First I tried to merge (IJC 5.3.4) a sdf and csv with a common column of compound ID and it worked fine following the procedure but some data not mapped were not integrated which is normal as far as I have understood the "help" on the subject. I therefore tried the other way around, import first the csv (which contains more data than the sdf) and try to map the sdf BUT when doing that I need to import the csv as a database and when I import the sdf INTO the database, the "structure" element of my sdf becomes text field. Is there a way out ?
Not directly, but there are some tricks you could try.
The most obvious would be to add the "missing" IDs to your structure table with empty structures.
e..g if you have 21 "missing" IDs create 21 empty structures and enter the missing IDs.Then you could merge successsfully.
If you have a large number that are mising then this will be painful and you will need to get more creative, but you can probably fill the missing values using copy and paste if you can generate a list of those.
Or you could create a file (CSV) with those missing IDs and empty values for the strucutres and then import them into the structure table as extra rows, then do the merge.
And to help with finding the missing IDs you might want to use the list management feature of IJC. Import the CSV file and generate a list of all the IDs. Then generate another list of the IDs that are present in the structure table from the SD file. Then use the list logic functions to generate a list of that is present in the CSV file but not the SD file.
I'm sure there are also other approaches too that might work.