Jesse_Chen
960 lines
38 KiB
TypeScript
960 lines
38 KiB
TypeScript
import assert from "node:assert/strict";
|
|
import test from "node:test";
|
|
|
|
import {
|
|
mapStreamChunkToActivity,
|
|
mapStreamChunkToPhase,
|
|
safePublicEvent,
|
|
streamToolNames,
|
|
} from "../src/lib/rectification-agentic/v9/stream-mapping.ts";
|
|
import { PUBLIC_RECTIFICATION_TOOLS } from "../src/lib/rectification-agentic/v9/public-receipt.ts";
|
|
import { runV9AgentTurn, type V9AgentRunOptions } from "../src/lib/rectification-agentic/v9/agent-run.ts";
|
|
import {
|
|
CASE_ID,
|
|
SESSION_ID,
|
|
TURN_ID,
|
|
USER_ID,
|
|
dossierFixture,
|
|
fakeAccounting,
|
|
receiptHandlers,
|
|
} from "./rectification-v9-test-support.ts";
|
|
import { RECTIFICATION_SKILL_NAME } from "../src/lib/rectification-agentic/v9/case-status.ts";
|
|
import { safeToolErrorCode } from "../src/lib/rectification-agentic/v9/tool-service.ts";
|
|
import {
|
|
createRectificationActivityReceiptState,
|
|
receiptFromRectificationActivityState,
|
|
reduceRectificationActivityReceipt,
|
|
} from "../src/lib/rectification-activity-receipt.ts";
|
|
|
|
type StreamChunk = {
|
|
type: string;
|
|
payload?: {
|
|
toolName?: unknown;
|
|
text?: unknown;
|
|
args?: unknown;
|
|
error?: unknown;
|
|
result?: unknown;
|
|
output?: unknown;
|
|
};
|
|
object?: unknown;
|
|
};
|
|
|
|
function chunk(type: string, payload?: Record<string, unknown>): StreamChunk {
|
|
return { type, ...(payload ? { payload } : {}) };
|
|
}
|
|
|
|
test("fullStream chunks map to the allowlisted NDJSON phases only", () => {
|
|
assert.equal(mapStreamChunkToPhase(chunk("start") as never), null);
|
|
assert.equal(
|
|
mapStreamChunkToPhase(chunk("tool-call", { toolName: "skill", args: { name: RECTIFICATION_SKILL_NAME } }) as never),
|
|
null,
|
|
);
|
|
assert.deepEqual(
|
|
mapStreamChunkToPhase(chunk("tool-result", { toolName: "skill" }) as never),
|
|
{ type: "skill.bound" },
|
|
);
|
|
assert.deepEqual(
|
|
mapStreamChunkToPhase(chunk("tool-call", { toolName: "rectification-compare-candidates" }) as never),
|
|
{ type: "candidates.comparing", tool: "rectification-compare-candidates" },
|
|
);
|
|
assert.deepEqual(
|
|
mapStreamChunkToPhase(chunk("tool-result", {
|
|
toolName: "rectification-compare-candidates",
|
|
result: {
|
|
executed_methods: [
|
|
"d1-rashi",
|
|
"vimshottari-dasha",
|
|
"internal-secret-technique",
|
|
"d1-rashi",
|
|
],
|
|
birth_context: { birth_date: "1997-08-08" },
|
|
event_contribution_matrix: { secret: true },
|
|
rule_ids: ["internal-rule"],
|
|
},
|
|
}) as never),
|
|
{
|
|
type: "candidates.updated",
|
|
tool: "rectification-compare-candidates",
|
|
methods: ["d1-rashi", "vimshottari-dasha"],
|
|
},
|
|
);
|
|
assert.equal(
|
|
mapStreamChunkToPhase(chunk("tool-call", { toolName: "rectification-read-case" }) as never),
|
|
null,
|
|
);
|
|
assert.deepEqual(
|
|
mapStreamChunkToPhase(chunk("tool-result", {
|
|
toolName: "rectification-read-case",
|
|
result: { birth_context: { birth_date: "1997-08-08", latitude: 36.4 } },
|
|
}) as never),
|
|
{ type: "case.loaded", tool: "rectification-read-case" },
|
|
);
|
|
assert.deepEqual(
|
|
mapStreamChunkToPhase(chunk("text-delta", { text: "你好" }) as never),
|
|
{ type: "answer.delta", text: "你好" },
|
|
);
|
|
assert.equal(mapStreamChunkToPhase(chunk("finish") as never), null);
|
|
assert.equal(mapStreamChunkToPhase(chunk("error", { error: new Error("boom") }) as never), null);
|
|
assert.equal(mapStreamChunkToPhase(chunk("abort") as never), null);
|
|
});
|
|
|
|
test("every public rectification tool maps its real lifecycle to public activity", () => {
|
|
for (const tool of PUBLIC_RECTIFICATION_TOOLS) {
|
|
assert.deepEqual(
|
|
mapStreamChunkToActivity(chunk("tool-call", {
|
|
toolName: tool,
|
|
args: { caseId: CASE_ID, birth_date: "1997-08-08", scores: [99] },
|
|
}) as never),
|
|
{ type: "tool.activity", tool, status: "started" },
|
|
);
|
|
assert.deepEqual(
|
|
mapStreamChunkToActivity(chunk("tool-result", {
|
|
toolName: tool,
|
|
result: {
|
|
executed_methods: ["d1-rashi", "private-method", "d1-rashi"],
|
|
birth_context: { birth_date: "1997-08-08" },
|
|
scores: [99],
|
|
},
|
|
}) as never),
|
|
{ type: "tool.activity", tool, status: "completed", methods: ["d1-rashi"] },
|
|
);
|
|
assert.deepEqual(
|
|
mapStreamChunkToActivity(chunk("tool-error", {
|
|
toolName: tool,
|
|
args: { caseId: CASE_ID, scores: [99] },
|
|
error: new Error("private provider error"),
|
|
}) as never),
|
|
{ type: "tool.activity", tool, status: "failed" },
|
|
);
|
|
}
|
|
assert.equal(mapStreamChunkToActivity(chunk("tool-call", { toolName: "skill" }) as never), null);
|
|
assert.equal(mapStreamChunkToActivity(chunk("tool-error", { toolName: "private-tool" }) as never), null);
|
|
