dsh-token-total 1.1.2 → 1.1.4

package/README.md

@@ -15,6 +15,7 @@ DSH（DeepSeek Harness）Web 插件：**Token 用量统计看板** + **单会话
 ### 统计看板
 
 入口在左侧边栏的插件区（与原生「插件」页并列的独立页面），点开即为看板，分上中下三部分。
+![统计看板](assets/image.png)
 
 #### 上部
 
@@ -40,15 +41,18 @@ DSH（DeepSeek Harness）Web 插件：**Token 用量统计看板** + **单会话
 
 会话输入框下方原生的「用时与速度 / Token 用量」两枚胶囊被接管并增强：
 
+![单会话统计](assets/image-1.png)
+
 - **Token 胶囊**显示的是**整棵会话树**（主智能体 + 全部子智能体）的合计，点开弹窗分为「主智能体 / 子智能体 / 会话合计」三行，并给出子智能体会话数。
 - 子智能体是独立持久化的会话，通过日志头的 `parentSession` 串成树；插件按需重放这些日志汇总（同一套 `(turn, step)` 去重规则，且跳过 fork 继承来的事件前缀，避免与父会话重复计数）。
 - **会话 Token 明细始终开启**：未命中缓存、缓存读 / 写、输出分别列出；底部胶囊的缓存命中率始终显示两位小数。
-- **流式吞吐速度始终开启**：始终显示当前的实时解码速度——输出过程中按已到达的流式增量实测（每步开始约 0.2 秒内出首个读数）；一步结算后保持该步最后的实测值，工具执行、步间空档也不回落到会话均值，只有从未产生过输出的会话才显示会话累计的精确值。
+- **流式吞吐速度始终开启**：输出过程中显示当前这一步的实时解码速度——按已到达的流式增量实测（每步开始约 0.2 秒内出首个读数）；一旦没有步骤在流（一步刚结算、工具执行中、步间空档），就交回本轮会话的加权平均解码速度（Σ 输出 token ÷ Σ 解码耗时，与会话「用量」页里的平均速率同口径），而不是停在最后那次短窗口读数上——那正是底部 57 tok/s 与请求记录 156 token/秒看起来互相矛盾的原因。
 - 不再提供底部增强功能的开关；旧版本保存的关闭设置不再影响显示。
 
 ### 会话「用量」页
 
 在每个会话原生的「对话 / 轨迹」旁新增 **用量** tab；沿用会话切换与视图导航，不另造顶部栏或输入框。
+![会话用量页](assets/image-2.png)
 
 - **两张实时卡**（形制参考 [magpie](https://github.com/yetone/magpie)）并排位于子代理／模型明细之下、请求记录之上，回答「提示词里现在有什么」与「整场会话到底花了什么」两个不同的问题：
   - 左卡「**当前上下文**」：`已用 / 模型窗口 tokens` 大字附占用百分比，标题行右侧是当前供应商 · 模型；下方一条**整条即模型窗口**的堆叠条——有颜色的部分是已用的七类（按类别占比分色），尾部是还没用上的余量，所以条本身就是占用率；再下面按统一图例列出系统提示词、用户消息、技能注入、工具结果、工具定义、注入内容、助手消息各自的占比与估算 token（输出不占窗口，故不列）。鼠标悬浮可读精确 token 数与占上下文比例；模型窗口未知时只写已用数。

