fix: 添加对 ephemeral 5m 和 1h 令牌的支持，优化费用计算逻辑

.gitignore

@@ -247,3 +247,5 @@ web/apiStats/
 
 # Admin SPA build files
 web/admin-spa/dist/
+
+.serena/

src/models/redis.js

@@ -1084,6 +1084,9 @@ class RedisClient {
     pipeline.hincrby(modelDaily, 'cacheReadTokens', finalCacheReadTokens)
     pipeline.hincrby(modelDaily, 'allTokens', totalTokens)
     pipeline.hincrby(modelDaily, 'requests', 1)
+    // 详细缓存类型统计
+    pipeline.hincrby(modelDaily, 'ephemeral5mTokens', ephemeral5mTokens)
+    pipeline.hincrby(modelDaily, 'ephemeral1hTokens', ephemeral1hTokens)
 
     // 按模型统计 - 每月
     pipeline.hincrby(modelMonthly, 'inputTokens', finalInputTokens)
@@ -1092,6 +1095,9 @@ class RedisClient {
     pipeline.hincrby(modelMonthly, 'cacheReadTokens', finalCacheReadTokens)
     pipeline.hincrby(modelMonthly, 'allTokens', totalTokens)
     pipeline.hincrby(modelMonthly, 'requests', 1)
+    // 详细缓存类型统计
+    pipeline.hincrby(modelMonthly, 'ephemeral5mTokens', ephemeral5mTokens)
+    pipeline.hincrby(modelMonthly, 'ephemeral1hTokens', ephemeral1hTokens)
 
     // API Key级别的模型统计 - 每日
     pipeline.hincrby(keyModelDaily, 'inputTokens', finalInputTokens)
@@ -1136,6 +1142,9 @@ class RedisClient {
     pipeline.hincrby(keyModelAlltime, 'cacheCreateTokens', finalCacheCreateTokens)
     pipeline.hincrby(keyModelAlltime, 'cacheReadTokens', finalCacheReadTokens)
     pipeline.hincrby(keyModelAlltime, 'requests', 1)
+    // 详细缓存类型统计
+    pipeline.hincrby(keyModelAlltime, 'ephemeral5mTokens', ephemeral5mTokens)
+    pipeline.hincrby(keyModelAlltime, 'ephemeral1hTokens', ephemeral1hTokens)
     // 费用统计
     if (realCost > 0) {
       pipeline.hincrby(keyModelAlltime, 'realCostMicro', Math.round(realCost * 1000000))
@@ -1152,6 +1161,9 @@ class RedisClient {
     pipeline.hincrby(hourly, 'cacheReadTokens', finalCacheReadTokens)
     pipeline.hincrby(hourly, 'allTokens', totalTokens)
     pipeline.hincrby(hourly, 'requests', 1)
+    // 详细缓存类型统计
+    pipeline.hincrby(hourly, 'ephemeral5mTokens', ephemeral5mTokens)
+    pipeline.hincrby(hourly, 'ephemeral1hTokens', ephemeral1hTokens)
 
     // 按模型统计 - 每小时
     pipeline.hincrby(modelHourly, 'inputTokens', finalInputTokens)
@@ -1160,6 +1172,9 @@ class RedisClient {
     pipeline.hincrby(modelHourly, 'cacheReadTokens', finalCacheReadTokens)
     pipeline.hincrby(modelHourly, 'allTokens', totalTokens)
     pipeline.hincrby(modelHourly, 'requests', 1)
+    // 详细缓存类型统计
+    pipeline.hincrby(modelHourly, 'ephemeral5mTokens', ephemeral5mTokens)
+    pipeline.hincrby(modelHourly, 'ephemeral1hTokens', ephemeral1hTokens)
 