assert.equal(mapStreamChunkToActivity(chunk("text-delta", { text: "x" }) as never), null);
|
|
});
|
|
|
|
test("reasoning, raw payloads, provider metadata and step internals never map", () => {
|
|
assert.equal(mapStreamChunkToPhase(chunk("reasoning-start", { id: "r1" }) as never), null);
|
|
assert.equal(mapStreamChunkToPhase(chunk("reasoning-delta", { text: "内部推理" }) as never), null);
|
|
assert.equal(mapStreamChunkToPhase(chunk("reasoning-end") as never), null);
|
|
assert.equal(mapStreamChunkToPhase(chunk("raw", { payload: { secret: true } }) as never), null);
|
|
assert.equal(mapStreamChunkToPhase(chunk("step-start", { messageId: "m1" }) as never), null);
|
|
assert.equal(mapStreamChunkToPhase(chunk("step-finish", { output: { text: "x" } }) as never), null);
|
|
assert.equal(mapStreamChunkToPhase(chunk("response-metadata", { signature: "s" }) as never), null);
|
|
assert.equal(mapStreamChunkToPhase(chunk("source", { title: "t" }) as never), null);
|
|
assert.equal(mapStreamChunkToPhase(chunk("file", { mimeType: "text/plain" }) as never), null);
|
|
});
|
|
|
|
test("streamToolNames exposes only allowlisted rectification tools", () => {
|
|
assert.deepEqual(streamToolNames(chunk("tool-call", { toolName: "rectification-read-case" }) as never), ["rectification-read-case"]);
|
|
assert.deepEqual(streamToolNames(chunk("tool-call", { toolName: "skill" }) as never), []);
|
|
assert.deepEqual(streamToolNames(chunk("tool-call", { toolName: "rectification-gate" }) as never), []);
|
|
assert.deepEqual(streamToolNames(chunk("text-delta", { text: "x" }) as never), []);
|
|
});
|
|
|
|
test("safePublicEvent drops anything outside the allowlist", () => {
|
|
assert.deepEqual(safePublicEvent({ type: "answer.delta", text: "你好" }), { type: "answer.delta", text: "你好" });
|
|
assert.deepEqual(safePublicEvent({ type: "skill.loaded" }), { type: "skill.loaded" });
|
|
assert.deepEqual(
|
|
safePublicEvent({ type: "run.completed", turnId: TURN_ID }),
|
|
{ type: "run.completed", turnId: TURN_ID },
|
|
);
|
|
assert.deepEqual(
|
|
safePublicEvent({ type: "answer.delta", text: "你好", turnId: TURN_ID }),
|
|
{ type: "answer.delta", text: "你好" },
|
|
);
|
|
assert.deepEqual(
|
|
safePublicEvent({ type: "run.completed", turnId: "not-a-uuid" }),
|
|
{ type: "run.completed" },
|
|
);
|
|
assert.deepEqual(
|
|
safePublicEvent({
|
|
type: "case.loaded",
|
|
tool: "rectification-read-case",
|
|
text: "1997-08-08 河北省邯郸市",
|
|
methods: ["d1-rashi"],
|
|
birth_context: { latitude: 36.4 },
|
|
}),
|
|
{ type: "case.loaded", tool: "rectification-read-case" },
|
|
);
|
|
assert.deepEqual(
|
|
safePublicEvent({
|
|
type: "tool.activity",
|
|
tool: "rectification-compare-candidates",
|
|
status: "completed",
|
|
methods: ["d10-dashamsa", "private-method", "d10-dashamsa"],
|
|
args: { caseId: CASE_ID },
|
|
error: "private provider error",
|
|
birth_context: { birth_date: "1997-08-08" },
|
|
scores: [99],
|
|
}),
|
|
{
|
|
type: "tool.activity",
|
|
tool: "rectification-compare-candidates",
|
|
status: "completed",
|
|
methods: ["d10-dashamsa"],
|
|
},
|
|
);
|
|
assert.deepEqual(
|
|
safePublicEvent({
|
|
type: "tool.activity",
|
|
tool: "rectification-read-case",
|
|
status: "started",
|
|
methods: ["d1-rashi"],
|
|
result: { private: true },
|
|
}),
|
|
{ type: "tool.activity", tool: "rectification-read-case", status: "started" },
|
|
);
|
|
assert.deepEqual(
|
|
safePublicEvent({
|
|
type: "tool.activity",
|
|
tool: "rectification-read-case",
|
|
status: "failed",
|
|
error: "private provider error",
|
|
}),
|
|
{ type: "tool.activity", tool: "rectification-read-case", status: "failed" },
|
|
);
|
|
assert.equal(safePublicEvent({ type: "tool.activity", tool: "private-tool", status: "started" }), null);
|
|
assert.equal(safePublicEvent({ type: "tool.activity", tool: "rectification-read-case", status: "pending" }), null);
|
|
assert.deepEqual(
|
|
safePublicEvent({
|
|
type: "candidates.updated",
|
|
tool: "rectification-compare-candidates",
|
|
methods: ["d10-dashamsa", "private-method", "d10-dashamsa"],
|
|
rule_ids: ["private-rule"],
|
|
}),
|
|
{
|
|
type: "candidates.updated",
|
|
tool: "rectification-compare-candidates",
|
|
methods: ["d10-dashamsa"],
|
|
},
|
|
);
|
|
assert.equal(safePublicEvent({ type: "provider.reasoning", text: "内部" }), null);
|
|
assert.equal(safePublicEvent({ type: "tool.payload", text: "秘密" }), null);
|
|
assert.equal(safePublicEvent({ type: "raw" }), null);
|
|
assert.equal(safePublicEvent(null), null);
|
|
assert.deepEqual(
|
|
safePublicEvent({ type: "error", code: "skill_identity_unverifiable", message: "请先采用当前 Skill" }),
|
|
{ type: "error", code: "skill_identity_unverifiable", message: "请先采用当前 Skill" },
|
|
);
|
|
assert.equal(safePublicEvent({ type: "error", code: "private_error", message: "secret" }), null);
|
|
});
|
|
|