package/lib/client.js

@@ -31531,7 +31531,8 @@ window.__ModuleLoader__.load({
 				self,
 				subagents: tree?.subagents ?? { ...ZERO },
 				total: tree === null ? self : combinedTotals(tree, self),
-				subagentSessions: tree?.subagentSessions ?? 0
+				subagentSessions: tree?.subagentSessions ?? 0,
+				...tree?.coverage ? { coverage: tree.coverage } : {}
 			};
 		}
 		function hasAnyTokens(totals) {
@@ -31661,14 +31662,14 @@ window.__ModuleLoader__.load({
 		* cannot see.
 		*
 		* Streaming throughput replaces the gauge pill's speed
-		* figure with a live estimate of what is happening now. The figure is the
-		* streaming step's measured rate while one streams and the LAST measured rate
-		* once it settles — a session average describes the whole session's past, and
-		* it cannot match the pace of the answer on screen, which is what users read
-		* as the pill disagreeing with the stream in front of them. The official
-		* session figure only shows on a session that has never streamed. That
-		* subscription lives in
-		* its own child component so stream deltas never re-render the pills. The live
+		* figure with a live estimate of what is happening now, and hands the slot back
+		* to the session's decode average when nothing is streaming. While a step
+		* streams the live rate is the only reading that can match the pace of the
+		* answer on screen; once the step settles it becomes a figure from a window
+		* nobody is watching any more, so the pill returns to the same quotient the
+		* usage view quotes (ΣoutputTokens / ΣdecodeMs) instead of freezing on the last
+		* measurement. That subscription lives in its own child component, so stream
+		* deltas never re-render the pills. The live
 		* figure itself is measured by a bounded sliding window over observed token
 		* growth (stats-line-core's StreamingRateSampler): the earlier cumulative
 		* quotient charged a whole step's text to a window that had just opened, so a
@@ -31754,14 +31755,23 @@ window.__ModuleLoader__.load({
 		* The gauge pill's always-on streaming speed slot.
 		*
 		* It subscribes to `legacy.partial` itself so per-frame deltas re-render only
-		* this child, not the surrounding pills. The figure it carries is the CURRENT
-		* stream's measured rate, and it stays that rate once the step settles:
-		* tool execution and the gap between steps happen inside the same answer, and
-		* the session average this slot used to fall back to there is a figure from
-		* the whole session's past — a number that cannot match the pace of the answer
-		* on screen, which is what reads as "not the real-time speed". The official
-		* session figure is only a session's opening state, shown until the first
-		* window measures anything.
+		* this child, not the surrounding pills. Two figures share the slot, chosen by
+		* whether a step is streaming right now:
+		*
+		* - while one is (`partial != null`), the measured rate of THAT stream — a
+		*   bounded sliding window over observed token growth (stats-line-core's
+		*   StreamingRateSampler), the only reading that can track the pace of the
+		*   answer currently on screen;
+		* - otherwise the session's own decode average, ΣoutputTokens / ΣdecodeMs —
+		*   the same quotient `decodeRateOf` quotes in the usage view. An idle composer
+		*   must not freeze on whatever the last window happened to measure: a settled
+		*   thirty-step turn kept showing its final second-long reading (57 tok/s)
+		*   beside the usage view's 156 tok/s for the same work, and the two read as a
+		*   contradiction rather than as two scopes of one number.
+		*
+		* The measured reading carries the step it was taken in, so a new step cannot
+		* inherit the previous one's figure before its own window publishes; nothing is
+		* held across the gap — the gap is exactly when the average applies.
 		*
 		* One window covers a whole step, across every output kind the accumulator
 		* carries — reasoning, answer text, and Tool-call arguments — because the
@@ -31772,7 +31782,8 @@ window.__ModuleLoader__.load({
 		* state, so no render reads a clock or mutates the window: a render React starts
 		* and then discards must not decide what the next committed one displays. The
 		* rate itself comes from stats-line-core's StreamingRateSampler — a bounded
-		* window over observed token growth, smoothed, and held while nothing arrives.
