fix: 分页问题
This commit is contained in:
07akioni
@@ -9,6 +9,7 @@ import {
|
||||
createAssistantMessage,
|
||||
createToolResultMessage,
|
||||
createUserMessage,
|
||||
isTokenDelta,
|
||||
} from '@deepseek-ai/dsh-llm/message'
|
||||
import { CallId } from '@deepseek-ai/dsh-llm/brand'
|
||||
import type {
|
||||
@@ -861,6 +862,76 @@ function tokenUsageOf(log: readonly SessionEvent[]): FixtureTokenUsageProjection
|
||||
return totals
|
||||
}
|
||||
|
||||
/** Fixture parallel of session-stats' whole-log counting and wall-time fold. */
|
||||
function sessionStatsOf(log: readonly SessionEvent[]): {
|
||||
turns: number
|
||||
steps: number
|
||||
llmMs: number
|
||||
toolMs: number
|
||||
ttftMs: number
|
||||
ttftSteps: number
|
||||
decodeMs: number
|
||||
decodeTokens: number
|
||||
} {
|
||||
const value = { turns: 0, steps: 0, llmMs: 0, toolMs: 0, ttftMs: 0, ttftSteps: 0, decodeMs: 0, decodeTokens: 0 }
|
||||
let lastTurn: number | null = null
|
||||
let openStep: { turn: number; step: number; startTime: number; firstTokenTime: number | null } | null = null
|
||||
const pendingCalls = new Map<string, number>()
|
||||
for (const event of log) {
|
||||
switch (event.type) {
|
||||
case 'step/start':
|
||||
openStep = { turn: event.data.turn, step: event.data.step, startTime: event.time, firstTokenTime: null }
|
||||
break
|
||||
case 'assistant/chunk':
|
||||
if (openStep !== null && openStep.turn === event.data.turn && openStep.step === event.data.step
|
||||
&& openStep.firstTokenTime === null && isTokenDelta(event.data.chunk)) {
|
||||
openStep.firstTokenTime = event.time
|
||||
}
|
||||
break
|
||||
case 'assistant/message': {
|
||||
if (openStep === null || openStep.turn !== event.data.turn || openStep.step !== event.data.step) break
|
||||
value.llmMs += Math.max(0, event.time - openStep.startTime)
|
||||
if (openStep.firstTokenTime !== null) {
|
||||
value.ttftMs += Math.max(0, openStep.firstTokenTime - openStep.startTime)
|
||||
value.ttftSteps += 1
|
||||
const outputTokens = event.data.usage?.outputTokens
|
||||
if (typeof outputTokens === 'number' && Number.isFinite(outputTokens) && outputTokens >= 0) {
|
||||
value.decodeMs += Math.max(0, event.time - openStep.firstTokenTime)
|
||||
value.decodeTokens += outputTokens
|
||||
}
|
||||
}
|
||||
openStep = null
|
||||
break
|
||||
}
|
||||
case 'tool/call':
|
||||
pendingCalls.set(event.data.callId, event.time)
|
||||
break
|
||||
case 'tool/result': {
|
||||
const callId = event.data.message.source.callId
|
||||
const dispatched = pendingCalls.get(callId)
|
||||
if (dispatched === undefined) break
|
||||
pendingCalls.delete(callId)
|
||||
value.toolMs += Math.max(0, event.time - dispatched)
|
||||
break
|
||||
}
|
||||
case 'step/end':
|
||||
if (event.data.turn !== lastTurn) {
|
||||
value.turns += 1
|
||||
lastTurn = event.data.turn
|
||||
}
|
||||
value.steps += 1
|
||||
openStep = null
|
||||
break
|
||||
case 'turn/end':
|
||||
pendingCalls.clear()
|
||||
break
|
||||
default:
|
||||
break
|
||||
}
|
||||
}
|
||||
return value
|
||||
}
|
||||
|
||||
interface FixtureRequestContext {
|
||||
provider: string
|
||||
model: string
|
||||
@@ -979,6 +1050,8 @@ function projectionValuesOf(log: readonly SessionEvent[]): Record<string, unknow
|
||||
values['contextPressure'] = contextPressureOf(log)
|
||||
// Always present (token-meter composed): heuristic request composition.
|
||||
values['contextBreakdown'] = contextBreakdownOf(log)
|
||||
// Always present (session-stats unit composed): whole-log turn/step counts.
|
||||
values['sessionStats'] = sessionStatsOf(log)
|
||||
return values
|
||||
}
|
||||
|
||||
@@ -1014,6 +1087,17 @@ function projectionFramesOf(id: SessionId, log: readonly SessionEvent[], event:
|
||||
seq: event.seq,
|
||||
})
|
||||
}
|
||||
// The stats fold's view advances on message assembly and tool settlement
|
||||
// (wall times) and on step close (counts).