|
type FakeStreamResult = {
|
|
fullStream: AsyncIterable<{ type: string; payload?: Record<string, unknown> }>;
|
|
totalUsage?: Promise<{ inputTokens?: number; outputTokens?: number }>;
|
|
};
|
|
|
|
function fakeAgentStream(chunks: Array<{ type: string; payload?: Record<string, unknown> }>) {
|
|
return {
|
|
stream: async () => ({
|
|
fullStream: (async function* () {
|
|
for (const item of chunks) yield item;
|
|
})(),
|
|
totalUsage: Promise.resolve({ inputTokens: 10, outputTokens: 20 }),
|
|
}) as FakeStreamResult,
|
|
getSkill: async () => ({ name: RECTIFICATION_SKILL_NAME, instructions: "skill" }),
|
|
};
|
|
}
|
|
|
|
function runOptions(overrides: Partial<V9AgentRunOptions> = {}) {
|
|
const emitted: Array<{ type: string; text?: string; turnId?: string }> = [];
|
|
const billing = { reserved: 0, completed: 0, released: 0 };
|
|
const accounting = fakeAccounting({
|
|
...receiptHandlers,
|
|
get_agentic_rectification_case_dossier: () => dossierFixture(),
|
|
append_agentic_rectification_turn: () => ({ turn_id: TURN_ID }),
|
|
finalize_agentic_rectification_turn: () => ({ turn_id: TURN_ID, status: "completed", idempotent: false }),
|
|
});
|
|
const optionsValue: V9AgentRunOptions = {
|
|
userId: USER_ID,
|
|
caseId: CASE_ID,
|
|
sessionId: SESSION_ID,
|
|
requestId: "aaaaaaaa-bbbb-4ccc-8ddd-eeeeeeeeeeee",
|
|
action: "evidence",
|
|
message: "2016年9月离开家去北京工作",
|
|
modelName: "gpt-4o-mini",
|
|
accounting: accounting.client,
|
|
billing: {
|
|
reserve: async () => { billing.reserved += 1; return { success: true, status: 200 }; },
|
|
complete: async () => { billing.completed += 1; return true; },
|
|
release: async () => { billing.released += 1; return true; },
|
|
},
|
|
emit: (event) => { emitted.push(event); },
|
|
buildAgent: async () => fakeAgentStream([]) as never,
|
|
...overrides,
|
|
};
|
|
return { options: optionsValue, emitted, billing };
|
|
}
|
|
|
|
test("request-aware append receives the caller request id", async () => {
|
|
const accounting = fakeAccounting({
|
|
...receiptHandlers,
|
|
get_agentic_rectification_case_dossier: () => dossierFixture(),
|
|
append_agentic_rectification_turn: () => ({ turn_id: TURN_ID }),
|
|
finalize_agentic_rectification_turn: () => ({ turn_id: TURN_ID, status: "completed", idempotent: false }),
|
|
});
|
|
const { options } = runOptions({
|
|
accounting: accounting.client,
|
|
buildAgent: async () => fakeAgentStream([
|
|
chunk("tool-result", { toolName: "skill" }),
|
|
chunk("tool-result", { toolName: "rectification-read-case" }),
|
|
chunk("text-delta", { text: "已记录" }),
|
|
chunk("finish"),
|
|
]) as never,
|
|
});
|
|
|
|
await runV9AgentTurn(options);
|
|
|
|
const append = accounting.calls.find((call) => call.fn === "append_agentic_rectification_turn");
|
|
assert.equal(append?.args.p_request_id, options.requestId);
|
|
});
|
|
|
|
test("completed request replay returns persisted truth without rebuilding or settling", async () => {
|
|
let buildCount = 0;
|
|
const accounting = fakeAccounting({
|
|
...receiptHandlers,
|
|
get_agentic_rectification_case_dossier: () => dossierFixture({ status: "confirmed" }),
|
|
append_agentic_rectification_turn: () => ({
|
|
turn_id: TURN_ID,
|
|
status: "completed",
|
|
assistant_message: "这是已持久化的回答",
|
|
successful_attempt_id: "aaaaaaaa-aaaa-4aaa-8aaa-aaaaaaaaaaaa",
|
|
should_execute: false,
|
|
already_in_progress: false,
|
|
idempotent: true,
|
|
}),
|
|
});
|
|
const { options, emitted, billing } = runOptions({
|
|
accounting: accounting.client,
|
|
buildAgent: async () => {
|
|
buildCount += 1;
|
|
return fakeAgentStream([]) as never;
|
|
},
|
|
});
|
|
|
|
const result = await runV9AgentTurn(options);
|
|
|
|
assert.equal(result.ok, true);
|
|
assert.equal(result.turnId, TURN_ID);
|
|
assert.equal(result.answerText, "这是已持久化的回答");
|
|
assert.equal(buildCount, 0);
|
|
assert.deepEqual(billing, { reserved: 1, completed: 0, released: 0 });
|
|
assert.deepEqual(emitted, [
|
|
{ type: "run.started" },
|
|
{ type: "answer.delta", text: "这是已持久化的回答" },
|
|
{ type: "run.completed", turnId: TURN_ID },
|
|
]);
|
|
assert.equal(
|
|
accounting.calls.some((call) => call.fn === "create_agentic_rectification_run_attempt"),
|
|
false,
|
|
);
|
|
});
|
|
|
|
test("pending request replay never rebuilds or releases the original reservation", async () => {
|
|
let buildCount = 0;
|
|
const accounting = fakeAccounting({
|
|
...receiptHandlers,
|
|
get_agentic_rectification_case_dossier: () => dossierFixture(),
|
|
append_agentic_rectification_turn: () => ({
|
|
turn_id: TURN_ID,
|
|
status: "pending",
|
|
assistant_message: null,
|
|
successful_attempt_id: null,
|
|
should_execute: false,
|
|
already_in_progress: true,
|
|
idempotent: true,
|
|
}),
|
|
});
|
|
const { options, billing } = runOptions({
|
|
