docs: Token limits implementation status

Co-authored-by: Qwen-Coder <qwen-coder@alibabacloud.com>
2026-03-23 16:22:55 -03:00
parent 585e436538
commit 83e3375ecf
1 changed files with 155 additions and 0 deletions
@@ -0,0 +1,155 @@
+# Token Limits - Estado de Implementación
+
+## ✅ Phase 1: Completado (Database + API)
+
+### Base de Datos
+- [x] Columnas `monthly_token_limit` y `token_limit_reset_day` en users
+- [x] Vista `ai_usage_monthly`
+- [x] Función `check_token_limit()`
+- [x] Función `get_user_usage_stats()`
+
+### API Endpoints
+- [x] `PUT /admin/users/{user_id}/token-limit`
+- [x] `GET /admin/users/{user_id}/token-usage`
+- [x] `GET /admin/users/{user_id}/token-limit/check`
+
+### Common Library
+- [x] Módulo `token_limits`
+- [x] Función `check_ai_token_limit()`
+
+---
+
+## 🔄 Phase 2: En Progreso (Enforce + Dashboard + Alertas)
+
+### 1. Enforce Automático en Handlers de IA
+- [ ] `generate_quiz` - 2000 tokens estimados
+- [ ] `generate_course` - 5000 tokens estimados
+- [ ] `generate_hotspots` - 2000 tokens estimados
+- [ ] `generate_role_play` - 2500 tokens estimados
+- [ ] `summarize_lesson` - 1500 tokens estimados
+- [ ] `chat_with_tutor` (LMS) - 1000 tokens estimados
+- [ ] `evaluate_audio` (LMS) - 1500 tokens estimados
+
+**Código a agregar en cada handler:**
+```rust
+// Check token limit before proceeding (estimate X tokens)
+if let Err(_) = common::token_limits::check_ai_token_limit(&pool, claims.sub, 2000).await {
+    return Err(StatusCode::TOO_MANY_REQUESTS);
+}
+```
+
+### 2. Dashboard UI para Admins
+- [ ] Página `/admin/token-usage` en Studio
+- [ ] Tabla de usuarios con uso de tokens
+- [ ] Gráfico de uso por día/semana/mes
+- [ ] Input para editar límites por usuario
+- [ ] Alertas visuales (>80%, >90%, 100%)
+
+**Componentes necesarios:**
+- `TokenUsageDashboard.tsx` - Página principal
+- `UserTokenRow.tsx` - Fila de tabla de usuario
+- `TokenLimitEditor.tsx` - Editor de límites
+- `UsageChart.tsx` - Gráfico de uso
+
+### 3. Sistema de Alertas
+- [ ] Trigger en `ai_usage_logs` INSERT
+- [ ] Notificaciones al alcanzar 80%, 90%, 100%
+- [ ] Email opcional al usuario
+- [ ] Registro en tabla `notifications`
+
+**Trigger SQL:**
+```sql
+CREATE OR REPLACE FUNCTION check_token_limit_alert()
+RETURNS TRIGGER AS $$
+DECLARE
+    v_limit INTEGER;
+    v_used BIGINT;
+    v_percentage NUMERIC;
+BEGIN
+    -- Get user limit
+    SELECT monthly_token_limit INTO v_limit
+    FROM users WHERE id = NEW.user_id;
+    
+    IF v_limit > 0 THEN
+        -- Calculate current usage
+        SELECT SUM(tokens_used) INTO v_used
+        FROM ai_usage_logs
+        WHERE user_id = NEW.user_id
+          AND created_at >= DATE_TRUNC('month', NOW());
+        
+        v_percentage := (v_used::NUMERIC / v_limit * 100);
+        
+        -- Send notification at thresholds
+        IF v_percentage >= 80 AND v_percentage < 90 THEN
+            -- Send 80% alert
+        ELSIF v_percentage >= 90 AND v_percentage < 100 THEN
+            -- Send 90% alert
+        ELSIF v_percentage >= 100 THEN
+            -- Send 100% alert
+        END IF;
+    END IF;
+    
+    RETURN NEW;
+END;
+$$ LANGUAGE plpgsql;
+```
+
+---
+
+## 📊 Métricas de Uso
+
+### Queries Útiles
+
+```sql
+-- Usuarios que excedieron 80% de su límite
+SELECT 
+    u.email,
+    u.monthly_token_limit,
+    SUM(au.tokens_used) as used,
+    ROUND((SUM(au.tokens_used)::NUMERIC / u.monthly_token_limit * 100), 2) as percentage
+FROM users u
+JOIN ai_usage_logs au ON u.id = au.user_id
+WHERE au.created_at >= DATE_TRUNC('month', NOW())
+  AND u.monthly_token_limit > 0
+GROUP BY u.id
+HAVING SUM(au.tokens_used)::NUMERIC / u.monthly_token_limit > 0.8
+ORDER BY percentage DESC;
+
+-- Uso por endpoint
+SELECT 
+    endpoint,
+    COUNT(*) as requests,
+    SUM(tokens_used) as total_tokens,
+    AVG(tokens_used) as avg_tokens
+FROM ai_usage_logs
+WHERE created_at >= DATE_TRUNC('month', NOW())
+GROUP BY endpoint
+ORDER BY total_tokens DESC;
+
+-- Top 10 usuarios por uso
+SELECT 
+    u.email,
+    SUM(au.tokens_used) as total_tokens,
+    COUNT(*) as requests
+FROM users u
+JOIN ai_usage_logs au ON u.id = au.user_id
+WHERE au.created_at >= DATE_TRUNC('month', NOW())
+GROUP BY u.id
+ORDER BY total_tokens DESC
+LIMIT 10;
+```
+
+---
+
+## 🎯 Próximos Pasos
+
+1. **Enforce Automático**: Agregar check en handlers de IA (CMS + LMS)
+2. **Dashboard UI**: Crear componentes React para admin
+3. **Alertas**: Implementar trigger y notificaciones
+4. **Testing**: Verificar que los límites se respetan
+5. **Documentación**: Actualizar guías de uso
+
+---
+
+**Fecha**: 2026-03-23
+**Estado**: Phase 1 ✅ Completado, Phase 2 🔄 En Progreso