|
||||
if (type === 'assistant/message' || type === 'tool/result' || type === 'step/end') {
|
||||
frames.push({
|
||||
type: 'session/projection',
|
||||
sessionId: id,
|
||||
key: 'sessionStats',
|
||||
value: sessionStatsOf(log),
|
||||
seq: event.seq,
|
||||
})
|
||||
}
|
||||
if (frames.length > 0) return frames
|
||||
if (type === 'session/title') {
|
||||
const values = projectionValuesOf(log)
|
||||
|
||||
@@ -164,6 +164,10 @@ describe('createFixtureApi', () => {
|
||||
toolsTokens: 0,
|
||||
messageTokens: 0,
|
||||
},
|
||||
// Session-stats unit composed: no figure accrues on the empty log.
|
||||
sessionStats: {
|
||||
turns: 0, steps: 0, llmMs: 0, toolMs: 0, ttftMs: 0, ttftSteps: 0, decodeMs: 0, decodeTokens: 0,
|
||||
},
|
||||
} },
|
||||
})
|
||||
})
|
||||
@@ -353,7 +357,7 @@ describe('createFixtureApi', () => {
|
||||
const envelopes: RpcRequest<MuxFrame>[] = []
|
||||
for await (const envelope of api.events.mux(req({}), abort.signal)) {
|
||||
envelopes.push(envelope)
|
||||
if (envelopes.length >= 11) abort.abort()
|
||||
if (envelopes.length >= 12) abort.abort()
|
||||
}
|
||||
return envelopes
|
||||
}
|
||||
@@ -374,10 +378,12 @@ describe('createFixtureApi', () => {
|
||||
value: { systemTokens: 0, toolsTokens: 0 },
|
||||
})
|
||||
expect((first[8]?.payload as { value: { messageTokens: number } }).value.messageTokens).toBeGreaterThan(0)
|
||||
expect(first[9]?.payload).toMatchObject({ type: 'approval/requested', toolName: 'dangerous_tool' })
|
||||
expect(second[9]?.rpcId).toBe(first[9]?.rpcId) // stable rpcId across replays (host replay semantics)
|
||||
expect(first[10]?.payload).toMatchObject({ type: 'question/requested', sessionId: 'fx-alpha' })
|
||||
expect(second[10]?.rpcId).toBe(first[10]?.rpcId)
|
||||
expect(first[9]?.payload).toMatchObject({ type: 'session/projection', sessionId: 'fx-alpha', key: 'sessionStats' })
|
||||
expect((first[9]?.payload as { value: { turns: number; steps: number } }).value.steps).toBeGreaterThan(0)
|
||||
expect(first[10]?.payload).toMatchObject({ type: 'approval/requested', toolName: 'dangerous_tool' })
|
||||
expect(second[10]?.rpcId).toBe(first[10]?.rpcId) // stable rpcId across replays (host replay semantics)
|
||||
expect(first[11]?.payload).toMatchObject({ type: 'question/requested', sessionId: 'fx-alpha' })
|
||||
expect(second[11]?.rpcId).toBe(first[11]?.rpcId)
|
||||
})
|
||||
|
||||
it('steer with no replay in flight falls through to a fresh queued turn; non-text blocks stringify empty', async () => {
|
||||
|
||||
@@ -2,9 +2,14 @@
|
||||
// history fold derive AssistantTiming from the same step/start -> first token
|
||||
// delta -> assistant/message sequence.
|
||||
|
||||
import { isTokenDelta } from '@deepseek-ai/dsh-llm/message'
|
||||
import type { SessionEvent } from '@deepseek-ai/dsh-session/types'
|
||||
import type { AssistantTiming } from './conversation.ts'
|
||||
|
||||
// The first-token predicate lives with the StreamChunk vocabulary in dsh-llm;
|
||||
// re-exported here so Chat Definitions keep their client-runtime import.
|
||||
export { isTokenDelta } from '@deepseek-ai/dsh-llm/message'
|
||||
|
||||
/** Pre-finalize timing boundaries for one assistant step (start + first token). */
|
||||
export interface AssistantStepMetadata {
|
||||
stepStartTime: number | null
|
||||
@@ -21,24 +26,6 @@ export function assistantStepKey(turn: number, step: number): string {
|
||||
return `${turn}\u0000${step}`
|
||||
}
|
||||
|
||||
/**
|
||||
* Whether a chunk carries visible model output (first-token boundary). Empty
|
||||
* deltas (heartbeats, empty tool-call frames) do not count as a first token.
|
||||
* @param chunk - the assistant/chunk payload.
|
||||
* @returns true when the chunk contains a non-empty text/reasoning/tool delta.
