add claude proxy

fix:maxTokens
fix empty models
2025-07-30 11:18:47 +08:00 · 2025-05-04 02:52:04 +08:00 · 2025-04-22 02:52:56 +08:00 · 2025-04-22 02:16:28 +08:00 · 2025-04-22 02:08:32 +08:00 · 2025-04-22 01:50:50 +08:00
32 changed files with 406 additions and 273 deletions
@@ -0,0 +1,8 @@
+version: '3.9'
+
+services:
+  adminer:
+    image: adminer
+    restart: always
+    ports:
+      - 8080:8080
@@ -4,6 +4,7 @@ services:
  mariadb:
    image: mariadb
    container_name: mysql
+    restart: unless-stopped
    ports:
      - "3306:3306"
    volumes:
@@ -17,9 +18,4 @@ services:
      MYSQL_DATABASE: openteam
      MYSQL_USER: openteam
      MYSQL_PASSWORD: openteam
-
-  # adminer:
-  #   image: adminer
-  #   restart: always
-  #   ports:
-  #     - 8080:8080
+    
@@ -20,8 +20,3 @@ services:
    volumes:
      - $PWD/pgdata:/var/lib/postgresql/data

-  # adminer:
-  #   image: adminer
-  #   restart: always
-  #   ports:
-  #     - 8080:8080
@@ -0,0 +1,10 @@
+version: '3.7'
+services:
+  sqlite-web:
+    image: vaalacat/sqlite-web
+    ports:
+        - 8800:8080
+    volumes:
+        - $PWD/db:/data
+    environment:
+        - SQLITE_DATABASE=openteam.db
@@ -1,5 +1,5 @@
 <template>
-    <div class="flex flex-col items-center space-y-4 p-6 bg-base-100 rounded-xl shadow-lg max-w-sm mx-auto backdrop-blur-xl glass">
+    <div class="flex flex-col items-center space-y-4 p-6 bg-base-100 rounded-xl shadow-lg mt-2 max-w-sm mx-auto backdrop-blur-xl glass">

        <div class="p-3 bg-white rounded-lg shadow-inner cursor-pointer" @click="toggleQRCode">
            <qrcode-vue :value="currentValue" :size="size" level="H" />
@@ -11,7 +11,7 @@ if (import.meta.env.DEV) { // Vite 的方式判断开发环境

 const service = axios.create({
  baseURL: baseURL, 
-  timeout: 5000,   
+  timeout: 6000,   
  headers: {
    'Content-Type': 'application/json',
  },
@@ -12,17 +12,6 @@
        </p>
      </div>

-      <div v-if="error" role="alert" class="alert shadow-lg bg-rose-100">
-        <svg xmlns="http://www.w3.org/2000/svg" class="h-6 w-6 shrink-0 stroke-current" fill="none" viewBox="0 0 24 24">
-          <path stroke-linecap="round" stroke-linejoin="round" stroke-width="2"
-            d="M10 14l2-2m0 0l2-2m-2 2l-2-2m2 2l2 2m7-2a9 9 0 11-18 0 9 9 0 0118 0z" />
-        </svg>
-        <div>
-          <span>Error! {{ error }}</span>
-        </div>
-        <button class="btn btn-sm" @click="error = null">X</button>
-      </div>
-
      <div class="card border border-base-300/40 shadow-sm">
        <form @submit.prevent="createApiKey" class="card-body space-y-5 p-3 sm:p-8">
          <div class="space-y-4">
@@ -100,13 +89,13 @@
                    class="input input-sm input-bordered w-full" />
                </div>

-                <div class="form-control">
+                <!-- <div class="form-control">
                  <label for="deployment_name" class="label">
                    <span class="label-text">Deployment Name</span>
                  </label>
                  <input id="deployment_name" type="text" v-model="newApiKey.deployment_name"
                    placeholder="Deployment Name" class="input input-sm input-bordered w-full" />
-                </div>
+                </div> -->

                <div class="form-control">
                  <label for="api_secret" class="label">
@@ -147,7 +136,7 @@
                  <!-- <textarea id="support_models" v-model="newApiKey.support_models_text"
                    placeholder='["model1", "model2"]' class="textarea textarea-sm textarea-bordered w-full"></textarea> -->
                  <el-input-tag v-model="newApiKey.support_models_array" :trigger="'Enter'" clearable
-                    placeholder="Please input" @change="onchange_supportmodel"/>
+                    placeholder="Please input" @change="onchange_supportmodel" />
                </div>

                <div class="form-control">
@@ -165,10 +154,23 @@
              </div>
            </div>
          </div>
+
+          <div v-if="error" role="alert" class="alert shadow-lg bg-rose-100">
+            <svg xmlns="http://www.w3.org/2000/svg" class="h-6 w-6 shrink-0 stroke-current" fill="none"
+              viewBox="0 0 24 24">
+              <path stroke-linecap="round" stroke-linejoin="round" stroke-width="2"
+                d="M10 14l2-2m0 0l2-2m-2 2l-2-2m2 2l2 2m7-2a9 9 0 11-18 0 9 9 0 0118 0z" />
+            </svg>
+            <div>
+              <span>{{ error }}</span>
+            </div>
+            <button class="btn btn-sm" @click="error = null">X</button>
+          </div>
          
          <div class="flex justify-end pt-4 items-center gap-2">
            <button @click="cancel" class="btn btn-sm btn-outline">Cancel</button>
-            <button type="submit" class="btn btn-outline btn-sm px-4 text-sm font-medium btn-success" :disabled="!isFormValid">
+            <button type="submit" class="btn btn-outline btn-sm px-4 text-sm font-medium btn-success"
+              :disabled="!isFormValid">
              Create
            </button>
          </div>
@@ -200,12 +202,12 @@ const newApiKey = ref({
  active: true,
  endpoint: '',
  resource_name: '',
-  deployment_name: '',
+  // deployment_name: '',
  api_secret: '',
  model_prefix: '',
  model_alias: '',
  parameters: '{}',
-  support_models: '',
+  support_models: '[]',
  support_models_array: [],
 })

@@ -217,12 +219,12 @@ const resetNewApiKey = () => {
    active: true,
    endpoint: '',
    resource_name: '',
-    deployment_name: '',
+    // deployment_name: '',
    api_secret: '',
    model_prefix: '',
    model_alias: '',
    parameters: '{}',
-    support_models: '',
+    support_models: '[]',
    support_models_array: [],
  }
 }
@@ -249,7 +251,7 @@ const apiKeyImageMap = {
  'gemini': '/assets/gemini.svg',
  'azure': '/assets/azure.svg',
  'github': '/assets/github.svg'
-  
+
 };

 const apiKeyImageUrl = (keytype) => {
@@ -261,7 +263,7 @@ const createApiKey = async () => {
    setToast('Please fill in all required fields (Name, Type, API Key).', 'error')
    return
  }
-  
+
  try {
    try {
      if (!Array.isArray(newApiKey.value.support_models_array)) {
@@ -79,13 +79,13 @@
                    class="input input-sm input-bordered w-full" />
                </div>

-                <div class="form-control">
+                <!-- <div class="form-control">
                  <label for="deployment_name" class="label">
                    <span class="label-text">Deployment Name</span>
                  </label>
                  <input id="deployment_name" type="text" v-model="key.deployment_name"
                    placeholder="Deployment Name" class="input input-sm input-bordered w-full" />
-                </div>
+                </div> -->

