dsh-free-search 0.9.12 → 0.9.13

package/README.md

@@ -307,7 +307,8 @@ provider: bing              # ddg / ddg-lite / bing / searxng / anysearch / exa
 fallbackOn: [quota, auth, bot-wall, transient, invalid-response, unknown]   # 允许触发回退的失败类别；[] = 任何失败都立即中止（默认全部）
 lang: zh                    # 搜索语言池/本地化（zh / en / ru …，不再被设置页强行改写）
 #   lang 决定：Bing 的 mkt/Accept-Language、Wikipedia 子域（ja → ja.wikipedia.org）、地理编码语言；
-#   我们自己渲染的文案只有中文/英文两套，其它语言（ja/ru/de…）一律走英文，避免回答前先塞一段中文（#92）。
+#   我们自己渲染的文案只有中文/英文两套，其它语言（ja/ru/de…）一律走英文，避免回答前先塞一段中文（#92）。
+#   web_fetch（fs-quality-fetch）的 Accept-Language 同样跟随 lang，不再固定中文（#92）。
 uiLang: zh                  # 设置页界面语言（zh / en）；留空则跟随 lang
 bingMarket: zh-CN           # Bing 市场；留空 = 自动（不发送 mkt）
 fetchProviderTakeover: true # false = 保留你显式 pin 的 fetchProvider: http
@@ -489,7 +490,7 @@ Search engine test:
 
 #### 离线自检断言（tools/）
 
-仓库自带 **24 个**零依赖断言脚本（合计 **555 项**），**不需要 API key、不联网、不需要测试框架**，Node ≥ 20 即可跑：
+仓库自带 **24 个**零依赖断言脚本（合计 **558 项**），**不需要 API key、不联网、不需要测试框架**，Node ≥ 20 即可跑：
 
 | 脚本 | 覆盖 | 运行（项数） |
 |---|---|---|
@@ -509,7 +510,7 @@ Search engine test:
 | `assert-killtree-win32.mjs` | Windows 子进程树终止（#74）：不再依赖 `taskkill` | `node tools/assert-killtree-win32.mjs`（8） |
 | `assert-killtree-portable.mjs` | 跨平台 `killTree` 便携版契约 | `node tools/assert-killtree-portable.mjs`（5） |
 | `assert-parallel-mcp.mjs` | Parallel MCP 引擎：25 s 超时/调用方取消覆盖**响应体读取**、共享 SSE 解析器（通知/多行帧）、6 类错误路径、REST 精确时间过滤 | `node tools/assert-parallel-mcp.mjs`（15） |
-| `assert-output-language.mjs` | 语言分支（#92）：`searchLanguage` 支持 zh/en/ja/ru/de/es/fr/ko 其余回退 en、`outputLanguage` 只输出 zh/en、天气文案 ja→英文、置信度说明随语言、失败备注随语言 | `node tools/assert-output-language.mjs`（5） |
+| `assert-output-language.mjs` | 语言分支（#92）：`searchLanguage` 支持 zh/en/ja/ru/de/es/fr/ko 其余回退 en、`outputLanguage` 只输出 zh/en、天气文案 ja→英文、置信度说明随语言、失败备注随语言 | `node tools/assert-output-language.mjs`（8） |
 | `assert-uncertainty.mjs` | 输出契约：0/1 源为 low、2 源与同站为 medium、3 个不同域为 high、时间过滤无日期、失败备注、聚合站提示 | `node tools/assert-uncertainty.mjs`（8） |
 | `assert-multikey.mjs` | 多 key 字段：单 key 不变、逗号/分号/换行分隔、空白项丢弃、空输入为空数组 | `node tools/assert-multikey.mjs`（3） |
 | `assert-firecrawl-keyless.mjs` | Firecrawl keyless 出口判定：403 + VPN/proxy 文案才算拒绝、普通 403 与 401/429/500 不算、冷却初始未触发 | `node tools/assert-firecrawl-keyless.mjs`（5） |
@@ -844,7 +845,8 @@ provider: bing              # ddg / ddg-lite / bing / searxng / anysearch / exa
 fallbackOn: [quota, auth, bot-wall, transient, invalid-response, unknown]   # failure classes allowed to trigger fallback; [] = abort on any failure (default: all)
 lang: zh                    # search language pools / localization (zh / en / ru …; no longer rewritten by the settings page)
 #   lang drives Bing's mkt/Accept-Language, the Wikipedia subdomain (ja → ja.wikipedia.org) and the geocoding language;
-#   text this plugin renders itself only exists in Chinese and English, so other languages (ja/ru/de…) get English rather than a wall of Chinese (#92).
+#   text this plugin renders itself only exists in Chinese and English, so other languages (ja/ru/de…) get English rather than a wall of Chinese (#92).
+#   web_fetch (fs-quality-fetch) sends the same Accept-Language instead of a fixed Chinese one (#92).
 uiLang: zh                  # settings UI language (zh / en); empty = follow lang
 bingMarket: zh-CN           # Bing market; empty = automatic (no mkt sent)
 fetchProviderTakeover: true # false = keep the fetchProvider: http you pinned yourself
@@ -1026,7 +1028,7 @@ Restart `dsh web` after switching to apply changes.
 
