Using sqlldr when source data column is 4000 chars

Question

I'm trying to load some data using sqlldr.
The table looks like this:
col1 number(10) primary key
col2 varchar2(100)
col3 varchar2(4000)
col4 varchar2(10)
col5 varchar2(1)
... and some more columns ...
For current purposes, I only need to load columns col1 through col3. The other columns will be NULL.
The source text data looks like this (tab-delimited) ...
col1-text<<<TAB>>>col2-text<<<TAB>>>col3-text
more-col3-text
more-col3-text
more-col3-text
more-col3-text
more-col3-text
END-OF-RECORD
There's nothing special about the source data for col1 and col2.
But the data for col3 is (usually) much longer than 4000 chars, so I just need to truncate it to fit varchar2(4000), right?
The control file looks like this ...
LOAD DATA
INFILE 'load.dat' "str 'END-OF-RECORD'"
TRUNCATE
INTO TABLE my_table
FIELDS TERMINATED BY "\t"
OPTIONALLY ENCLOSED BY '"'
TRAILING NULLCOLS
col1 "trim(:col1)",
col2 "trim(:col2)",
col3 char(10000) "substr(:col3,1,4000)"
I made the column 3 specification char(10000) to allow sqlldr to read text longer than 4000 chars.
And the subsequent directive is meant to truncate it to 4000 chars (to fit in the table column).
But I get this error ...
Record 1: Rejected - Error on table COL3.
ORA-01461: can bind a LONG value only for insert into a LONG column
The only solution I found was ugly.
I changed the control file to this ...
col3 char(4000) "substr(:col3,1,4000)"
And then I hand-edited (truncated) the source data for column 3 to be shorter than 4000 chars.
Painful and tedious!
Is there a way around this difficulty?
Note: I cannot use a CLOB for col3. There's no option to change the app, so col3 must remain varchar2(4000).

Answer 1

You can load the data into a staging table with a clob column, then insert into your target table using substr, as demonstated below. I have truncated the data display to save space.
-- load.dat:
1     col2-text     col3-text
more-col3-text
more-col3-text
more-col3-text
more-col3-text
more-col3-text
XYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYY
YYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYY
END-OF-RECORD-- test.ctl:
LOAD DATA
INFILE 'load.dat' "str 'END-OF-RECORD'"
TRUNCATE
INTO TABLE staging
FIELDS TERMINATED BY X'09'
OPTIONALLY ENCLOSED BY '"'
TRAILING NULLCOLS
col1 "trim(:col1)",
col2 "trim(:col2)",
col3 char(10000)
SCOTT@orcl_11gR2> create table staging
  2    (col1 varchar2(10),
  3       col2 varchar2(100),
  4       col3 clob)
  5  /
Table created.
SCOTT@orcl_11gR2> host sqlldr scott/tiger control=test.ctl log=test.log
SCOTT@orcl_11gR2> select * from staging
  2  /
COL1
COL2
COL3
1
col2-text
col3-text
more-col3-text
more-col3-text
more-col3-text
more-col3-text
more-col3-text
XYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYY
YYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYY
1 row selected.
SCOTT@orcl_11gR2> create table my_table
  2    (col1 varchar2(10) primary key,
  3       col2 varchar2(100),
  4       col3 varchar2(4000),
  5       col4 varchar2(10),
  6       col5 varchar2(1))
  7  /
Table created.
SCOTT@orcl_11gR2> insert into my_table (col1, col2, col3)
  2  select col1, col2, substr (col3, 1, 4000) from staging
  3  /
1 row created.
SCOTT@orcl_11gR2> select * from my_table
  2  /
COL1
COL2
COL3
COL4       C
1
col2-text
col3-text
more-col3-text
more-col3-text
more-col3-text
more-col3-text
more-col3-text
XYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYY
YYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYY
1 row selected.