                <div class="form-control">
                  <label for="api_secret" class="label">
@@ -126,7 +126,7 @@
            <!-- <td class="text-xs dark:text-white">{{ key.apikey }}</td> -->
            <!-- <td class="text-xs dark:text-white">{{ key.endpoint }}</td> -->
            <td>
-              <div class="flex gap-1">
+              <div class="flex gap-2">
                <div class="lg:tooltip lg:tooltip-top lg:tooltip-open" data-tip="预览">
                  <button class="btn btn-ghost btn-xs btn-square " @click="viewKey(key)">
                    <EyeIcon class="w-4 h-4 dark:text-white" />
@@ -24,86 +24,81 @@
    </div>

    <!-- Table -->
-    <div class="card card-bordered bg-base-100 shadow-sm mt-6" v-if="user">
+    <div class="card card-bordered bg-base-100 shadow-sm mt-6">

-      <div class="card-body">
-        <h3 class="card-title text-lg">Tokens</h3>
-        <div v-if="user.tokens">
+      <div class="card bg-base-100 shadow-xs overflow-x-auto dark:bg-base-200" v-if="user">
+        <table class="table table-sm w-full">
+          <thead>
+            <tr class="text-xs text-base-content/70 uppercase bg-base-200">
+              <th class="px-2 py-3">Token</th>
+              <th class="px-2 py-3">Status</th>
+              <!-- <th class="px-2 py-3">Key</th> -->
+              <th class="px-2 py-3">Expired</th>
+              <th class="px-2 py-3">Quota</th>
+              <th class="px-2 py-3">Used</th>
+              <th class="text-right px-2 py-3"></th>
+            </tr>
+          </thead>
+          <tbody>
+            <tr v-for="token in user.tokens" :key="token.id" class="hover">
+              <td class="font-mono text-xs px-2 py-3">{{ token.name }}</td>
+              <td>
+                <input type="checkbox" class="toggle toggle-xs"
+                  :class="token.active ? 'toggle-success' : 'toggle-error'" v-model="token.active"
+                  @change="updateStatus(token)" />
+              </td>
+              <!-- <td class="font-mono text-xs px-2 py-3">{{ token.key }}</td> -->
+              <td class="px-2 py-3">{{ token.expired_at == 0 ? 'Never' : unixToDate(token.expired_at) }}</td>
+              <td class="px-2 py-3">
+                <template v-if="token.unlimited_quota">
+                  <Infinity />
+                </template>
+                <template v-else>{{ token.quota }}</template>
+              </td>
+              <td class="px-2 py-3">{{ token.used_quota }}</td>
+              <td class="text-right px-1 py-3">
+                <div class="flex items-center gap-2.5">
+                  <div class="lg:tooltip lg:tooltip-top lg:tooltip-open pt-1" data-tip="预览">
+                    <button class="btn btn-ghost btn-xs btn-square" onclick="" @click="viewToken(token)">
+                      <EyeIcon class="w-5 h-5 dark:text-white" />
+                    </button>
+                  </div>

-        </div>
+                  <div class="md:tooltip" data-tip="clean used">
+                    <button class="btn btn-ghost btn-xs btn-square text-sky-300 mt-1" @click="cleanUsedToken(token)"
+                      aria-label="Revoke token">
+                      <Eraser class="w-5 h-5" />
+                    </button>
+                  </div>

-        <p v-else class="text-center text-base-content/70 py-4">No tokens found</p>
-      </div>
-    </div>
-
-    <div class="card bg-base-100 shadow-xs overflow-x-auto dark:bg-base-200" v-if="user">
-      <table class="table table-sm w-full">
-        <thead>
-          <tr class="text-xs text-base-content/70 uppercase bg-base-200">
-            <th class="px-2 py-3">Token</th>
-            <th class="px-2 py-3">Status</th>
-            <!-- <th class="px-2 py-3">Key</th> -->
-            <th class="px-2 py-3">Expired</th>
-            <th class="px-2 py-3">Quota</th>
-            <th class="px-2 py-3">Used</th>
-            <th class="text-right px-2 py-3"></th>
-          </tr>
-        </thead>
-        <tbody>
-          <tr v-for="token in user.tokens" :key="token.id" class="hover">
-            <td class="font-mono text-xs px-2 py-3">{{ token.name }}</td>
-            <td>
-              <input type="checkbox" class="toggle toggle-xs" :class="token.active ? 'toggle-success' : 'toggle-error'"
-                v-model="token.active" @change="updateStatus(token)" />
-            </td>
-            <!-- <td class="font-mono text-xs px-2 py-3">{{ token.key }}</td> -->
-            <td class="px-2 py-3">{{ token.expired_at == 0 ? 'Never' : unixToDate(token.expired_at) }}</td>
-            <td class="px-2 py-3">
-              <template v-if="token.unlimited_quota">
-                <Infinity />
-              </template>
-              <template v-else>{{ token.quota }}</template>
-            </td>
-            <td class="px-2 py-3">{{ token.used_quota }}</td>
-            <td class="text-right px-1 py-3">
-              <div class="flex items-center gap-2.5">
-                <div class="lg:tooltip lg:tooltip-top lg:tooltip-open pt-1" data-tip="预览">
-                  <button class="btn btn-ghost btn-xs btn-square" onclick="" @click="viewToken(token)">
-                    <EyeIcon class="w-5 h-5 dark:text-white" />
-                  </button>
-                </div>
-                <br />
-                <div class="md:tooltip" data-tip="clean used">
-                  <button class="btn btn-ghost btn-xs btn-square text-sky-300 mt-1" @click="cleanUsedToken(token)"
+                  <button v-if="token.name !== 'default'"
+                    class="btn btn-ghost btn-xs btn-square text-error items-center" @click="confirmRevokeToken(token)"
                    aria-label="Revoke token">
-                    <Eraser class="w-5 h-5" />
+                    <TrashIcon class="w-5 h-5" />
                  </button>
                </div>
-
-                <button v-if="token.name !== 'default'" class="btn btn-ghost btn-xs btn-square text-error items-center"
-                  @click="confirmRevokeToken(token)" aria-label="Revoke token">
-                  <TrashIcon class="w-5 h-5" />
-                </button>
-              </div>
-            </td>
-          </tr>
-        </tbody>
-      </table>
-      <dialog id="myToken" class="modal" ref="tokenRef">
-        <div class="modal-box  px-0 sm:px-8">
-          <form method="dialog">
-            <button class="btn btn-sm btn-circle btn-ghost absolute right-2 top-2">✕</button>
+              </td>
+            </tr>
+          </tbody>
+        </table>
+        <dialog id="myToken" class="modal" ref="tokenRef">
+          <div class="modal-box  px-0 sm:px-8">
+            <form method="dialog">
+              <button class="btn btn-sm btn-circle btn-ghost absolute right-2 top-2">✕</button>
+            </form>
+            <QRCodeCard :value="qrCodeValue" :size="120" />
+          </div>
+          <form method="dialog" class="modal-backdrop">
+            <button>关闭</button>
          </form>
-          token
-          <QRCodeCard 
-          :value="qrCodeValue" 
-          :size="120" />
-        </div>
-        <form method="dialog" class="modal-backdrop">
-          <button>关闭</button>
-        </form>
-      </dialog>
+        </dialog>
+      </div>
+
+      <p v-else class="text-center text-base-content/70 py-4">No tokens found</p>
    </div>
+    <!-- </div> -->
+
+