 #### Offline self-check assertions (`tools/`)
 
-**Twenty-four** zero-dependency assertion scripts ship with the repo (**555 assertions** total): **no API keys, no network, no test framework**, Node ≥ 20:
+**Twenty-four** zero-dependency assertion scripts ship with the repo (**558 assertions** total): **no API keys, no network, no test framework**, Node ≥ 20:
 
 | Script | Covers | Run (count) |
 |---|---|---|
@@ -1046,7 +1048,7 @@ Restart `dsh web` after switching to apply changes.
 | `assert-killtree-win32.mjs` | Windows process-tree termination (#74): no longer relies on `taskkill` | `node tools/assert-killtree-win32.mjs` (8) |
 | `assert-killtree-portable.mjs` | portable `killTree` contract | `node tools/assert-killtree-portable.mjs` (5) |
 | `assert-parallel-mcp.mjs` | Parallel MCP engine: the 25 s timeout / caller abort covers the **body read**, shared SSE parser (notifications, multiline frames), six error paths, REST precise time filtering | `node tools/assert-parallel-mcp.mjs` (15) |
-| `assert-output-language.mjs` | language branches (#92): `searchLanguage` passes zh/en/ja/ru/de/es/fr/ko through and falls back to en, `outputLanguage` only returns zh/en, weather text is English for ja, and both confidence and failure notes follow the language | `node tools/assert-output-language.mjs` (5) |
+| `assert-output-language.mjs` | language branches (#92): `searchLanguage` passes zh/en/ja/ru/de/es/fr/ko through and falls back to en, `outputLanguage` only returns zh/en, weather text is English for ja, and both confidence and failure notes follow the language | `node tools/assert-output-language.mjs` (8) |
 | `assert-uncertainty.mjs` | output contract: 0/1 source is low, two sources or a single site is medium, three distinct domains is high, time filter without dates, failure notes, aggregator caveat | `node tools/assert-uncertainty.mjs` (8) |
 | `assert-multikey.mjs` | multi-key fields: a single key is unchanged, comma/semicolon/newline separate keys, blank entries are dropped, empty input yields none | `node tools/assert-multikey.mjs` (3) |
 | `assert-firecrawl-keyless.mjs` | Firecrawl keyless egress refusal: only a 403 with the VPN/proxy wording counts, plain 403 and 401/429/500 do not, and the cooldown starts inactive | `node tools/assert-firecrawl-keyless.mjs` (5) |

package/lib/client.js

@@ -426,7 +426,7 @@ window.__ModuleLoader__.load({
     };
     const tt = (lang) => I18N[String(lang ?? "").trim().toLowerCase() === "zh" ? "zh" : "en"];
     // 当前插件版本（与 lib/index.js 的 PLUGIN_VERSION 保持一致；跟随上游 v0.7.5）
-    const PLUGIN_VERSION = "0.9.12";
+    const PLUGIN_VERSION = "0.9.13";
     const PLUGIN_REPO_URL = "https://github.com/DDDMUC/dsh-free-search";
     // 右侧浮动目录（#85）：条目顺序即页面顺序，anchor 是页面里对应 field 的 id。
     const NAV_ITEMS = [

package/lib/fetch_provider.js

@@ -274,11 +274,11 @@ export function httpsGetFollow(url, { headers = {}, timeoutMs = 20000, signal }
 }
 
