1353 lines
45 KiB
TypeScript
1353 lines
45 KiB
TypeScript
/**
|
||
* Provider 请求形态测试矩阵（v0.6.4 P2-6）
|
||
*
|
||
* 此前 ollama（599 行）/ anthropic（532 行）两个最复杂的适配器零测试 —— 恰好也是
|
||
* 本轮审计中缺陷密度最高的文件。本文件通过 mock fetch 记录真实请求体，
|
||
* 锁定以下契约：
|
||
*
|
||
* Anthropic:
|
||
* A1 消息转换（system 顶层 / user-assistant-tool 三角色映射 / 孤立 tool_result 过滤）
|
||
* A2 max_tokens 按模型钳制（引擎默认 63488 → sonnet 64000 / opus 32000）
|
||
* A3 thinking 预算下限保护（小 maxTokens 场景 budget≥1024 且 < max_tokens，此前 API 400）
|
||
* A4 thinking 开启时不传 temperature；关闭时显式传递
|
||
*
|
||
* Ollama:
|
||
* O1 options 映射（num_predict=numTokens、num_ctx=contextLength、stop、top_p）
|
||
* O2 think 参数 effort 映射（low→"low"、xhigh→"xhigh"、max/true→true）与未配置时缺省
|
||
* O3 图片归一化（data URI 剥前缀；无 URL 触发下载分支时零网络请求）
|
||
*
|
||
* Agnes:
|
||
* G1 思考模式对称性 —— thinkingEnabled=false 必须显式发送 enable_thinking:false
|
||
*/
|
||
|
||
import { describe, it, expect, vi } from 'vitest';
|
||
|
||
vi.mock('electron-log', () => ({
|
||
default: { info: vi.fn(), warn: vi.fn(), error: vi.fn(), debug: vi.fn() },
|
||
}));
|
||
|
||
import { AnthropicAdapter } from '../anthropic.adapter';
|
||
import { OllamaAdapter } from '../ollama.adapter';
|
||
import { MimoAdapter } from '../mimo.adapter';
|
||
import { AgnesAdapter } from '../agnes-ai.adapter';
|
||
import { DeepSeekAdapter } from '../deepseek.adapter';
|
||
import { OpenAIAdapter } from '../openai.adapter';
|
||
import type { MetonaRequest } from '../../types';
|
||
|
||
/** 安装全局 fetch 捕获器：记录每次请求体并返回一个三家协议都能解析的合成响应 */
|
||
function captureFetch(): { bodies: Array<Record<string, unknown>> } {
|
||
const bodies: Array<Record<string, unknown>> = [];
|
||
// 兼容三家的非流式解析所需的最小字段集：
|
||
// OpenAI 兼容(agnes): choices[].message/finish_reason；Anthropic: content[]/usage/stop_reason；
|
||
// Ollama: message/done/prompt_eval_count/eval_count
|
||
const genericBody = {
|
||
id: 'cmpl-test',
|
||
object: 'chat.completion',
|
||
created: Date.now(),
|
||
model: 'test-model',
|
||
choices: [{ index: 0, message: { role: 'assistant', content: 'ok' }, finish_reason: 'stop' }],
|
||
content: [],
|
||
usage: {
|
||
prompt_tokens: 3,
|
||
completion_tokens: 2,
|
||
total_tokens: 5,
|
||
input_tokens: 3,
|
||
output_tokens: 2,
|
||
prompt_eval_count: 3,
|
||
eval_count: 2,
|
||
},
|
||
stop_reason: 'end_turn',
|
||
message: { role: 'assistant', content: 'ok' },
|
||
done: true,
|
||
};
|
||
const fetchMock = vi.fn(async (_url: string | URL, init?: RequestInit) => {
|
||
bodies.push(JSON.parse(String(init?.body ?? '{}')) as Record<string, unknown>);
|
||
return new Response(JSON.stringify(genericBody), {
|
||
status: 200,
|
||
headers: { 'Content-Type': 'application/json' },
|
||
});
|
||
});
|
||
vi.stubGlobal('fetch', fetchMock);
|
||
return { bodies };
|
||
}
|
||
|
||
function makeRequest(overrides?: Partial<MetonaRequest>): MetonaRequest {
|
||
return {
|
||
meta: {
|
||
sessionId: 's1',
|
||
iteration: 1,
|
||
requestId: 'r1',
|
||
timestamp: Date.now(),
|
||
agentVersion: 'test',
|
||
},
|
||
systemPrompt: {
|
||
roleDefinition: 'You are Metona.',
|
||
outputConstraints: 'Be concise.',
|
||
safetyGuidelines: 'Stay safe.',
|
||
dynamicReminders: '',
|
||
},
|
||