  </div>
 </template>
@@ -205,22 +200,6 @@ const viewToken = (token) => {

 const qrCodeValue = ref('');

-// 监听showTokenModel的变化，控制模态框的显示和隐藏
-// watch(showTokenModel, (newValue) => {
-//   const dialog = tokenRef.value;
-//   if (dialog) { 
-//     if (newValue) {
-//       if (!dialog.hasAttribute('open')) {
-//         dialog.showModal(); 
-//       }
-//     } else {
-//       if (dialog.hasAttribute('open')) {
-//         dialog.close(); 
-//       }
-//     }
-//   }
-// });
-

 // 关闭模态框
 const modalRef = ref(null);
@@ -125,13 +125,13 @@
              <div class="flex gap-1">
                <div class="lg:tooltip lg:tooltip-top lg:tooltip-open" data-tip="预览">
                  <button class="btn btn-ghost btn-xs btn-square " @click="viewUser(user)">
-                    <EyeIcon class="w-3.5 h-3.5 dark:text-white" />
+                    <EyeIcon class="w-4 h-4 dark:text-white" />
                  </button>
                </div>
                <div class="lg:tooltip lg:tooltip-top lg:tooltip-open" data-tip="删除" v-if="user.role < 20">
                  <button class="btn btn-ghost btn-xs btn-square text-error hover:bg-error/30"
                    @click="confirmDeleteUser(user)">
-                    <TrashIcon class="w-3.5 h-3.5 dark:text-white" />
+                    <TrashIcon class="w-4 h-4 dark:text-white" />
                  </button>
                </div>
              </div>
@@ -28,7 +28,7 @@ func (a Api) CreateApiKey(c *gin.Context) {
 	}
 	if slice.Contain([]string{"openai", "azure", "claude"}, *newkey.ApiType) {
 		sma, err := utils.FetchKeyModel(a.db, newkey)
-		if err == nil {
+		if err == nil && len(sma) > 0 {
 			newkey.SupportModelsArray = sma
 			var buf = new(bytes.Buffer)
 			json.NewEncoder(buf).Encode(sma) //nolint:errcheck
@@ -4,15 +4,23 @@ import (
 	"fmt"
 	"net/http"
 	"opencatd-open/internal/dto"
+	"opencatd-open/internal/model"
 	"opencatd-open/llm"
 	"opencatd-open/llm/claude/v2"
 	"opencatd-open/llm/google/v2"
 	"opencatd-open/llm/openai_compatible"
+	"opencatd-open/pkg/tokenizer"

 	"github.com/gin-gonic/gin"
 )

 func (h *Proxy) ChatHandler(c *gin.Context) {
+	user := c.MustGet("user").(*model.User)
+	if user == nil {
+		dto.WrapErrorAsOpenAI(c, 401, "Unauthorized")
+		return
+	}
+
 	var chatreq llm.ChatRequest
 	if err := c.ShouldBindJSON(&chatreq); err != nil {
 		c.JSON(http.StatusBadRequest, gin.H{"error": err.Error()})
@@ -35,10 +43,10 @@ func (h *Proxy) ChatHandler(c *gin.Context) {
 		fallthrough
 	default:
 		llm, err = openai_compatible.NewOpenAICompatible(h.apikey)
-		if err != nil {
-			dto.WrapErrorAsOpenAI(c, 500, fmt.Errorf("create llm client error: %w", err).Error())
-			return
-		}
+	}
+	if err != nil {
+		dto.WrapErrorAsOpenAI(c, 500, fmt.Errorf("create llm client error: %w", err).Error())
+		return
 	}

 	if !chatreq.Stream {
@@ -57,4 +65,13 @@ func (h *Proxy) ChatHandler(c *gin.Context) {
 			c.SSEvent("", data)
 		}
 	}
+
+	llmusage := llm.GetTokenUsage()
+	llmusage.User = user
+	llmusage.TokenID = c.GetInt64("token_id")
+	cost := tokenizer.Cost(llmusage.Model, llmusage.PromptTokens+llmusage.ToolsTokens, llmusage.CompletionTokens)
+
+	h.SendUsage(llmusage)
+	defer fmt.Println("cost:", cost, "prompt_tokens:", llmusage.PromptTokens, "completion_tokens:", llmusage.CompletionTokens, "total_tokens:", llmusage.TotalTokens)
+
 }
@@ -0,0 +1,58 @@
+package controller
+
+import (
+	"encoding/json"
+	"fmt"
+	"net/http"
+	"opencatd-open/internal/dto"
+
+	"github.com/gin-gonic/gin"
+)
+
+func (p *Proxy) HandleModels(c *gin.Context) {
+	models, err := p.getModelCache()
+	if err != nil {
+		dto.Fail(c, http.StatusBadGateway, err.Error())
+		return
+	}
+	type _model struct {
+		ID string `json:"id"`
+	}
+	var ms []_model
+	for _, model := range models {
+		ms = append(ms, _model{ID: model})
+	}
+	dto.Success(c, ms)
+}
+
+func (p *Proxy) setModelCache() error {
+	apikeys, err := p.apiKeyDao.FindKeys(nil)
+	models := make(map[string]bool)
+	if err == nil && len(apikeys) > 0 {
+		for _, k := range apikeys {
+			if len(k.SupportModelsArray) > 0 {
+				for _, sm := range k.SupportModelsArray {
+					models[sm] = true
+				}
+			} else {
+				var sma []string
+				json.Unmarshal([]byte(*k.SupportModels), &sma) // nolint:errCheck
+				for _, sm := range sma {
+					models[sm] = true
+				}
+			}
+		}
+	} else {
+		return fmt.Errorf("empty data")
+	}
+	var support_models []string
+	for m, _ := range models {
+		support_models = append(support_models, m)
+	}
+	return p.cache.Set("models", support_models)
+}
+
+func (p *Proxy) getModelCache() ([]string, error) {
+	models, err := p.cache.Get("models")
+	return models.([]string), err
+}
@@ -12,17 +12,20 @@ import (
 	"net/url"
 	"opencatd-open/internal/dao"
 	"opencatd-open/internal/model"
+	"opencatd-open/internal/utils"
+	"opencatd-open/llm"
 	"opencatd-open/pkg/config"
+	"opencatd-open/pkg/tokenizer"
 	"os"
 	"strings"
 	"sync"
 	"time"

+	"github.com/bluele/gcache"
 	"github.com/gin-gonic/gin"
 	"github.com/lib/pq"
 	"github.com/tidwall/gjson"
 	"gorm.io/gorm"
-	"gorm.io/gorm/clause"
 )

 type Proxy struct {
@@ -30,9 +33,10 @@ type Proxy struct {
 	cfg        *config.Config
 	db         *gorm.DB
 	wg         *sync.WaitGroup
-	usageChan  chan *model.Usage // 用于异步处理的channel
+	usageChan  chan *llm.TokenUsage // 用于异步处理的channel
 	apikey     *model.ApiKey
 	httpClient *http.Client
+	cache      gcache.Cache

 	userDAO       *dao.UserDAO
 	apiKeyDao     *dao.ApiKeyDAO
@@ -52,13 +56,15 @@ func NewProxy(ctx context.Context, cfg *config.Config, db *gorm.DB, wg *sync.Wai
 			client.Transport = tr
 		}
 	}
+
 	np := &Proxy{
 		ctx:           ctx,
 		cfg:           cfg,
 		db:            db,
 		wg:            wg,
 		httpClient:    client,
-		usageChan:     make(chan *model.Usage, cfg.UsageChanSize),
+		cache:         gcache.New(1).Build(),
+		usageChan:     make(chan *llm.TokenUsage, cfg.UsageChanSize),
 		userDAO:       userDAO,
 		apiKeyDao:     apiKeyDAO,
 		tokenDAO:      tokenDAO,
@@ -68,7 +74,7 @@ func NewProxy(ctx context.Context, cfg *config.Config, db *gorm.DB, wg *sync.Wai

 	go np.ProcessUsage()
 	go np.ScheduleTask()
-
+	np.setModelCache()
 	return np
 }