accounting: accounting.client,
|
|
buildAgent: async () => {
|
|
buildCount += 1;
|
|
return fakeAgentStream([]) as never;
|
|
},
|
|
});
|
|
|
|
await assert.rejects(
|
|
runV9AgentTurn(options),
|
|
(error: unknown) => error instanceof Error
|
|
&& error.message.includes("agentic_rectification_turn_in_progress"),
|
|
);
|
|
assert.equal(buildCount, 0);
|
|
assert.deepEqual(billing, { reserved: 1, completed: 0, released: 0 });
|
|
});
|
|
|
|
for (const finalizedStatus of ["failed", "retryable"] as const) {
|
|
test(`${finalizedStatus} request replay does not rebuild and releases only the current claim`, async () => {
|
|
let buildCount = 0;
|
|
const accounting = fakeAccounting({
|
|
...receiptHandlers,
|
|
get_agentic_rectification_case_dossier: () => dossierFixture(),
|
|
append_agentic_rectification_turn: () => ({
|
|
turn_id: TURN_ID,
|
|
status: finalizedStatus,
|
|
assistant_message: null,
|
|
successful_attempt_id: null,
|
|
should_execute: false,
|
|
already_in_progress: false,
|
|
idempotent: true,
|
|
}),
|
|
});
|
|
const { options, billing } = runOptions({
|
|
accounting: accounting.client,
|
|
buildAgent: async () => {
|
|
buildCount += 1;
|
|
return fakeAgentStream([]) as never;
|
|
},
|
|
});
|
|
|
|
await assert.rejects(
|
|
runV9AgentTurn(options),
|
|
(error: unknown) => error instanceof Error
|
|
&& error.message.includes("agentic_rectification_turn_already_finalized"),
|
|
);
|
|
assert.equal(buildCount, 0);
|
|
assert.deepEqual(billing, { reserved: 1, completed: 0, released: 1 });
|
|
});
|
|
}
|
|
|
|
test("answer deltas stream in order and reasoning is never forwarded", async () => {
|
|
const { options, emitted } = runOptions({
|
|
buildAgent: async () => fakeAgentStream([
|
|
chunk("start"),
|
|
chunk("tool-call", { toolName: "skill", args: { name: RECTIFICATION_SKILL_NAME } }),
|
|
chunk("tool-result", { toolName: "skill" }),
|
|
chunk("tool-call", { toolName: "rectification-read-case", args: { caseId: CASE_ID } }),
|
|
chunk("tool-result", { toolName: "rectification-read-case" }),
|
|
chunk("reasoning-start", { id: "r1" }),
|
|
chunk("reasoning-delta", { text: "我应该先……" }),
|
|
chunk("reasoning-end"),
|
|
chunk("text-delta", { text: "好的，" }),
|
|
chunk("text-delta", { text: "先确认一下：" }),
|
|
chunk("raw", { payload: { tool_args: { secret: true } } }),
|
|
chunk("finish"),
|
|
]) as never,
|
|
});
|
|
const result = await runV9AgentTurn(options);
|
|
assert.equal(result.ok, true);
|
|
const deltas = emitted.filter((event) => event.type === "answer.delta");
|
|
assert.deepEqual(deltas, [
|
|
{ type: "answer.delta", text: "好的，" },
|
|
{ type: "answer.delta", text: "先确认一下：" },
|
|
]);
|
|
assert.deepEqual(
|
|
emitted.find((event) => event.type === "run.completed"),
|
|
{ type: "run.completed", turnId: TURN_ID },
|
|
);
|
|
assert.equal(emitted.some((event) => String(event.type).includes("reasoning")), false);
|
|
assert.equal(emitted.some((event) => String(event.type).includes("raw")), false);
|
|
});
|
|
|
|
test("half-failure never becomes settled history and releases usage", async () => {
|
|
const { options, emitted, billing } = runOptions({
|
|
buildAgent: async () => fakeAgentStream([
|
|
chunk("start"),
|
|
chunk("tool-call", { toolName: "skill", args: { name: RECTIFICATION_SKILL_NAME } }),
|
|
chunk("tool-result", { toolName: "skill" }),
|
|
chunk("tool-call", { toolName: "rectification-read-case", args: { caseId: CASE_ID } }),
|
|
chunk("tool-result", { toolName: "rectification-read-case" }),
|
|
chunk("text-delta", { text: "正在计算，" }),
|
|
chunk("error", { error: new Error("provider failure") }),
|
|
]) as never,
|
|
});
|
|
const result = await runV9AgentTurn(options);
|
|
assert.equal(result.ok, false);
|
|
assert.equal(result.turnStatus, "retryable");
|
|
assert.equal(billing.released, 1);
|
|
assert.equal(billing.completed, 0);
|
|
assert.equal(emitted.some((event) => event.type === "run.failed"), true);
|
|
assert.equal(emitted.some((event) => event.type === "run.completed"), false);
|
|
});
|
|
|
|
test("browser disconnect aborts the run, finalizes retryable and releases usage", async () => {
|
|
const controller = new AbortController();
|
|
const aborted = new Promise<void>((resolve) => {
|
|
controller.signal.addEventListener("abort", () => resolve(), { once: true });
|
|
});
|
|
const hanging = (async function* () {
|
|
yield chunk("start");
|
|
yield chunk("tool-call", { toolName: "skill", args: { name: RECTIFICATION_SKILL_NAME } });
|
|
yield chunk("tool-result", { toolName: "skill" });
|
|
yield chunk("tool-call", { toolName: "rectification-read-case", args: { caseId: CASE_ID } });
|
|
yield chunk("tool-result", { toolName: "rectification-read-case" });
|
|
yield chunk("text-delta", { text: "你好，" });
|
|
// The provider stream hangs until the client disconnects.
