To handle empty file/only hd tr file

[ajithaselvan]

Hi,

My source is fixed width file with below format.
H00014700
Ddewuuwq
T00014720

I have below scenarios for rejections. If input file falls under any one below, then I have to update another table with file name and reject reason

1. Empty file (zero rows)
2. Only header and trailer
3. Only Detail
4. More than one header/trailer

any idea how to handle?

[ArndW]

There isn't any 1-click type of solution for this, unfortunately.

There are various ways to handle this, either from the command line using utilities such as awk or via a subroutine call which uses BASIC or a call to C-code. But if you wish to do this exclusively inside a DataStage PX job you would need to read the file and then sort and aggregate on your first 1-letter column.

- If there are more than 1 "H" rows, error
- If there are more than 1 "T" rows, error
- If the "H" row is not row number 1 then error
- If the "T" row is not the highest row number, then error
- If the "D" row count is 0, then error

[chulett]

Also realize that you cannot recognize an empty file inside a job to process it and note that somehow. That one you'll need to check separately. Elsewhere.

[rameshrr3]

External source stage, call:

Code: Select all

grep -c ^D #srcfile# 

This shoudl give zero if zero detail records

Similarly

Code: Select all

grep -c ^H

and

Code: Select all

grep -c^T 

for your file, these will give you counts of headers and trailers into your jobs data stream - use a funnel to combine these streams ( There could be a creative way in unix itself , using echo & awk to call all 3 greps in a single line) and parse it in a transformer
maybe like :

Code: Select all

echo "DETAIL" `grep -c ^D file.txt` '\n' "HEADER" `grep -c ^H file.txt` '\n' "TRAILER" `grep -c ^T file.txt`

I would advise you not to waste more time on development using sorts aggregators etc - when shell one-liners do it much faster.

[qt_ky]

The unix test -s tests if the specified file exists and has a size greater than zero.

Code: Select all

if [ -s /path/file ] then ...

[ssnegi]

Code: Select all

touch filename.txt | wc -l

This will create an empty file if it does not exist and then do a line count.
empty file will return 0.

[chulett]

Why? There's no need to create a file that isn't needed nor a need to count all of the rows in it. The test Eric posted will tell you in a simple true/false manner if the file exists and is not empty, with true moving on to more complex tests and false causing a reject. If you need to differentiate between 'exists' and 'empty' do a simple test for existence first, then test for empty as noted.