+		* window over observed token growth, smoothed, and held while nothing arrives
+		* inside the step it belongs to.
 		*/
 		function LiveSpeed({ useChat, scale, fallback, t }) {
 			const partial = useChat((s) => s.legacy.partial);
@@ -31788,13 +31799,17 @@ window.__ModuleLoader__.load({
 			const scaleRef = (0, react.useRef)(scale);
 			const [beat, setBeat] = (0, react.useState)(0);
 			(0, react.useLayoutEffect)(() => {
+				if (key === null) return;
 				if (scaleRef.current !== scale) {
 					scaleRef.current = scale;
 					rate.reset();
 				}
 				rate.observe(tokens, Date.now());
 				const next = rate.ratePerSecond();
-				if (next !== null) setReading((previous) => previous === next ? previous : next);
+				if (next !== null) setReading((previous) => previous !== null && previous.key === key && previous.value === next ? previous : {
+					key,
+					value: next
+				});
 			}, [
 				tokens,
 				key,
@@ -31812,7 +31827,8 @@ window.__ModuleLoader__.load({
 					clearInterval(handle);
 				};
 			}, [streaming]);
-			return /* @__PURE__ */ (0, react_jsx_runtime.jsx)(SpeedFigure, { text: reading === null ? fallback : t("stats.tokensPerSecond", { throughput: formatTokensPerSecond(reading) }) });
+			const measured = reading !== null && reading.key === key ? reading.value : null;
+			return /* @__PURE__ */ (0, react_jsx_runtime.jsx)(SpeedFigure, { text: measured === null ? fallback : t("stats.tokensPerSecond", { throughput: formatTokensPerSecond(measured) }) });
 		}
 		function TimePill({ stats, t, dialog, useChat, scale, dock }) {
 			const { open, setOpen, rootRef, panelRef, pos } = useStatDialog(dialog);
@@ -31886,9 +31902,11 @@ window.__ModuleLoader__.load({
 		}
 		function UsagePill({ usage, tree, t, dialog, dock }) {
 			const { open, setOpen, rootRef, panelRef, pos } = useStatDialog(dialog);
-			const billed = sessionUsageForDisplay(usage, tree?.tree ?? null).total;
+			const display = sessionUsageForDisplay(usage, tree?.tree ?? null);
+			const billed = display.total;
 			const total = billedTotal(billed);
-			const totalText = t("message.turnUsage.count", { count: formatTokensCompact(total) });
+			const incomplete = display.coverage?.complete === false;
+			const totalText = (incomplete ? "≥ " : "") + t("message.turnUsage.count", { count: formatTokensCompact(total) });
 			const cacheHit = cacheHitPercent(billed);
 			const cacheHitText = cacheHit !== null ? t("stats.cacheHit", { percent: cacheHit }) : null;
 			const exactCount = (value) => t("message.turnUsage.count", { count: formatTokens(value) });
@@ -31901,7 +31919,8 @@ window.__ModuleLoader__.load({
 					className: StatsLineEnhanced_module_default.pill,
 					"aria-haspopup": "dialog",
 					"aria-expanded": open,
-					"aria-label": cacheHitText === null ? totalText : totalText + " · " + cacheHitText,
+					"aria-label": (incomplete ? t("stats.partial") + " · " : "") + (cacheHitText === null ? totalText : totalText + " · " + cacheHitText),
+					title: incomplete ? t("stats.incomplete", { count: display.coverage.unreadableSessions.length }) : void 0,
 					onClick: () => {
 						setOpen(!open);
 					},
@@ -31928,13 +31947,18 @@ window.__ModuleLoader__.load({
 								children: [/* @__PURE__ */ (0, react_jsx_runtime.jsx)(DatabaseIcon, {}), t("stats.dialog.usageTitle")]
 							}), /* @__PURE__ */ (0, react_jsx_runtime.jsx)("span", {
 								className: StatsLineEnhanced_module_default.dialogTitleValue,
-								children: exactCount(total)
+								children: (incomplete ? "≥ " : "") + exactCount(total)
 							})]
 						}),
 						/* @__PURE__ */ (0, react_jsx_runtime.jsx)("div", {
 							className: StatsLineEnhanced_module_default.dialogTitleRule,
 							"aria-hidden": true
 						}),
+						incomplete && /* @__PURE__ */ (0, react_jsx_runtime.jsx)("p", {
+							role: "status",
+							"data-session-usage-partial": true,
+							children: t("stats.incomplete", { count: display.coverage.unreadableSessions.length })
+						}),
 						/* @__PURE__ */ (0, react_jsx_runtime.jsxs)("dl", {
 							className: StatsLineEnhanced_module_default.dialogDetails,
 							"data-session-stats-usage": true,
@@ -31967,7 +31991,8 @@ window.__ModuleLoader__.load({
 			const stats = (0, react.useMemo)(() => projected ?? deriveStats(settledNodes ?? []), [projected, settledNodes]);
 			const tokenScale = (0, react.useMemo)(() => measuredTokenScale(settledNodes ?? []) ?? 1, [settledNodes]);