|
||||
*/
|
||||
export function isTokenDelta(chunk: SessionEvent<'assistant/chunk'>['data']['chunk']): boolean {
|
||||
switch (chunk.type) {
|
||||
case 'text-delta':
|
||||
case 'reasoning-delta':
|
||||
return chunk.text !== ''
|
||||
case 'tool-call-delta':
|
||||
return chunk.argumentsDelta !== '' || chunk.name !== undefined
|
||||
default:
|
||||
return false
|
||||
}
|
||||
}
|
||||
|
||||
/**
|
||||
* Fold one event into the per-step timing index: step/start opens the entry,
|
||||
* the first non-empty token delta stamps first-token time once. Other event
|
||||
|
||||
@@ -2,5 +2,5 @@
|
||||
# side as of the last confirmed-consistent state. Both languages carry equal authority;
|
||||
# after editing either side, bring the other along and re-record with:
|
||||
# pnpm run verify-translation-pairing --write packages/client/ui-conversation/README.md
|
||||
README.md: ed8f888d35693ecaa2667ea432b462f4bb3369cf
|
||||
README.zh.md: e6a2dd0b545b66ab01b213b5ebc937e22af8ac1a
|
||||
README.md: 6d0f3d0a088ad580d55d524699a7bf8d2806b724
|
||||
README.zh.md: 70f828d29512eb5e25a873a8fce2b4ffad6380fc
|
||||
|
||||
@@ -38,7 +38,7 @@ Per-session UI state for selection and the active view lives in the declared cha
|
||||
|
||||
The composer bar declares session-scoped single seats for `'conversation.input.plan'` (right of the local access-mode control) and `'conversation.input.model'` (immediately before the pending indicator and send/stop controls), plus list slots for overlay, dock, left, and right input extensions. Feature packages own each control and its state; ui-conversation supplies placement, the `locked` owner prop, and the standard slot shares. The leading plus button is a Command launcher, not an attachment surface: it asks the session's `SlashController` to open only the `/` trigger's `command` source over the current textarea selection, while ui-slash's existing `MenuView` remains the sole floating menu and pick path. No file row, file input, upload protocol, or second menu component is introduced. While the `plan` projection's effective target is plan mode, InputBar swaps its textarea placeholder to the plan-task wording, localized through the `conversation` locale namespace this package registers (the `placeholder.plan` / `hint.plan` keys) and shared verbatim with the claimed `/plan` command hint (a host-folded value read through the standard-kit `useProjection`; owner-supplied placeholders win). A pending composer takeover remains mounted when another conversation view is active so the blocked agent can still receive its answer; without a pending interaction, the active-session composer belongs to Chat. The composer-bar slot itself is `session-maybe`: with no current session the same bar keeps message actions inert (machine faces absent, `disabled` owner prop), while the whole dashed card opens the existing Workspace picker by pointer and the read-only textarea opens it through Enter or Space. Disabled controls release pointer events to the card, and the card contains `pointerdown` so the open picker's outside-close cannot race a reopen. The bar never swaps in a parallel tree, so the textarea DOM survives Workspace selection; strict-session control seats stay empty until a session exists.
|
||||
|
||||
The chat stats line takes its token accounting from the generic token-meter `tokenUsage` projection read through the standard-kit `useProjection`: billed input is uncached input plus cache reads and writes; cache hit divides cache reads by that total. Visible nodes supply only the turn and step counts plus the LLM and tool wall times, which are window-scoped facts about what is on screen rather than accounting; durable token and context groups remain visible when compaction leaves no assistant node in the loaded window. The same window fold averages each recorded step's TTFT and divides sampled output tokens by their summed decode spans into a latency/throughput group localized through the `conversation` locale namespace (`TTFT avg … · … tok/s` in English); a step missing a timing boundary or a usage sample drops out of those figures instead of skewing them. The turn-count, step-count, duration, cache, and token labels use the same namespace. Each settled turn additionally appends hover-revealed `TTFT {s}s · {tps} tok/s` labels to its assistant footer after the `Ran for` duration — the turn's first-step TTFT and its turn-aggregate decode throughput — gated on the turn's timing being in the loaded window (a contiguous log suffix, so an in-window turn carries every one of its steps) and omitting whichever figure is unrecorded. A deployment without token-meter drops the token groups; when the line overflows, it elides with an ellipsis and a delayed hover tooltip carries the full text only while actually clipped. Context occupancy renders as the composer's trailing ContextMeter: a 14px occupancy ring after the model seat, fed by `contextPressure` and rendered only once both a numerator and a route capacity are known, that click-opens a panel pairing the `percent used` header and `~used / capacity` figures with a color-segmented bar and `~`-prefixed heuristic composition rows (system prompt, tools, messages) from the `contextBreakdown` projection. The ring and header read `projectedTokens` — the provider sample carried forward over the surface's movement since — so a compaction registers immediately instead of after a further turn; the composition rows stay wholly heuristic and therefore still do not sum to the header ([rationale](../../llm/token-meter/README.md)). Occupancy is deliberately an approximation: numerator and capacity are independent last-wins projection fields, not one atomic request observation.
