Improve (date)_range just like filters (#970)

Don't get dissuaded by the nr of changed lines, there's almost nothing
happening here.

`(Date)_range` and `filters` are basically the same aggregation. I
recently improved `filters` a bit, but didn't do anything to
`(date)_range`.
This PR makes all those combinators work in the same way, so there are
two issues left - #971
(easier) and #944 (a bit
harder). Afterwards they will all fully work.

Author: trzysiek

quesma/queryparser/pancake_sql_query_generation_test.go

@@ -53,8 +53,9 @@ func TestPancakeQueryGeneration(t *testing.T) {
 			if filters(test.TestName) {
 				t.Skip("Fix filters")
 			}
-			if test.TestName == "complex sum_bucket. Reproduce: Visualize -> Vertical Bar: Metrics: Sum Bucket (Bucket: Date Histogram, Metric: Average), Buckets: X-Asis: Histogram(file:opensearch-visualize/pipeline_agg_req,nr:22)" {
-				t.Skip("error: filter(s)/range/dataRange aggregation must be the last bucket aggregation")
+
+			if test.TestName == "Line, Y-axis: Min, Buckets: Date Range, X-Axis: Terms, Split Chart: Date Histogram(file:kibana-visualize/agg_req,nr:9)" {
+				t.Skip("Date range is broken, fix in progress (PR #971)")
 			}
 
 			if test.TestName == "Terms with order by top metrics(file:kibana-visualize/agg_req,nr:8)" {
@@ -66,10 +67,6 @@ func TestPancakeQueryGeneration(t *testing.T) {
 				t.Skip("Was skipped before. Wrong key in max_bucket, should be an easy fix")
 			}
 
-			if test.TestName == "complex sum_bucket. Reproduce: Visualize -> Vertical Bar: Metrics: Sum Bucket (Bucket: Date Histogram, Metric: Average), Buckets: X-Asis: Histogram(file:opensearch-visualize/pipeline_agg_req,nr:24)" {
-				t.Skip("Was skipped before, no expected results")
-			}
-
 			// TODO: add test for filter(s) both at the beginning and end of aggregation tree
 
 			fmt.Println("i:", i, "test:", test.TestName)

quesma/queryparser/pancake_transformer.go

@@ -408,48 +408,61 @@ func (a *pancakeTransformer) aggregationTreeToPancakes(topLevel pancakeAggregati
 		// TODO: if both top_hits/top_metrics, and filters, it probably won't work...
 		// Care: order of these two functions is unfortunately important.
 		// Should be fixed after this TODO
-		newFiltersPancakes := a.createFiltersPancakes(&newPancake)
+		newCombinatorPancakes := a.createCombinatorPancakes(&newPancake)
 		additionalTopHitPancakes, err := a.createTopHitAndTopMetricsPancakes(&newPancake)
 		if err != nil {
 			return nil, err
 		}
 
 		pancakeResults = append(pancakeResults, additionalTopHitPancakes...)
-		pancakeResults = append(pancakeResults, newFiltersPancakes...)
+		pancakeResults = append(pancakeResults, newCombinatorPancakes...)
 	}
 
 	return
 }
 
 // createFiltersPancakes only does something, if first layer aggregation is Filters.
 // It creates new pancakes for each filter in that aggregation, and updates `pancake` to have only first filter.
-func (a *pancakeTransformer) createFiltersPancakes(pancake *pancakeModel) (newPancakes []*pancakeModel) {
+func (a *pancakeTransformer) createCombinatorPancakes(pancake *pancakeModel) (newPancakes []*pancakeModel) {
 	if len(pancake.layers) == 0 || pancake.layers[0].nextBucketAggregation == nil {
 		return
 	}
 
 	firstLayer := pancake.layers[0]
-	filters, isFilters := firstLayer.nextBucketAggregation.queryType.(bucket_aggregations.Filters)
-	canSimplyAddFilterToWhereClause := len(firstLayer.currentMetricAggregations) == 0 && len(firstLayer.currentPipelineAggregations) == 0
-	areNewPancakesReallyNeeded := len(pancake.layers) > 1 // if there is only one layer, it's better to get it done with combinators.
+	combinator, isCombinator := firstLayer.nextBucketAggregation.queryType.(bucket_aggregations.CombinatorAggregationInterface)
+	if !isCombinator {
+		return
+	}
+
+	noMoreBucket := len(pancake.layers) <= 1 || (len(pancake.layers) == 2 && pancake.layers[1].nextBucketAggregation == nil)
+	noMetricOnFirstLayer := len(firstLayer.currentMetricAggregations) == 0 && len(firstLayer.currentPipelineAggregations) == 0
+	canSimplyAddCombinatorToWhereClause := noMoreBucket && noMetricOnFirstLayer
+	if canSimplyAddCombinatorToWhereClause {
+		return
+	}
 
-	if !isFilters || !canSimplyAddFilterToWhereClause || !areNewPancakesReallyNeeded || len(filters.Filters) == 0 {
+	areNewPancakesReallyNeeded := len(pancake.layers) > 1 // if there is only one layer above combinator, it easily can be done with 1 pancake, no need for more
+	groups := combinator.CombinatorGroups()
+	if !areNewPancakesReallyNeeded || len(groups) == 0 {
 		return
 	}
 
-	// First create N-1 new pancakes, each with different filter
-	for i := 1; i < len(filters.Filters); i++ {
+	combinatorSplit := combinator.CombinatorSplit()
+	combinatorGroups := combinator.CombinatorGroups()
+	// First create N-1 new pancakes [1...N), each with different filter
+	// (important to update the first (0th) pancake at the end)
+	for i := 1; i < len(groups); i++ {
 		newPancake := pancake.Clone()
 		bucketAggr := newPancake.layers[0].nextBucketAggregation.ShallowClone()
-		bucketAggr.queryType = filters.NewFiltersSingleFilter(i)
+		bucketAggr.queryType = combinatorSplit[i]
 		newPancake.layers[0] = newPancakeModelLayer(&bucketAggr)
-		newPancake.whereClause = model.And([]model.Expr{newPancake.whereClause, filters.Filters[i].Sql.WhereClause})
+		newPancake.whereClause = model.And([]model.Expr{newPancake.whereClause, combinatorGroups[i].WhereClause})
 		newPancakes = append(newPancakes, newPancake)
 	}
 
-	// Then update original to have 1 filter as well
-	pancake.layers[0].nextBucketAggregation.queryType = filters.NewFiltersSingleFilter(0)
-	pancake.whereClause = model.And([]model.Expr{pancake.whereClause, filters.Filters[0].Sql.WhereClause})
+	// Update original
+	pancake.layers[0].nextBucketAggregation.queryType = combinatorSplit[0]
+	pancake.whereClause = model.And([]model.Expr{pancake.whereClause, combinatorGroups[0].WhereClause})
 
 	return
 }