惯性聚合 高效追踪和阅读你感兴趣的博客、新闻、科技资讯
阅读原文 在惯性聚合中打开

推荐订阅源

M
MIT News - Artificial intelligence
雷峰网
雷峰网
OSCHINA 社区最新新闻
OSCHINA 社区最新新闻
Last Week in AI
Last Week in AI
奇客Solidot–传递最新科技情报
奇客Solidot–传递最新科技情报
阮一峰的网络日志
阮一峰的网络日志
月光博客
月光博客
博客园 - Franky
腾讯CDC
T
Tailwind CSS Blog
Recent Announcements
Recent Announcements
V
V2EX
N
Netflix TechBlog - Medium
量子位
Jina AI
Jina AI
Y
Y Combinator Blog
The GitHub Blog
The GitHub Blog
G
Google Developers Blog
爱范儿
爱范儿
博客园 - 叶小钗
D
Docker
MongoDB | Blog
MongoDB | Blog
D
DataBreaches.Net
T
The Blog of Author Tim Ferriss

Recent Commits to openclaw:main

test: merge chat side-result checks · openclaw/openclaw@ddd2c2a test: merge cron history checks · openclaw/openclaw@f7eb746 test: merge responsive navigation shell checks · openclaw/openclaw@c2e4b47 docs(changelog): add codex oauth fixes · openclaw/openclaw@628e6cd test: merge navigation routing cases · openclaw/openclaw@5d8cecb Tests: mock channel registry bundled fallback · openclaw/openclaw@2b08233 Secrets: avoid broad web search discovery for single plugin config · openclaw/openclaw@a464f59 test: merge config view browser checks · openclaw/openclaw@20cf511 fix(status): align oauth health with runtime · openclaw/openclaw@eed7116 feat: add macOS screen snapshots for monitor preview (#67954) thanks … · openclaw/openclaw@f377db1 fix: report shared auth scopes in hello-ok (#67810) thanks @BunsDev · openclaw/openclaw@0b6c39b Auto-reply: avoid eager bundled route fallback · openclaw/openclaw@3ea1bf4 Tests: narrow session binding contract setup · openclaw/openclaw@54e4e16 fix(macOS): enable undo/redo in webchat composer text input (#34962) · openclaw/openclaw@00951dc Tests: speed up channel setup promotion · openclaw/openclaw@82b529a Docs: refresh agent instructions · openclaw/openclaw@5775fe2 fix(auth): serialize OAuth refresh across agents to fix #26322 (#67876) · openclaw/openclaw@8e79080 test: allow ollama public surface boundary test · openclaw/openclaw@7d4f1a6 Docs: add test performance guardrails · openclaw/openclaw@89706d3 Tests: restore context-engine usage proof · openclaw/openclaw@e4c4f95 Tests: slim context engine runtime coverage · openclaw/openclaw@74c198f ci: retry failed custom checkouts · openclaw/openclaw@0ee5baf test: trim duplicate provider auth onboarding cases · openclaw/openclaw@1ffc02e matrix: fix sessions_spawn --thread subagent session spawning (#67643) · openclaw/openclaw@1ce2596 test: reduce auth choice fixture churn · openclaw/openclaw@857b9cd test: mock health status config boundaries · openclaw/openclaw@9d5ab4a test: mock onboard config io boundary · openclaw/openclaw@299694d test: mock legacy state plugin boundaries · openclaw/openclaw@2713089 test: mock channel install boundaries · openclaw/openclaw@b945248 test: mock doctor preview channel boundaries · openclaw/openclaw@b1a3ad4
fix(gateway): report omitted chat-history messages in tru...
ZengWen-DT · 2026-06-26 · via Recent Commits to openclaw:main
1+

// Real-behavior proof that the chat.history budget pipeline emits the

2+

// `payload.large` / `truncated` diagnostic whenever older history is omitted,

3+

// and that the omitted count reflects unique source messages (a message that is

4+

// first replaced and then trimmed is not double-counted). These run the real

5+

// production helpers and capture the real diagnostic event bus output.

6+

import { describe, expect, it } from "vitest";

7+

import { onDiagnosticEvent } from "../../infra/diagnostic-events.js";

8+

import type { DiagnosticPayloadLargeEvent } from "../../infra/diagnostic-events.js";

9+

import { capArrayByJsonBytes } from "../session-utils.js";

10+

import {

11+

enforceChatHistoryFinalBudget,

12+

replaceOversizedChatHistoryMessages,

13+

reportOmittedChatHistory,

14+

} from "./chat.js";

15+16+

type Captured = DiagnosticPayloadLargeEvent[];

17+18+

// Mirrors the production sequence in handleChatHistoryRequest: replace oversized

19+

// messages, cap the array by byte budget, enforce the final budget, then report

20+

// omissions. Captures any emitted `payload.large` diagnostic event.

