Files
openclaw/extensions/deepseek/index.test.ts
NVIDIAN 797d574dfd fix(deepseek): expose V4 max thinking levels (#73008)
Merged via squash.

Prepared head SHA: ef561a59de
Co-authored-by: ai-hpc <183861985+ai-hpc@users.noreply.github.com>
Co-authored-by: hxy91819 <8814856+hxy91819@users.noreply.github.com>
Reviewed-by: @hxy91819
2026-04-30 23:34:05 +08:00

466 lines
14 KiB
TypeScript

import type { Context, Model } from "@mariozechner/pi-ai";
import { createAssistantMessageEventStream } from "@mariozechner/pi-ai";
import {
registerSingleProviderPlugin,
resolveProviderPluginChoice,
} from "openclaw/plugin-sdk/plugin-test-runtime";
import { buildOpenAICompletionsParams } from "openclaw/plugin-sdk/provider-transport-runtime";
import { describe, expect, it } from "vitest";
import { runSingleProviderCatalog } from "../test-support/provider-model-test-helpers.js";
import deepseekPlugin from "./index.js";
import { createDeepSeekV4ThinkingWrapper } from "./stream.js";
type OpenAICompletionsModel = Model<"openai-completions">;
type PayloadCapture = {
payload?: Record<string, unknown>;
};
function deepSeekV4Model(id: "deepseek-v4-flash" | "deepseek-v4-pro"): OpenAICompletionsModel {
return {
provider: "deepseek",
id,
name: id === "deepseek-v4-flash" ? "DeepSeek V4 Flash" : "DeepSeek V4 Pro",
api: "openai-completions",
baseUrl: "https://api.deepseek.com",
reasoning: true,
input: ["text"],
cost: { input: 0, output: 0, cacheRead: 0, cacheWrite: 0 },
contextWindow: 1_000_000,
maxTokens: 384_000,
compat: {
supportsUsageInStreaming: true,
supportsReasoningEffort: true,
maxTokensField: "max_tokens",
},
} as OpenAICompletionsModel;
}
function createPayloadCapturingStream(capture: PayloadCapture) {
return (
streamModel: OpenAICompletionsModel,
streamContext: Context,
options?: { onPayload?: (payload: unknown, model: unknown) => unknown },
) => {
capture.payload = buildOpenAICompletionsParams(streamModel, streamContext, {
reasoning: "high",
} as never);
options?.onPayload?.(capture.payload, streamModel);
const stream = createAssistantMessageEventStream();
queueMicrotask(() => stream.end());
return stream;
};
}
describe("deepseek provider plugin", () => {
it("registers DeepSeek with api-key auth wizard metadata", async () => {
const provider = await registerSingleProviderPlugin(deepseekPlugin);
const resolved = resolveProviderPluginChoice({
providers: [provider],
choice: "deepseek-api-key",
});
expect(provider.id).toBe("deepseek");
expect(provider.label).toBe("DeepSeek");
expect(provider.envVars).toEqual(["DEEPSEEK_API_KEY"]);
expect(provider.auth).toHaveLength(1);
expect(resolved).not.toBeNull();
expect(resolved?.provider.id).toBe("deepseek");
expect(resolved?.method.id).toBe("api-key");
});
it("builds the static DeepSeek model catalog", async () => {
const provider = await registerSingleProviderPlugin(deepseekPlugin);
const catalogProvider = await runSingleProviderCatalog(provider);
expect(catalogProvider.api).toBe("openai-completions");
expect(catalogProvider.baseUrl).toBe("https://api.deepseek.com");
expect(catalogProvider.models?.map((model) => model.id)).toEqual([
"deepseek-v4-flash",
"deepseek-v4-pro",
"deepseek-chat",
"deepseek-reasoner",
]);
expect(catalogProvider.models?.find((model) => model.id === "deepseek-v4-flash")).toMatchObject(
{
reasoning: true,
contextWindow: 1_000_000,
maxTokens: 384_000,
compat: expect.objectContaining({
supportsReasoningEffort: true,
maxTokensField: "max_tokens",
}),
},
);
expect(
catalogProvider.models?.find((model) => model.id === "deepseek-reasoner")?.reasoning,
).toBe(true);
});
it("owns OpenAI-compatible replay policy", async () => {
const provider = await registerSingleProviderPlugin(deepseekPlugin);
expect(provider.buildReplayPolicy?.({ modelApi: "openai-completions" } as never)).toMatchObject(
{
sanitizeToolCallIds: true,
toolCallIdMode: "strict",
validateGeminiTurns: true,
validateAnthropicTurns: true,
},
);
});
it("advertises max thinking levels for DeepSeek V4 models only", async () => {
const provider = await registerSingleProviderPlugin(deepseekPlugin);