@@ -77,9 +83,13 @@ func (p *Proxy) HandleProxy(c *gin.Context) {
 		p.ChatHandler(c)
 		return
 	}
+	if strings.HasPrefix(c.Request.URL.Path, "/v1/messages") {
+		p.ProxyClaude(c)
+		return
+	}
 }

-func (p *Proxy) SendUsage(usage *model.Usage) {
+func (p *Proxy) SendUsage(usage *llm.TokenUsage) {
 	select {
 	case p.usageChan <- usage:
 	default:
@@ -135,46 +145,90 @@ func (p *Proxy) ProcessUsage() {
 	}
 }

-func (p *Proxy) Do(usage *model.Usage) error {
+func (p *Proxy) Do(llmusage *llm.TokenUsage) error {
 	err := p.db.Transaction(func(tx *gorm.DB) error {
+		now := time.Now()
+		today, _ := time.Parse("2006-01-02", now.Format("2006-01-02"))
+
+		cost := tokenizer.Cost(llmusage.Model, llmusage.PromptTokens, llmusage.CompletionTokens)
+		token, err := p.tokenDAO.GetByID(p.ctx, llmusage.TokenID)
+		if err != nil {
+			return err
+		}
+
+		usage := &model.Usage{
+			UserID:           llmusage.User.ID,
+			TokenID:          llmusage.TokenID,
+			Date:             now,
+			Model:            llmusage.Model,
+			Stream:           llmusage.Stream,
+			PromptTokens:     llmusage.PromptTokens,
+			CompletionTokens: llmusage.CompletionTokens,
+			TotalTokens:      llmusage.TotalTokens,
+			Cost:             fmt.Sprintf("%.8f", cost),
+		}
 		// 1. 记录使用记录
 		if err := tx.WithContext(p.ctx).Create(usage).Error; err != nil {
 			return fmt.Errorf("create usage error: %w", err)
 		}

-		// 2. 更新每日统计（upsert 操作）
-		dailyUsage := model.DailyUsage{
-			UserID:           usage.UserID,
-			TokenID:          usage.TokenID,
-			Capability:       usage.Capability,
-			Date:             time.Date(usage.Date.Year(), usage.Date.Month(), usage.Date.Day(), 0, 0, 0, 0, usage.Date.Location()),
-			Model:            usage.Model,
-			Stream:           usage.Stream,
-			PromptTokens:     usage.PromptTokens,
-			CompletionTokens: usage.CompletionTokens,
-			TotalTokens:      usage.TotalTokens,
-			Cost:             usage.Cost,
-		}
-
-		// 使用 OnConflict 实现 upsert
-		if err := tx.WithContext(p.ctx).Clauses(clause.OnConflict{
-			Columns: []clause.Column{{Name: "user_id"}, {Name: "token_id"}, {Name: "capability"}, {Name: "date"}}, // 唯一键
-			DoUpdates: clause.Assignments(map[string]interface{}{
-				"prompt_tokens":     gorm.Expr("prompt_tokens + ?", usage.PromptTokens),
-				"completion_tokens": gorm.Expr("completion_tokens + ?", usage.CompletionTokens),
-				"total_tokens":      gorm.Expr("total_tokens + ?", usage.TotalTokens),
-				"cost":              gorm.Expr("cost + ?", usage.Cost),
-			}),
-		}).Create(&dailyUsage).Error; err != nil {
-			return fmt.Errorf("upsert daily usage error: %w", err)
+		// 2. 更新每日统计
+		var dailyUsage model.DailyUsage
+		result := tx.WithContext(p.ctx).Where("user_id = ? and date = ?", llmusage.User.ID, today).First(&dailyUsage)
+		if result.RowsAffected == 0 {
+			dailyUsage.UserID = llmusage.User.ID
+			dailyUsage.TokenID = llmusage.TokenID
+			dailyUsage.Date = today
+			dailyUsage.Model = llmusage.Model
+			dailyUsage.Stream = llmusage.Stream
+			dailyUsage.PromptTokens = llmusage.PromptTokens
+			dailyUsage.CompletionTokens = llmusage.CompletionTokens
+			dailyUsage.TotalTokens = llmusage.TotalTokens
+			dailyUsage.Cost = fmt.Sprintf("%.8f", cost)
+			if err := tx.WithContext(p.ctx).Create(&dailyUsage).Error; err != nil {
+				return fmt.Errorf("create daily usage error: %w", err)
+			}
+		} else {
+			if err := tx.WithContext(p.ctx).Model(&model.DailyUsage{}).Where("user_id = ? and date = ?", llmusage.User.ID, today).
+				Updates(map[string]interface{}{
+					"prompt_tokens":     gorm.Expr("prompt_tokens + ?", llmusage.PromptTokens),
+					"completion_tokens": gorm.Expr("completion_tokens + ?", llmusage.CompletionTokens),
+					"total_tokens":      gorm.Expr("total_tokens + ?", llmusage.TotalTokens),
+				}).Error; err != nil {
+				return fmt.Errorf("update daily usage error: %w", err)
+			}
 		}

 		// 3. 更新用户额度
-		if err := tx.WithContext(p.ctx).Model(&model.User{}).Where("id = ?", usage.UserID).Updates(map[string]interface{}{
-			"quota":      gorm.Expr("quota - ?", usage.Cost),
-			"used_quota": gorm.Expr("used_quota + ?", usage.Cost),
-		}).Error; err != nil {
-			return fmt.Errorf("update user quota and used_quota error: %w", err)
+		if *llmusage.User.UnlimitedQuota {
+			if err := tx.WithContext(p.ctx).Model(&model.User{}).Where("id = ?", llmusage.User.ID).Updates(map[string]interface{}{
+				"used_quota": gorm.Expr("used_quota + ?", fmt.Sprintf("%.8f", cost)),
+			}).Error; err != nil {
+				return fmt.Errorf("update user quota and used_quota error: %w", err)
+			}
+		} else {
+			if err := tx.WithContext(p.ctx).Model(&model.User{}).Where("id = ?", llmusage.User.ID).Updates(map[string]interface{}{
+				"quota":      gorm.Expr("quota - ?", fmt.Sprintf("%.8f", cost)),
+				"used_quota": gorm.Expr("used_quota + ?", fmt.Sprintf("%.8f", cost)),
+			}).Error; err != nil {
+				return fmt.Errorf("update user quota and used_quota error: %w", err)
+			}
+		}
+
+		//4 . 更新token额度
+		if *token.UnlimitedQuota {
+			if err := tx.WithContext(p.ctx).Model(&model.Token{}).Where("id = ?", llmusage.TokenID).Updates(map[string]interface{}{
+				"used_quota": gorm.Expr("used_quota + ?", fmt.Sprintf("%.8f", cost)),
+			}).Error; err != nil {
+				return fmt.Errorf("update token quota and used_quota error: %w", err)
+			}
+		} else {
+			if err := tx.WithContext(p.ctx).Model(&model.Token{}).Where("id = ?", llmusage.TokenID).Updates(map[string]interface{}{
+				"quota":      gorm.Expr("quota - ?", fmt.Sprintf("%.8f", cost)),
+				"used_quota": gorm.Expr("used_quota + ?", fmt.Sprintf("%.8f", cost)),
+			}).Error; err != nil {
+				return fmt.Errorf("update token quota and used_quota error: %w", err)
+			}
 		}