messages: [{ role: 'user', content: 'hi', timestamp: Date.now() }],
|
||
params: { maxTokens: 63_488, temperature: 0, stream: false },
|
||
...overrides,
|
||
};
|
||
}
|
||
|
||
// ===== Anthropic =====
|
||
|
||
describe('AnthropicAdapter — 请求体契约', () => {
|
||
it('A1: system 拼为顶层字段；tool 结果映射为 user 角色 tool_result 块', async () => {
|
||
const adapter = new AnthropicAdapter({
|
||
provider: 'anthropic',
|
||
baseURL: 'http://a.test',
|
||
apiKey: 'k',
|
||
defaultModel: 'claude-sonnet-4-5',
|
||
});
|
||
const { bodies } = captureFetch();
|
||
await adapter.send(
|
||
makeRequest({
|
||
messages: [
|
||
{ role: 'user', content: 'read it', timestamp: Date.now() },
|
||
{
|
||
role: 'assistant',
|
||
content: null,
|
||
toolCalls: [
|
||
{
|
||
id: 'tc_1',
|
||
name: 'read_file',
|
||
args: { path: 'a.txt' },
|
||
iteration: 1,
|
||
timestamp: Date.now(),
|
||
},
|
||
],
|
||
timestamp: Date.now(),
|
||
},
|
||
{
|
||
role: 'tool',
|
||
content: null,
|
||
toolResult: {
|
||
toolCallId: 'tc_1',
|
||
toolName: 'read_file',
|
||
result: 'data',
|
||
success: true,
|
||
durationMs: 1,
|
||
timestamp: Date.now(),
|
||
},
|
||
timestamp: Date.now(),
|
||
},
|
||
// 孤立 tool_result（前面没有对应 tool_use）应被过滤
|
||
{
|
||
role: 'tool',
|
||
content: null,
|
||
toolResult: {
|
||
toolCallId: 'tc_orphan',
|
||
toolName: 'x',
|
||
result: '',
|
||
success: true,
|
||
durationMs: 1,
|
||
timestamp: Date.now(),
|
||
},
|
||
timestamp: Date.now(),
|
||
},
|
||
{ role: 'user', content: 'next?', timestamp: Date.now() },
|
||
],
|
||
}),
|
||
);
|
||
|
||
const body = bodies[0];
|
||
// v0.7.3 P1-1: system 转为块数组并打 cache_control 断言（稳定前缀 prompt cache）
|
||
const system = body.system as Array<{
|
||
type: string;
|
||
text: string;
|
||
cache_control: { type: string };
|
||
}>;
|
||
expect(Array.isArray(system)).toBe(true);
|
||
expect(system[0].text).toContain('You are Metona.');
|
||
expect(system[0].cache_control).toEqual({ type: 'ephemeral' });
|
||
expect(Array.isArray(body.messages)).toBe(true);
|
||
const msgs = body.messages as Array<{ role: string; content: Array<Record<string, unknown>> }>;
|
||
// tool_use 的 assistant 消息存在且携带 id/name
|
||
const assistantToolMsg = msgs.find((m) => m.role === 'assistant');
|
||
expect(assistantToolMsg?.content[0]).toMatchObject({
|
||
type: 'tool_use',
|
||
id: 'tc_1',
|
||
name: 'read_file',
|
||
});
|
||
// tool 结果以 user 角色 tool_result 形态出现且配对 id 正确；孤立者被丢弃
|
||
const toolResultBlocks = msgs.flatMap((m) => m.content.filter((c) => c.type === 'tool_result'));
|
||
expect(toolResultBlocks).toHaveLength(1);
|
||
expect(toolResultBlocks[0].tool_use_id).toBe('tc_1');
|
||
});
|
||
|
||
it('A2: max_tokens 原样透传（v0.8.1：模型钳制已废除，设置面板是唯一上限来源）', async () => {
|
||
const sonnet = new AnthropicAdapter({
|
||
provider: 'anthropic',
|
||
baseURL: 'http://a.test',
|
||
apiKey: 'k',
|
||
defaultModel: 'claude-sonnet-4-5',
|
||
});
|
||
const opus = new AnthropicAdapter({
|
||
provider: 'anthropic',
|
||
baseURL: 'http://a.test',
|
||
apiKey: 'k',
|
||
defaultModel: 'claude-opus-4-1',
|
||
});
|
||
const { bodies } = captureFetch();
|
||
await sonnet.send(makeRequest());
|
||
await opus.send(makeRequest());
|
||
// v0.8.1: 设置面板「最大输出上限」对一切模型原样透传，无任何按模型钳制