     // API Key级别的模型统计 - 每小时
     pipeline.hincrby(keyModelHourly, 'inputTokens', finalInputTokens)
@@ -1168,6 +1183,9 @@ class RedisClient {
     pipeline.hincrby(keyModelHourly, 'cacheReadTokens', finalCacheReadTokens)
     pipeline.hincrby(keyModelHourly, 'allTokens', totalTokens)
     pipeline.hincrby(keyModelHourly, 'requests', 1)
+    // 详细缓存类型统计
+    pipeline.hincrby(keyModelHourly, 'ephemeral5mTokens', ephemeral5mTokens)
+    pipeline.hincrby(keyModelHourly, 'ephemeral1hTokens', ephemeral1hTokens)
     // 费用统计
     if (realCost > 0) {
       pipeline.hincrby(keyModelHourly, 'realCostMicro', Math.round(realCost * 1000000))
@@ -1235,18 +1253,24 @@ class RedisClient {
     pipeline.hincrby('usage:global:total', 'cacheCreateTokens', finalCacheCreateTokens)
     pipeline.hincrby('usage:global:total', 'cacheReadTokens', finalCacheReadTokens)
     pipeline.hincrby('usage:global:total', 'allTokens', totalTokens)
+    pipeline.hincrby('usage:global:total', 'ephemeral5mTokens', ephemeral5mTokens)
+    pipeline.hincrby('usage:global:total', 'ephemeral1hTokens', ephemeral1hTokens)
     pipeline.hincrby(globalDaily, 'requests', 1)
     pipeline.hincrby(globalDaily, 'inputTokens', finalInputTokens)
     pipeline.hincrby(globalDaily, 'outputTokens', finalOutputTokens)
     pipeline.hincrby(globalDaily, 'cacheCreateTokens', finalCacheCreateTokens)
     pipeline.hincrby(globalDaily, 'cacheReadTokens', finalCacheReadTokens)
     pipeline.hincrby(globalDaily, 'allTokens', totalTokens)
+    pipeline.hincrby(globalDaily, 'ephemeral5mTokens', ephemeral5mTokens)
+    pipeline.hincrby(globalDaily, 'ephemeral1hTokens', ephemeral1hTokens)
     pipeline.hincrby(globalMonthly, 'requests', 1)
     pipeline.hincrby(globalMonthly, 'inputTokens', finalInputTokens)
     pipeline.hincrby(globalMonthly, 'outputTokens', finalOutputTokens)
     pipeline.hincrby(globalMonthly, 'cacheCreateTokens', finalCacheCreateTokens)
     pipeline.hincrby(globalMonthly, 'cacheReadTokens', finalCacheReadTokens)
     pipeline.hincrby(globalMonthly, 'allTokens', totalTokens)
+    pipeline.hincrby(globalMonthly, 'ephemeral5mTokens', ephemeral5mTokens)
+    pipeline.hincrby(globalMonthly, 'ephemeral1hTokens', ephemeral1hTokens)
     pipeline.expire(globalDaily, 86400 * 32)
     pipeline.expire(globalMonthly, 86400 * 365)
 

src/routes/admin/apiKeys.js

@@ -1289,6 +1289,8 @@ async function calculateKeyStats(keyId, timeRange, startDate, endDate) {
         outputTokens: 0,
         cacheCreateTokens: 0,
         cacheReadTokens: 0,
+        ephemeral5mTokens: 0,
+        ephemeral1hTokens: 0,
         requests: 0
       })
     }
@@ -1300,6 +1302,10 @@ async function calculateKeyStats(keyId, timeRange, startDate, endDate) {
       parseInt(data.totalCacheCreateTokens) || parseInt(data.cacheCreateTokens) || 0
     stats.cacheReadTokens +=
       parseInt(data.totalCacheReadTokens) || parseInt(data.cacheReadTokens) || 0
+    stats.ephemeral5mTokens +=
+      parseInt(data.totalEphemeral5mTokens) || parseInt(data.ephemeral5mTokens) || 0
+    stats.ephemeral1hTokens +=
+      parseInt(data.totalEphemeral1hTokens) || parseInt(data.ephemeral1hTokens) || 0
     stats.requests += parseInt(data.totalRequests) || parseInt(data.requests) || 0
 
     totalRequests += parseInt(data.totalRequests) || parseInt(data.requests) || 0
@@ -1318,15 +1324,22 @@ async function calculateKeyStats(keyId, timeRange, startDate, endDate) {
     cacheCreateTokens += stats.cacheCreateTokens
     cacheReadTokens += stats.cacheReadTokens
 
