Automatic type detection when creating document from InputStream in Java

Question

Hello all,
I'm trying to write a batch load program in Java, that connects to an Informix database, reads from a large document table the equivalent of BLOB fields (in Informix, a TEXT field) to a Java InputStream, and then, creates a document in iFS from this stream.
Almost everything works, but since I have no indication of what document type the stream is, I have no clue of what name should the newly created document in iFS have.
Knowing that the original documents can only be Word docs or TIFF images, is there any way for iFS to autodetect the document type, so it can be created with the right extension (in my case, .doc or .tif)? I'm thinking about some API class that peeks inside the stream and do some kind of autodetection based on the doc header or something, but I really don't know if it exists.
Any help will be greatly appreciated. Thanks!
Luis Derani ([email protected])
DBA - ALESP
Brazil

Answer 1

Hi Luis, (funny, first time I ever see a 'Luis' like me in a Forum, with 's' and all),
I have a column indicating the type in the original database, but of about 220000 documents, 40000 have a null in this field (talk about strong constraint enforcement - the field does not have any constraint at all!).
I made a size checking and found that the tif images have around 100-150Kb, and most of the docs are much smaller, so I´ll make a guess and create the docs based on size. Once they are created as documents in iFS, I think it will be easier to detect and correct the type using some windows-based tool.
Thanks for your reply!
Luis Derani ([email protected])
DBA - ALESP
Brazil