 // 完整浏览器请求头（Chrome 128 形态）：部分反爬站（如百度系）对精简头直接 403，完整头可通过。
+// Accept-Language 不在这里：它跟随 lang（#92），由调用方逐次传入，否则日语环境抓到的仍是中文版页面。
 // 注意：声明了 Accept-Encoding 就必须解压（httpsGetFollow 已支持 gzip/br/deflate）。
 const BROWSERISH = {
   "User-Agent": "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/128.0.0.0 Safari/537.36",
   Accept: "text/html,application/xhtml+xml,application/xml;q=0.9,image/avif,image/webp,image/apng,*/*;q=0.8",
-  "Accept-Language": "zh-CN,zh;q=0.9,en;q=0.8",
   "Accept-Encoding": "gzip, deflate, br",
   "sec-ch-ua": '"Chromium";v="128", "Not;A=Brand";v="24", "Google Chrome";v="128"',
   "sec-ch-ua-mobile": "?0",
@@ -290,7 +290,11 @@ const BROWSERISH = {
   "Upgrade-Insecure-Requests": "1",
 };
 
-export function createFetchProvider({ resolveApiKey, logger }) {
+/** 请求头：Accept-Language 跟随调用方（#92），其余固定为浏览器形态。 */
+export function browserishHeaders(acceptLanguage = "en-US,en;q=0.9") {
+  return { ...BROWSERISH, "Accept-Language": String(acceptLanguage ?? "").trim() || "en-US,en;q=0.9" };
+}
+export function createFetchProvider({ resolveApiKey, logger, acceptLanguage = () => "en-US,en;q=0.9" }) {
   return {
     id: "fs-quality-fetch",
     /** 廉价本地可用性检查（seam 契约要求，禁止网络调用）。 */
@@ -302,7 +306,7 @@ export function createFetchProvider({ resolveApiKey, logger }) {
 
       // T1：直连（文本类内容；二进制交给后层——exa 取正文，PDF 等仍由 firecrawl 处理）
       try {
-        const r = await httpsGetFollow(url, { headers: BROWSERISH, signal });
+        const r = await httpsGetFollow(url, { headers: browserishHeaders(acceptLanguage()), signal });
         if (r.status < 400 && TEXTUAL.test(r.contentType) && r.text.length >= T1_MIN_CHARS) {
           return { url: r.url, statusCode: r.status, body: { kind: "html", content: r.text }, truncated: false };
         }

package/lib/index.js

@@ -164,6 +164,13 @@ const USER_AGENT =
 const ACCEPT_LANG = "zh-CN,zh;q=0.9,en;q=0.8";
 // 当前请求生效的 Accept-Language：由 lang（或显式 bingMarket）决定，搜索前设置（#92）。
 let activeAcceptLanguage = ACCEPT_LANG;
+/** 由配置推导 Accept-Language：显式 bingMarket 优先，否则跟随 lang（未知回退 en）。 */
+export function acceptLanguageFor(cfg) {
+  const market = cfg?.bingMarket;
+  if (market && MARKET_TO_LANG[market]) return MARKET_TO_LANG[market];
+  return LANG_PROFILES[searchLanguage(configLanguage(cfg))].acceptLang;
+}
+
 export function setActiveAcceptLanguage(value) {
   if (typeof value === "string" && value.trim() !== "") activeAcceptLanguage = value.trim();
 }
@@ -180,6 +187,18 @@ export function searchLanguage(lang) {
   return Object.prototype.hasOwnProperty.call(LANG_PROFILES, tag) ? tag : "en";
 }
 
