dsh-layered-memory 0.8.6 → 0.8.8

package/dist/config.d.ts

@@ -4,6 +4,7 @@
  * 默认数据目录：$DSH_HOME/memory（用官方 dshHomePath 解析，DSH_HOME 缺省 ~/.dsh）。
  */
 import Schema from '@deepseek-ai/schemastery';
+import type { StaticFallbackEntry } from './contract.js';
 import type { ExtractMode } from './types.js';
 /**
  * 蒸馏思考档位全词汇表（唯一事实源）：'' = 自动（模型默认档 → high），
@@ -96,10 +97,19 @@ export interface MemoryConfig {
         provider: string;
         /** 蒸馏用的模型；留空用当前默认选择。 */
         model: string;
+        /** 回退链（#31）：主路由失败（报错/掐断/网络异常/空输出）后按序降级的备用路由，
+         *  条目顺序即优先级。与主路由完全相同的条目自动跳过；provider/model 缺失的条目剔除；
+         *  条目 reasoningEffort 非空时覆盖全局档位（未配置运行时链 distillChain 时，
+         *  旧档位键 reasoningEffort 的整体接管——含给条目盖章——仍对存量值生效）；
+         *  空数组（缺省）= 单路由行为不变。 */
+        fallbacks?: StaticFallbackEntry[];
         /** 单次蒸馏调用的输出 token 上限（推理模型的 reasoning 与正文共享该预算）。 */
         maxTokens: number;
         /** 蒸馏调用的思考档位；空串不传（跟随模型默认）。 */
         reasoningEffort: string;
+        /** 运行时主路由显式档位（distillChain[0].reasoningEffort 经 effectiveCfg 注入）；
+         *  非静态 schema——'' = 跟随全局静态 reasoningEffort，仅供链模式传递主路由档位。 */
+        primaryEffort?: string;
         temperature: number;
         /** 单次蒸馏调用的用户 prompt 字符预算（≈token 数，按中文 1 字≈1 token 保守估算）。 */
         maxInputChars: number;
@@ -114,6 +124,10 @@ export interface MemoryConfig {
             l3: number;
         }>;
     };
+    tokenCost: {
+        /** token_cost 明细保留天数；写入时滚动清理更早行。0 = 永久保留。 */
+        retentionDays: number;
+    };
     /** 是否注册模型可调用的记忆工具。 */
     tools: boolean;
     /** 注册 bench 控制服务（dsh-memory-bench，进程内 rebuild 触发面）。
@@ -171,7 +185,7 @@ export declare const memorySchema: Schema<Schemastery.ObjectS<{
         timeoutMs: Schema<number, number>;
         includePersona: Schema<boolean, boolean>;
         includeSceneNav: Schema<boolean, boolean>;
-        strategy: Schema<"keyword" | "embedding" | "hybrid", "keyword" | "embedding" | "hybrid">;
+        strategy: Schema<"hybrid" | "keyword" | "embedding", "hybrid" | "keyword" | "embedding">;
         scoreThreshold: Schema<number, number>;
         decayHalfLifeDays: Schema<number, number>;
     }>, Schemastery.ObjectT<{
@@ -182,7 +196,7 @@ export declare const memorySchema: Schema<Schemastery.ObjectS<{
         timeoutMs: Schema<number, number>;
         includePersona: Schema<boolean, boolean>;
         includeSceneNav: Schema<boolean, boolean>;
-        strategy: Schema<"keyword" | "embedding" | "hybrid", "keyword" | "embedding" | "hybrid">;
+        strategy: Schema<"hybrid" | "keyword" | "embedding", "hybrid" | "keyword" | "embedding">;
         scoreThreshold: Schema<number, number>;
         decayHalfLifeDays: Schema<number, number>;
     }>>;
@@ -212,6 +226,15 @@ export declare const memorySchema: Schema<Schemastery.ObjectS<{
     llm: Schema<Schemastery.ObjectS<{
         provider: Schema<string, string>;
         model: Schema<string, string>;
+        fallbacks: Schema<({
+            provider?: string | null | undefined;
+            model?: string | null | undefined;
+            reasoningEffort?: "" | "off" | "none" | "minimal" | "low" | "medium" | "high" | "xhigh" | "max" | null | undefined;
+        } & import("@deepseek-ai/cosmokit").Dict)[], Schemastery.ObjectT<{
+            provider: Schema<string, string>;
+            model: Schema<string, string>;
+            reasoningEffort: Schema<"" | "off" | "none" | "minimal" | "low" | "medium" | "high" | "xhigh" | "max", "" | "off" | "none" | "minimal" | "low" | "medium" | "high" | "xhigh" | "max">;
+        }>[]>;
         maxTokens: Schema<number, number>;
