It is a bad idea to bypass the API altogether and directly query SOLR. From your question, I see no reason for this.
Need to index more data than what Tridion has already indexed? If not, can you just search using the API?
If you do, you might consider introducing a special search indexing handler for additional data. Although this is not well documented at the moment, the creation seems quite simple (implementing ISearchIndexingHandler and updating the configuration of CM and SOLR). The benefit will be that your data can also be found using the standard Tridion search.
source share