@angelky7/dsh-browser-control 1.7.9 → 1.7.10

package/README.en.md

@@ -64,8 +64,8 @@ In one line: **for "use MY browser" tasks (logged-in Bilibili, Zhihu, admin pane
 
 | File | Purpose |
 |---|---|
-| [DSH-Browser-Control-v1.7.2.zip](https://github.com/angelky7/dsh-browser-control/releases/download/v1.7.2/DSH-Browser-Control-v1.7.2.zip) | Chrome extension (unzip and load; **the extension no longer changes with feature updates from v1.7.2 on**: v1.7.3–v1.7.9 do not touch the extension — skip this if v1.7.2 is already loaded) |
-| [dsh-browser-control-plugin-v1.7.9.zip](https://github.com/angelky7/dsh-browser-control/releases/download/v1.7.9/dsh-browser-control-plugin-v1.7.9.zip) | dsh plugin (offline fallback; online installs use Option A/B) |
+| [DSH-Browser-Control-v1.7.2.zip](https://github.com/angelky7/dsh-browser-control/releases/download/v1.7.2/DSH-Browser-Control-v1.7.2.zip) | Chrome extension (unzip and load; **the extension no longer changes with feature updates from v1.7.2 on**: v1.7.3–v1.7.10 need no extension reinstall; the v1.7.10 session fix is plugin-side only (extension clipping improvements ship with the next extension update)) |
+| [dsh-browser-control-plugin-v1.7.10.zip](https://github.com/angelky7/dsh-browser-control/releases/download/v1.7.10/dsh-browser-control-plugin-v1.7.10.zip) | dsh plugin (offline fallback; online installs use Option A/B) |
 
 ## Install the Chrome extension (30 seconds)
 
@@ -104,7 +104,7 @@ npm install @angelky7/dsh-browser-control
 
 ```bash
 # Straight from GitHub
-dsh plugin --profile web add "github:angelky7/dsh-browser-control#v1.7.9"
+dsh plugin --profile web add "github:angelky7/dsh-browser-control#v1.7.10"
 
 # Local checkout for debugging (note: the explicit file: prefix is required)
 dsh plugin --profile web add "file:D:\path\to\dsh-browser-control"

package/README.md

@@ -64,8 +64,8 @@ Chrome 扩展（CDP 驱动）
 
 | 文件 | 说明 |
 |---|---|
-| [DSH-Browser-Control-v1.7.2.zip](https://github.com/angelky7/dsh-browser-control/releases/download/v1.7.2/DSH-Browser-Control-v1.7.2.zip) | Chrome 扩展（解压后加载；**扩展从 v1.7.2 起不再随功能更新**：v1.7.3–v1.7.9 都没有动扩展，装过 v1.7.2 的不用重装） |
-| [dsh-browser-control-plugin-v1.7.9.zip](https://github.com/angelky7/dsh-browser-control/releases/download/v1.7.9/dsh-browser-control-plugin-v1.7.9.zip) | dsh 插件（离线兜底，在线装直接用方式 A/B） |
+| [DSH-Browser-Control-v1.7.2.zip](https://github.com/angelky7/dsh-browser-control/releases/download/v1.7.2/DSH-Browser-Control-v1.7.2.zip) | Chrome 扩展（解压后加载；**扩展从 v1.7.2 起不再随功能更新**：v1.7.3–v1.7.10 的功能更新都不需要重装扩展；v1.7.10 的会话修复只改了插件（扩展侧的截断优化随下一次扩展更新）） |
+| [dsh-browser-control-plugin-v1.7.10.zip](https://github.com/angelky7/dsh-browser-control/releases/download/v1.7.10/dsh-browser-control-plugin-v1.7.10.zip) | dsh 插件（离线兜底，在线装直接用方式 A/B） |
 
 ## 安装 Chrome 扩展（30 秒）
 
@@ -104,7 +104,7 @@ npm install @angelky7/dsh-browser-control
 
 ```bash
 # 直接从 GitHub 安装
-dsh plugin --profile web add "github:angelky7/dsh-browser-control#v1.7.9"
+dsh plugin --profile web add "github:angelky7/dsh-browser-control#v1.7.10"
 
