From d0439b5597d6df8503c085563535fb8a8f253136 Mon Sep 17 00:00:00 2001
From: zhuofeng <zhuofeng@yunshan.net.cn>
Date: Tue, 10 Sep 2024 19:39:48 +0800
Subject: [PATCH] feat: filter profiler

---
 server/querier/profile/model/model.go     |  1 +
 server/querier/profile/service/profile.go | 55 ++++++++++++++++++++++-
 2 files changed, 54 insertions(+), 2 deletions(-)

diff --git a/server/querier/profile/model/model.go b/server/querier/profile/model/model.go
index 8871b3cbc78..4edc2ff310d 100644
--- a/server/querier/profile/model/model.go
+++ b/server/querier/profile/model/model.go
@@ -43,6 +43,7 @@ type ProfileTreeNode struct {
 	ParentNodeID int
 	SelfValue    int
 	TotalValue   int
+	Depth        int
 }
 
 type Debug struct {
diff --git a/server/querier/profile/service/profile.go b/server/querier/profile/service/profile.go
index f54b9e83092..b024ecb3f89 100644
--- a/server/querier/profile/service/profile.go
+++ b/server/querier/profile/service/profile.go
@@ -213,6 +213,7 @@ func GenerateProfile(args model.Profile, cfg *config.QuerierConfig, where string
 
 				if preNodeID >= 0 {
 					nodes[preNodeID].ParentNodeID = nodeID
+					nodes[preNodeID].Depth = nodes[nodeID].Depth + 1
 				} else {
 					// remember the entier stack
 					stackToNodeID[profileLocationCompress] = nodeID // compressed stack
@@ -236,14 +237,43 @@ func GenerateProfile(args model.Profile, cfg *config.QuerierConfig, where string
 	for i := range result.FunctionValues.Values {
 		result.FunctionValues.Values[i] = []int{0, 0}
 	}
+
 	result.NodeValues.Values = make([][]int, 0, len(nodes))
+	maxDepth := 0
+	// 0.5%
+	threshold := nodes[0].TotalValue / 200
+
 	for _, node := range nodes {
 		locationID := node.LocationID
 		result.FunctionValues.Values[locationID][0] += node.SelfValue
 		result.FunctionValues.Values[locationID][1] += node.TotalValue
-		result.NodeValues.Values = append(result.NodeValues.Values, []int{locationID, node.ParentNodeID, node.SelfValue, node.TotalValue})
 	}
 
+	deleteNodeIDs := []int{}
+	for i, node := range nodes {
+		if node.TotalValue >= threshold || node.Depth < 6 {
+			continue
+		}
+		if node.ParentNodeID > 0 {
+			nodes[node.ParentNodeID].SelfValue = nodes[node.ParentNodeID].SelfValue + node.TotalValue
+		}
+		deleteNodeIDs = append(deleteNodeIDs, i)
+	}
+
+	mapping := getMapping(len(nodes), deleteNodeIDs)
+
+	for _, node := range nodes {
+		if node.TotalValue >= threshold || node.Depth < 6 {
+			result.NodeValues.Values = append(result.NodeValues.Values, []int{node.LocationID, mapping[node.ParentNodeID], node.SelfValue, node.TotalValue})
+		}
+
+		if node.Depth > maxDepth {
+			maxDepth = node.Depth
+		}
+	}
+
+	log.Infof("total nodes len=%d,total value is %d, threshold is %d, valid node len=%d, maxDepth=%d", len(nodes), nodes[0].TotalValue, threshold, len(result.NodeValues.Values), maxDepth)
+
 	result.Functions = locations
 	locationTypes := GetLocationType(locations, result.FunctionValues.Values, args.ProfileEventType)
 	result.FunctionTypes = locationTypes
@@ -270,12 +300,14 @@ func updateAllParentNodes(nodes []model.ProfileTreeNode, thisNodeID, selfValue,
 	thisNode.SelfValue += selfValue
 	thisNode.TotalValue += totalValue
 
+	depth := 0
 	// parent nodes
 	for thisNode.ParentNodeID >= 0 {
+		depth++
 		thisNode = &nodes[thisNode.ParentNodeID]
 		thisNode.TotalValue += totalValue
 	}
-
+	nodes[thisNodeID].Depth = depth
 }
 
 func CutKernelFunction(profileLocationByteSlice []byte, maxKernelStackDepth int, sep string) ([]byte, bool) {
@@ -343,3 +375,22 @@ func GetLocationType(locations []string, locationValues [][]int, profileEventTyp
 	}
 	return locationTypes
 }
+
+func getMapping(arrLen int, indicesToRemove []int) map[int]int {
+	removed := make(map[int]struct{})
+	for _, index := range indicesToRemove {
+		removed[index] = struct{}{}
+	}
+
+	mapping := make(map[int]int)
+	newIndex := 0
+
+	for i := 0; i < arrLen; i++ {
+		if _, found := removed[i]; !found {
+			mapping[i] = newIndex
+			newIndex++
+		}
+	}
+
+	return mapping
+}