|
||
expect(bodies[0].max_tokens).toBe(63_488);
|
||
expect(bodies[1].max_tokens).toBe(63_488);
|
||
});
|
||
|
||
it('A3: 小 maxTokens 时 thinking budget 不跌破协议下限 1024（v0.6.4 边界加固）', async () => {
|
||
const adapter = new AnthropicAdapter({
|
||
provider: 'anthropic',
|
||
baseURL: 'http://a.test',
|
||
apiKey: 'k',
|
||
defaultModel: 'claude-haiku-4-5',
|
||
});
|
||
const { bodies } = captureFetch();
|
||
await adapter.send(
|
||
makeRequest({
|
||
params: {
|
||
maxTokens: 1500,
|
||
temperature: 0,
|
||
stream: false,
|
||
thinkingEnabled: true,
|
||
thinkingEffort: 'low',
|
||
},
|
||
}),
|
||
);
|
||
const body = bodies[0];
|
||
const thinking = body.thinking as { type: string; budget_tokens: number };
|
||
// max_tokens 被抬升到安全下限，budget 落在 [1024, max_tokens/2] 区间内
|
||
expect(body.max_tokens as number).toBeGreaterThanOrEqual(2048);
|
||
expect(thinking.budget_tokens).toBeGreaterThanOrEqual(1024);
|
||
expect(thinking.budget_tokens).toBeLessThanOrEqual((body.max_tokens as number) / 2);
|
||
});
|
||
|
||
it('A4: thinking 开启不传 temperature；关闭时显式传递', async () => {
|
||
const adapter = new AnthropicAdapter({
|
||
provider: 'anthropic',
|
||
baseURL: 'http://a.test',
|
||
apiKey: 'k',
|
||
defaultModel: 'claude-sonnet-4-5',
|
||
});
|
||
const { bodies } = captureFetch();
|
||
await adapter.send(
|
||
makeRequest({
|
||
params: { maxTokens: 4096, temperature: 0.7, stream: false, thinkingEnabled: true },
|
||
}),
|
||
);
|
||
expect(bodies[0].temperature).toBeUndefined();
|
||
expect(bodies[0].thinking).toBeDefined();
|
||
|
||
await adapter.send(
|
||
makeRequest({
|
||
params: { maxTokens: 4096, temperature: 0.7, stream: false, thinkingEnabled: false },
|
||
}),
|
||
);
|
||
expect(bodies[1].temperature).toBe(0.7);
|
||
expect(bodies[1].thinking).toBeUndefined();
|
||
});
|
||
});
|
||
|
||
// ===== Ollama =====
|
||
|
||
describe('OllamaAdapter — 请求体契约', () => {
|
||
function makeOllama(): OllamaAdapter {
|
||
return new OllamaAdapter({
|
||
provider: 'ollama',
|
||
baseURL: 'http://localhost:11434',
|
||
defaultModel: 'qwen3',
|
||
});
|
||
}
|
||
|
||
it('O1: options 映射 num_predict/num_ctx/stop/top_p/temperature', async () => {
|
||
const adapter = makeOllama();
|
||
const { bodies } = captureFetch();
|
||
await adapter.send(
|
||
makeRequest({
|
||
params: {
|
||
maxTokens: 8192,
|
||
temperature: 0.3,
|
||
topP: 0.9,
|
||
stream: false,
|
||
contextLength: 16384,
|
||
stopSequences: ['STOP'],
|
||
},
|
||
}),
|
||
);
|
||
const options = bodies[0].options as Record<string, unknown>;
|
||
expect(options.num_predict).toBe(8192);
|
||
expect(options.num_ctx).toBe(16384);
|
||
expect(options.temperature).toBe(0.3);
|
||
expect(options.top_p).toBe(0.9);
|
||
expect(options.stop).toEqual(['STOP']);
|
||
});
|
||
|
||
it('O2: think 参数 effort 映射（low→"low"、xhigh→"xhigh"、max/true→true）；未开启思考时缺省', async () => {
|
||
const adapter = makeOllama();
|
||
const { bodies } = captureFetch();
|
||
|
||
await adapter.send(
|
||
makeRequest({
|
||
params: {
|
||
maxTokens: 4096,
|
||
temperature: 0,
|
||
stream: false,
|
||
thinkingEnabled: true,
|
||
thinkingEffort: 'low',
|
||
},
|
||
}),
|
||
);
|
||
expect(bodies[0].think).toBe('low');
