verify file format before run

[samp]

Hi,

i would like to test if my sequential file in source (csv separated is good before run job.

i explain.

i would like to check if all my columns have got datas.
my goal is to isolate if it is possible the bad line (ex: wrong date) before running the job in order to eliminate the potential abort of the job.

and i would like to isolate the bad lines.

should i read my csv as one column and separate in a transformer in n columns and test them one by one?

or

is there an an option in the sequential witch can tell me if the file is corrupted or not?

[gateleys]

Most of the file format checking is something I do via a routine. Only when the routine reports no exception do I go ahead.

However, you may try this sequence of steps -

1. Read them as a single column.
2. Exclude non-numeric columns that may contain commas (if file is comma-delimited) by checking if the commas are contained within a pair of quotes. This way you get the actual number of delimiters (here, comma).
3. In the Xfmr, check if the delimiter count (obtained in step 2) = expected num of cols - 1. If not, report as exception.
4. In the Xfmr, split the row into columns.
5. Regarding dates, I think you meant wrongly "formatted" date, and not "wrong date". I expect to get my dates in D4/YMD format, and hence, perform to check if they satisfy the condition, else raise an exception.