 		return nil
@@ -184,10 +238,9 @@ func (p *Proxy) Do(usage *model.Usage) error {

 func (p *Proxy) SelectApiKey(model string) error {
 	akpikeys, err := p.apiKeyDao.FindApiKeysBySupportModel(p.db, model)
-
 	if err != nil || len(akpikeys) == 0 {
 		if strings.HasPrefix(model, "gpt") || strings.HasPrefix(model, "o1") || strings.HasPrefix(model, "o3") || strings.HasPrefix(model, "o4") {
-			keys, err := p.apiKeyDao.FindKeys(map[string]any{"apitype = ?": "openai"})
+			keys, err := p.apiKeyDao.FindKeys(map[string]any{"active = ?": true, "apitype = ?": "openai"})
 			if err != nil {
 				return err
 			}
@@ -195,7 +248,7 @@ func (p *Proxy) SelectApiKey(model string) error {
 		}

 		if strings.HasPrefix(model, "gemini") {
-			keys, err := p.apiKeyDao.FindKeys(map[string]any{"apitype = ?": "gemini"})
+			keys, err := p.apiKeyDao.FindKeys(map[string]any{"active = ?": true, "apitype = ?": "gemini"})
 			if err != nil {
 				return err
 			}
@@ -203,7 +256,7 @@ func (p *Proxy) SelectApiKey(model string) error {
 		}

 		if strings.HasPrefix(model, "claude") {
-			keys, err := p.apiKeyDao.FindKeys(map[string]any{"apitype = ?": "claude"})
+			keys, err := p.apiKeyDao.FindKeys(map[string]any{"active = ?": true, "apitype = ?": "claude"})
 			if err != nil {
 				return err
 			}
@@ -270,7 +323,10 @@ func (p *Proxy) ScheduleTask() {
 			select {
 			case <-time.After(time.Duration(p.cfg.TaskTimeInterval) * time.Minute):
 				p.updateSupportModel()
-
+			case <-time.After(time.Hour * 12):
+				if err := p.setModelCache(); err != nil {
+					fmt.Println("refrash model cache err:", err)
+				}
 			case <-p.ctx.Done():
 				fmt.Println("schedule task done")
 				return
@@ -287,6 +343,9 @@ func (p *Proxy) getOpenAISupportModels(apikey model.ApiKey) ([]string, error) {
 	var supportModels []string
 	var req *http.Request
 	if *apikey.ApiType == "azure" {
+		if strings.HasSuffix(*apikey.Endpoint, "/") {
+			apikey.Endpoint = utils.ToPtr(strings.TrimSuffix(*apikey.Endpoint, "/"))
+		}
 		req, _ = http.NewRequest("GET", *apikey.Endpoint+azureModelsUrl, nil)
 		req.Header.Set("Content-Type", "application/json")
 		req.Header.Set("api-key", *apikey.ApiKey)
@@ -0,0 +1,14 @@
+package controller
+
+import (
+	"fmt"
+	"io"
+
+	"github.com/gin-gonic/gin"
+)
+
+func (p *Proxy) ProxyClaude(c *gin.Context) {
+	fmt.Println(c.Request.URL.String())
+	data, _ := io.ReadAll(c.Request.Body)
+	fmt.Println(string(data))
+}
@@ -1,7 +1,6 @@
 package controller

 import (
-	"fmt"
 	"net/http"
 	"opencatd-open/internal/dto"
 	"opencatd-open/internal/model"
@@ -142,7 +141,7 @@ func (a Api) CreateUser(c *gin.Context) {
 		dto.Fail(c, 400, err.Error())
 		return
 	}
-	fmt.Printf("user:%+v\n", user)
+
 	err = a.userService.Create(c, &user)
 	if err != nil {
 		dto.Fail(c, http.StatusInternalServerError, err.Error())
@@ -90,7 +90,7 @@ func (a Api) ResetToken(c *gin.Context) {
 		dto.Fail(c, http.StatusNotFound, "token not found")
 		return
 	}
-	token.UsedQuota = utils.ToPtr(int64(0))
+	token.UsedQuota = utils.ToPtr(float64(0))

 	err = a.tokenService.UpdateToken(c, token)
 	if err != nil {
@@ -3,6 +3,7 @@ package dao
 import (
 	"errors"
 	"opencatd-open/internal/model"
+	"opencatd-open/internal/utils"
 	"opencatd-open/pkg/config"

 	"gorm.io/gorm"
@@ -38,6 +39,9 @@ func (dao *ApiKeyDAO) Create(apiKey *model.ApiKey) error {
 	if apiKey == nil {
 		return errors.New("apiKey is nil")
 	}
+	if len(*apiKey.SupportModels) < 2 {
+		apiKey.SupportModels = utils.ToPtr("[]")
+	}
 	return dao.db.Create(apiKey).Error
 }

@@ -87,14 +91,21 @@ func (dao *ApiKeyDAO) FindApiKeysBySupportModel(db *gorm.DB, modelName string) (
 	var apiKeys []model.ApiKey
 	switch dao.cfg.DB_Type {
 	case "mysql":
-		return nil, errors.New("not support")
+		err := db.Raw(`
+		SELECT *
+		FROM apikeys
+		WHERE active = true 
+      	AND JSON_CONTAINS(support_models, ?, '$')`, modelName).
+			Scan(&apiKeys).Error
+		return apiKeys, err
 	case "postgres":
 		return nil, errors.New("not support")
 	}
-	err := db.Model(&model.ApiKey{}).
-		Joins("CROSS JOIN JSON_EACH(apikeys.support_models)").
-		Where("value = ?", modelName).
-		Find(&apiKeys).Error
+	err := db.Raw(`
+    SELECT a.*
+    FROM apikeys a
+    JOIN json_each(a.support_models) AS je ON je.value = ?
+    WHERE a.active = true`, modelName).Scan(&apiKeys).Error
 	return apiKeys, err
 }