+/**
+ * 配置里生效的语言标签：未设置时按 schema 默认 zh（#92）。
+ * searchLanguage() 只负责「未知值回退 en」，两者不要混用。
+ */
+export function configLanguage(cfg) {
+  const value = cfg?.lang;
+  if (value === undefined || value === null) return "zh";
+  const tag = String(value).trim().toLowerCase();
+  return tag === "" ? "zh" : tag;
+}
+
+
 export function outputLanguage(lang) {
   return searchLanguage(lang) === "zh" ? "zh" : "en";
 }
@@ -380,7 +399,7 @@ function routeEngines(query, { timeRange, disabledEngines = [], fallbackOrder =
 }
 
 // 当前插件版本（发布时与 package.json 同步；跟随上游 v0.7.5，不在 PR 里 bump）
-const PLUGIN_VERSION = "0.9.12";
+const PLUGIN_VERSION = "0.9.13";
 // 检查更新的 npm registry 元数据地址（dsh-free-search 是 npmjs 上的公开包）
 const NPM_REGISTRY_URL = "https://registry.npmjs.org/dsh-free-search/latest";
 const PLUGIN_NPM_URL = "https://www.npmjs.com/package/dsh-free-search";
@@ -742,13 +761,13 @@ function extractBingUrl(href) {
 async function searchBing(query, maxResults, options, signal) {
   // mkt：仅显式 bingMarket 才发。默认不发——mkt=zh-CN 在部分地区/线路会返回空 SERP 容器
   // （#71：海外 IP + mkt=zh-CN → b_algo=0），Bing 会按 Accept-Language 自行本地化。
-  const profile = LANG_PROFILES[options?.lang] ?? LANG_PROFILES.zh;
+  const profile = LANG_PROFILES[searchLanguage(configLanguage(options))];
   const params = new URLSearchParams({ q: query });
   if (options?.bingMarket) params.set("mkt", options.bingMarket);
   // Accept-Language：显式 bingMarket 时按 market 反推（避免中文 header 污染），否则用 lang 档案
   const acceptLang = options?.bingMarket
-    ? (MARKET_TO_LANG[options.bingMarket] ?? ACCEPT_LANG)
-    : (profile.acceptLang ?? ACCEPT_LANG);
+    ? (MARKET_TO_LANG[options.bingMarket] ?? activeAcceptLanguage)
+    : (profile.acceptLang ?? activeAcceptLanguage);
   const adlt = options?.safeSearch ?? "off";
   if (adlt === "off") params.set("adlt", "off");
   else if (adlt === "moderate") params.set("adlt", "moderate");
@@ -1383,7 +1402,7 @@ function sliceBalancedObject(text, start) {
 // json 字段 murl(真实播放地址) / vt(标题) / du(时长) / purl(详情页)（字段语义以实测为准）。
 async function searchBingVideos(query, maxResults, options, signal) {
   const params = new URLSearchParams({ q: query, count: String(Math.max(Number(maxResults) || 10, 1)), first: "1", mmasync: "1" });
-  const market = options?.bingMarket ?? (LANG_PROFILES[options?.lang] ?? LANG_PROFILES.zh).market;
+  const market = options?.bingMarket ?? LANG_PROFILES[searchLanguage(configLanguage(options))].market;
   if (market) params.set("mkt", market);
   const html = await fetchVideoHtml(`${BING_VIDEOS_ASYNC_URL}?${params}`, signal);
   if (/captcha|unusual traffic|robot check/i.test(html.slice(0, 5000))) {
@@ -3822,10 +3841,8 @@ function apply(ctx, config) {
       // #72：本次请求生效的域名黑名单（模块级活动值；各结果咽喉点过滤读取它）
       setActiveBlockedDomains(cfg.blockedDomains);
       // #92：Accept-Language 跟随 lang（显式 bingMarket 优先），不再固定中文
-      setActiveAcceptLanguage(
-        (cfg.bingMarket && MARKET_TO_LANG[cfg.bingMarket]) || LANG_PROFILES[searchLanguage(cfg.lang)].acceptLang
-      );
-      setActiveOutputLanguage(outputLanguage(cfg.lang));
+      setActiveAcceptLanguage(acceptLanguageFor(cfg));
+      setActiveOutputLanguage(outputLanguage(configLanguage(cfg)));
       const disabledEngines = normalizeDisabledEngines(cfg.disabledEngines);
       const fallbackOnSet = new Set(normalizeFallbackOn(cfg.fallbackOn));
       const disabledSet = new Set(disabledEngines);
@@ -4371,7 +4388,12 @@ function apply(ctx, config) {
   // web_fetch 质量链接管：注册 fs-quality-fetch（直连 → exa contents → firecrawl → crawl4ai），
   // 未设置 fetchProviderId、或仍是出厂默认 http 时接管；已被显式指向其他 provider 时不抢占。
   // 与 web_search 的接管规则同构：运行时兜底，不依赖 bundle patch 是否落地。
-  const fetchProvider = createFetchProvider({ resolveApiKey, logger });
+  const fetchProvider = createFetchProvider({
+    resolveApiKey,
+    logger,
+    // #92：web_fetch 不经过 searchOnce，所以这里按当前配置现算，而不是复用上次搜索留下的值。
+    acceptLanguage: () => acceptLanguageFor(current()),
+  });
   registerOnce(() => ctx.web.registerFetchProvider(fetchProvider), fetchProvider, "fetch provider");
 