-    const costResult = CostCalculator.calculateCost(
+    const costUsage = {
-      {
+      input_tokens: stats.inputTokens,
-        input_tokens: stats.inputTokens,
+      output_tokens: stats.outputTokens,
-        output_tokens: stats.outputTokens,
+      cache_creation_input_tokens: stats.cacheCreateTokens,
-        cache_creation_input_tokens: stats.cacheCreateTokens,
+      cache_read_input_tokens: stats.cacheReadTokens
-        cache_read_input_tokens: stats.cacheReadTokens
+    }
-      },
+
-      model
+    // 如果有 ephemeral 5m/1h 拆分数据，添加 cache_creation 子对象以实现精确计费
-    )
+    if (stats.ephemeral5mTokens > 0 || stats.ephemeral1hTokens > 0) {
+      costUsage.cache_creation = {
+        ephemeral_5m_input_tokens: stats.ephemeral5mTokens,
+        ephemeral_1h_input_tokens: stats.ephemeral1hTokens
+      }
+    }
+
+    const costResult = CostCalculator.calculateCost(costUsage, model)
     totalCost += costResult.costs.total
   }
 

src/routes/admin/dashboard.js

@@ -472,7 +472,9 @@ router.get('/model-stats', authenticateAdmin, async (req, res) => {
           outputTokens: 0,
           cacheCreateTokens: 0,
           cacheReadTokens: 0,
-          allTokens: 0
+          allTokens: 0,
+          ephemeral5mTokens: 0,
+          ephemeral1hTokens: 0
         }
 
         stats.requests += parseInt(data.requests) || 0
@@ -481,6 +483,8 @@ router.get('/model-stats', authenticateAdmin, async (req, res) => {
         stats.cacheCreateTokens += parseInt(data.cacheCreateTokens) || 0
         stats.cacheReadTokens += parseInt(data.cacheReadTokens) || 0
         stats.allTokens += parseInt(data.allTokens) || 0
+        stats.ephemeral5mTokens += parseInt(data.ephemeral5mTokens) || 0
+        stats.ephemeral1hTokens += parseInt(data.ephemeral1hTokens) || 0
 
         modelStatsMap.set(normalizedModel, stats)
       }
@@ -497,6 +501,14 @@ router.get('/model-stats', authenticateAdmin, async (req, res) => {
         cache_read_input_tokens: stats.cacheReadTokens
       }
 
+      // 如果有 ephemeral 5m/1h 拆分数据，添加 cache_creation 子对象以实现精确计费
+      if (stats.ephemeral5mTokens > 0 || stats.ephemeral1hTokens > 0) {
+        usage.cache_creation = {
+          ephemeral_5m_input_tokens: stats.ephemeral5mTokens,
+          ephemeral_1h_input_tokens: stats.ephemeral1hTokens
+        }
+      }
+
       // 计算费用
       const costData = CostCalculator.calculateCost(usage, model)
 

src/routes/admin/usageStats.js

@@ -786,6 +786,8 @@ router.get('/usage-trend', authenticateAdmin, async (req, res) => {
           const modelOutputTokens = parseInt(data.outputTokens) || 0
           const modelCacheCreateTokens = parseInt(data.cacheCreateTokens) || 0
           const modelCacheReadTokens = parseInt(data.cacheReadTokens) || 0
+          const modelEphemeral5mTokens = parseInt(data.ephemeral5mTokens) || 0
+          const modelEphemeral1hTokens = parseInt(data.ephemeral1hTokens) || 0
           const modelRequests = parseInt(data.requests) || 0
 