|
||||
The chat stats line takes its token accounting from the generic token-meter `tokenUsage` projection read through the standard-kit `useProjection`: billed input is uncached input plus cache reads and writes; cache hit divides cache reads by that total. The turn and step counts, the LLM and tool wall times, and the latency/throughput group all ride the whole-log `sessionStats` projection (host-folded from step boundaries, first-token chunks, tool pairs, and assembled messages), so paging and compaction cannot change any strip figure; an assembly without that unit falls back to the window fold over visible nodes, whose fields mirror the projection's. The strip averages each recorded step's TTFT and divides sampled output tokens by their summed decode spans into a latency/throughput group localized through the `conversation` locale namespace (`TTFT avg … · … tok/s` in English); a step missing a timing boundary or a usage sample drops out of those figures instead of skewing them, and durable count, token, and context groups remain visible when compaction leaves no assistant node in the loaded window. The turn-count, step-count, duration, cache, and token labels use the same namespace. Each settled turn additionally appends hover-revealed `TTFT {s}s · {tps} tok/s` labels to its assistant footer after the `Ran for` duration — the turn's first-step TTFT and its turn-aggregate decode throughput — gated on the turn's timing being in the loaded window (a contiguous log suffix, so an in-window turn carries every one of its steps) and omitting whichever figure is unrecorded. A deployment without token-meter drops the token groups; when the line overflows, it elides with an ellipsis and a delayed hover tooltip carries the full text only while actually clipped. Context occupancy renders as the composer's trailing ContextMeter: a 14px occupancy ring after the model seat, fed by `contextPressure` and rendered only once both a numerator and a route capacity are known, that click-opens a panel pairing the `percent used` header and `~used / capacity` figures with a color-segmented bar and `~`-prefixed heuristic composition rows (system prompt, tools, messages) from the `contextBreakdown` projection. The ring and header read `projectedTokens` — the provider sample carried forward over the surface's movement since — so a compaction registers immediately instead of after a further turn; the composition rows stay wholly heuristic and therefore still do not sum to the header ([rationale](../../llm/token-meter/README.md)). Occupancy is deliberately an approximation: numerator and capacity are independent last-wins projection fields, not one atomic request observation.
|
||||
|
||||
`src/client/` is organized by domain. `contract/` is the shared face for slot declarations, composed props, and cross-domain types; `skeleton/`, `chat/`, `input/`, `queue/`, and `settings/` keep their implementations internal, while `apply.ts` is their assembly point. The `/client` exports contain only loader entries, service classes, and contract types; components and store factories reach the page through slot registrations.
|
||||
|
||||
@@ -54,7 +54,7 @@ None; this package neither assembles nor sends a provider request.
|
||||
|
||||
## Known Limitations and Deferred Work
|
||||
|
||||
- **Stats-line durations and speeds cover the in-window flow only** — LLM and tool wall times plus the TTFT and throughput averages fold the snapshot's assistant `timing` and tool call/result pairs, so nodes outside the loaded event window (older history) are not counted.
|
||||
- **The stats-line fallback fold covers the in-window flow only** — without the `sessionStats` projection (an assembly that does not mount the unit), every figure folds the snapshot's assistant `timing` and tool call/result pairs, so nodes outside the loaded event window (older history) are not counted and the numbers grow per loaded page.
|
||||
- **The details panel has no entry point** — `ChatViewInjected.openDetails` is implemented but uncalled, so the raw selected-call display is unreachable in the assembled application. There is no Input/Output/Metadata switch, Prev/Next stepping, or trajectory deep link.
|
||||
- **Assistant per-message paging is a reserved slot** — drawn in the design, not implemented. The finalized content IconActions row (copy / clock / branch) ships under the last content-text assistant of each turn that has ended; mid-turn narration, Think-only nodes, and every node of a turn still producing steps stay chrome-free. Branch stays disabled unless that message is also the last transcript node of a completed turn; when enabled, it forks through that turn, increments the inherited title on the client, and opens the child. A fork or rename failure leaves the source selected ([decision](../../../.agents/notes/implemented/bug-fix/2026-08-02-message-fork-actions-require-completed-turn-tail.md)).
|
||||
- **Sent user messages cannot be edited** — user bubbles retain clock and copy; branch lives only under assistant answers ([decision](../../../.agents/notes/implemented/simplification/2026-08-06-user-bubbles-drop-the-branch-action.md)). Editing returns with the capability behind it: a client mutation over a settled user message, plus the host behavior for the turn that already consumed it ([decision](../../../.agents/notes/implemented/simplification/2026-07-31-drop-user-message-edit-stub.md)).
