I'm new to elasticsearch (Awesome product by the way )
I am using the FS river plugin to index files on a filesystem.
I want to add customs fields inferred from other fields of the indexed file.
Let me illustrate this by an example :
Suppose we index a file named test.pdf.
FS river will create a document with a _source field containing a field named filename.
I'd like to add a new field named type (to _source or whatever) that will use the filename to determine the filetype (using the extension)
So at the end, the returned document would have a filename field (test.pdf) and a type field (pdf) infered from the filename.
I can already do this on query with the scripts_fields but this mean an overhead on every request. Is there a way to transfert that logic on indexing ?
Thanks in advance,
P.S: I know rivers are now deprecated, I just want to index a filesystem. What are the alternatives in the new version of elasticsearch. The documentation isn't really explicit about indexing a filesystem.