@@ -212,11 +212,7 @@ func (d *DailyUsageDAO) UpsertDailyUsage(ctx context.Context, usage *model.Usage
 		return db.Clauses(clause.OnConflict{
 			Columns: []clause.Column{
 				{Name: "user_id"},
-				{Name: "token_id"},
-				{Name: "capability"},
 				{Name: "date"},
-				{Name: "model"},
-				{Name: "stream"},
 			},
 			DoUpdates: clause.Assignments(updateColumns),
 		}).Create(dailyUsage).Error
@@ -231,11 +227,7 @@ func (d *DailyUsageDAO) UpsertDailyUsage(ctx context.Context, usage *model.Usage
 		return db.Clauses(clause.OnConflict{
 			Columns: []clause.Column{
 				{Name: "user_id"},
-				{Name: "token_id"},
-				{Name: "capability"},
 				{Name: "date"},
-				{Name: "model"},
-				{Name: "stream"},
 			},
 			DoUpdates: clause.Assignments(updateColumns),
 		}).Create(dailyUsage).Error
@@ -244,8 +236,8 @@ func (d *DailyUsageDAO) UpsertDailyUsage(ctx context.Context, usage *model.Usage
 	default:
 		return db.Transaction(func(tx *gorm.DB) error {
 			var existing model.DailyUsage
-			err := tx.Where("user_id = ? AND token_id = ? AND capability = ? AND date = ? AND model = ? AND stream = ?",
-				usage.UserID, usage.TokenID, usage.Capability, date, usage.Model, usage.Stream).
+			err := tx.Where("user_id = ? AND date = ?",
+				usage.UserID, date).
 				First(&existing).Error

 			if err == gorm.ErrRecordNotFound {
@@ -3,14 +3,14 @@ package model
 import "github.com/lib/pq" //pq.StringArray

 type ApiKey_PG struct {
-	ID                 int64          `gorm:"column:id;primaryKey;autoIncrement" json:"id,omitempty"`
-	Name               *string        `gorm:"column:name;not null;unique;index:idx_apikey_name" json:"name,omitempty"`
-	ApiType            *string        `gorm:"column:apitype;not null;index:idx_apikey_apitype" json:"type,omitempty"`
-	ApiKey             *string        `gorm:"column:apikey;not null;index:idx_apikey_apikey" json:"apikey,omitempty"`
-	Active             *bool          `gorm:"column:active;default:true" json:"active,omitempty"`
-	Endpoint           *string        `gorm:"column:endpoint" json:"endpoint,omitempty"`
-	ResourceNmae       *string        `gorm:"column:resource_name" json:"resource_name,omitempty"`
-	DeploymentName     *string        `gorm:"column:deployment_name" json:"deployment_name,omitempty"`
+	ID           int64   `gorm:"column:id;primaryKey;autoIncrement" json:"id,omitempty"`
+	Name         *string `gorm:"column:name;not null;unique;index:idx_apikey_name" json:"name,omitempty"`
+	ApiType      *string `gorm:"column:apitype;not null;index:idx_apikey_apitype" json:"type,omitempty"`
+	ApiKey       *string `gorm:"column:apikey;not null;index:idx_apikey_apikey" json:"apikey,omitempty"`
+	Active       *bool   `gorm:"column:active;default:true" json:"active,omitempty"`
+	Endpoint     *string `gorm:"column:endpoint" json:"endpoint,omitempty"`
+	ResourceNmae *string `gorm:"column:resource_name" json:"resource_name,omitempty"`
+	// DeploymentName     *string        `gorm:"column:deployment_name" json:"deployment_name,omitempty"`
 	ApiSecret          *string        `gorm:"column:api_secret" json:"api_secret,omitempty"`
 	ModelPrefix        *string        `gorm:"column:model_prefix" json:"model_prefix,omitempty"`
 	ModelAlias         *string        `gorm:"column:model_alias" json:"model_alias,omitempty"`
@@ -26,14 +26,14 @@ func (ApiKey_PG) TableName() string {
 }

 type ApiKey struct {
-	ID                 int64    `gorm:"column:id;primaryKey;autoIncrement" json:"id,omitempty"`
-	Name               *string  `gorm:"column:name;not null;unique;index:idx_apikey_name" json:"name,omitempty"`
-	ApiType            *string  `gorm:"column:apitype;not null;index:idx_apikey_apitype" json:"type,omitempty"`
-	ApiKey             *string  `gorm:"column:apikey;not null;index:idx_apikey_apikey" json:"apikey,omitempty"`
-	Active             *bool    `gorm:"column:active;default:true" json:"active,omitempty"`
-	Endpoint           *string  `gorm:"column:endpoint" json:"endpoint,omitempty"`
-	ResourceNmae       *string  `gorm:"column:resource_name" json:"resource_name,omitempty"`
-	DeploymentName     *string  `gorm:"column:deployment_name" json:"deployment_name,omitempty"`
+	ID           int64   `gorm:"column:id;primaryKey;autoIncrement" json:"id,omitempty"`
+	Name         *string `gorm:"column:name;not null;unique;index:idx_apikey_name" json:"name,omitempty"`
+	ApiType      *string `gorm:"column:apitype;not null;index:idx_apikey_apitype" json:"type,omitempty"`
+	ApiKey       *string `gorm:"column:apikey;not null;index:idx_apikey_apikey" json:"apikey,omitempty"`
+	Active       *bool   `gorm:"column:active;default:true" json:"active,omitempty"`
+	Endpoint     *string `gorm:"column:endpoint" json:"endpoint,omitempty"`
+	ResourceNmae *string `gorm:"column:resource_name" json:"resource_name,omitempty"`
+	// DeploymentName     *string  `gorm:"column:deployment_name" json:"deployment_name,omitempty"`
 	AccessKey          *string  `gorm:"column:access_key" json:"access_key,omitempty"`
 	SecretKey          *string  `gorm:"column:secret_key" json:"secret_key,omitempty"`
 	ModelPrefix        *string  `gorm:"column:model_prefix" json:"model_prefix,omitempty"`
@@ -2,19 +2,19 @@ package model

 // 用户的token
 type Token struct {
-	ID             int64  `gorm:"column:id;primaryKey;autoIncrement" json:"id,omitempty"`
-	UserID         int64  `gorm:"column:user_id;not null;index:idx_token_user_id" json:"userid,omitempty"`
-	Name           string `gorm:"column:name;not null;index:idx_token_name" json:"name,omitempty" binding:"required,min=1,max=20"`
-	Key            string `gorm:"column:key;not null;uniqueIndex:idx_token_key;comment:token key" json:"key,omitempty"`
-	Active         *bool  `gorm:"column:active;default:true" json:"active,omitempty"`                   //
-	Quota          *int64 `gorm:"column:quota;type:bigint;default:0" json:"quota,omitempty"`            // default 0
-	UnlimitedQuota *bool  `gorm:"column:unlimited_quota;default:true" json:"unlimited_quota,omitempty"` // set Quota 1  unlimited
-	UsedQuota      *int64 `gorm:"column:used_quota;type:bigint;default:0" json:"used_quota,omitempty"`
-	ExpiredAt      *int64 `gorm:"column:expired_at;type:bigint;default:0" json:"expired_at,omitempty"`
-	NeverExpired   *bool  `gorm:"column:never_expires;type:bigint;" json:"never_expires,omitempty"`
-	CreatedAt      int64  `gorm:"column:created_at;type:bigint;autoCreateTime" json:"created_at,omitempty"`
-	LastUsedAt     int64  `gorm:"column:lastused_at;type:bigint;autoUpdateTime" json:"lastused_at,omitempty"`
-	User           *User  `gorm:"foreignKey:UserID;constraint:OnUpdate:CASCADE,OnDelete:CASCADE" json:"-"`
+	ID             int64    `gorm:"column:id;primaryKey;autoIncrement" json:"id,omitempty"`
+	UserID         int64    `gorm:"column:user_id;not null;index:idx_token_user_id" json:"userid,omitempty"`
+	Name           string   `gorm:"column:name;not null;index:idx_token_name" json:"name,omitempty" binding:"required,min=1,max=20"`
+	Key            string   `gorm:"column:key;not null;uniqueIndex:idx_token_key;comment:token key" json:"key,omitempty"`
+	Active         *bool    `gorm:"column:active;default:true" json:"active,omitempty"`                   //
+	Quota          *float64 `gorm:"column:quota;type:bigint;default:0" json:"quota,omitempty"`            // default 0
+	UnlimitedQuota *bool    `gorm:"column:unlimited_quota;default:true" json:"unlimited_quota,omitempty"` // set Quota 1  unlimited
+	UsedQuota      *float64 `gorm:"column:used_quota;type:bigint;default:0" json:"used_quota,omitempty"`
+	ExpiredAt      *int64   `gorm:"column:expired_at;type:bigint;default:0" json:"expired_at,omitempty"`
+	NeverExpired   *bool    `gorm:"column:never_expires;type:bigint;" json:"never_expires,omitempty"`
+	CreatedAt      int64    `gorm:"column:created_at;type:bigint;autoCreateTime" json:"created_at,omitempty"`
+	LastUsedAt     int64    `gorm:"column:lastused_at;type:bigint;autoUpdateTime" json:"lastused_at,omitempty"`
+	User           *User    `gorm:"foreignKey:UserID;constraint:OnUpdate:CASCADE,OnDelete:CASCADE" json:"-"`
 }

