOADER lets you load multiple data files at once into a single table. But all the data files should be of the same format.
Here is a working example:
Say you have a table named EMP which has the below structure:
Column | Data Type |
emp_num | Number |
emp_name | Varchar2(25) |
department_num | Number |
department_name | Varchar2(25) |
You are trying to load the below comma delimited data files named eg.dat and eg1.dat:
eg.dat:
7369,SMITH,7902,Accounting
7499,ALLEN,7698,Sales
7521,WARD,7698,Accounting
7566,JONES,7839,Sales
7654,MARTIN,7698,Accounting
eg1.dat:
1234,Tom,2345,Accounting
3456,Berry,8976,Accounting
The Control file should be built as below:
LOAD DATA
INFILE ‘eg.dat’ — File 1
INFILE ‘eg1.dat’ — File 2
APPEND
INTO TABLE emp
FIELDS TERMINATED BY “,”
( emp_num, emp_name, department_num, department_name )
2)Load a single file into multiple tables:
SQL Loader lets you load a single data file into multiple tables using “INTO TABLE” clause.
Here is a working example:
Say you have two tables named EMP and DEPT which have the below structure:
Table | Column | Data Type |
EMP | emp_num | Number |
EMP | emp_name | Varchar2(25) |
DEPT | department_num | Number |
DEPT | department_name | Varchar2(25) |
You are trying to load the below comma delimited data file named eg.dat which has columns Emp_num and emp_name that need to be loaded into table EMP and columns department_num and department_name that need to be loaded into table DEPT using a single CTL file here.
eg.dat:
7369,SMITH,7902,Accounting
7499,ALLEN,7698,Sales
7521,WARD,7698,Accounting
7566,JONES,7839,Sales
7654,MARTIN,7698,Accounting
The Control file should be built as below:
LOAD DATA
INFILE ‘eg.dat’
APPEND
INTO TABLE emp
FIELDS TERMINATED BY “,”
( emp_num, emp_name )
INTO TABLE dept
FIELDS TERMINATED BY “,”
(department_num, department_name)
You can further use WHEN clause to selectively load the records into the tables which will be explained later in this article.
3)Skip a column while loading using “FILLER” and Load field in the delimited data file into two different columns in a table using “POSITION”
SQL LOADER lets to skip unwanted fields in the data file by using the “FILLER” clause. Filler was introduced in Oracle 8i.
SQL LOADER also lets you load the same field into two different columns of the table.
If the data file is position based, loading the same field into two different columns is pretty straight forward. You can use Position (start_pos:end_pos) keyword
If the data file is a delimited file and it has a header included in it, then this can be achieved by referring the field preceded with “:” eg description “(:emp_name)”.
If the data file is delimited file without a header included in it, Position (start_pos:end_pos) or “(:field)” will not work. This can be achieved using POSITION (1) clause which takes you to the beginning of the record.
Here is a Working Example:
The requirement here is to load the field emp_name in the data field into two columns – emp_name and description of the table EMP. Here is the Working Example:
Say you have a table named EMP which has the below structure:
Column | Data Type |
emp_num | Number |
emp_name | Varchar2(25) |
description | Varchar2(25) |
department_num | Number |
department_name | Varchar2(25) |
You are trying to load the below comma delimited data file named eg.dat which has 4 fields that need to be loaded into 5 columns of the table EMP.
eg.dat:
7369,SMITH,7902,Accounting
7499,ALLEN,7698,Sales
7521,WARD,7698,Accounting
7566,JONES,7839,Sales
7654,MARTIN,7698,Accounting
Control File:
LOAD DATA
INFILE ‘eg.dat’
APPEND
INTO TABLE emp
FIELDS TERMINATED BY “,”
(emp_num,
emp_name,
desc_skip FILLER POSITION(1),
description,
department_num,
department_name)
Explanation on how SQL LOADER processes the above CTL file:
The first field in the data file is loaded into column emp_num of table EMP
The second field in the data file is loaded into column emp_name of table EMP
The field desc_skip enables SQL LOADER to start scanning the same record it is at from the beginning because of the clause POSITION(1) . SQL LOADER again reads the first delimited field and skips it as directed by “FILLER” keyword.
Now SQL LOADER reads the second field again and loads it into description column of the table EMP.
SQL LOADER then reads the third field in the data file and loads into column department_num of table EMP
Finally the fourth field is loaded into column department_name of table EMP.
4)Usage of BOUNDFILLER
BOUNDFILLER is available with Oracle 9i and above and can be used if the skipped column’s value will be required later again.
