Community
Participate
Working Groups
Build Identifier: M20090211-1700 The flatfile UI accurately parses the csv file I gave it but when it tries to preview the data, fails with an invalid csv exception. The reason it fails is that only the first 3 fields are treated as required by the logger which creates the file. The trailing 3 are treated as optional, and are not always present. It doesn't place trailing commas there. The easiest and most awesome solution would be if the flatfile adaptor could treat any missing trailing columns as NULL, the way Oracles SQL*Loader uses "TRAILING NULLCOLS" in its CSV specifications. I am setting the severity to "Major feature is broken" because SQL*Loader, which was written in like the 1890's, does it so BIRT should be able to handle it too... :-) Reproducible: Always Steps to Reproduce: 1. Create a CSV file with the following 2 lines: One,Two,Three,Four,Five,Six 1,2,3,4, 2. Create a flat file dataset 3. Try to preview it
I looked for the source to try to help out but I cannot find it. The page http://www.eclipse.org/birt/phoenix/build/buildInstructions.php indicates that the code should be in org.eclipse.birt.report.data.oda.flatfile, but when I navigate to http://dev.eclipse.org/viewcvs/index.cgi/source/?root=BIRT_Project, there is no such package!! So, sorry - I'dda helped but I can't find the code.
flatfile is now part of Dtp project under /cvsroot/datatools/connectivity in CVS. It is a valid enhancement for flatfile.
Add "Use trailing null columns" checkbox to flatfile datasource select folder page. End users can dis/enable trailing null columns function by de/selecting the checkbox. If trailing null columns is disabled, the flat file driver behavior is consistent with that prior to this enhancement. Default is disabled. If trailing null columns is enabled, null value will be returned for the missing columns in the data row in flatfile.
