On Aug 24, 2007, at 5:29 PM, Wu, Daniel wrote:
Theoretically and practically, embedded solution will be faster than
going through http/xml. I would like to see solr has some sort of
document source adapter architecture which will iterate through all
the
documents available in the document source. This way, if the
documents
come from database for example, it can be simply a sql query in the
solr
configuration file.
In Ruby land we already have this sort of mechanism:
source = YourCustomDataSource.new
mapping = {
:title = :custom_data_source_title,
:author = :custom_data_source_author
# ...
}
indexer = Solr::Indexer.new(source, mapping)
indexer.index
The only requirements are that YourCustomDataSource returns documents
successively via the #each method, and that each of those documents
have an attribute accessor #[] method.
This already works with CSV files, another Solr server, or an array
of objects with the current solr-ruby library.
It is my dream to see JRuby blend into Solr to facilitate this sort
of thing "in process".
Erik
