Processing complex flat file

Question

I have a requirement to load complex (hierarchical) flat file to Oracle database. The flat file has multiple (different) records. Each record is fixed positional.Here is an example
1 123455 6777888
2566773 4444 test
466666 7844 tyeww
477777 7844 tyeww
5213
Records are identified by the first character in the line. (In this example we have record type 1, record type 2, record type 4 and record type 5). Also the records are hierarchical.
Record Type 1
---Record Type 2 *
------Record Type 4 *
---Record Type 5
When the data is loaded to database it should be normalized. That means I have to take some value from Record Type 1, Record Type 2 and Record Type 4.
Can ODI handle this? do I have to write some script? If so How?

Answer 1

OK,
The following technique will help you:
1) Create a database sequence
2) create a interface to load a temp table (I named MY_TEMP_TABLE) with 1 "varchar' column so long as the biggest record (I named col_record), 1 column to store the sequence (I named col_seq) and 1 column to storage the parent line (that one should accept null values) (I named col_parent)
3) Map the record to col_record at source
4) Map the sequence to the col_seq at staging area
5) at col_parent column, put the following code and map it to staging area:
case when substr(src.col_record, 1,1) 1
then (select max(col_sequence)
from MY_TEMP_TABLE mtt
where substr(mtt.col_record, 1,1) < substr(src.col_record, 1,1)
else
null
end
PS: src.col_record is the column from your text file
This code will create the right hierarchy to the file.
after that, you can you the first technique that I told you, dragging and drooping this MY_TEMP_TABLE so many time as records types at a new interface, each one with the right filter, and create the joins.
You could even, add one more column to load the record type and work with it instead the "substr" command in the above select and at filter in the new interface.
Did I miss something? Does it solve the problem?