|
||||
|
||||
@@ -38,7 +38,7 @@ Host 带 placement 的 `session/queue` 快照也会携带待处理 steering。Qu
|
||||
|
||||
输入栏为 `'conversation.input.plan'`（位于本地 access 模式控件右侧）和 `'conversation.input.model'`（渲染在 pending 指示器与发送／停止控件之前）声明会话作用域的单实例 seat，并为 overlay、dock、left 和 right 输入扩展声明列表 slot。各功能包拥有相应控件及其状态；ui-conversation 提供放置位置、`locked` owner prop 和标准 slot share。前置加号按钮是 Command launcher，而非附件入口：它要求当前会话的 `SlashController` 基于 textarea 当前 selection，只打开 `/` trigger 的 `command` source，同时 ui-slash 既有的 `MenuView` 仍是唯一的浮层菜单与 pick 路径。不引入 File 行、file input、上传协议或第二套菜单组件。当 `plan` 投影的有效目标为 plan mode 时，InputBar 将文本框 placeholder 切换为 plan 任务措辞，经本包注册的 `conversation` locale 命名空间（`placeholder.plan` / `hint.plan` 键）本地化，并与已认领 `/plan` 命令的提示逐字共用同一份文案（经标准套件 `useProjection` 读取的 host 折叠值；owner 提供的 placeholder 优先）。另一个会话视图活跃时，待处理的 composer 接管仍保持挂载，使被阻塞的 agent（智能体）仍能收到回答；没有待处理交互时，活跃会话的 composer 归 Chat 所有。composer bar slot 本身为 `session-maybe`：没有当前会话时，同一个 bar 会让消息操作保持不可交互（machine face 均缺席、`disabled` owner prop），整张虚线卡片可经指针打开现有 Workspace picker，只读 textarea 也可通过 Enter 或 Space 打开。禁用控件会把指针事件交给卡片，卡片也会拦下 `pointerdown`，避免已打开 picker 的外点关闭与重新打开发生竞态。它不会换入一棵平行树，因此选择 Workspace 时 textarea DOM 不会被销毁；严格会话作用域的控件 seat 在会话存在之前保持为空。
|
||||
|
||||
聊天统计行的 token 账目来自经标准套件 `useProjection` 读取的通用 token-meter 投影 `tokenUsage`：计费输入为未缓存输入、缓存读取与缓存写入之和；缓存命中率以缓存读取除以该总量。可见节点只提供轮次与步骤计数，以及 LLM（大语言模型）和工具的墙钟时间：这些是关于「屏幕上有什么」的窗口作用域事实，而非账目；压缩（compaction）使已加载窗口不再包含 assistant 节点时，持久 token 与上下文分组仍保持可见。同一次窗口折算还会把每个有完整记录的步骤的 TTFT（首 token 延迟）取平均，并用采样到的输出 token 数除以其解码时长之和，得到经 `conversation` locale 命名空间本地化的延迟／吞吐分组（中文为 `首 token 平均 … · … tok/s`）；缺少某个 timing 边界或 usage 采样的步骤会直接退出这些数字，而不是让它们失真。轮次计数、步骤计数、耗时、缓存与 token 各项的标签也使用同一命名空间。每个已结算轮次还会在其 assistant footer 的 `用时` 之后追加 hover 才显示的 `首 token {s}秒 · {tps} tok/s` 标签——即该轮次首个步骤的 TTFT 与轮次聚合的解码吞吐——仅当该轮次的 timing 位于已加载窗口内才显示（窗口是日志的连续后缀，因此窗口内的轮次必然带着它的全部步骤），未记录的数字会各自省略。未组合 token-meter 的部署会整组省略 token 分组；统计行过长时以省略号截断，仅在内容真的被裁切时由延迟 hover tooltip 承载完整文本。上下文占用率渲染为 composer 尾部的 ContextMeter：模型座位之后的一枚 14px 占用圆环，由 `contextPressure` 供数，仅当分子与路由容量都已知时才渲染；点击弹出的面板把「已用百分比」标题与 `~已用 / 容量` 数字，与来自 `contextBreakdown` 投影、带 `~` 前缀的启发式组成明细行（系统提示词、工具、对话消息）及分色分段进度条并列。圆环与标题读取 `projectedTokens`——把提供方样本沿此后表层的增减推进到当下——因此压缩会立刻反映出来，而不必再等一整轮；组成明细行仍是纯启发式，因此加起来依然不等于标题数字（[原理](../../llm/token-meter/README.md)）。占用率是刻意为之的近似值：分子与容量是两个相互独立的「后写覆盖」投影字段，并非同一次请求的原子观测。
|
||||
聊天统计行的 token 账目来自经标准套件 `useProjection` 读取的通用 token-meter 投影 `tokenUsage`：计费输入为未缓存输入、缓存读取与缓存写入之和；缓存命中率以缓存读取除以该总量。轮次与步骤计数、LLM（大语言模型）与工具墙钟时间、以及延迟／吞吐分组都来自全日志的 `sessionStats` 投影（Host 端从步边界、首 token chunk、工具配对与已组装消息折算），因此分页与压缩都无法改变统计条的任何数字；未组合该单元的装配回退为对可见节点做窗口折算，其字段与投影一一对应。统计条把每个有完整记录的步骤的 TTFT（首 token 延迟）取平均，并用采样到的输出 token 数除以其解码时长之和，得到经 `conversation` locale 命名空间本地化的延迟／吞吐分组（中文为 `首 token 平均 … · … tok/s`）；缺少某个 timing 边界或 usage 采样的步骤会直接退出这些数字，而不是让它们失真；压缩（compaction）使已加载窗口不再包含 assistant 节点时，持久计数、token 与上下文分组仍保持可见。轮次计数、步骤计数、耗时、缓存与 token 各项的标签也使用同一命名空间。每个已结算轮次还会在其 assistant footer 的 `用时` 之后追加 hover 才显示的 `首 token {s}秒 · {tps} tok/s` 标签——即该轮次首个步骤的 TTFT 与轮次聚合的解码吞吐——仅当该轮次的 timing 位于已加载窗口内才显示（窗口是日志的连续后缀，因此窗口内的轮次必然带着它的全部步骤），未记录的数字会各自省略。未组合 token-meter 的部署会整组省略 token 分组；统计行过长时以省略号截断，仅在内容真的被裁切时由延迟 hover tooltip 承载完整文本。上下文占用率渲染为 composer 尾部的 ContextMeter：模型座位之后的一枚 14px 占用圆环，由 `contextPressure` 供数，仅当分子与路由容量都已知时才渲染；点击弹出的面板把「已用百分比」标题与 `~已用 / 容量` 数字，与来自 `contextBreakdown` 投影、带 `~` 前缀的启发式组成明细行（系统提示词、工具、对话消息）及分色分段进度条并列。圆环与标题读取 `projectedTokens`——把提供方样本沿此后表层的增减推进到当下——因此压缩会立刻反映出来，而不必再等一整轮；组成明细行仍是纯启发式，因此加起来依然不等于标题数字（[原理](../../llm/token-meter/README.md)）。占用率是刻意为之的近似值：分子与容量是两个相互独立的「后写覆盖」投影字段，并非同一次请求的原子观测。