const resolveThinkingProfile = provider.resolveThinkingProfile!;
const expectedV4Levels = ["off", "minimal", "low", "medium", "high", "xhigh", "max"];
expect(
resolveThinkingProfile({
provider: "deepseek",
modelId: "deepseek-v4-pro",
} as never)?.levels.map((level) => level.id),
).toEqual(expectedV4Levels);
expect(
resolveThinkingProfile({
provider: "deepseek",
modelId: "deepseek-v4-flash",
} as never)?.defaultLevel,
).toBe("high");
expect(
resolveThinkingProfile({
provider: "deepseek",
modelId: "deepseek-v4-flash",
} as never)?.levels.map((level) => level.id),
).toEqual(expectedV4Levels);
expect(
resolveThinkingProfile({ provider: "deepseek", modelId: "deepseek-chat" } as never),
).toBe(undefined);
expect(
resolveThinkingProfile({ provider: "deepseek", modelId: "deepseek-reasoner" } as never),
).toBe(undefined);
});
it("maps thinking levels to DeepSeek V4 payload controls", async () => {
let capturedPayload: Record<string, unknown> | undefined;
const baseStreamFn = (
_model: Model<"openai-completions">,
_context: Context,
options?: { onPayload?: (payload: unknown) => unknown },
) => {
capturedPayload = {
model: "deepseek-v4-pro",
reasoning_effort: "high",
};
options?.onPayload?.(capturedPayload);
const stream = createAssistantMessageEventStream();
queueMicrotask(() => stream.end());
return stream;
};
const wrapThinkingOff = createDeepSeekV4ThinkingWrapper(baseStreamFn as never, "off");
expect(wrapThinkingOff).toBeDefined();
await wrapThinkingOff?.(
{
provider: "deepseek",
id: "deepseek-v4-pro",
api: "openai-completions",
} as never,
{ messages: [] } as never,
{},
);
expect(capturedPayload).toMatchObject({ thinking: { type: "disabled" } });
expect(capturedPayload).not.toHaveProperty("reasoning_effort");
const wrapThinkingXhigh = createDeepSeekV4ThinkingWrapper(baseStreamFn as never, "xhigh");
expect(wrapThinkingXhigh).toBeDefined();
await wrapThinkingXhigh?.(
{
provider: "deepseek",
id: "deepseek-v4-pro",
api: "openai-completions",
} as never,
{ messages: [] } as never,
{},
);
expect(capturedPayload).toMatchObject({
thinking: { type: "enabled" },
reasoning_effort: "max",
});
});
it("preserves replayed reasoning_content when DeepSeek V4 thinking is enabled", async () => {
const capture: PayloadCapture = {};
const model = deepSeekV4Model("deepseek-v4-flash");
const context = {
messages: [
{ role: "user", content: "hi", timestamp: 1 },
{
role: "assistant",
api: "openai-completions",
provider: "deepseek",
model: "deepseek-v4-flash",
content: [
{
type: "thinking",
thinking: "call reasoning",
thinkingSignature: "reasoning_content",
},
{ type: "toolCall", id: "call_1", name: "read", arguments: {} },
],
usage: {
input: 0,
output: 0,
cacheRead: 0,
cacheWrite: 0,
totalTokens: 0,
cost: { input: 0, output: 0, cacheRead: 0, cacheWrite: 0, total: 0 },
},
stopReason: "toolUse",
timestamp: 2,
},
{
role: "toolResult",
toolCallId: "call_1",
toolName: "read",
content: [{ type: "text", text: "ok" }],
isError: false,
timestamp: 3,
},
],
tools: [
{
name: "read",
description: "Read data",
parameters: { type: "object", properties: {}, required: [], additionalProperties: false },
},
],
} as Context;
const baseStreamFn = createPayloadCapturingStream(capture);
const wrapThinkingHigh = createDeepSeekV4ThinkingWrapper(baseStreamFn as never, "high");
expect(wrapThinkingHigh).toBeDefined();
await wrapThinkingHigh?.(model, context, {});
expect(capture.payload).toMatchObject({
thinking: { type: "enabled" },
reasoning_effort: "high",
});
expect((capture.payload?.messages as Array<Record<string, unknown>>)[1]).toMatchObject({
role: "assistant",
reasoning_content: "call reasoning",
tool_calls: [
{
id: "call_1",
type: "function",
function: {
name: "read",
arguments: "{}",
},
},
],
});
});
it("adds blank reasoning_content for replayed tool calls from non-DeepSeek turns", async () => {
const capture: PayloadCapture = {};
const model = deepSeekV4Model("deepseek-v4-pro");
const context = {
messages: [
{ role: "user", content: "hi", timestamp: 1 },
{
role: "assistant",
api: "openai-completions",
provider: "openai",
model: "gpt-5.4",
content: [{ type: "toolCall", id: "call_1", name: "read", arguments: {} }],
