feat: Add database table for model context length configuration (#477)

* feat: add database table for model context length configuration - Add model_context table with provider/model/context_limit fields - Implement UPSERT endpoint for model context configuration - Add priority lookup: database > config.yaml > custom_providers > cache - Add frontend click-to-edit UI in ChatInput with tooltip - Add i18n support for context editing dialog (all 8 locales) - Use context_limit field consistently across frontend and backend Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com> * fix: use useMessage() composable instead of window.$message in ChatInput - Remove incorrect NMessage import (not a component) - Use useMessage() composable from naive-ui - Replace window.$message?.xxx() with message.xxx() - Fixes TypeScript build errors Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com> --------- Co-authored-by: Claude Sonnet 4.6 <noreply@anthropic.com>
2026-05-06 15:05:44 +08:00
parent f338aeea18
commit 479e1feef6
14 changed files with 406 additions and 3 deletions
@@ -5,6 +5,8 @@ import { readConfigYaml, writeConfigYaml, fetchProviderModels, buildModelGroups,
 import { buildProviderModelMap, PROVIDER_PRESETS } from '../../shared/providers'
 import { getCopilotModelsDetailed, resolveCopilotOAuthToken, type CopilotModelMeta } from '../../services/hermes/copilot-models'
 import { readAppConfig } from '../../services/app-config'
+import { getDb } from '../../db'
+import { MODEL_CONTEXT_TABLE } from '../../db/hermes/schemas'

 const PROVIDER_MODEL_CATALOG = buildProviderModelMap()

@@ -236,3 +238,126 @@ export async function setConfigModel(ctx: any) {
    ctx.body = { error: err.message }
  }
 }
+
+/**
+ * 设置模型上下文配置（UPSERT：存在则更新，不存在则插入）
+ * 支持路径参数和查询参数两种方式
+ */
+export async function updateModelContext(ctx: any) {
+  // 支持两种方式：
+  // 1. 路径参数: /api/hermes/model-context/:provider/:model
+  // 2. 查询参数: /api/hermes/model-context?provider=xxx&model=xxx
+  let provider: string | undefined
+  let model: string | undefined
+
+  // 优先从路径参数获取
+  if (ctx.params.provider && ctx.params.model) {
+    provider = ctx.params.provider
+    model = ctx.params.model
+  } else {
+    // 从查询参数获取
+    const query = ctx.query as { provider?: string; model?: string }
+    provider = query.provider
+    model = query.model
+  }
+
+  // 如果没有参数，从请求体获取
+  if (!provider || !model) {
+    const body = ctx.request.body as { provider?: string; model?: string; context_limit?: number }
+    provider = body.provider
+    model = body.model
+  }
+
+  const { context_limit } = ctx.request.body as { context_limit: number }
+
+  if (!provider || !model || !context_limit) {
+    ctx.status = 400
+    ctx.body = { error: 'Missing required fields: provider, model, context_limit' }
+    return
+  }
+
+  if (typeof context_limit !== 'number' || context_limit <= 0) {
+    ctx.status = 400
+    ctx.body = { error: 'Context limit must be a positive number' }
+    return
+  }
+
+  try {
+    const db = getDb()
+    if (!db) {
+      ctx.status = 500
+      ctx.body = { error: 'Database not available' }
+      return
+    }
+
+    // 使用 REPLACE 实现 UPSERT：存在则替换，不存在则插入
+    db.prepare(
+      `REPLACE INTO ${MODEL_CONTEXT_TABLE} (provider, model, context_limit) VALUES (?, ?, ?)`
+    ).run(provider, model, context_limit)
+
+    // 查询并返回更新后的数据
+    const row = db.prepare(
+      `SELECT id, provider, model, context_limit FROM ${MODEL_CONTEXT_TABLE} WHERE provider = ? AND model = ?`
+    ).get(provider, model) as { id: number; provider: string; model: string; context_limit: number }
+
+    ctx.body = {
+      success: true,
+      data: row
+    }
+  } catch (err: any) {
+    ctx.status = 500
+    ctx.body = { error: err.message }
+  }
+}
+
+/**
+ * 查询模型上下文配置
+ */
+export async function getModelContext(ctx: any) {
+  // 支持两种方式：
+  // 1. 路径参数: /api/hermes/model-context/:provider/:model
+  // 2. 查询参数: /api/hermes/model-context?provider=xxx&model=xxx
+  let provider: string | undefined
+  let model: string | undefined
+
+  // 优先从路径参数获取
+  if (ctx.params.provider && ctx.params.model) {
+    provider = ctx.params.provider
+    model = ctx.params.model
+  } else {
+    // 从查询参数获取
+    const query = ctx.query as { provider?: string; model?: string }
+    provider = query.provider
+    model = query.model
+  }
+
+  if (!provider || !model) {
+    ctx.status = 400
+    ctx.body = { error: 'Missing provider or model parameter' }
+    return
+  }
+
+  try {
+    const db = getDb()
+    if (!db) {
+      ctx.status = 500
+      ctx.body = { error: 'Database not available' }
+      return
+    }
+
+    const row = db.prepare(
+      `SELECT id, provider, model, context_limit FROM ${MODEL_CONTEXT_TABLE} WHERE provider = ? AND model = ?`
+    ).get(provider, model) as { id: number; provider: string; model: string; context_limit: number } | undefined
+
+    if (!row) {
+      ctx.status = 404
+      ctx.body = { error: 'Model context not found' }
+      return
+    }
+
+    ctx.body = { data: { ...row, limit: row.context_limit } }
+  } catch (err: any) {
+    ctx.status = 500
+    ctx.body = { error: err.message }
+  }
+}
@@ -89,6 +89,21 @@ export const COMPRESSION_SNAPSHOT_SCHEMA: Record<string, string> = {
  updated_at: 'INTEGER NOT NULL',
 }