         reasoningEffort: Schema<"" | "off" | "none" | "minimal" | "low" | "medium" | "high" | "xhigh" | "max", "" | "off" | "none" | "minimal" | "low" | "medium" | "high" | "xhigh" | "max">;
         temperature: Schema<number, number>;
@@ -220,12 +243,26 @@ export declare const memorySchema: Schema<Schemastery.ObjectS<{
     }>, Schemastery.ObjectT<{
         provider: Schema<string, string>;
         model: Schema<string, string>;
+        fallbacks: Schema<({
+            provider?: string | null | undefined;
+            model?: string | null | undefined;
+            reasoningEffort?: "" | "off" | "none" | "minimal" | "low" | "medium" | "high" | "xhigh" | "max" | null | undefined;
+        } & import("@deepseek-ai/cosmokit").Dict)[], Schemastery.ObjectT<{
+            provider: Schema<string, string>;
+            model: Schema<string, string>;
+            reasoningEffort: Schema<"" | "off" | "none" | "minimal" | "low" | "medium" | "high" | "xhigh" | "max", "" | "off" | "none" | "minimal" | "low" | "medium" | "high" | "xhigh" | "max">;
+        }>[]>;
         maxTokens: Schema<number, number>;
         reasoningEffort: Schema<"" | "off" | "none" | "minimal" | "low" | "medium" | "high" | "xhigh" | "max", "" | "off" | "none" | "minimal" | "low" | "medium" | "high" | "xhigh" | "max">;
         temperature: Schema<number, number>;
         maxInputChars: Schema<number, number>;
         timeoutMs: Schema<number, number>;
     }>>;
+    tokenCost: Schema<Schemastery.ObjectS<{
+        retentionDays: Schema<number, number>;
+    }>, Schemastery.ObjectT<{
+        retentionDays: Schema<number, number>;
+    }>>;
     tools: Schema<boolean, boolean>;
     benchControl: Schema<boolean, boolean>;
 }>, Schemastery.ObjectT<{
@@ -279,7 +316,7 @@ export declare const memorySchema: Schema<Schemastery.ObjectS<{
         timeoutMs: Schema<number, number>;
         includePersona: Schema<boolean, boolean>;
         includeSceneNav: Schema<boolean, boolean>;
-        strategy: Schema<"keyword" | "embedding" | "hybrid", "keyword" | "embedding" | "hybrid">;
+        strategy: Schema<"hybrid" | "keyword" | "embedding", "hybrid" | "keyword" | "embedding">;
         scoreThreshold: Schema<number, number>;
         decayHalfLifeDays: Schema<number, number>;
     }>, Schemastery.ObjectT<{
@@ -290,7 +327,7 @@ export declare const memorySchema: Schema<Schemastery.ObjectS<{
         timeoutMs: Schema<number, number>;
         includePersona: Schema<boolean, boolean>;
         includeSceneNav: Schema<boolean, boolean>;
-        strategy: Schema<"keyword" | "embedding" | "hybrid", "keyword" | "embedding" | "hybrid">;
+        strategy: Schema<"hybrid" | "keyword" | "embedding", "hybrid" | "keyword" | "embedding">;
         scoreThreshold: Schema<number, number>;
         decayHalfLifeDays: Schema<number, number>;
     }>>;
@@ -320,6 +357,15 @@ export declare const memorySchema: Schema<Schemastery.ObjectS<{
     llm: Schema<Schemastery.ObjectS<{
         provider: Schema<string, string>;
         model: Schema<string, string>;
+        fallbacks: Schema<({
+            provider?: string | null | undefined;
+            model?: string | null | undefined;
+            reasoningEffort?: "" | "off" | "none" | "minimal" | "low" | "medium" | "high" | "xhigh" | "max" | null | undefined;
+        } & import("@deepseek-ai/cosmokit").Dict)[], Schemastery.ObjectT<{
+            provider: Schema<string, string>;
+            model: Schema<string, string>;
+            reasoningEffort: Schema<"" | "off" | "none" | "minimal" | "low" | "medium" | "high" | "xhigh" | "max", "" | "off" | "none" | "minimal" | "low" | "medium" | "high" | "xhigh" | "max">;