 # 本地目录调试（注意：必须显式 file: 前缀）
 dsh plugin --profile web add "file:D:\path\to\dsh-browser-control"

package/lib/index.js

@@ -1284,6 +1284,38 @@ var SessionRegistry = class {
 	}
 };
 //#endregion
+//#region lib/types/text.js
+/**
+* UTF-16 截断守卫（issue #26）。
+*
+* 页面文本按**码元** `slice(0, n)` 会切出孤立代理项（半个 emoji）。`JSON.stringify`
+* 把它写成裸 `\ud83d`，模型的 JSON 解析器要求代理对成对出现，于是**整个请求体**
+* 被 400 拒收；坏字符进了会话历史就每轮重放，只能换会话。
+*
+* 两条规则：
+*  1. 切点落在高位代理项上时退一格——宁可少一个字，不留半个字符；
+*  2. `toWellFormed()` 兜底页面自己提供的坏字符（孤立代理项 → U+FFFD）。
+*/
+/** 把任意字符串变成 well-formed UTF-16（孤立代理项 → U+FFFD）。 */
+function toWellFormed(text) {
+	const fn = text.toWellFormed;
+	if (typeof fn === "function") return fn.call(text);
+	return text.replace(/[\uD800-\uDBFF](?![\uDC00-\uDFFF])|(?<![\uD800-\uDBFF])[\uDC00-\uDFFF]/g, "�");
+}
+/**
+* 按码元上限截断，但不留下半个代理对。
+* @param text - 原文本。
+* @param maxChars - 码元上限（与既有 offset/limit 语义一致）；≤0 返回空串。
+*/
+function sliceWellFormed(text, maxChars) {
+	if (maxChars <= 0) return "";
+	if (text.length <= maxChars) return toWellFormed(text);
+	let end = maxChars;
+	const unit = text.charCodeAt(end - 1);
+	if (unit >= 55296 && unit <= 56319) end -= 1;
+	return toWellFormed(text.slice(0, Math.max(0, end)));
+}
+//#endregion
 //#region lib/types/index.js
 /**
 * Browser-bridge plugin: one local WebSocket endpoint the DSH Browser Control
@@ -1361,11 +1393,23 @@ function defineTool$1(options) {
 	const execute = ((args, exec) => {
 		const agent = exec.agent;
 		const sessionId = typeof agent?.id === "string" && agent.id.length > 0 ? agent.id : void 0;
-		return TOOL_SESSION.run(sessionId, () => inner(args, exec));
+		return TOOL_SESSION.run(sessionId, () => inner(args, exec)).catch((error) => {
+			if (error instanceof Error) error.message = toWellFormed(error.message);
+			throw error;
+		});
 	});
+	const rawRender = options.output?.render;
+	const output = rawRender === void 0 ? options.output : {
+		...options.output,
+		render: (args, value) => rawRender(args, value).map((part) => part.type === "text" && typeof part.text === "string" ? {
+			...part,
+			text: toWellFormed(part.text)
+		} : part)
+	};
 	return defineTool({
 		...options,
-		execute
+		execute,
+		output
 	});
 }
 /** Cordis plugin name used by loader diagnostics. */
@@ -1783,9 +1827,10 @@ function errorMessage(error) {
 */
 const MAX_RENDERED_CHARS = 4e3;
 function clampRendered(text, maxChars = MAX_RENDERED_CHARS) {
-	if (text.length <= maxChars) return text;
-	const omitted = text.length - maxChars;
-	return `${text.slice(0, maxChars)}\n… [renderedTruncated omittedChars=${omitted} limitChars=${maxChars}] narrow the query and call again to see the rest`;
+	const safe = toWellFormed(text);
+	if (safe.length <= maxChars) return safe;
+	const cut = sliceWellFormed(safe, maxChars);
+	return `${cut}\n… [renderedTruncated omittedChars=${safe.length - cut.length} limitChars=${maxChars}] narrow the query and call again to see the rest`;
 }
 /** Model-facing budget for one snapshot listing; `limit` narrows the listing. */
 const SNAPSHOT_RENDER_MAX_CHARS = 8e3;
@@ -1793,8 +1838,8 @@ const SNAPSHOT_RENDER_MAX_CHARS = 8e3;
 const LOG_RENDER_MAX_CHARS = 6e3;
 /** Collapse one page-supplied field to a single line inside a rendered listing. */
 function flattenOneLine(value, maxChars) {
-	const flat = value.replace(/\s+/g, " ").trim();
-	return flat.length <= maxChars ? flat : `${flat.slice(0, maxChars)}…`;
+	const flat = toWellFormed(value.replace(/\s+/g, " ").trim());
+	return flat.length <= maxChars ? flat : `${sliceWellFormed(flat, maxChars)}…`;
 }
 /** `HH:MM:SS` in UTC from an epoch-milliseconds stamp. */
 function formatClock(epochMs) {
@@ -2129,7 +2174,7 @@ function applyBrowserTools(ctx, controller) {
 				tabId: args.tabId
 			}, exec.signal);
 			const full = raw.content ?? "";
-			const content = full.slice(offset, offset + limit);
+			const content = toWellFormed(full.slice(offset, offset + limit));
 			return {
 				tabId: raw.tabId,
 				mode,
@@ -4440,7 +4485,7 @@ function registerSessionsTool(ctx, controller) {
 			},
 			render: (_args, value) => {
 				const v = value;
-				const short = (id) => id.length <= 18 ? id : `${id.slice(0, 16)}…`;
+				const short = (id) => id.length <= 18 ? id : `${sliceWellFormed(id, 16)}…`;
 				if (v.releasedTabId !== void 0) return [{
 					type: "text",
 					text: `Sessions: released your working tab ${v.releasedTabId} — later calls without tabId follow the browser's active tab again.`