|
|
await aborted;
|
|
})();
|
|
const { options, billing } = runOptions({
|
|
signal: controller.signal,
|
|
buildAgent: async () => ({
|
|
stream: async () => ({
|
|
fullStream: hanging,
|
|
totalUsage: Promise.resolve({ inputTokens: 10, outputTokens: 20 }),
|
|
}),
|
|
getSkill: async () => ({ name: RECTIFICATION_SKILL_NAME, instructions: "skill" }),
|
|
}) as never,
|
|
});
|
|
const pending = runV9AgentTurn(options);
|
|
setTimeout(() => controller.abort(), 30);
|
|
const result = await pending;
|
|
assert.equal(result.ok, false);
|
|
assert.equal(result.errorCode, "stream_aborted");
|
|
assert.equal(billing.released, 1);
|
|
});
|
|
|
|
test("empty stream fails closed without completing billing", async () => {
|
|
const { options, emitted, billing } = runOptions({
|
|
buildAgent: async () => fakeAgentStream([
|
|
chunk("start"),
|
|
chunk("tool-call", { toolName: "skill", args: { name: RECTIFICATION_SKILL_NAME } }),
|
|
chunk("tool-result", { toolName: "skill" }),
|
|
chunk("tool-call", { toolName: "rectification-read-case", args: { caseId: CASE_ID } }),
|
|
chunk("tool-result", { toolName: "rectification-read-case" }),
|
|
chunk("finish"),
|
|
]) as never,
|
|
});
|
|
const result = await runV9AgentTurn(options);
|
|
assert.equal(result.ok, false);
|
|
assert.equal(result.errorCode, "empty_stream");
|
|
assert.equal(billing.released, 1);
|
|
assert.equal(emitted.some((event) => event.type === "run.completed"), false);
|
|
});
|
|
|
|
test("legacy Skill identity fails before billing reservation", async () => {
|
|
const accounting = fakeAccounting({
|
|
...receiptHandlers,
|
|
get_agentic_rectification_case_dossier: () => dossierFixture(),
|
|
get_agentic_rectification_skill_identity: () => {
|
|
throw new Error("agentic_rectification_legacy_skill_identity_unverifiable");
|
|
},
|
|
});
|
|
const { options, billing } = runOptions({ accounting: accounting.client });
|
|
await assert.rejects(
|
|
runV9AgentTurn(options),
|
|
(error: unknown) => error instanceof Error
|
|
&& error.message.includes("agentic_rectification_legacy_skill_identity_unverifiable"),
|
|
);
|
|
assert.equal(billing.reserved, 0);
|
|
});
|
|
|
|
test("Skill identity tool failures retain their safe public codes", () => {
|
|
assert.equal(
|
|
safeToolErrorCode(new Error("agentic_rectification_legacy_skill_identity_unverifiable")),
|
|
"legacy_skill_identity_unverifiable",
|
|
);
|
|
assert.equal(
|
|
safeToolErrorCode(new Error("agentic_rectification_skill_identity_missing")),
|
|
"skill_identity_missing",
|
|
);
|
|
});
|
|
|
|
test("execution receipts are persisted per turn (phases + tools)", async () => {
|
|
const accounting = fakeAccounting({
|
|
...receiptHandlers,
|
|
get_agentic_rectification_case_dossier: () => dossierFixture(),
|
|
append_agentic_rectification_turn: () => ({ turn_id: TURN_ID }),
|
|
finalize_agentic_rectification_turn: () => ({ turn_id: TURN_ID, status: "completed", idempotent: false }),
|
|
});
|
|
const { options } = runOptions({
|
|
accounting: accounting.client,
|
|
buildAgent: async () => fakeAgentStream([
|
|
chunk("start"),
|
|
chunk("tool-call", { toolName: "skill", args: { name: RECTIFICATION_SKILL_NAME } }),
|
|
chunk("tool-result", { toolName: "skill" }),
|
|
chunk("tool-call", { toolName: "rectification-read-case", args: { caseId: CASE_ID } }),
|
|
chunk("tool-result", { toolName: "rectification-read-case" }),
|
|
chunk("text-delta", { text: "你好" }),
|
|
chunk("finish"),
|
|
]) as never,
|
|
});
|
|
const result = await runV9AgentTurn(options);
|
|
assert.equal(result.ok, true);
|
|
const phases = accounting.calls
|
|
.filter((call) => call.fn === "insert_agentic_rectification_run_phase")
|
|
.map((call) => call.args.p_phase);
|
|
assert.ok(phases.includes("run.started"));
|
|
assert.ok(phases.includes("skill.bound"));
|
|
assert.ok(phases.includes("case.loaded"));
|
|
assert.ok(phases.includes("intent.classified"));
|
|
assert.ok(phases.includes("answer.composed"));
|
|
assert.ok(phases.includes("billing.settled"));
|
|
assert.ok(phases.includes("run.completed"));
|
|
// answer.delta is never persisted per-delta.
