Hi, I am using SOLR 7.5 in the cloud mode. I want to create a custom function similar to 'strdist' that works on multivalued fields (multiValued=true) and finds the highest matching score. Yes, I know the potential performance issues, but in my usecase this would bring a huge benefit.
There is not much information on how to work with multiValued fields, but I have found a piece of code that might be useful. It's how SOLR standard functions are registered: https://github.com/apache/lucene-solr/blob/master/solr/core/src/java/org/apache/solr/search/ValueSourceParser.java The interesting part for me starts in line 424, when the 'field' function is registered. It optionally accepts a multivalue field for min/max calculation. If the 2nd argument is 'min' or 'max' it tries to resolve the field as SchemaField. SchemaField f = fp.getReq().getSchema().getField(fieldName); Now the questions are: 1. Is this the path I should follow? If not - are there any other ways? 2. How to retrieve all the actual *String *or *Text *values from a multivalue field, not just a single value? Some kind of a table or set of values. How? 3. Does cloud mode change anything here? In my case the whole index is on a single machine, but there are several replicas. Best regards, Dariusz Wojtas