+        }>[]>;
         maxTokens: Schema<number, number>;
         reasoningEffort: Schema<"" | "off" | "none" | "minimal" | "low" | "medium" | "high" | "xhigh" | "max", "" | "off" | "none" | "minimal" | "low" | "medium" | "high" | "xhigh" | "max">;
         temperature: Schema<number, number>;
@@ -328,12 +374,26 @@ export declare const memorySchema: Schema<Schemastery.ObjectS<{
     }>, Schemastery.ObjectT<{
         provider: Schema<string, string>;
         model: Schema<string, string>;
+        fallbacks: Schema<({
+            provider?: string | null | undefined;
+            model?: string | null | undefined;
+            reasoningEffort?: "" | "off" | "none" | "minimal" | "low" | "medium" | "high" | "xhigh" | "max" | null | undefined;
+        } & import("@deepseek-ai/cosmokit").Dict)[], Schemastery.ObjectT<{
+            provider: Schema<string, string>;
+            model: Schema<string, string>;
+            reasoningEffort: Schema<"" | "off" | "none" | "minimal" | "low" | "medium" | "high" | "xhigh" | "max", "" | "off" | "none" | "minimal" | "low" | "medium" | "high" | "xhigh" | "max">;
+        }>[]>;
         maxTokens: Schema<number, number>;
         reasoningEffort: Schema<"" | "off" | "none" | "minimal" | "low" | "medium" | "high" | "xhigh" | "max", "" | "off" | "none" | "minimal" | "low" | "medium" | "high" | "xhigh" | "max">;
         temperature: Schema<number, number>;
         maxInputChars: Schema<number, number>;
         timeoutMs: Schema<number, number>;
     }>>;
+    tokenCost: Schema<Schemastery.ObjectS<{
+        retentionDays: Schema<number, number>;
+    }>, Schemastery.ObjectT<{
+        retentionDays: Schema<number, number>;
+    }>>;
     tools: Schema<boolean, boolean>;
     benchControl: Schema<boolean, boolean>;
 }>>;

package/dist/config.js

@@ -11,6 +11,7 @@ import { dshHomePath } from '@deepseek-ai/dsh-home-paths';
  * schema（config/settings）、运行时解析（settings.resolveSettings）与 RPC
  * 写入门（stats.settings-set）共用，勿在别处再抄字面量表。
  */
+// satisfies 反向锁定：词汇表扩词必须同步契约的 EffortChoice 联合（host 与 TS 化的 client 共用）
 export const EFFORT_CHOICES = ['', 'off', 'none', 'minimal', 'low', 'medium', 'high', 'xhigh', 'max'];
 export const memorySchema = Schema.object({
     dataDir: Schema.string().default(''),
@@ -68,6 +69,13 @@ export const memorySchema = Schema.object({
     llm: Schema.object({
         provider: Schema.string().default(''),
         model: Schema.string().default(''),
+        // 回退链（#31）：主路由失败后按序降级；每条路由各享全额 timeoutMs（慢 TTFT 模型的
+        // 回退位正是要给它留足首包时间，共享预算会让回退链失效）；条目档位经能力钳制后发送
+        fallbacks: Schema.array(Schema.object({
+            provider: Schema.string().default(''),
+            model: Schema.string().default(''),
+            reasoningEffort: Schema.union([...EFFORT_CHOICES]).default(''),
+        })).default([]),
         // 推理模型（如 v4-flash）的 reasoning 计入输出预算：预算不足会被思考吃光导致正文 0 字符。
         // 0.8.0 起各蒸馏层显式传分层预算（见 llm.ts LAYER_MAX_TOKENS_*），本值为未分层调用的兜底总闸
         maxTokens: Schema.number().min(1024).max(1_000_000).default(65_536),
@@ -80,6 +88,10 @@ export const memorySchema = Schema.object({
         maxInputChars: Schema.number().min(1000).max(1_000_000).default(700_000),
         timeoutMs: Schema.number().min(1000).max(600_000).default(120_000),
     }),
+    // token_cost 明细保留期（写入时滚动清理；0 = 永久保留）。成本看板的「近 N 天」窗口上限也取此值
+    tokenCost: Schema.object({
+        retentionDays: Schema.number().min(0).max(3650).default(365),
+    }),
     tools: Schema.boolean().default(true),
     benchControl: Schema.boolean().default(false),
 });