|
||||
|
||||
`src/client/` 按领域组织。`contract/` 是 slot 声明、组合 props 与跨领域类型的共享表层；`skeleton/`、`chat/`、`input/`、`queue/` 和 `settings/` 保持内部实现，`apply.ts` 是它们的组装点。`/client` 导出表层只包含 loader entry、service class 和 contract 类型；组件与 store factory 经 slot 注册抵达页面。
|
||||
|
||||
@@ -54,7 +54,7 @@ Host 带 placement 的 `session/queue` 快照也会携带待处理 steering。Qu
|
||||
|
||||
## 已知限制与暂缓事项
|
||||
|
||||
- **统计行的耗时与速率只覆盖窗口内消息流**：LLM 与工具墙钟时间以及 TTFT 与吞吐平均值由快照的 assistant `timing` 与工具 call/result 配对折算，落在已加载事件窗口之外的节点（更早的历史）不计入。
|
||||
- **统计行的回退折算只覆盖窗口内消息流**：未组合 `sessionStats` 投影单元的装配中，所有数字由快照的 assistant `timing` 与工具 call/result 配对折算，落在已加载事件窗口之外的节点（更早的历史）不计入，数字随加载页数增长。
|
||||
- **详情面板没有入口**：`ChatViewInjected.openDetails` 虽已实现却无人调用，因此以原始形式显示已选择调用的那部分在组装后的应用中不可达。没有 Input/Output/Metadata 切换、Prev/Next 步进，也没有 trajectory 深链接。
|
||||
- **assistant 逐消息分页是预留 slot**：设计中已有图稿，尚未实现。已定稿的内容 IconActions 行（复制／时钟／分支）只挂在每个已结束轮次中最后一条带 text 内容的 assistant 下；轮次中间的叙述、纯 Think 节点，以及仍在产出步骤的轮次里的所有节点都不带 chrome。除非该消息同时也是已完成轮次的最后一个 transcript 节点，否则分支保持禁用；启用后，它会 fork 到该轮次末尾，在 client 端递增继承标题并打开子会话。fork 或改名失败时源会话保持选中（[决策](../../../.agents/notes/implemented/bug-fix/2026-08-02-message-fork-actions-require-completed-turn-tail.md)）。
|
||||
- **已发送的 user 消息无法编辑**：user 气泡保留时钟和复制；分支只存在于 assistant 回答之下（[决策](../../../.agents/notes/implemented/simplification/2026-08-06-user-bubbles-drop-the-branch-action.md)）。编辑功能要与其背后的能力一起回归：既需要针对已定稿 user 消息的 client 变更，也需要 host 侧对已经消费过它的轮次给出行为（[决策](../../../.agents/notes/implemented/simplification/2026-07-31-drop-user-message-edit-stub.md)）。
|
||||
|
||||
@@ -70,6 +70,7 @@
|
||||
"@deepseek-ai/dsh-compact": "workspace:^",
|
||||
"@deepseek-ai/dsh-invariants": "workspace:^",
|
||||
"@deepseek-ai/dsh-llm-retry": "workspace:^",
|
||||
"@deepseek-ai/dsh-session-stats": "workspace:^",
|
||||
"@deepseek-ai/dsh-token-meter": "workspace:^",
|
||||
"@deepseek-ai/dsh-tools": "workspace:^",
|
||||
"react": "^18.2.0"
|
||||
@@ -98,6 +99,7 @@
|
||||
"@deepseek-ai/dsh-permission": "workspace:^",
|
||||
"@deepseek-ai/dsh-plan-mode": "workspace:^",
|
||||
"@deepseek-ai/dsh-session-projection": "workspace:^",
|
||||
"@deepseek-ai/dsh-session-stats": "workspace:^",
|
||||
"@deepseek-ai/dsh-token-meter": "workspace:^",
|
||||
"@deepseek-ai/dsh-tool-todo": "workspace:^",
|
||||
"@deepseek-ai/dsh-tools": "workspace:^",
|
||||
|
||||
@@ -6,6 +6,8 @@ import { Fragment, memo, useLayoutEffect, useMemo, useRef, useState } from 'reac
|
||||
import { Tooltip } from '@deepseek-ai/dsh-client-ui-primitives'
|
||||
import type { ConversationSnapshot, UseProjection } from '@deepseek-ai/dsh-client-runtime/client'
|
||||
import type { SnapshotSelectorHook } from '@deepseek-ai/dsh-client-ui-slots'
|
||||
// Type-only: merges the sessionStats key into SessionProjectionMap for useProjection.