 func (Token) TableName() string {
@@ -1,11 +1,7 @@
 package model

 import (
-	"net/http"
-	"opencatd-open/store"
 	"time"
-
-	"github.com/gin-gonic/gin"
 )

 type Usage struct {
@@ -16,9 +12,9 @@ type Usage struct {
 	Date             time.Time `gorm:"column:date;autoCreateTime;index:idx_date"`
 	Model            string    `gorm:"column:model"`
 	Stream           bool      `gorm:"column:stream"`
-	PromptTokens     float64   `gorm:"column:prompt_tokens"`
-	CompletionTokens float64   `gorm:"column:completion_tokens"`
-	TotalTokens      float64   `gorm:"column:total_tokens"`
+	PromptTokens     int       `gorm:"column:prompt_tokens"`
+	CompletionTokens int       `gorm:"column:completion_tokens"`
+	TotalTokens      int       `gorm:"column:total_tokens"`
 	Cost             string    `gorm:"column:cost"`
 }

@@ -28,47 +24,18 @@ func (Usage) TableName() string {

 type DailyUsage struct {
 	ID               int64     `gorm:"column:id;primaryKey;autoIncrement"`
-	UserID           int64     `gorm:"column:user_id;uniqueIndex:idx_daily_unique,priority:1"`
-	TokenID          int64     `gorm:"column:token_id;index:idx_daily_token_id"`
-	Capability       string    `gorm:"column:capability;uniqueIndex:idx_daily_unique,priority:2;comment:模型能力"`
+	UserID           int64     `gorm:"column:user_id;uniqueIndex:idx_daily_unique,priority:1"` // uniqueIndex:idx_daily_unique,priority:1
+	TokenID          int64     `gorm:"column:token_id;uniqueIndex:idx_daily_unique,priority:2"`
+	Capability       string    `gorm:"column:capability;index:idx_daily_usage_capability;comment:模型能力"`
 	Date             time.Time `gorm:"column:date;autoCreateTime;uniqueIndex:idx_daily_unique,priority:3"`
 	Model            string    `gorm:"column:model"`
 	Stream           bool      `gorm:"column:stream"`
-	PromptTokens     float64   `gorm:"column:prompt_tokens"`
-	CompletionTokens float64   `gorm:"column:completion_tokens"`
-	TotalTokens      float64   `gorm:"column:total_tokens"`
+	PromptTokens     int       `gorm:"column:prompt_tokens"`
+	CompletionTokens int       `gorm:"column:completion_tokens"`
+	TotalTokens      int       `gorm:"column:total_tokens"`
 	Cost             string    `gorm:"column:cost"`
 }

 func (DailyUsage) TableName() string {
 	return "daily_usages"
 }
-
-func HandleUsage(c *gin.Context) {
-	fromStr := c.Query("from")
-	toStr := c.Query("to")
-	getMonthStartAndEnd := func() (start, end string) {
-		loc, _ := time.LoadLocation("Local")
-		now := time.Now().In(loc)
-
-		year, month, _ := now.Date()
-
-		startOfMonth := time.Date(year, month, 1, 0, 0, 0, 0, loc)
-		endOfMonth := startOfMonth.AddDate(0, 1, 0)
-
-		start = startOfMonth.Format("2006-01-02")
-		end = endOfMonth.Format("2006-01-02")
-		return
-	}
-	if fromStr == "" || toStr == "" {
-		fromStr, toStr = getMonthStartAndEnd()
-	}
-
-	usage, err := store.QueryUsage(fromStr, toStr)
-	if err != nil {
-		c.JSON(http.StatusForbidden, gin.H{"error": err.Error()})
-		return
-	}
-
-	c.JSON(200, usage)
-}
@@ -52,9 +52,6 @@ func (s *ApiKeyServiceImpl) UpdateApiKey(ctx context.Context, apikey *model.ApiK
 	if apikey.ResourceNmae != nil {
 		_key.ResourceNmae = apikey.ResourceNmae
 	}
-	if apikey.DeploymentName != nil {
-		_key.DeploymentName = apikey.DeploymentName
-	}
 	if apikey.AccessKey != nil {
 		_key.AccessKey = apikey.AccessKey
 	}
@@ -8,12 +8,13 @@ import (
 	"opencatd-open/internal/model"
 	"os"
 	"strings"
+	"time"

 	"github.com/tidwall/gjson"
 	"gorm.io/gorm"
 )

-var client = &http.Client{}
+var client = &http.Client{Timeout: 2 * time.Second}

 func init() {
 	if os.Getenv("LOCAL_PROXY") != "" {
@@ -29,13 +30,13 @@ func FetchKeyModel(db *gorm.DB, key *model.ApiKey) ([]string, error) {
 	var err error
 	if *key.ApiType == "openai" || *key.ApiType == "azure" {
 		supportModels, err = FetchOpenAISupportModels(db, key)
-		if err != nil {
-			fmt.Println(err)
-		}
 	}
 	if *key.ApiType == "claude" {
 		supportModels, err = FetchClaudeSupportModels(db, key)
 	}
+	if err != nil {
+		fmt.Println(err)
+	}
 	return supportModels, err
 }

@@ -47,6 +48,9 @@ func FetchOpenAISupportModels(db *gorm.DB, apikey *model.ApiKey) ([]string, erro
 	var supportModels []string
 	var req *http.Request
 	if *apikey.ApiType == "azure" {
+		if strings.HasSuffix(*apikey.Endpoint, "/") {
+			apikey.Endpoint = ToPtr(strings.TrimSuffix(*apikey.Endpoint, "/"))
+		}
 		req, _ = http.NewRequest("GET", *apikey.Endpoint+azureModelsUrl, nil)
 		req.Header.Set("Content-Type", "application/json")
 		req.Header.Set("api-key", *apikey.ApiKey)
@@ -94,7 +94,9 @@ func (c *Claude) Chat(ctx context.Context, chatReq llm.ChatRequest) (*llm.ChatRe
 	if chatReq.MaxTokens > 0 {
 		maxTokens = chatReq.MaxTokens
 	} else {
-		if strings.Contains(chatReq.Model, "sonnet") || strings.Contains(chatReq.Model, "haiku") {
+		if strings.Contains(chatReq.Model, "3-7") {
+			maxTokens = 64000
+		} else if strings.Contains(chatReq.Model, "3-5") {
 			maxTokens = 8192
 		} else {
 			maxTokens = 4096
@@ -111,6 +113,9 @@ func (c *Claude) Chat(ctx context.Context, chatReq llm.ChatRequest) (*llm.ChatRe
 		return nil, err
 	}