|
|
assert.ok(!phases.includes("answer.delta"));
|
|
assert.deepEqual(result.toolsUsed, ["rectification-read-case"]);
|
|
});
|
|
|
|
const SECOND_ATTEMPT_ID = "bbbbbbbb-bbbb-4bbb-8bbb-bbbbbbbbbbbb";
|
|
|
|
type AttemptFailure = "stream_aborted" | "stream_unfinished" | "empty_stream";
|
|
|
|
function attemptStream(
|
|
chunks: StreamChunk[],
|
|
usage: { inputTokens: number; outputTokens: number },
|
|
) {
|
|
return {
|
|
stream: async () => ({
|
|
fullStream: (async function* () {
|
|
for (const item of chunks) yield item;
|
|
})(),
|
|
totalUsage: Promise.resolve(usage),
|
|
}),
|
|
getSkill: async () => ({ name: RECTIFICATION_SKILL_NAME, instructions: "skill" }),
|
|
};
|
|
}
|
|
|
|
function failedAttemptChunks(errorCode: AttemptFailure): StreamChunk[] {
|
|
const chunks: StreamChunk[] = [
|
|
chunk("start"),
|
|
chunk("tool-call", { toolName: "skill", args: { name: RECTIFICATION_SKILL_NAME } }),
|
|
chunk("tool-result", { toolName: "skill" }),
|
|
chunk("tool-call", { toolName: "rectification-read-case", args: { caseId: CASE_ID } }),
|
|
chunk("tool-result", { toolName: "rectification-read-case" }),
|
|
chunk("tool-call", { toolName: "rectification-set-focus", args: { caseId: CASE_ID } }),
|
|
chunk("tool-result", { toolName: "rectification-set-focus" }),
|
|
chunk("text-delta", { text: errorCode === "empty_stream" ? " " : "失败 attempt 的半截文本" }),
|
|
];
|
|
if (errorCode === "stream_aborted") {
|
|
chunks.push(chunk("error", { error: new Error("provider stream aborted") }));
|
|
} else if (errorCode === "empty_stream") {
|
|
chunks.push(chunk("finish"));
|
|
}
|
|
return chunks;
|
|
}
|
|
|
|
function successfulAttemptChunks(): StreamChunk[] {
|
|
return [
|
|
chunk("start"),
|
|
chunk("tool-call", { toolName: "skill", args: { name: RECTIFICATION_SKILL_NAME } }),
|
|
chunk("tool-result", { toolName: "skill" }),
|
|
chunk("tool-call", { toolName: "rectification-read-case", args: { caseId: CASE_ID } }),
|
|
chunk("tool-result", { toolName: "rectification-read-case" }),
|
|
chunk("tool-call", { toolName: "rectification-record-evidence-batch", args: { caseId: CASE_ID } }),
|
|
chunk("tool-result", { toolName: "rectification-record-evidence-batch" }),
|
|
chunk("text-delta", { text: "第二次 attempt 成功" }),
|
|
chunk("finish"),
|
|
];
|
|
}
|
|
|
|
for (const failureCode of ["stream_aborted", "stream_unfinished", "empty_stream"] as const) {
|
|
test(`${failureCode} attempt is isolated and the second successful attempt exclusively commits public truth`, async () => {
|
|
let buildCount = 0;
|
|
const completedUsage: Array<{ inputTokens: number; outputTokens: number; durationMs: number }> = [];
|
|
const accounting = fakeAccounting({
|
|
...receiptHandlers,
|
|
get_agentic_rectification_case_dossier: () => dossierFixture(),
|
|
append_agentic_rectification_turn: () => ({ turn_id: TURN_ID }),
|
|
finalize_agentic_rectification_turn: () => ({ turn_id: TURN_ID, status: "completed", idempotent: false }),
|
|
});
|
|
const { options, emitted, billing } = runOptions({
|
|
accounting: accounting.client,
|
|
billing: {
|
|
reserve: async () => { billing.reserved += 1; return { success: true, status: 200 }; },
|
|
complete: async (usage) => { completedUsage.push(usage); billing.completed += 1; return true; },
|
|
release: async () => { billing.released += 1; return true; },
|
|
},
|
|
buildAgent: async () => {
|
|
buildCount += 1;
|
|
return buildCount === 1
|
|
? attemptStream(failedAttemptChunks(failureCode), { inputTokens: 901, outputTokens: 902 }) as never
|
|
: attemptStream(successfulAttemptChunks(), { inputTokens: 31, outputTokens: 17 }) as never;
|
|
},
|
|
});
|
|
|
|
const result = await runV9AgentTurn(options);
|
|
|
|
assert.equal(buildCount, 2);
|
|
assert.equal(result.ok, true);
|
|
assert.equal(result.answerText, "第二次 attempt 成功");
|
|
assert.deepEqual(result.toolsUsed, [
|
|
"rectification-read-case",
|
|
"rectification-record-evidence-batch",
|
|
]);
|
|
assert.deepEqual(
|
|
emitted.filter((event) => event.type === "answer.delta"),
|
|
[{ type: "answer.delta", text: "第二次 attempt 成功" }],
|
|
);
|
|
assert.equal(
|
|
emitted.some((event) => event.type === "answer.delta" && event.text?.includes("失败 attempt")),
|
|
false,
|
|
);
|
|
assert.equal(
|
|
emitted.some((event) =>
|
|
event.type === "tool.activity"
|
|
&& (event as { tool?: string }).tool === "rectification-set-focus"),
|
|
false,
|
|
);
|
|
assert.equal(
|
|
emitted.some((event) =>
|
|
event.type === "tool.activity"
|
|
&& (event as { tool?: string }).tool === "rectification-record-evidence-batch"),
|
|
true,
|
|
);
|
|
assert.deepEqual(completedUsage.map(({ inputTokens, outputTokens }) => ({ inputTokens, outputTokens })), [
|
|
{ inputTokens: 31, outputTokens: 17 },
|
|
]);
|
|
assert.deepEqual(billing, { reserved: 1, completed: 1, released: 0 });
|
|
|
|
const attemptFinalizations = accounting.calls
|
|
.filter((call) => call.fn === "finalize_agentic_rectification_run_attempt")
|
|
.map((call) => ({
|
|
attemptId: call.args.p_attempt_id,
|
|
status: call.args.p_status,
|
|
errorCode: call.args.p_error_code,
|
|
usage: call.args.p_usage,
|
|
}));
|
|
assert.deepEqual(attemptFinalizations, [
|
|
{
|
|
attemptId: "aaaaaaaa-aaaa-4aaa-8aaa-aaaaaaaaaaaa",
|
|
status: "retryable",
|
|
errorCode: failureCode,
|
|
usage: { inputTokens: 0, outputTokens: 0 },
|
|
},
|
|
{
|
|
attemptId: SECOND_ATTEMPT_ID,
|
|
status: "completed",
