I should be able to pipe JSON/XML/CSV or whatever is possible at the /update/* to a command which in turn uses SolrJ to send the docs to the correct leader in native format.
In the following examples , all connection details of the cluster is put into a file called solrj.properties
The behavior of the script would be exactly similar to the behavior if I were to post the request directly to solr to the specified qt . Everything parameter the requesthandler accepts would be accepted as a -<param-name>=<param-value> format. The same things could be put into a properties file called indexer.properties and be passed as a -p parameter. The script would expect the following extra properties zk.url for cloud or solr.url for standalone.