webrequest_tag_dataframe_to_hive_pseudocode.scala

val webrequest = spark.table("wmf.webrequest").where(...)

// tags is an array right?  need to concat these somehow, but you get the point.
// Need just distinct list of tags
val tags = webrequest.select("tags").distinct

val tagGroupsDfs = tags.map(tag => (tag, webrequest.where(s"${tag} in tags")) )

val partDfs = tagGroupDfs.map(t => {
  tag = t._1
  df = t._2
  PartitionedDataFrame(df, HivePartition(year,month,day,hour,tag=tag))
})

partDfs.foreach { partDf => DataFrameToHive(..., partDf, ...) }