package/lib/types/index.js

@@ -29,6 +29,7 @@ import { policyRefusal } from "./policy.js";
 import { AUDIT_MAX_ENTRIES, AuditLog, summarizeAuditParams } from "./audit.js";
 import { Journal, readJournal } from "./journal.js";
 import { SessionRegistry } from "./sessions.js";
+import { sliceWellFormed, toWellFormed } from "./text.js";
 /** Navigation wait ceiling (same as browser_wait); the bridge gets this plus BRIDGE_MARGIN_MS. */
 const NAV_MAX_MS = 120_000;
 /**
@@ -76,12 +77,28 @@ function defineTool(options) {
     const execute = ((args, exec) => {
         const agent = exec.agent;
         const sessionId = typeof agent?.id === 'string' && agent.id.length > 0 ? agent.id : undefined;
-        return TOOL_SESSION.run(sessionId, () => inner(args, exec));
+        return TOOL_SESSION.run(sessionId, () => inner(args, exec)).catch((error) => {
+            // 错误文本同样进入模型请求体：页面异常可能带孤立代理项（issue #26）。
+            // 原地改 message 而不是包一层，保留 BridgeError 的 code/effectState。
+            if (error instanceof Error)
+                error.message = toWellFormed(error.message);
+            throw error;
+        });
     });
+    // 模型可见文本的最后一道路障（issue #26）：任何 render 产出的 text 在交给宿主前
+    // 都修成 well-formed。扩展侧旧版本还在用码元截断，这里必须兜住；否则半个 emoji
+    // 会让整个请求体 400，且坏字符会随会话历史每轮重放。
+    const rawRender = options.output?.render;
+    const output = rawRender === undefined
+        ? options.output
+        : {
+            ...options.output,
+            render: (args, value) => rawRender(args, value).map((part) => (part.type === 'text' && typeof part.text === 'string' ? { ...part, text: toWellFormed(part.text) } : part)),
+        };
     // Call the host helper through a non-generic point: re-instantiating its
     // recursive InferObject types from inside a generic wrapper is what TS chokes on.
     const hostDefine = defineHostTool;
-    return hostDefine({ ...options, execute });
+    return hostDefine({ ...options, execute, output });
 }
 // Re-exported so the policy tables stay testable from the published bundle
 // (verify/verify_policy.mjs reconciles them against the extension's COMMANDS).
@@ -506,10 +523,14 @@ function errorMessage(error) {
  */
 const MAX_RENDERED_CHARS = 4_000;
 function clampRendered(text, maxChars = MAX_RENDERED_CHARS) {
-    if (text.length <= maxChars)
-        return text;
-    const omitted = text.length - maxChars;
-    return `${text.slice(0, maxChars)}\n… [renderedTruncated omittedChars=${omitted} limitChars=${maxChars}] narrow the query and call again to see the rest`;
+    // 页面文本可能自带孤立代理项，也可能是上一次按码元切出来的；先修好再判断长度。
+    const safe = toWellFormed(text);
+    if (safe.length <= maxChars)
+        return safe;
+    const cut = sliceWellFormed(safe, maxChars);
+    // 报数按**实际展示**的长度算：退掉的半个代理对算进 omitted，可见 + omitted = 总长。
+    const omitted = safe.length - cut.length;
+    return `${cut}\n… [renderedTruncated omittedChars=${omitted} limitChars=${maxChars}] narrow the query and call again to see the rest`;
 }
 /** Model-facing budget for one snapshot listing; `limit` narrows the listing. */
 const SNAPSHOT_RENDER_MAX_CHARS = 8_000;
@@ -517,8 +538,8 @@ const SNAPSHOT_RENDER_MAX_CHARS = 8_000;
 const LOG_RENDER_MAX_CHARS = 6_000;
 /** Collapse one page-supplied field to a single line inside a rendered listing. */
 function flattenOneLine(value, maxChars) {
-    const flat = value.replace(/\s+/g, ' ').trim();
-    return flat.length <= maxChars ? flat : `${flat.slice(0, maxChars)}…`;
+    const flat = toWellFormed(value.replace(/\s+/g, ' ').trim());
+    return flat.length <= maxChars ? flat : `${sliceWellFormed(flat, maxChars)}…`;
 }
 /** `HH:MM:SS` in UTC from an epoch-milliseconds stamp. */
 function formatClock(epochMs) {
@@ -835,7 +856,9 @@ function applyBrowserTools(ctx, controller) {
             const limit = Math.min(READ_WINDOW_MAX_CHARS, Math.max(1, Math.floor(Number(args.limit ?? READ_WINDOW_MAX_CHARS)) || READ_WINDOW_MAX_CHARS));
             const raw = await controller.execute('content', args.tabId === undefined ? { mode } : { mode, tabId: args.tabId }, exec.signal);
             const full = raw.content ?? '';
-            const content = full.slice(offset, offset + limit);
+            // 码元窗口可能切在代理对中间；修成 well-formed，长度不变（U+FFFD 也是一码元），
+            // offset/truncated 的算术保持原语义。
+            const content = toWellFormed(full.slice(offset, offset + limit));
             return {
                 tabId: raw.tabId,
                 mode,
@@ -2458,7 +2481,7 @@ function registerSessionsTool(ctx, controller) {
             },
             render: (_args, value) => {
                 const v = value;
-                const short = (id) => (id.length <= 18 ? id : `${id.slice(0, 16)}…`);
+                const short = (id) => (id.length <= 18 ? id : `${sliceWellFormed(id, 16)}…`);
                 if (v.releasedTabId !== undefined) {
                     return [{ type: 'text', text: `Sessions: released your working tab ${v.releasedTabId} — later calls without tabId follow the browser's active tab again.` }];
                 }

