Merge pull request #960 from kubecost/AjayTripathy-improve-alloc-perf

Ajay tripathy improve alloc perf

pkg/kubecost/allocation.go

@@ -121,7 +121,7 @@ func (pv *PVAllocations) Clone() PVAllocations {
 		return nil
 	}
 	apv := *pv
-	clonePV := PVAllocations{}
+	clonePV := make(map[PVKey]*PVAllocation, len(apv))
 	for k, v := range apv {
 		clonePV[k] = &PVAllocation{
 			ByteHours: v.ByteHours,
@@ -1679,17 +1679,17 @@ func (as *AllocationSet) Clone() *AllocationSet {
 	as.RLock()
 	defer as.RUnlock()
 
-	allocs := map[string]*Allocation{}
+	allocs := make(map[string]*Allocation, len(as.allocations))
 	for k, v := range as.allocations {
 		allocs[k] = v.Clone()
 	}
 
-	externalKeys := map[string]bool{}
+	externalKeys := make(map[string]bool, len(as.externalKeys))
 	for k, v := range as.externalKeys {
 		externalKeys[k] = v
 	}
 
-	idleKeys := map[string]bool{}
+	idleKeys := make(map[string]bool, len(as.idleKeys))
 	for k, v := range as.idleKeys {
 		idleKeys[k] = v
 	}

pkg/kubecost/allocationprops.go

@@ -133,13 +133,13 @@ func (p *AllocationProperties) Clone() *AllocationProperties {
 	}
 	clone.Services = services
 
-	labels := make(map[string]string)
+	labels := make(map[string]string, len(p.Labels))
 	for k, v := range p.Labels {
 		labels[k] = v
 	}
 	clone.Labels = labels
 
-	annotations := make(map[string]string)
+	annotations := make(map[string]string, len(p.Annotations))
 	for k, v := range p.Annotations {
 		annotations[k] = v
 	}

pkg/kubecost/asset.go

@@ -2502,7 +2502,7 @@ func (as *AssetSet) Clone() *AssetSet {
 		aggregateBy = append([]string{}, as.aggregateBy...)
 	}
 
-	assets := map[string]Asset{}
+	assets := make(map[string]Asset, len(as.assets))
 	for k, v := range as.assets {
 		assets[k] = v.Clone()
 	}

pkg/kubecost/config_test.go

@@ -38,7 +38,7 @@ func TestLabelConfig_Map(t *testing.T) {
 
 func TestLabelConfig_GetExternalAllocationName(t *testing.T) {
 	// Make sure that AWS's Glue/Athena column formatting is supported
-	glueFormattedLabel := cloudutil.ConvertToGlueColumnFormat("Non__GlueFormattedLabel")
+	glueFormattedLabel := cloudutil.ConvertToGlueColumnFormat("___Non_&GlueFormattedLabel___&")
 
 	labels := map[string]string{
 		"kubens":                      "kubecost-staging",

pkg/util/cloudutil/aws.go

@@ -1,8 +1,9 @@
 package cloudutil
 
 import (
-	"regexp"
+	"fmt"
 	"strings"
+	"unicode"
 
 	"github.com/kubecost/cost-model/pkg/log"
 )
@@ -14,27 +15,34 @@ import (
 // https://docs.aws.amazon.com/awsaccountbilling/latest/aboutv2/run-athena-sql.html
 // It returns a string containing the column name in proper column name format and length.
 func ConvertToGlueColumnFormat(columnName string) string {
-	log.Debugf("Converting string \"%s\" to proper AWS Glue column name.", columnName)
-
-	// An underscore is added in front of uppercase letters
-	capitalUnderscore := regexp.MustCompile(`[A-Z]`)
-	final := capitalUnderscore.ReplaceAllString(columnName, `_$0`)
-
-	// Any non-alphanumeric characters are replaced with an underscore
-	noSpacePunc := regexp.MustCompile(`[\s]{1,}|[^A-Za-z0-9]`)
-	final = noSpacePunc.ReplaceAllString(final, "_")
-
-	// Duplicate underscores are removed
-	noDupUnderscore := regexp.MustCompile(`_{2,}`)
-	final = noDupUnderscore.ReplaceAllString(final, "_")
-
-	// Any leading and trailing underscores are removed
-	noFrontUnderscore := regexp.MustCompile(`(^\_|\_$)`)
-	final = noFrontUnderscore.ReplaceAllString(final, "")
-
-	// Uppercase to lowercase
-	final = strings.ToLower(final)
+	fmt.Printf("Calling convert on %s: ", columnName)
+	var sb strings.Builder
+	var prev rune
+	for i, r := range columnName {
+		if unicode.IsUpper(r) && prev != '_' && i != 0 {
+			sb.WriteRune('_')
+		}
+		if !unicode.IsLetter(r) && !unicode.IsNumber(r) {
+			if prev != '_' && i != 0 && i != (len(columnName)-1) {
+				sb.WriteRune('_')
+			}
+			prev = '_'
+			continue
+		}
+		if r == '_' {
+			if prev == '_' || i == 0 || i == len(columnName)-1 {
+				prev = '_'
+				continue
+			}
+		}
+		sb.WriteRune(unicode.ToLower(r))
+		prev = r
+	}
 
+	final := sb.String()
+	if prev == '_' { // string any trailing '_'
+		final = final[:len(final)-1]
+	}
 	// Longer column name than expected - remove _ left to right
 	allowedColLen := 128
 	underscoreToRemove := len(final) - allowedColLen
@@ -47,7 +55,7 @@ func ConvertToGlueColumnFormat(columnName string) string {
 	if len(final) > allowedColLen {
 		final = final[:allowedColLen]
 	}
-
+	fmt.Println(final)
 	log.Debugf("Column name being returned: \"%s\". Length: \"%d\".", final, len(final))
 
 	return final