|
|
errorCode: null,
|
|
usage: { inputTokens: 31, outputTokens: 17 },
|
|
},
|
|
]);
|
|
|
|
const finalizedTurn = accounting.calls.find((call) => call.fn === "finalize_agentic_rectification_turn");
|
|
assert.deepEqual(finalizedTurn?.args, {
|
|
p_user_id: USER_ID,
|
|
p_case_id: CASE_ID,
|
|
p_turn_id: TURN_ID,
|
|
p_attempt_id: SECOND_ATTEMPT_ID,
|
|
p_status: "completed",
|
|
p_assistant_message: "第二次 attempt 成功",
|
|
p_successful_attempt_id: SECOND_ATTEMPT_ID,
|
|
});
|
|
const settledPhases = accounting.calls.filter((call) =>
|
|
call.fn === "insert_agentic_rectification_run_phase"
|
|
&& (call.args.p_phase === "billing.settled" || call.args.p_phase === "run.completed"));
|
|
assert.deepEqual(settledPhases.map((call) => call.args.p_attempt_id), [
|
|
SECOND_ATTEMPT_ID,
|
|
SECOND_ATTEMPT_ID,
|
|
]);
|
|
const billingSettledIndex = accounting.calls.findIndex((call) =>
|
|
call.fn === "insert_agentic_rectification_run_phase"
|
|
&& call.args.p_phase === "billing.settled");
|
|
const runCompletedIndex = accounting.calls.findIndex((call) =>
|
|
call.fn === "insert_agentic_rectification_run_phase"
|
|
&& call.args.p_phase === "run.completed");
|
|
const successfulAttemptFinalizeIndex = accounting.calls.findIndex((call) =>
|
|
call.fn === "finalize_agentic_rectification_run_attempt"
|
|
&& call.args.p_status === "completed");
|
|
const completedTurnFinalizeIndex = accounting.calls.findIndex((call) =>
|
|
call.fn === "finalize_agentic_rectification_turn"
|
|
&& call.args.p_status === "completed");
|
|
assert.ok(billingSettledIndex < runCompletedIndex);
|
|
assert.ok(runCompletedIndex < successfulAttemptFinalizeIndex);
|
|
assert.ok(successfulAttemptFinalizeIndex < completedTurnFinalizeIndex);
|
|
});
|
|
}
|
|
|
|
test("a failed set-focus cannot complete a question turn even when the agent emits answer text", async () => {
|
|
let buildCount = 0;
|
|
const accounting = fakeAccounting({
|
|
...receiptHandlers,
|
|
get_agentic_rectification_case_dossier: () => dossierFixture(),
|
|
append_agentic_rectification_turn: () => ({ turn_id: TURN_ID }),
|
|
finalize_agentic_rectification_turn: (_fn, args) => ({
|
|
turn_id: TURN_ID,
|
|
status: args.p_status,
|
|
idempotent: false,
|
|
}),
|
|
});
|
|
const failedFocusAttempt = () => attemptStream([
|
|
chunk("start"),
|
|
chunk("tool-call", { toolName: "skill", args: { name: RECTIFICATION_SKILL_NAME } }),
|
|
chunk("tool-result", { toolName: "skill" }),
|
|
chunk("tool-call", { toolName: "rectification-read-case", args: { caseId: CASE_ID } }),
|
|
chunk("tool-result", { toolName: "rectification-read-case" }),
|
|
chunk("tool-call", { toolName: "rectification-set-focus", args: { caseId: CASE_ID } }),
|
|
chunk("tool-error", { toolName: "rectification-set-focus", error: new Error("invalid_focus") }),
|
|
chunk("text-delta", { text: "主问题：请确认这段经历发生在哪个月？" }),
|
|
chunk("finish"),
|
|
], { inputTokens: 41, outputTokens: 23 });
|
|
const { options, emitted, billing } = runOptions({
|
|
accounting: accounting.client,
|
|
buildAgent: async () => {
|
|
buildCount += 1;
|
|
return failedFocusAttempt() as never;
|
|
},
|
|
});
|
|
|
|
const result = await runV9AgentTurn(options);
|
|
|
|
assert.equal(buildCount, 2);
|
|
assert.equal(result.ok, false);
|
|
assert.equal(result.turnStatus, "retryable");
|
|
assert.equal(result.errorCode, "focus_persistence_failed");
|
|
assert.equal(result.answerText, "");
|
|
assert.deepEqual(result.toolsUsed, ["rectification-read-case", "rectification-set-focus"]);
|
|
assert.deepEqual(billing, { reserved: 1, completed: 0, released: 1 });
|
|
assert.equal(emitted.some((event) => event.type === "answer.delta"), false);
|
|
assert.equal(emitted.some((event) => event.type === "run.completed"), false);
|
|
assert.equal(
|
|
emitted.some((event) => event.type === "tool.activity"
|
|
&& (event as { tool?: string; status?: string }).tool === "rectification-set-focus"
|
|
&& (event as { tool?: string; status?: string }).status === "failed"),
|
|
true,
|
|
);
|
|
assert.equal(emitted.at(-1)?.type, "run.failed");
|
|
|
|
const attemptFinalizations = accounting.calls
|
|
.filter((call) => call.fn === "finalize_agentic_rectification_run_attempt")
|
|
.map((call) => ({
|
|
status: call.args.p_status,
|
|
errorCode: call.args.p_error_code,
|
|
usage: call.args.p_usage,
|
|
}));
|
|
assert.deepEqual(attemptFinalizations, [
|
|
{ status: "retryable", errorCode: "focus_persistence_failed", usage: { inputTokens: 0, outputTokens: 0 } },
|
|
{ status: "retryable", errorCode: "focus_persistence_failed", usage: { inputTokens: 0, outputTokens: 0 } },
|
|
]);
|
|
assert.equal(
|
|
accounting.calls.some((call) => call.fn === "insert_agentic_rectification_run_phase"
|
|
&& call.args.p_phase === "run.completed"),
|
|
false,
|
|
);
|
|
const finalizedTurn = accounting.calls.find((call) => call.fn === "finalize_agentic_rectification_turn");
|
|
assert.deepEqual(finalizedTurn?.args, {
|
|
p_user_id: USER_ID,
|
|
p_case_id: CASE_ID,
|
|
p_turn_id: TURN_ID,
|
|
p_attempt_id: SECOND_ATTEMPT_ID,
|
|
p_status: "retryable",
|
|
p_assistant_message: null,
|
|
p_successful_attempt_id: null,
|
|
});
|
|
});
|
|
|
|
test("a same-attempt set-focus retry that finishes completed can commit the question turn", async () => {
|
|
const accounting = fakeAccounting({
|
|
...receiptHandlers,
|
|
get_agentic_rectification_case_dossier: () => dossierFixture(),
|
|
