Oh, FWIW, content data of type VARCHAR should also work. Karl
On Fri, Jul 29, 2011 at 6:05 AM, Karl Wright <[email protected]> wrote: > I believe the end-user documentation talks about this to some extent. > Nevertheless, the JDBC handler is designed to pull all the necessary > information for a document, including the content data, out of a > single database table. So it presumes the content is stored as either > CLOB data or BLOB data in one column of the table. > > The url field is necessary because that is what ManifoldCF uses for > the "id" in the target search engine. It needs this to be able to > remove or replace the document in the target on subsequent job runs. > It might as well be a URL because it presumes that the search user > will need some way to get to the content of the indexed document. > > Hope that answers your question. > > Karl > > 2011/7/29 Shinichiro Abe <[email protected]>: >> Hello. >> >> I used JDBC Repository Connection and created >> the following view table[1] on postgesql. >> I set the default setting at Queries tab in job lists. >> I run the job, then on the Solr, only urlfield was indexed as id field. >> >> 1)I also want to index datafield. What is needed to set? >> Can I use it like solr dataimporthandler? >> For example, can it index datafield1, datafield2, datafield…? >> >> 2)Why ingesting datafield need to know if url is valid in source code? >> I want to index datafield without urlfield. >> >> My usage may be wrong, I assumed that string data of datafield is indexed as >> contents. >> I want to know what kind of table Data-query assume. >> >> [1]view:documenttable >> | idfield | versionfield | urlfield | datafield >> | modifydatefield >> | char varying | char varying | char varying | char varying | bigint >> -------------------------------------------------------------------------- >> | 1 | 1 | file:///dummy/1| test >> string | 1 >> | 2 | 1 | file:///dummy/2| test >> info | 1 >> >> Thank you, >> Shinichiro Abe >
