chaotic data fields in .sdf interfere with InChI generation?

Hi - got the following (I admit, nastily formatted) .sdfile with a lot of chaotic literature refs in it. I want it simultaneously parsed into both stereosensitive & stereoindependent smiles and InChIs, according to MultiConvert.java. Well - it fails at the first entry, reporting:





See error log:/tmp/temp_java_inchi26380prb


FATAL - exception java.lang.IllegalArgumentException: Could not export to InChi format.


At molecule Cc1c(SC2CCCCC2)n(COCCO)c(=O)[nH]c1=O





Unfortunately, there is no such log in /tmp/temp_java/whatever!





Interestingly, if I either molconvert the input.SDF to smiles OR simply wash all those data fields out, (such as in the attached inputmol.SDF) then MultiConvert does... a proper job. So there is nothing wrong with the molecules as such. Some of that bibliographic text might have caused some collateral damage to the system????

Thanks for the bug report!


We are checking it.

The problem with the sdfile was that it has empty lines in the data fields.


Please check www.mdl.com/downloads/public/ctfile/ctfile.pdf for the correct sdfile format.





There was a bug in the exceptions of the InChI conversion. Thanks for the bug report it will be corrected in the next bugfix release (5.0.1).