feat(default): add real rankings data

model/perf_metric.go

@@ -13,11 +13,13 @@ type PerfMetric struct {
 	ModelName      string `json:"model_name" gorm:"size:128;uniqueIndex:idx_perf_model_group_bucket,priority:1"`
 	Group          string `json:"group" gorm:"column:group;size:64;uniqueIndex:idx_perf_model_group_bucket,priority:2"`
 	BucketTs       int64  `json:"bucket_ts" gorm:"uniqueIndex:idx_perf_model_group_bucket,priority:3;index:idx_perf_bucket_ts"`
-	RequestCount   int64  `json:"request_count" gorm:"default:0"`
-	SuccessCount   int64  `json:"success_count" gorm:"default:0"`
-	TotalLatencyMs int64  `json:"total_latency_ms" gorm:"default:0"`
-	TtftSumMs      int64  `json:"ttft_sum_ms" gorm:"default:0"`
-	TtftCount      int64  `json:"ttft_count" gorm:"default:0"`
+	RequestCount   int64  `json:"-" gorm:"default:0"`
+	SuccessCount   int64  `json:"-" gorm:"default:0"`
+	TotalLatencyMs int64  `json:"-" gorm:"default:0"`
+	TtftSumMs      int64  `json:"-" gorm:"default:0"`
+	TtftCount      int64  `json:"-" gorm:"default:0"`
+	OutputTokens   int64  `json:"-" gorm:"default:0"`
+	GenerationMs   int64  `json:"-" gorm:"default:0"`
 }
 
 func (PerfMetric) TableName() string {
@@ -40,6 +42,8 @@ func UpsertPerfMetric(metric *PerfMetric) error {
 			"total_latency_ms": gorm.Expr("total_latency_ms + ?", metric.TotalLatencyMs),
 			"ttft_sum_ms":      gorm.Expr("ttft_sum_ms + ?", metric.TtftSumMs),
 			"ttft_count":       gorm.Expr("ttft_count + ?", metric.TtftCount),
+			"output_tokens":    gorm.Expr("output_tokens + ?", metric.OutputTokens),
+			"generation_ms":    gorm.Expr("generation_ms + ?", metric.GenerationMs),
 		}),
 	}).Create(metric).Error
 }

model/usedata_rankings.go

@@ -0,0 +1,66 @@
+package model
+
+import (
+	"fmt"
+
+	"github.com/QuantumNous/new-api/common"
+	"gorm.io/gorm"
+)
+
+type RankingQuotaTotal struct {
+	ModelName   string `json:"model_name"`
+	TotalTokens int64  `json:"total_tokens"`
+}
+
+type RankingQuotaBucket struct {
+	ModelName string `json:"model_name"`
+	Bucket    int64  `json:"bucket"`
+	Tokens    int64  `json:"tokens"`
+}
+
+func GetRankingQuotaTotals(startTime int64, endTime int64) ([]RankingQuotaTotal, error) {
+	var rows []RankingQuotaTotal
+	query := DB.Table("quota_data").
+		Select("model_name, sum(token_used) as total_tokens").
+		Where("model_name <> ''").
+		Group("model_name").
+		Having("sum(token_used) > 0").
+		Order("total_tokens DESC")
+	query = applyRankingQuotaTimeRange(query, startTime, endTime)
+	err := query.Find(&rows).Error
+	return rows, err
+}
+
+func GetRankingQuotaBuckets(startTime int64, endTime int64, bucketSize int64) ([]RankingQuotaBucket, error) {
+	if bucketSize <= 0 {
+		bucketSize = 3600
+	}
+	bucketExpr := rankingBucketExpr(bucketSize)
+	var rows []RankingQuotaBucket
+	query := DB.Table("quota_data").
+		Select(fmt.Sprintf("model_name, %s as bucket, sum(token_used) as tokens", bucketExpr)).
+		Where("model_name <> ''").
+		Group(fmt.Sprintf("model_name, %s", bucketExpr)).
+		Having("sum(token_used) > 0").
+		Order("bucket ASC")
+	query = applyRankingQuotaTimeRange(query, startTime, endTime)
+	err := query.Find(&rows).Error
+	return rows, err
+}
+
+func rankingBucketExpr(bucketSize int64) string {
+	if common.UsingMySQL {
+		return fmt.Sprintf("FLOOR(created_at / %d) * %d", bucketSize, bucketSize)
+	}
+	return fmt.Sprintf("(created_at / %d) * %d", bucketSize, bucketSize)
+}
+
+func applyRankingQuotaTimeRange(query *gorm.DB, startTime int64, endTime int64) *gorm.DB {
+	if startTime > 0 {
+		query = query.Where("created_at >= ?", startTime)
+	}
+	if endTime > 0 {
+		query = query.Where("created_at <= ?", endTime)
+	}
+	return query
+}