+// ============================================================================
+// Model Context (model-context.ts)
+// ============================================================================
+
+export const MODEL_CONTEXT_TABLE = 'model_context'
+
+export const MODEL_CONTEXT_SCHEMA: Record<string, string> = {
+  id: 'INTEGER PRIMARY KEY AUTOINCREMENT',
+  provider: 'TEXT NOT NULL',
+  model: 'TEXT NOT NULL',
+  context_limit: 'INTEGER NOT NULL',
+}
+
+export const MODEL_CONTEXT_INDEX = 'CREATE UNIQUE INDEX IF NOT EXISTS idx_model_context_provider_model ON model_context(provider, model)'
+
 // ============================================================================
 // Group Chat (services/hermes/group-chat/index.ts)
 // ============================================================================
@@ -469,6 +484,13 @@ export function initAllHermesTables(retryCount = 0): void {
    // Compression snapshot
    syncTable(COMPRESSION_SNAPSHOT_TABLE, COMPRESSION_SNAPSHOT_SCHEMA)

+    // Model context
+    syncTable(MODEL_CONTEXT_TABLE, MODEL_CONTEXT_SCHEMA, {
+      indexes: {
+        idx_model_context_provider_model: MODEL_CONTEXT_INDEX,
+      }
+    })
+
    // Group chat - basic tables
    syncTable(GC_ROOMS_TABLE, GC_ROOMS_SCHEMA)
    syncTable(GC_MESSAGES_TABLE, GC_MESSAGES_SCHEMA)
@@ -6,3 +6,9 @@ export const modelRoutes = new Router()
 modelRoutes.get('/api/hermes/available-models', ctrl.getAvailable)
 modelRoutes.get('/api/hermes/config/models', ctrl.getConfigModels)
 modelRoutes.put('/api/hermes/config/model', ctrl.setConfigModel)
+
+// Model context routes
+modelRoutes.get('/api/hermes/model-context', ctrl.getModelContext)
+modelRoutes.get('/api/hermes/model-context/:provider/:model', ctrl.getModelContext)
+modelRoutes.put('/api/hermes/model-context/:provider/:model', ctrl.updateModelContext)
+modelRoutes.put('/api/hermes/model-context', ctrl.updateModelContext)
@@ -2,6 +2,8 @@ import { resolve, join } from 'path'
 import { homedir } from 'os'
 import { readFileSync, existsSync, statSync } from 'fs'
 import yaml from 'js-yaml'
+import { getDb } from '../../db'
+import { MODEL_CONTEXT_TABLE } from '../../db/hermes/schemas'

 const HERMES_BASE = resolve(homedir(), '.hermes')
 const MODELS_DEV_CACHE = resolve(HERMES_BASE, 'models_dev_cache.json')
@@ -234,6 +236,25 @@ function lookupContextFromCache(modelName: string, provider: string | null): num
 *   3. models_dev_cache.json, scoped to model.provider when configured
 *   4. DEFAULT_CONTEXT_LENGTH (200K hardcoded fallback)
 */
+/**
+ * 从数据库 model_context 表查找上下文长度（最高优先级）
+ */
+function lookupContextFromDatabase(modelName: string, provider: string | null): number | null {
+  const db = getDb()
+  if (!db) return null
+
+  try {
+    // 尝试精确匹配 provider 和 model
+    const row = db
+      .prepare(`SELECT context_limit FROM ${MODEL_CONTEXT_TABLE} WHERE provider = ? AND model = ?`)
+      .get(provider || 'default', modelName) as { context_limit: number } | undefined
+
+    return row?.context_limit || null
+  } catch {
+    return null
+  }
+}
+
 export function getModelContextLength(profile?: string): number {
  const profileDir = getProfileDir(profile)
  const config = loadConfig(profileDir)
@@ -242,12 +263,17 @@ export function getModelContextLength(profile?: string): number {
  const model = getDefaultModel(config)
  if (!model) return DEFAULT_CONTEXT_LENGTH

+  const provider = getDefaultProvider(config)
+
+  // 0. Database model_context table (highest priority)
+  const dbCtx = lookupContextFromDatabase(model, provider)
+  if (dbCtx && dbCtx > 0) return dbCtx
+
  // 1. Global context_length override in config.yaml
  const configCtx = getConfigContextLength(config)
  if (configCtx && configCtx > 0) return configCtx

  // 2. Custom provider context_length
-  const provider = getDefaultProvider(config)
  const customCtx = lookupCustomProviderContextLength(config, model, provider)
  if (customCtx && customCtx > 0) return customCtx