           dayInputTokens += modelInputTokens
@@ -800,6 +802,15 @@ router.get('/usage-trend', authenticateAdmin, async (req, res) => {
             cache_creation_input_tokens: modelCacheCreateTokens,
             cache_read_input_tokens: modelCacheReadTokens
           }
+
+          // 如果有 ephemeral 5m/1h 拆分数据，添加 cache_creation 子对象以实现精确计费
+          if (modelEphemeral5mTokens > 0 || modelEphemeral1hTokens > 0) {
+            modelUsage.cache_creation = {
+              ephemeral_5m_input_tokens: modelEphemeral5mTokens,
+              ephemeral_1h_input_tokens: modelEphemeral1hTokens
+            }
+          }
+
           const modelCostResult = CostCalculator.calculateCost(modelUsage, model)
           dayCost += modelCostResult.costs.total
         }
@@ -948,6 +959,8 @@ router.get('/api-keys/:keyId/model-stats', authenticateAdmin, async (req, res) =
               outputTokens: 0,
               cacheCreateTokens: 0,
               cacheReadTokens: 0,
+              ephemeral5mTokens: 0,
+              ephemeral1hTokens: 0,
               allTokens: 0
             })
           }
@@ -957,6 +970,8 @@ router.get('/api-keys/:keyId/model-stats', authenticateAdmin, async (req, res) =
           stats.outputTokens += parseInt(data.outputTokens) || 0
           stats.cacheCreateTokens += parseInt(data.cacheCreateTokens) || 0
           stats.cacheReadTokens += parseInt(data.cacheReadTokens) || 0
+          stats.ephemeral5mTokens += parseInt(data.ephemeral5mTokens) || 0
+          stats.ephemeral1hTokens += parseInt(data.ephemeral1hTokens) || 0
           stats.allTokens += parseInt(data.allTokens) || 0
         }
       }
@@ -992,6 +1007,8 @@ router.get('/api-keys/:keyId/model-stats', authenticateAdmin, async (req, res) =
             outputTokens: 0,
             cacheCreateTokens: 0,
             cacheReadTokens: 0,
+            ephemeral5mTokens: 0,
+            ephemeral1hTokens: 0,
             allTokens: 0
           })
         }
@@ -1001,6 +1018,8 @@ router.get('/api-keys/:keyId/model-stats', authenticateAdmin, async (req, res) =
         stats.outputTokens += parseInt(data.outputTokens) || 0
         stats.cacheCreateTokens += parseInt(data.cacheCreateTokens) || 0
         stats.cacheReadTokens += parseInt(data.cacheReadTokens) || 0
+        stats.ephemeral5mTokens += parseInt(data.ephemeral5mTokens) || 0
+        stats.ephemeral1hTokens += parseInt(data.ephemeral1hTokens) || 0
         stats.allTokens += parseInt(data.allTokens) || 0
       }
     }
@@ -1016,6 +1035,14 @@ router.get('/api-keys/:keyId/model-stats', authenticateAdmin, async (req, res) =
         cache_read_input_tokens: stats.cacheReadTokens
       }
 
+      // 如果有 ephemeral 5m/1h 拆分数据，添加 cache_creation 子对象以实现精确计费
+      if (stats.ephemeral5mTokens > 0 || stats.ephemeral1hTokens > 0) {
+        usage.cache_creation = {
+          ephemeral_5m_input_tokens: stats.ephemeral5mTokens,
+          ephemeral_1h_input_tokens: stats.ephemeral1hTokens
+        }
+      }
+
       // 使用CostCalculator计算费用
       const costData = CostCalculator.calculateCost(usage, model)
 
@@ -1424,6 +1451,16 @@ router.get('/account-usage-trend', authenticateAdmin, async (req, res) => {
               cache_read_input_tokens: parseInt(modelData.cacheReadTokens) || 0
             }
 
