We have been working on a search index that contains Archived Web Pages that has been collected over a number of years. This can result in the same page(url) being collected on many dates. The problem that we faced is that we wanted to group results by Site(domain) but this left us with the same page being found many times so we needed a second level of grouping.
I have extended the SOLR 5.5.3 grouping code to allow for 2 level grouping, through discussions with some of the people that are involved with archiving websites it was requested that the code be shared with the SOLR developers. I have made the code public on github SOLR Grouping<https://github.com/nla/solr-grouping>. When extending the SOLR grouping code I tried to keep the code generic so that it could possible used elsewhere but I did not try to make all existing features work, only focusing on the parts that we needed for our system. Along the way I found a couple of bugs that I fixed in this code (1. Integer overflow in holding the total record count & 2. Not searching all shards when performing the second phase of the query(get all records within a group)). Ian Caldwell National Library of Australia