|
||||
import type {} from '@deepseek-ai/dsh-session-stats/client'
|
||||
import type { ContextPressureProjection, TokenUsageProjection } from '@deepseek-ai/dsh-token-meter/client'
|
||||
import type { ComposerBarProps } from '../contract/slots.ts'
|
||||
import { formatTokensPerSecond } from './message-chrome.ts'
|
||||
@@ -30,14 +32,16 @@ interface WindowStats {
|
||||
}
|
||||
|
||||
/**
|
||||
* Fold assistant and tool-result nodes into the window-scoped display totals.
|
||||
* Fold assistant and tool-result nodes into window-scoped display totals —
|
||||
* the FALLBACK for assemblies without the `sessionStats` projection.
|
||||
*
|
||||
* Counts and wall times describe the loaded window on purpose — they answer
|
||||
* "what is on screen". Token accounting deliberately does NOT come from here:
|
||||
* the window is paged and compaction rewrites it, so billing rides the durable
|
||||
* `tokenUsage` projection instead.
|
||||
* Every displayed figure rides that durable whole-log projection (and token
|
||||
* accounting rides `tokenUsage`) because the window is paged and compaction
|
||||
* rewrites it; this fold answers "what is on screen" only when no projection
|
||||
* value is served. Its field names deliberately mirror the projection's so
|
||||
* the two swap wholesale.
|
||||
* @param nodes - snapshot nodes.
|
||||
* @returns visible counts and summed wall times.
|
||||
* @returns fallback counts and summed wall times.
|
||||
*/
|
||||
export function deriveStats(nodes: ConversationSnapshot['nodes']): WindowStats {
|
||||
const turns = new Set<number>()
|
||||
@@ -158,8 +162,12 @@ export interface StatsLineProps {
|
||||
|
||||
export const StatsLine = memo(function StatsLine({ useSession, useProjection, t }: StatsLineProps) {
|
||||
const settledNodes = useSession(s => s.chat.legacy.nodes)
|
||||
const stats = useMemo(() => deriveStats(settledNodes), [settledNodes])
|
||||
const windowStats = useMemo(() => deriveStats(settledNodes), [settledNodes])
|
||||
const usage = useProjection('tokenUsage')
|
||||
// Every figure rides the durable sessionStats projection, so paging and
|
||||
// compaction cannot change any of them; an assembly without the unit falls
|
||||
// back to the window-scoped fold wholesale (same field names).
|
||||
const stats = useProjection('sessionStats') ?? windowStats
|
||||
// Pipe-separated groups (figma stats strip); a group with no data drops out whole.
|
||||
const groups: string[] = []
|
||||
if (stats.steps > 0) {
|
||||
@@ -168,7 +176,6 @@ export const StatsLine = memo(function StatsLine({ useSession, useProjection, t
|
||||
if (stats.llmMs > 0) durations.push(t('stats.llm', { duration: formatDuration(stats.llmMs) }))
|
||||
if (stats.toolMs > 0) durations.push(t('stats.toolCall', { duration: formatDuration(stats.toolMs) }))
|
||||
if (durations.length > 0) groups.push(durations.join(' · '))
|
||||
// Window-scoped like the wall times above: averages describe loaded steps.
|
||||
const speeds: string[] = []
|
||||
if (stats.ttftSteps > 0) {
|
||||
speeds.push(t('stats.ttftAverage', { duration: formatDuration(stats.ttftMs / stats.ttftSteps) }))
|
||||
|
||||
@@ -98,9 +98,11 @@ describe('deriveStats', () => {
|
||||
])
|
||||
expect(stats.turns).toBe(2)
|
||||
expect(stats.steps).toBe(3)
|
||||
// Window-scoped by design: the paged window is not an accounting source, so
|
||||
// the fold exposes no billing fields (billing rides the projection);
|
||||
// decodeTokens is a throughput input, not a billed total.
|
||||
// The window fold's counts are only the fallback for assemblies without
|
||||
// the sessionStats projection; the paged window is not an accounting
|
||||
// source either, so the fold exposes no billing fields (billing rides the
|
||||
// tokenUsage projection); decodeTokens is a throughput input, not a
|
||||
// billed total.
|
||||
expect(Object.keys(stats).sort()).toEqual(
|
||||
['decodeMs', 'decodeTokens', 'llmMs', 'steps', 'toolMs', 'ttftMs', 'ttftSteps', 'turns'],
|
||||
)
|
||||
@@ -169,6 +171,14 @@ describe('formatters', () => {
|
||||
describe('StatsLine', () => {
|
||||
const USAGE = { uncachedInputTokens: 10, outputTokens: 5, cacheReadTokens: 90, cacheWriteTokens: 0 }
|
||||
|
||||
/** A whole-log sessionStats value: zeros plus overrides. */
|
||||
function sessionStats(overrides: Record<string, number>): Record<string, number> {
|
||||
return {
|
||||
turns: 0, steps: 0, llmMs: 0, toolMs: 0, ttftMs: 0, ttftSteps: 0, decodeMs: 0, decodeTokens: 0,
|
||||
...overrides,
|
||||
}
|
||||
}
|
||||
|
||||
/** Stub the projection seat: a key-addressed table of whole values. */
|
||||
function projections(values: Record<string, unknown>): StatsLineProps['useProjection'] {
|
||||
return (key: string) => values[key]
|
||||
@@ -281,6 +291,58 @@ describe('StatsLine', () => {
|
||||
expect(view.container.textContent).toBe('1 turns · 1 steps')
|
||||
})
|
||||
|
||||
it('renders whole-session counts from the sessionStats projection over the paged window', () => {
|
||||
// The bug's acceptance at unit level: one loaded page must not scope the
|
||||
// counter — the durable projection's totals win over the window fold.