 			const tree = useSessionUsage(sessionId, usageSignature(usage, stats.steps));
-			const hasTokens = hasAnyTokens(sessionUsageForDisplay(usage, tree.tree).total);
+			const display = sessionUsageForDisplay(usage, tree.tree);
+			const hasTokens = hasAnyTokens(display.total) || display.coverage?.complete === false;
 			if (stats.steps === 0 && !hasTokens) return null;
 			return /* @__PURE__ */ (0, react_jsx_runtime.jsxs)("div", {
 				ref: markDockContainer,
@@ -32033,7 +32058,8 @@ window.__ModuleLoader__.load({
 		const StatsUsageEntry = (0, react.memo)(function StatsUsageEntry({ useProjection, t, sessionId }) {
 			const usage = useProjection("tokenUsage");
 			const tree = useSessionUsage(sessionId, usageSignature(usage, usage === void 0 ? 0 : 1));
-			if (!hasAnyTokens(sessionUsageForDisplay(usage, tree.tree).total)) return null;
+			const display = sessionUsageForDisplay(usage, tree.tree);
+			if (!(hasAnyTokens(display.total) || display.coverage?.complete === false)) return null;
 			return /* @__PURE__ */ (0, react_jsx_runtime.jsx)(UsagePill, {
 				usage,
 				tree,
@@ -32405,6 +32431,8 @@ window.__ModuleLoader__.load({
 			"stats.cacheMiss": "Cache miss",
 			"stats.dialog.title": "Time & speed",
 			"stats.dialog.usageTitle": "Token usage",
+			"stats.partial": "Partial usage",
+			"stats.incomplete": "{count} subagent session(s) could not be read; totals are incomplete.",
 			"stats.dialog.llmTime": "LLM time",
 			"stats.dialog.toolTime": "Tool time",
 			"stats.dialog.ttft": "Mean TTFT",
@@ -32493,6 +32521,8 @@ window.__ModuleLoader__.load({
 			"stats.cacheMiss": "未命中缓存",
 			"stats.dialog.title": "用时与速度",
 			"stats.dialog.usageTitle": "Token 用量",
+			"stats.partial": "部分用量",
+			"stats.incomplete": "有 {count} 个子会话暂不可读取，总计尚不完整。",
 			"stats.dialog.llmTime": "LLM 用时",
 			"stats.dialog.toolTime": "工具用时",
 			"stats.dialog.ttft": "首 token 平均",
@@ -32581,6 +32611,8 @@ window.__ModuleLoader__.load({
 			"stats.cacheMiss": "未命中快取",
 			"stats.dialog.title": "用時與速度",
 			"stats.dialog.usageTitle": "Token 用量",
+			"stats.partial": "部分用量",
+			"stats.incomplete": "有 {count} 個子會話暫不可讀取，總計尚不完整。",
 			"stats.dialog.llmTime": "LLM 用時",
 			"stats.dialog.toolTime": "工具用時",
 			"stats.dialog.ttft": "首 token 平均",
@@ -33829,6 +33861,15 @@ window.__ModuleLoader__.load({
 					className: SessionUsageView_module_default.scope,
 					children: t("scope")
 				}),
+				detail.coverage?.complete === false && /* @__PURE__ */ (0, react_jsx_runtime.jsxs)("div", {
+					className: SessionUsageView_module_default.error,
+					role: "status",
+					"data-session-usage-partial": true,
+					children: [t("partialCoverage", { count: String(detail.coverage.unreadableSessions.length) }), /* @__PURE__ */ (0, react_jsx_runtime.jsx)("span", {
+						title: detail.coverage.unreadableSessions.map((issue) => issue.sessionId).join("\n"),
+						children: t("partialTotals")
+					})]
+				}),
 				/* @__PURE__ */ (0, react_jsx_runtime.jsx)(Metrics, {
 					detail,
 					t
@@ -33957,6 +33998,8 @@ window.__ModuleLoader__.load({
 			noModels: "暂无模型用量",
 			scope: "本会话及所有子代理 · 历史日志累计",
 			note: "统计包含本会话与全部后代的独立请求、重试和压缩调用，不重复计算继承历史。首 token 与速度仅使用已记录的时间信息，缺失时显示 —。",
+			partialCoverage: "有 {count} 个子会话暂不可读取。",
+			partialTotals: "当前仅显示已读取用量，未将缺失用量视为零。",
 			failedLoad: "用量读取失败，已保留上次结果。",
 			retry: "重新读取",
 			completed: "已完成",
@@ -34065,6 +34108,8 @@ window.__ModuleLoader__.load({
 			noModels: "No model usage yet",
 			scope: "This session and all subagents · Historical log totals",
 			note: "Includes independent requests, retries and compaction calls from this session and all descendants, without counting inherited history twice. Timing uses recorded timestamps only; missing values appear as —.",
+			partialCoverage: "{count} subagent session(s) could not be read.",
+			partialTotals: "Only readable usage is shown; missing usage is not treated as zero.",
 			failedLoad: "Could not read usage. The last successful result is preserved.",
 			retry: "Retry",
 			completed: "Completed",
@@ -34173,6 +34218,8 @@ window.__ModuleLoader__.load({
 			noModels: "暫無模型用量",
 			scope: "本會話及所有子代理 · 歷史日誌累計",
 			note: "統計包含本會話與全部後代的獨立請求、重試和壓縮呼叫，不重複計算繼承歷史。首 token 與速度僅使用已記錄的時間資訊，缺失時顯示 —。",
+			partialCoverage: "有 {count} 個子會話暫不可讀取。",
+			partialTotals: "目前僅顯示已讀取用量，未將缺失用量視為零。",
 			failedLoad: "用量讀取失敗，已保留上次結果。",
 			retry: "重新讀取",
 			completed: "已完成",