+	if c.tokenUsage.Model == "" && resp.Model != "" {
+		c.tokenUsage.Model = string(resp.Model)
+	}
 	c.tokenUsage.PromptTokens += resp.Usage.InputTokens
 	c.tokenUsage.CompletionTokens += resp.Usage.OutputTokens
 	c.tokenUsage.TotalTokens += resp.Usage.InputTokens + resp.Usage.OutputTokens
@@ -110,6 +110,9 @@ func (g *Gemini) Chat(ctx context.Context, chatReq llm.ChatRequest) (*llm.ChatRe
 		return nil, err
 	}

+	if g.tokenUsage.Model == "" && response.ModelVersion != "" {
+		g.tokenUsage.Model = response.ModelVersion
+	}
 	if response.UsageMetadata != nil {
 		g.tokenUsage.PromptTokens += int(response.UsageMetadata.PromptTokenCount)
 		g.tokenUsage.CompletionTokens += int(response.UsageMetadata.CandidatesTokenCount)
@@ -13,7 +13,7 @@ type LLM interface {

 type llm struct {
 	ApiKey   *model.ApiKey
-	Usage    *model.Usage
+	Usage    *TokenUsage
 	tools    any   // TODO
 	Messages []any // TODO
 	llm      LLM
@@ -14,6 +14,8 @@ import (
 	"opencatd-open/llm"
 	"os"
 	"strings"
+
+	"github.com/sashabaranov/go-openai"
 )

 // https://learn.microsoft.com/en-us/azure/ai-services/openai/api-version-deprecation#latest-preview-api-releases
@@ -86,6 +88,9 @@ func (o *OpenAICompatible) Chat(ctx context.Context, chatReq llm.ChatRequest) (*
 		}
 		var buildurl string
 		if *o.ApiKey.Endpoint != "" {
+			if strings.HasSuffix(*o.ApiKey.Endpoint, "/") {
+				o.ApiKey.ApiKey = utils.ToPtr(strings.TrimSuffix(*o.ApiKey.Endpoint, "/"))
+			}
 			buildurl = fmt.Sprintf("%s/openai/deployments/%s/chat/completions?api-version=%s", *o.ApiKey.Endpoint, formatModel(chatReq.Model), AzureApiVersion)
 		} else {
 			buildurl = fmt.Sprintf("https://%s.openai.azure.com/openai/deployments/%s/chat/completions?api-version=%s", *o.ApiKey.ResourceNmae, formatModel(chatReq.Model), AzureApiVersion)
@@ -116,6 +121,9 @@ func (o *OpenAICompatible) Chat(ctx context.Context, chatReq llm.ChatRequest) (*
 		return nil, err
 	}

+	if o.tokenUsage.Model == "" && chatResp.Model != "" {
+		o.tokenUsage.Model = chatResp.Model
+	}
 	o.tokenUsage.PromptTokens = chatResp.Usage.PromptTokens
 	o.tokenUsage.CompletionTokens = chatResp.Usage.CompletionTokens
 	o.tokenUsage.TotalTokens = chatResp.Usage.TotalTokens
@@ -124,6 +132,7 @@ func (o *OpenAICompatible) Chat(ctx context.Context, chatReq llm.ChatRequest) (*

 func (o *OpenAICompatible) StreamChat(ctx context.Context, chatReq llm.ChatRequest) (chan *llm.StreamChatResponse, error) {
 	chatReq.Stream = true
+	chatReq.StreamOptions = &openai.StreamOptions{IncludeUsage: true}
 	dst, err := utils.StructToMap(chatReq)
 	if err != nil {
 		return nil, err
@@ -197,6 +206,7 @@ func (o *OpenAICompatible) StreamChat(ctx context.Context, chatReq llm.ChatReque
 					if err := json.Unmarshal(line, &streamResp); err != nil {
 						continue
 					}
+
 					if streamResp.Usage != nil {
 						o.tokenUsage.PromptTokens += streamResp.Usage.PromptTokens
 						o.tokenUsage.CompletionTokens += streamResp.Usage.CompletionTokens
@@ -2,6 +2,7 @@ package llm

 import (
 	"fmt"
+	"opencatd-open/internal/model"

 	"github.com/sashabaranov/go-openai"
 )
@@ -15,9 +16,13 @@ type StreamChatResponse openai.ChatCompletionStreamResponse
 type ChatMessage openai.ChatCompletionMessage

 type TokenUsage struct {
+	User             *model.User
+	TokenID          int64
+	Model            string `json:"model"`
+	Stream           bool
 	PromptTokens     int `json:"prompt_tokens"`
 	CompletionTokens int `json:"completion_tokens"`
-	ToolsTokens      int `json:"total_tokens"`
+	ToolsTokens      int `json:"tools_tokens"`
 	TotalTokens      int `json:"total_tokens"`
 }

@@ -94,6 +94,8 @@ func AuthLLM(db *gorm.DB) gin.HandlerFunc {
 		}

 		c.Set("user", token.User)
+		c.Set("user_id", token.User.ID)
+		c.Set("token_id", token.ID)
 		c.Set("authed", true)
 		// 可以在这里对 token 进行验证并检查权限

@@ -124,7 +124,7 @@ func SetRouter(cfg *config.Config, db *gorm.DB, web *embed.FS) {
 	{
 		// v1.POST("/v2/*proxypath", router.HandleProxy)
 		v1.POST("/*proxypath", proxy.HandleProxy)
-		// v1.GET("/models", dashboard.HandleModels)
+		v1.GET("/models", proxy.HandleModels)
 	}

 	idxFS, err := fs.Sub(web, "dist")
Author	SHA1	Message	Date
Sakurasan	4a68ff6162	add claude proxy	2025-07-30 11:18:47 +08:00
Sakurasan	6b2d78fe56	fix:maxTokens	2025-05-04 02:52:04 +08:00
Sakurasan	9c604460b1	fix empty models	2025-04-22 02:52:56 +08:00
Sakurasan	8d34f8d6fe	up	2025-04-22 02:16:28 +08:00
Sakurasan	6d1d0f3b6b	up	2025-04-22 02:08:32 +08:00
Sakurasan	24529189d9	fix daily usage	2025-04-22 01:50:50 +08:00
Sakurasan	000162b1b1	fix usage	2025-04-22 01:06:03 +08:00
Sakurasan	6662ea5e04	fix record usage	2025-04-22 00:48:24 +08:00
Sakurasan	5789d50e9e	update record usage	2025-04-21 23:59:30 +08:00
Sakurasan	ca3d89751d	fix stream usage	2025-04-21 22:48:28 +08:00
Sakurasan	2bc857cf88	add log	2025-04-21 21:59:13 +08:00
Sakurasan	a9ff7e1c94	add log	2025-04-21 21:50:29 +08:00
Sakurasan	51d4651c6c	up	2025-04-21 20:19:48 +08:00
Sakurasan	e112f3af12	collect usage	2025-04-21 19:10:27 +08:00
Sakurasan	73e53c2333	add models task	2025-04-21 01:40:06 +08:00
Sakurasan	470e49b850	support fetch models	2025-04-21 01:30:17 +08:00
Sakurasan	d426781e47	UP	2025-04-20 19:21:51 +08:00
Sakurasan	b80f0759a5	fix select key	2025-04-20 18:33:59 +08:00
Sakurasan	b83c6d9786	fix active key,suffix	2025-04-20 16:22:23 +08:00