It is possible in cascading.
Suppose your field names are (url,count).
Apply a function to add one more field named "domain" that contains value google if row contains the word google and discard url field.
Now if you don't require any other domains then filter them out.
So now you have two fields (domain,count) where domain contains only word google
Now use AggregateBy() , SumBy() functions of cascading.
SumBy any_name = new SumBy(field_name_to_sum , field_name_after_sum , dataType class);
Pipe result = new AggregateBy("name" , Pipe.pipes(sourcePipeName) , name_of_groupBy_field , number_of_SumBy_instances , name_of_sumBy_instance);
in your case it becomes
SumBy xyz = new SumBy(new Fields("count") , new Fields("combined_count") , Integer.class);
Pipe result = new AggregateBy("result" , Pipe.pipes(sourcePipeName) , new Fields("domain") , 1 , xyz);
So now result pipe contains a single row (google,count)
So the above code snippet will work similar to the below SQL Query.
select domain,sum(count) from source group by on domain;