package/lib/types/text.d.ts

@@ -0,0 +1,19 @@
+/**
+ * UTF-16 截断守卫（issue #26）。
+ *
+ * 页面文本按**码元** `slice(0, n)` 会切出孤立代理项（半个 emoji）。`JSON.stringify`
+ * 把它写成裸 `\ud83d`，模型的 JSON 解析器要求代理对成对出现，于是**整个请求体**
+ * 被 400 拒收；坏字符进了会话历史就每轮重放，只能换会话。
+ *
+ * 两条规则：
+ *  1. 切点落在高位代理项上时退一格——宁可少一个字，不留半个字符；
+ *  2. `toWellFormed()` 兜底页面自己提供的坏字符（孤立代理项 → U+FFFD）。
+ */
+/** 把任意字符串变成 well-formed UTF-16（孤立代理项 → U+FFFD）。 */
+export declare function toWellFormed(text: string): string;
+/**
+ * 按码元上限截断，但不留下半个代理对。
+ * @param text - 原文本。
+ * @param maxChars - 码元上限（与既有 offset/limit 语义一致）；≤0 返回空串。
+ */
+export declare function sliceWellFormed(text: string, maxChars: number): string;

package/lib/types/text.js

@@ -0,0 +1,36 @@
+/**
+ * UTF-16 截断守卫（issue #26）。
+ *
+ * 页面文本按**码元** `slice(0, n)` 会切出孤立代理项（半个 emoji）。`JSON.stringify`
+ * 把它写成裸 `\ud83d`，模型的 JSON 解析器要求代理对成对出现，于是**整个请求体**
+ * 被 400 拒收；坏字符进了会话历史就每轮重放，只能换会话。
+ *
+ * 两条规则：
+ *  1. 切点落在高位代理项上时退一格——宁可少一个字，不留半个字符；
+ *  2. `toWellFormed()` 兜底页面自己提供的坏字符（孤立代理项 → U+FFFD）。
+ */
+/** 把任意字符串变成 well-formed UTF-16（孤立代理项 → U+FFFD）。 */
+export function toWellFormed(text) {
+    const fn = text.toWellFormed;
+    if (typeof fn === 'function')
+        return fn.call(text);
+    // 老运行时兜底：只替换没有配对的代理项。
+    return text.replace(/[\uD800-\uDBFF](?![\uDC00-\uDFFF])|(?<![\uD800-\uDBFF])[\uDC00-\uDFFF]/g, '\uFFFD');
+}
+/**
+ * 按码元上限截断，但不留下半个代理对。
+ * @param text - 原文本。
+ * @param maxChars - 码元上限（与既有 offset/limit 语义一致）；≤0 返回空串。
+ */
+export function sliceWellFormed(text, maxChars) {
+    if (maxChars <= 0)
+        return '';
+    if (text.length <= maxChars)
+        return toWellFormed(text);
+    let end = maxChars;
+    const unit = text.charCodeAt(end - 1);
+    // 高位代理项（0xD800–0xDBFF）意味着下一码元才是配对的一半：退一格。
+    if (unit >= 0xd800 && unit <= 0xdbff)
+        end -= 1;
+    return toWellFormed(text.slice(0, Math.max(0, end)));
+}