21+

function runHistoryBudgetPipeline(params: {

22+

messages: unknown[];

23+

maxHistoryBytes: number;

24+

perMessageHardCap: number;

25+

}): { emittedCount: number; events: Captured; replacedCount: number; frontCapDropped: number } {

26+

const { messages, maxHistoryBytes, perMessageHardCap } = params;

27+

const events: Captured = [];

28+

const unsubscribe = onDiagnosticEvent((evt) => {

29+

if (evt.type === "payload.large") {

30+

events.push(evt);

31+

}

32+

});

33+

try {

34+

const replaced = replaceOversizedChatHistoryMessages({

35+

messages,

36+

maxSingleMessageBytes: perMessageHardCap,

37+

});

38+

const capped = capArrayByJsonBytes(replaced.messages, maxHistoryBytes).items;

39+

const bounded = enforceChatHistoryFinalBudget({ messages: capped, maxBytes: maxHistoryBytes });

40+

const emittedCount = reportOmittedChatHistory({

41+

originalMessages: messages,

42+

finalMessages: bounded.messages,

43+

normalizedBytes: Buffer.byteLength(JSON.stringify(messages), "utf8"),

44+

maxHistoryBytes,

45+

logDebug: () => {},

46+

});

47+

return {

48+

emittedCount,

49+

events,

50+

replacedCount: replaced.replacedCount,

51+

frontCapDropped: replaced.messages.length - capped.length,

52+

};

53+

} finally {

54+

unsubscribe();

55+

}

56+

}

57+58+

function textMessage(role: string, text: string): Record<string, unknown> {

59+

return { role, content: [{ type: "text", text }] };

60+

}

61+62+

describe("chat.history truncation logging (real diagnostic bus)", () => {

63+

it("emits a truncated diagnostic when history is trimmed to the last message", () => {

64+

const big = textMessage("user", "x".repeat(8000));

65+

const last = textMessage("assistant", "ok");

66+

const result = runHistoryBudgetPipeline({

67+

messages: [big, last],

68+

maxHistoryBytes: 2_000,

69+

perMessageHardCap: 2_000,

70+

});

71+72+

expect(result.events).toHaveLength(1);

73+

const event = result.events[0];

74+

expect(event.surface).toBe("gateway.chat.history");

75+

expect(event.action).toBe("truncated");

76+

expect(event.reason).toBe("chat_history_budget");

77+

expect(event.count).toBe(1);

78+

expect(result.emittedCount).toBe(1);

79+

});

80+81+

it("emits no diagnostic when nothing is omitted", () => {

82+

const result = runHistoryBudgetPipeline({

83+

messages: [textMessage("user", "hello"), textMessage("assistant", "hi")],

84+

maxHistoryBytes: 1_000_000,

85+

perMessageHardCap: 1_000_000,

86+

});

87+88+

expect(result.events).toHaveLength(0);

89+

expect(result.emittedCount).toBe(0);

90+

});

91+92+

it("counts a replaced-then-trimmed message once, not twice", () => {

93+

// `huge` is oversized so it is replaced with a small placeholder, then the

94+

// placeholder sits at the front and is dropped by the byte cap. The naive

95+

// sum of replacedCount + front-cap drops would count `huge` twice.

96+

const huge = textMessage("user", "h".repeat(8000));

97+

const big1 = textMessage("assistant", "a".repeat(2000));

98+

const big2 = textMessage("user", "b".repeat(2000));

99+

const last = textMessage("assistant", "ok");

100+

const messages = [huge, big1, big2, last];

101+102+

const result = runHistoryBudgetPipeline({

103+

messages,

104+

maxHistoryBytes: 4_000,

105+

perMessageHardCap: 3_000,

106+

});

107+108+

// Scenario preconditions: a message was replaced AND front-capped, so the

109+

// old additive count would have over-reported.

110+

expect(result.replacedCount).toBeGreaterThan(0);

111+

expect(result.frontCapDropped).toBeGreaterThan(0);

112+

const naiveAdditive = result.replacedCount + result.frontCapDropped;

113+114+

// The emitted count equals the number of original messages that lost their

115+

// verbatim representation, and is strictly less than the double-counting sum.

116+

expect(result.events).toHaveLength(1);

117+

expect(result.events[0].count).toBe(result.emittedCount);

118+

expect(result.emittedCount).toBe(2);

119+

expect(naiveAdditive).toBeGreaterThan(result.emittedCount);

120+

});

121+

});