|
||||
const { source } = makeSource({ nodes: [assistant(1, 1)] })
|
||||
const view = render(<StatsLine {...props(source, {
|
||||
tokenUsage: USAGE,
|
||||
sessionStats: sessionStats({ turns: 10, steps: 89 }),
|
||||
})} />)
|
||||
expect(view.container.textContent)
|
||||
.toBe('10 turns · 89 steps| Cache hit 90%| Input 100 tok · Output 5 tok')
|
||||
})
|
||||
|
||||
it('treats a defined zero-count projection as empty, not as fallback', () => {
|
||||
// A composed unit always serves the key; all-zero genuinely means no
|
||||
// closed step in the whole log, so nothing renders on a brand-new session.
|
||||
const empty = makeSource()
|
||||
const view = render(<StatsLine {...props(empty.source, {
|
||||
tokenUsage: { uncachedInputTokens: 0, outputTokens: 0, cacheReadTokens: 0, cacheWriteTokens: 0 },
|
||||
sessionStats: sessionStats({}),
|
||||
})} />)
|
||||
expect(view.container.textContent).toBe('')
|
||||
})
|
||||
|
||||
it('keeps the counts group over an empty visible window when the projection carries totals', () => {
|
||||
// Extends the durable-groups guarantee: full-session counts survive a
|
||||
// window that compaction (or paging) left without assistant nodes.
|
||||
const { source } = makeSource()
|
||||
const view = render(<StatsLine {...props(source, {
|
||||
tokenUsage: USAGE,
|
||||
sessionStats: sessionStats({ turns: 7, steps: 44 }),
|
||||
})} />)
|
||||
expect(view.container.textContent)
|
||||
.toBe('7 turns · 44 steps| Cache hit 90%| Input 100 tok · Output 5 tok')
|
||||
})
|
||||
|
||||
it('renders whole-log wall times and speeds from the projection, not the loaded window', () => {
|
||||
// The 加载更早 hazard beyond counts: LLM/tool durations and the TTFT and
|
||||
// throughput figures must not grow per loaded page either. An untimed
|
||||
// 1-node window renders the projection's whole-log figures verbatim.
|
||||
const { source } = makeSource({ nodes: [assistant(1, 1)] })
|
||||
const view = render(<StatsLine {...props(source, {
|
||||
tokenUsage: USAGE,
|
||||
sessionStats: sessionStats({
|
||||
turns: 200, steps: 200, llmMs: 100_000, toolMs: 62_000,
|
||||
ttftMs: 1_600, ttftSteps: 2, decodeMs: 3_000, decodeTokens: 60,
|
||||
}),
|
||||
})} />)
|
||||
expect(view.container.textContent).toBe(
|
||||
'200 turns · 200 steps| LLM 1m40s · Tool call 1m2s| TTFT avg 0.8s · 20 tok/s| Cache hit 90%| Input 100 tok · Output 5 tok',
|
||||
)
|
||||
})
|
||||
|
||||
it('omits cache hit when nothing was billed on the input side', () => {
|
||||
const { source } = makeSource({ nodes: [assistant(1, 1)] })
|
||||
const view = render(<StatsLine {...props(source, {
|
||||
|
||||
@@ -70,9 +70,11 @@ describe('render branch tails', () => {
|
||||
expect(view.container.querySelector('[data-state="ok"]')).not.toBeNull()
|
||||
})
|
||||
|
||||
it('StatsLine counts window nodes but drops every token group without a projection', () => {
|
||||
// Node `usage` is deliberately ignored: billing rides the durable
|
||||
// tokenUsage projection, so an absent projection leaves counts only.
|
||||
it('StatsLine falls back to window-node counts and drops every token group without projections', () => {
|
||||
// No sessionStats key → the window fold supplies the counts (the
|
||||
// assembly-without-the-unit fallback). Node `usage` is deliberately
|
||||
// ignored: billing rides the durable tokenUsage projection, so an absent
|
||||
// projection leaves counts only.
|
||||
const nodes = [
|
||||
{ kind: 'assistant', seq: 1, time: 1, turn: 1, step: 1, blocks: [] },
|
||||
{ kind: 'assistant', seq: 2, time: 2, turn: 1, step: 2, blocks: [], usage: { inputTokens: 4, outputTokens: 6 } },
|
||||
|
||||
@@ -44,6 +44,9 @@
|
||||
{
|
||||
"path": "../../session/session-projection"
|
||||
},
|
||||
{
|
||||
"path": "../../session/session-stats"
|
||||
},
|
||||
{
|
||||
"path": "../../llm/token-meter"
|
||||
},
|
||||
|
||||
Reference in New Issue
Block a user