+            // 如果有 ephemeral 5m/1h 拆分数据，添加 cache_creation 子对象以实现精确计费
+            const eph5m = parseInt(modelData.ephemeral5mTokens) || 0
+            const eph1h = parseInt(modelData.ephemeral1hTokens) || 0
+            if (eph5m > 0 || eph1h > 0) {
+              usage.cache_creation = {
+                ephemeral_5m_input_tokens: eph5m,
+                ephemeral_1h_input_tokens: eph1h
+              }
+            }
+
             const costResult = CostCalculator.calculateCost(usage, modelName)
             cost += costResult.costs.total
           }
@@ -1582,6 +1619,16 @@ router.get('/account-usage-trend', authenticateAdmin, async (req, res) => {
               cache_read_input_tokens: parseInt(modelData.cacheReadTokens) || 0
             }
 
+            // 如果有 ephemeral 5m/1h 拆分数据，添加 cache_creation 子对象以实现精确计费
+            const eph5m = parseInt(modelData.ephemeral5mTokens) || 0
+            const eph1h = parseInt(modelData.ephemeral1hTokens) || 0
+            if (eph5m > 0 || eph1h > 0) {
+              usage.cache_creation = {
+                ephemeral_5m_input_tokens: eph5m,
+                ephemeral_1h_input_tokens: eph1h
+              }
+            }
+
             const costResult = CostCalculator.calculateCost(usage, modelName)
             cost += costResult.costs.total
           }

src/routes/apiStats.js

@@ -270,7 +270,9 @@ router.post('/api/user-stats', async (req, res) => {
                 inputTokens: 0,
                 outputTokens: 0,
                 cacheCreateTokens: 0,
-                cacheReadTokens: 0
+                cacheReadTokens: 0,
+                ephemeral5mTokens: 0,
+                ephemeral1hTokens: 0
               })
             }
 
@@ -279,6 +281,8 @@ router.post('/api/user-stats', async (req, res) => {
             modelUsage.outputTokens += parseInt(data.outputTokens) || 0
             modelUsage.cacheCreateTokens += parseInt(data.cacheCreateTokens) || 0
             modelUsage.cacheReadTokens += parseInt(data.cacheReadTokens) || 0
+            modelUsage.ephemeral5mTokens += parseInt(data.ephemeral5mTokens) || 0
+            modelUsage.ephemeral1hTokens += parseInt(data.ephemeral1hTokens) || 0
           }
         }
 
@@ -291,6 +295,14 @@ router.post('/api/user-stats', async (req, res) => {
             cache_read_input_tokens: usage.cacheReadTokens
           }
 
+          // 如果有 ephemeral 5m/1h 拆分数据，添加 cache_creation 子对象以实现精确计费
+          if (usage.ephemeral5mTokens > 0 || usage.ephemeral1hTokens > 0) {
+            usageData.cache_creation = {
+              ephemeral_5m_input_tokens: usage.ephemeral5mTokens,
+              ephemeral_1h_input_tokens: usage.ephemeral1hTokens
+            }
+          }
+
           const costResult = CostCalculator.calculateCost(usageData, model)
           totalCost += costResult.costs.total
         }

src/services/accountBalanceService.js

@@ -607,6 +607,16 @@ class AccountBalanceService {
             cache_read_input_tokens: parseInt(data.cacheReadTokens || 0)
           }
 
+          // 如果有 ephemeral 5m/1h 拆分数据，添加 cache_creation 子对象以实现精确计费
+          const eph5m = parseInt(data.ephemeral5mTokens || 0)
+          const eph1h = parseInt(data.ephemeral1hTokens || 0)
+          if (eph5m > 0 || eph1h > 0) {
+            usage.cache_creation = {
+              ephemeral_5m_input_tokens: eph5m,
+              ephemeral_1h_input_tokens: eph1h
+            }
+          }
+
           const costResult = CostCalculator.calculateCost(usage, model)
           totalCost += costResult.costs.total || 0
         }