append_agentic_rectification_turn: () => ({ turn_id: TURN_ID }),
|
|
finalize_agentic_rectification_turn: () => ({ turn_id: TURN_ID, status: "completed", idempotent: false }),
|
|
});
|
|
const { options, emitted, billing } = runOptions({
|
|
accounting: accounting.client,
|
|
buildAgent: async () => attemptStream([
|
|
chunk("start"),
|
|
chunk("tool-call", { toolName: "skill", args: { name: RECTIFICATION_SKILL_NAME } }),
|
|
chunk("tool-result", { toolName: "skill" }),
|
|
chunk("tool-call", { toolName: "rectification-read-case", args: { caseId: CASE_ID } }),
|
|
chunk("tool-result", { toolName: "rectification-read-case" }),
|
|
chunk("tool-call", { toolName: "rectification-set-focus", args: { caseId: CASE_ID } }),
|
|
chunk("tool-error", { toolName: "rectification-set-focus", error: new Error("invalid_focus") }),
|
|
chunk("tool-call", { toolName: "rectification-set-focus", args: { caseId: CASE_ID } }),
|
|
chunk("tool-result", { toolName: "rectification-set-focus" }),
|
|
chunk("text-delta", { text: "主问题：请确认这段经历发生在哪个月？" }),
|
|
chunk("finish"),
|
|
], { inputTokens: 43, outputTokens: 29 }) as never,
|
|
});
|
|
|
|
const result = await runV9AgentTurn(options);
|
|
|
|
assert.equal(result.ok, true);
|
|
assert.equal(result.turnStatus, "completed");
|
|
assert.equal(result.errorCode, null);
|
|
assert.deepEqual(billing, { reserved: 1, completed: 1, released: 0 });
|
|
assert.deepEqual(
|
|
emitted.filter((event) => event.type === "answer.delta"),
|
|
[{ type: "answer.delta", text: "主问题：请确认这段经历发生在哪个月？" }],
|
|
);
|
|
assert.equal(emitted.some((event) => event.type === "run.completed"), true);
|
|
|
|
let receiptState = createRectificationActivityReceiptState();
|
|
const focusStatuses: string[] = [];
|
|
for (const event of emitted) {
|
|
const activity = event as { type: string; tool?: string; status?: string };
|
|
if (activity.type !== "tool.activity" || activity.tool !== "rectification-set-focus") continue;
|
|
if (activity.status !== "started" && activity.status !== "completed" && activity.status !== "failed") continue;
|
|
focusStatuses.push(activity.status);
|
|
receiptState = reduceRectificationActivityReceipt(receiptState, {
|
|
tool: "rectification-set-focus",
|
|
status: activity.status,
|
|
});
|
|
}
|
|
assert.deepEqual(focusStatuses, ["started", "failed", "started", "completed"]);
|
|
assert.deepEqual(receiptFromRectificationActivityState(receiptState), {
|
|
steps: ["rectification-set-focus"],
|
|
methods: [],
|
|
});
|
|
});
|
|
|
|
test("an unclaimed V10 attempt never starts the model", async () => {
|
|
let buildCount = 0;
|
|
const accounting = fakeAccounting({
|
|
...receiptHandlers,
|
|
get_agentic_rectification_case_dossier: () => dossierFixture(),
|
|
append_agentic_rectification_turn: () => ({ turn_id: TURN_ID }),
|
|
create_agentic_rectification_run_attempt: () => ({
|
|
attempt_id: "aaaaaaaa-aaaa-4aaa-8aaa-aaaaaaaaaaaa",
|
|
status: "started",
|
|
should_execute: false,
|
|
already_in_progress: true,
|
|
idempotent: true,
|
|
}),
|
|
});
|
|
const { options, billing } = runOptions({
|
|
accounting: accounting.client,
|
|
buildAgent: async () => {
|
|
buildCount += 1;
|
|
return fakeAgentStream([]) as never;
|
|
},
|
|
});
|
|
|
|
await assert.rejects(
|
|
runV9AgentTurn(options),
|
|
(error: unknown) => error instanceof Error
|
|
&& error.message.includes("agentic_rectification_attempt_in_progress"),
|
|
);
|
|
assert.equal(buildCount, 0);
|
|
assert.deepEqual(billing, { reserved: 1, completed: 0, released: 1 });
|
|
assert.equal(
|
|
accounting.calls.some((call) => call.fn === "finalize_agentic_rectification_run_attempt"),
|
|
false,
|
|
);
|
|
});
|
|
|
|
test("non-retryable attempt errors do not start a second attempt", async () => {
|
|
let buildCount = 0;
|
|
const accounting = fakeAccounting({
|
|
...receiptHandlers,
|
|
get_agentic_rectification_case_dossier: () => dossierFixture(),
|
|
append_agentic_rectification_turn: () => ({ turn_id: TURN_ID }),
|
|
finalize_agentic_rectification_turn: () => ({ turn_id: TURN_ID, status: "failed", idempotent: false }),
|
|
});
|
|
const { options, emitted, billing } = runOptions({
|
|
accounting: accounting.client,
|
|
buildAgent: async () => {
|
|
buildCount += 1;
|
|
throw new Error("provider contract violation");
|
|
},
|
|
});
|
|
|
|
const result = await runV9AgentTurn(options);
|
|
|
|
assert.equal(buildCount, 1);
|
|
assert.equal(result.ok, false);
|
|
assert.equal(result.turnStatus, "failed");
|
|
assert.equal(result.errorCode, "run_failed");
|
|
assert.deepEqual(billing, { reserved: 1, completed: 0, released: 1 });
|
|
assert.deepEqual(emitted, [{ type: "run.started" }, { type: "run.failed" }]);
|
|
assert.equal(
|
|
accounting.calls.filter((call) => call.fn === "create_agentic_rectification_run_attempt").length,
|
|
1,
|
|
);
|
|
const attemptFinalize = accounting.calls.find((call) => call.fn === "finalize_agentic_rectification_run_attempt");
|
|
assert.equal(attemptFinalize?.args.p_status, "failed");
|
|
assert.equal(attemptFinalize?.args.p_error_code, "run_failed");
|
|
const turnFinalize = accounting.calls.find((call) => call.fn === "finalize_agentic_rectification_turn");
|
|
assert.equal(turnFinalize?.args.p_attempt_id, attemptFinalize?.args.p_attempt_id);
|
|
assert.equal(turnFinalize?.args.p_assistant_message, null);
|
|
assert.equal(turnFinalize?.args.p_successful_attempt_id, null);
|
|
});
|