usage: {
input: 0,
output: 0,
cacheRead: 0,
cacheWrite: 0,
totalTokens: 0,
cost: { input: 0, output: 0, cacheRead: 0, cacheWrite: 0, total: 0 },
},
stopReason: "toolUse",
timestamp: 2,
},
{
role: "toolResult",
toolCallId: "call_1",
toolName: "read",
content: [{ type: "text", text: "ok" }],
isError: false,
timestamp: 3,
},
],
tools: [
{
name: "read",
description: "Read data",
parameters: { type: "object", properties: {}, required: [], additionalProperties: false },
},
],
} as Context;
const baseStreamFn = createPayloadCapturingStream(capture);
const wrapThinkingHigh = createDeepSeekV4ThinkingWrapper(baseStreamFn as never, "high");
expect(wrapThinkingHigh).toBeDefined();
await wrapThinkingHigh?.(model, context, {});
expect((capture.payload?.messages as Array<Record<string, unknown>>)[1]).toMatchObject({
role: "assistant",
reasoning_content: "",
tool_calls: [
{
id: "call_1",
type: "function",
function: {
name: "read",
arguments: "{}",
},
},
],
});
});
it("adds blank reasoning_content for replayed plain assistant messages", async () => {
const capture: PayloadCapture = {};
const model = deepSeekV4Model("deepseek-v4-pro");
const context = {
messages: [
{ role: "user", content: "hi", timestamp: 1 },
{
role: "assistant",
api: "openai-completions",
provider: "openai",
model: "gpt-5.4",
content: [{ type: "text", text: "Hello." }],
usage: {
input: 0,
output: 0,
cacheRead: 0,
cacheWrite: 0,
totalTokens: 0,
cost: { input: 0, output: 0, cacheRead: 0, cacheWrite: 0, total: 0 },
},
stopReason: "stop",
timestamp: 2,
},
{ role: "user", content: "next", timestamp: 3 },
],
} as Context;
const baseStreamFn = createPayloadCapturingStream(capture);
const wrapThinkingHigh = createDeepSeekV4ThinkingWrapper(baseStreamFn as never, "high");
expect(wrapThinkingHigh).toBeDefined();
await wrapThinkingHigh?.(model, context, {});
expect((capture.payload?.messages as Array<Record<string, unknown>>)[1]).toMatchObject({
role: "assistant",
content: "Hello.",
reasoning_content: "",
});
});
it("strips replayed reasoning_content when DeepSeek V4 thinking is disabled", async () => {
const capture: PayloadCapture = {};
const model = deepSeekV4Model("deepseek-v4-flash");
const context = {
messages: [
{ role: "user", content: "hi", timestamp: 1 },
{
role: "assistant",
api: "openai-completions",
provider: "deepseek",
model: "deepseek-v4-flash",
content: [
{
type: "thinking",
thinking: "call reasoning",
thinkingSignature: "reasoning_content",
},
{ type: "toolCall", id: "call_1", name: "read", arguments: {} },
],
usage: {
input: 0,
output: 0,
cacheRead: 0,
cacheWrite: 0,
totalTokens: 0,
cost: { input: 0, output: 0, cacheRead: 0, cacheWrite: 0, total: 0 },
},
stopReason: "toolUse",
timestamp: 2,
},
{
role: "toolResult",
toolCallId: "call_1",
toolName: "read",
content: [{ type: "text", text: "ok" }],
isError: false,
timestamp: 3,
},
],
tools: [
{
name: "read",
description: "Read data",
parameters: { type: "object", properties: {}, required: [], additionalProperties: false },
},
],
} as Context;
const baseStreamFn = createPayloadCapturingStream(capture);
const wrapThinkingNone = createDeepSeekV4ThinkingWrapper(
baseStreamFn as never,
"none" as never,
);
expect(wrapThinkingNone).toBeDefined();
await wrapThinkingNone?.(model, context, {});
expect(capture.payload).toMatchObject({ thinking: { type: "disabled" } });
expect(capture.payload).not.toHaveProperty("reasoning_effort");
expect((capture.payload?.messages as Array<Record<string, unknown>>)[1]).not.toHaveProperty(
"reasoning_content",
);
});
it("publishes configured DeepSeek models through plugin-owned catalog augmentation", async () => {
const provider = await registerSingleProviderPlugin(deepseekPlugin);
expect(
provider.augmentModelCatalog?.({
config: {
models: {
providers: {
deepseek: {
models: [
{
id: "deepseek-chat",
name: "DeepSeek Chat",
input: ["text"],
reasoning: false,
contextWindow: 65536,
},
],
},
},
},
},
} as never),
).toEqual([
{
provider: "deepseek",
id: "deepseek-chat",
name: "DeepSeek Chat",
input: ["text"],
reasoning: false,
contextWindow: 65536,
},
]);
});
});