Here is an example:
The requirement is to load first two fields concatenated with the third field as emp_num into table emp and Fourth field as Emp_name
Data File:
1,15,7369,SMITH
1,15,7499,ALLEN
1,15,7521,WARD
1,18,7566,JONES
1,20,7654,MARTIN
The requirement can be achieved using the below Control File:
LOAD DATA
INFILE ‘C:eg.dat’
APPEND
INTO TABLE EMP
FIELDS TERMINATED BY “,”
(
Rec_skip BOUNDFILLER,
tmp_skip BOUNDFILLER,
Emp_num “(:Rec_skip||:tmp_skip||:emp_num)”,
Emp_name
)
5)Load the same record twice into a single table:
SQL Loader lets you load record twice using POSITION clause but you have to take into account whether the constraints defined on the table allow you to insert duplicate rows.
Below is the Control file:
LOAD DATA
INFILE ‘eg.dat’
APPEND
INTO TABLE emp
FIELDS TERMINATED BY “,”
( emp_num, emp_name, department_num, department_name )
INTO TABLE emp
FIELDS TERMINATED BY “,”
(emp_num POSITION(1),emp_name,department_num,department_name)
SQL LOADER processes the above control file this way:
First “INTO TABLE” clause loads the 4 fields specified in the first line of the data file into the respective columns (emp_num, emp_name, department_num, department_name)
Field scanning does not start over from the beginning of the record when SQL LOADER encounters the second INTO TABLE clause in the CTL file. Instead, scanning continues where it left off. Statement “emp_num POSITION(1)” in the CTL file forces the SQL LOADER to read the same record from the beginning and loads the first field in the data file into emp_num column again. The remaining fields in the first record of the data file are again loaded into respective columns emp_name, department_num, department_name. Thus the same record can be loaded multiple times into the same table using “INTO TABLE” clause.
6)Using WHEN to selectively load the records into the table
WHEN clause can be used to direct SQL LOADER to load the record only when the condition specified in the WHEN clause is TRUE. WHEN statement can have any number of comparisons preceded by AND. SQL*Loader does not allow the use of OR in the WHEN clause.
Here is a working example which illustrates how to load the records into 2 tables EMP and DEPT based on the record type specified in the data file.
The below is delimited data file eg.dat which has the first field as the record type. The requirement here is to load all the records with record type = 1 into table EMP and all the records with record type = 2 into table DEPT and record with record type =3 which happens to be the trailer record should not be loaded.
1,7369,SMITH
2,7902,Accounting
1,7499,ALLEN
2,7698,Sales
1,7521,WARD
2,7698,Accounting
1,7566,JONES
2,7839,Sales
1,7654,MARTIN
2,7698,Accounting
3,10
Control File:
LOAD DATA
INFILE ‘eg.dat’
APPEND
INTO TABLE emp
WHEN (01) = ‘1’
FIELDS TERMINATED BY “,”
( rec_skip filler POSITION(1),emp_num , emp_name )
INTO TABLE dept
WHEN (01) = ‘2’
FIELDS TERMINATED BY “,”
(rec_skip filler POSITION(1),department_num,
department_name )
Let’s now see how SQL LOADER processes the CTL file:
SQL LOADER loads the records into table EMP only when first position (01) of the record which happens to be the record type is ‘1’ as directed by command
INTO TABLE emp
WHEN (01) = ‘1’
If condition When (01) = ‘1’ holds true for the current record, then SQL LOADER gets to the beginning of the record as directed by command POSITION(1) and skips the first field which is the record type.
It then loads the second field into emp_num and third field into emp_name column in the table EMP.
SQL LOADER loads the records into table DEPT only when first position (01) of the record which happens to be the record type is ‘2’ as directed by the commands –
INTO TABLE dept
WHEN (01) = ‘2’
If condition When (01) = ‘2’ holds true for the current record, then SQL LOADER gets to the beginning of the record as directed by command POSITION(1) and skips the first field which is the record type.
It then loads the second field into department_num and third field into department_name columns in the table DEPT.
The records with record type = ‘3’ are not loaded into any table.
Thus you can selectively loads the necessary records into various tables using WHEN clause.
7)Run SQLLDR from SQL PLUS
SQL LOADER can be invoked from SQL PLUS using “host” command as shown below:
host sqlldr userid= username/password@host control = C:eg.ctl log = eg.log
8)Default path for Discard, bad and log files
If bad and discard file paths are not specified in the CTL file and if this SQL Loader is registered as a concurrent program, then they will be created in the directory where the regular Concurrent programs’ output files reside. You can also find the paths where the discard and bad files have been created in the log file of the SQL LOADER concurrent request.