317 lines
13 KiB
TypeScript
317 lines
13 KiB
TypeScript
import assert from "node:assert/strict";
|
|
import test from "node:test";
|
|
|
|
import {
|
|
mapStreamChunkToPhase,
|
|
safePublicEvent,
|
|
streamToolNames,
|
|
} from "../src/lib/rectification-agentic/v9/stream-mapping.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";
|
|
|
|
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.deepEqual(
|
|
mapStreamChunkToPhase(chunk("tool-call", { toolName: "skill", args: { name: RECTIFICATION_SKILL_NAME } }) as never),
|
|
{ type: "skill.started" },
|
|
);
|
|
assert.deepEqual(
|
|
mapStreamChunkToPhase(chunk("tool-result", { toolName: "skill" }) as never),
|
|
{ type: "skill.loaded" },
|
|
);
|
|
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("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: "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: "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);
|
|
});
|
|
|
|
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 }> = [];
|
|
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("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("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.equal(emitted.some((event) => event.type === "run.completed"), true);
|
|
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("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("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("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("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.started"));
|
|
assert.ok(phases.includes("skill.loaded"));
|
|
assert.ok(phases.includes("case.loaded"));
|
|
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"]);
|
|
});
|