   // 接管必须先等 settings 服务就绪：rc.2 起 provider 选择是 config 的 volatile ref,
@@ -5049,7 +5071,7 @@ function apply(ctx, config) {
               );
             }
             const limit = Math.min(args.maxResults ?? 5, 10);
-            let result = await searchPlatform(platform, args.query, limit, undefined, current().lang);
+            let result = await searchPlatform(platform, args.query, limit, undefined, configLanguage(current()));
             if (platform === "vimeo" && (result.sources ?? []).length === 0) {
               // Vimeo 搜索页有时不返回结果：退回网页搜索（site: 过滤）
               try {
@@ -5194,7 +5216,7 @@ function apply(ctx, config) {
           output: {
             schema: WEATHER_OUTPUT_SCHEMA,
             render(args, value) {
-              const lang = outputLanguage(current().lang);
+              const lang = outputLanguage(configLanguage(current()));
               const c = value.current ?? {};
               const lines = [];
               lines.push(lang === "en" ? `Weather for ${value.location}:` : `${value.location} 天气：`);
@@ -5215,7 +5237,7 @@ function apply(ctx, config) {
           },
           async execute(args) {
             // 天气正文只提供 zh/en（回退 en），地理编码用真实搜索语言（ja 就查日文地名）。
-            const lang = searchLanguage(current().lang);
+            const lang = searchLanguage(configLanguage(current()));
             const location = String(args.location ?? "").trim();
             if (!location) throw new Error("weather: location is required");
             return sanitizeWeatherResult(await fetchWeather(location, args.days, lang, undefined));
@@ -5324,7 +5346,7 @@ function apply(ctx, config) {
                   sources: result.sources ?? [],
                   timeRange: Boolean(request.timeRange),
                   note: typeof result.content === "string" ? result.content : "",
-                  lang: current().lang,
+                  lang: configLanguage(current()),
                 }),
               },
             };
@@ -5424,7 +5446,7 @@ function apply(ctx, config) {
                 uncertainty: assessUncertainty({
                   sources: toolResult.sources ?? [],
                   note: typeof toolResult.content === "string" ? toolResult.content : "",
-                  lang: current().lang,
+                  lang: configLanguage(current()),
                 }),
               },
             };

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "dsh-free-search",
-  "version": "0.9.12",
+  "version": "0.9.13",
   "description": "Free web search for DeepSeek Harness: 27 engines (Bing/DuckDuckGo/AnySearch/SearXNG/Exa/Tavily/Keenable/Firecrawl/Parallel keyless; Zhihu global/site via MCP and Perplexity/SerpBase/DeepSeek (official + native tools/search)/You.com/Baidu/Kimi/Aliyun/Doubao with key; OpenAI/Gemini/Claude model built-in search, explicit-only) + auto smart routing + global engine enable/disable & fallback priority + editable zh/en language pools + multi mode (parallel full language-pool merge) + multi_search cross-source tool + failure-aware fallback (quota/auth cooldown, transient retry, fallbackOn) + time filtering + platform search + domain blocklist (blockedDomains) + web_fetch, with web settings UI.",
   "type": "module",
   "main": "lib/index.js",