package/package.json

@@ -1,7 +1,7 @@
 {
   "name": "@angelky7/dsh-browser-control",
   "description": "Local WebSocket bridge between dsh and the DSH Browser Control browser extension, with model-facing browser_* tools and a Settings-managed on/off switch",
-  "version": "1.7.9",
+  "version": "1.7.10",
   "type": "module",
   "main": "lib/index.js",
   "types": "lib/types/index.d.ts",
@@ -37,7 +37,7 @@
     "build": "tsc -p tsconfig.json && tsdown",
     "typecheck": "tsc -p tsconfig.json --noEmit",
     "check": "npm run typecheck && npm run build && publint",
-    "verify": "node verify/verify_config_patch.mjs && node verify/verify_peers.mjs && node verify/verify_surface.mjs && node verify/verify_tab_groups.mjs && node verify/verify_tabs_close_many.mjs && node verify/verify_cdp_retry.mjs && node verify/verify_snapshot_ax.mjs && node verify/verify_hard_lock.mjs && node verify/verify_policy.mjs && node verify/verify_cookies.mjs && node verify/verify_extension_e2e.mjs && node verify/verify_plugin.mjs legacy && node verify/verify_plugin.mjs conflict && node verify/verify_plugin.mjs new && node verify/verify_http.mjs && node verify/verify_errors.mjs && node verify/verify_server_hardening.mjs && node verify/verify_hold.mjs && node verify/verify_protocol.mjs && node verify/verify_bundle_http.mjs && node verify/verify_nav_report.mjs && node verify/verify_nav_landing.mjs && node verify/verify_render_contract.mjs && node verify/verify_press_events.mjs && node verify/verify_nav_timeout.mjs && node verify/verify_nav_waituntil.mjs && node verify/verify_archive.mjs && node verify/verify_windows.mjs && node verify/verify_history_nav.mjs && node verify/verify_metrics_throttle.mjs && node verify/verify_drag.mjs && node verify/verify_audit.mjs && node verify/verify_journal.mjs && node verify/verify_sessions.mjs && node verify/verify_snapshot_page.mjs && node verify/verify_compat_matrix.mjs"
+    "verify": "node verify/verify_config_patch.mjs && node verify/verify_peers.mjs && node verify/verify_surface.mjs && node verify/verify_tab_groups.mjs && node verify/verify_tabs_close_many.mjs && node verify/verify_cdp_retry.mjs && node verify/verify_snapshot_ax.mjs && node verify/verify_hard_lock.mjs && node verify/verify_policy.mjs && node verify/verify_cookies.mjs && node verify/verify_extension_e2e.mjs && node verify/verify_plugin.mjs legacy && node verify/verify_plugin.mjs conflict && node verify/verify_plugin.mjs new && node verify/verify_http.mjs && node verify/verify_errors.mjs && node verify/verify_server_hardening.mjs && node verify/verify_text_safety.mjs && node verify/verify_hold.mjs && node verify/verify_protocol.mjs && node verify/verify_bundle_http.mjs && node verify/verify_nav_report.mjs && node verify/verify_nav_landing.mjs && node verify/verify_render_contract.mjs && node verify/verify_press_events.mjs && node verify/verify_nav_timeout.mjs && node verify/verify_nav_waituntil.mjs && node verify/verify_archive.mjs && node verify/verify_windows.mjs && node verify/verify_history_nav.mjs && node verify/verify_metrics_throttle.mjs && node verify/verify_drag.mjs && node verify/verify_audit.mjs && node verify/verify_journal.mjs && node verify/verify_sessions.mjs && node verify/verify_snapshot_page.mjs && node verify/verify_compat_matrix.mjs"
   },
   "publishConfig": {
     "access": "public",