|
||
|
||
await adapter.send(
|
||
makeRequest({
|
||
params: {
|
||
maxTokens: 4096,
|
||
temperature: 0,
|
||
stream: false,
|
||
thinkingEnabled: true,
|
||
thinkingEffort: 'max',
|
||
},
|
||
}),
|
||
);
|
||
expect(bodies[1].think).toBe(true);
|
||
|
||
// v0.8.3: xhigh 原样透传（Qwen3 等思考模板原生档）；true → 布尔 true（模型默认档）
|
||
await adapter.send(
|
||
makeRequest({
|
||
params: {
|
||
maxTokens: 4096,
|
||
temperature: 0,
|
||
stream: false,
|
||
thinkingEnabled: true,
|
||
thinkingEffort: 'xhigh',
|
||
},
|
||
}),
|
||
);
|
||
expect(bodies[2].think).toBe('xhigh');
|
||
|
||
await adapter.send(
|
||
makeRequest({
|
||
params: {
|
||
maxTokens: 4096,
|
||
temperature: 0,
|
||
stream: false,
|
||
thinkingEnabled: true,
|
||
thinkingEffort: 'true',
|
||
},
|
||
}),
|
||
);
|
||
expect(bodies[3].think).toBe(true);
|
||
|
||
await adapter.send(
|
||
makeRequest({
|
||
params: { maxTokens: 4096, temperature: 0, stream: false, thinkingEnabled: false },
|
||
}),
|
||
);
|
||
expect(bodies[4].think).toBeUndefined();
|
||
});
|
||
|
||
it('O3: data URI 图片剥前缀转纯 base64 数组（无网络下载路径触发）', async () => {
|
||
const adapter = makeOllama();
|
||
const { bodies } = captureFetch();
|
||
await adapter.send(
|
||
makeRequest({
|
||
messages: [
|
||
{
|
||
role: 'user',
|
||
content: '看图',
|
||
images: [{ url: 'data:image/png;base64,iVBORw0KGgoAAAANSU', detail: 'auto' }],
|
||
timestamp: Date.now(),
|
||
},
|
||
],
|
||
}),
|
||
);
|
||
const messages = bodies[0].messages as Array<Record<string, unknown>>;
|
||
const userMsg = messages[messages.length - 1];
|
||
expect(userMsg.images).toEqual(['iVBORw0KGgoAAAANSU']);
|
||
});
|
||
});
|
||
|
||
// ===== MiMo providerOptions（v0.6.4 P4-3） =====
|
||
|
||
describe('MimoAdapter — 服务端能力扩展（providerOptions）', () => {
|
||
it('enableWebSearch 开启时附加 {type:web_search} 服务端工具', async () => {
|
||
const adapter = new MimoAdapter({
|
||
provider: 'mimo',
|
||
baseURL: 'http://m.test/v1',
|
||
apiKey: 'k',
|
||
defaultModel: 'mimo-v2.5',
|
||
providerOptions: { enableWebSearch: true },
|
||
});
|
||
const { bodies } = captureFetch();
|
||
await adapter.send(makeRequest());
|
||
const tools = bodies[0].tools as Array<Record<string, unknown>>;
|
||
expect(tools.some((tc) => (tc as { type?: string }).type === 'web_search')).toBe(true);
|
||
expect(bodies[0].tool_choice).toBe('auto');
|
||
});
|
||
|
||
it('responseFormatJson 开启时写入 response_format json_object；默认不写', async () => {
|
||
const on = new MimoAdapter({
|
||
provider: 'mimo',
|
||
baseURL: 'http://m.test/v1',
|
||
apiKey: 'k',
|
||
defaultModel: 'mimo-v2.5',
|
||
providerOptions: { responseFormatJson: true },
|
||
});
|
||
const off = new MimoAdapter({
|
||
provider: 'mimo',
|
||
baseURL: 'http://m.test/v1',
|
||
apiKey: 'k',
|
||
defaultModel: 'mimo-v2.5',
|
||
});
|
||
const { bodies } = captureFetch();
|
||
await on.send(makeRequest());
|
||
await off.send(makeRequest());
|
||
expect(bodies[0].response_format).toEqual({ type: 'json_object' });
|
||
expect(bodies[1].response_format).toBeUndefined();
|
||
});
|
||
});
|
||
|
||
// ===== Agnes =====
|
||
|
||
describe('AgnesAdapter — 思考模式对称性（v0.6.4）', () => {
|
||
it('G1: thinkingEnabled=false 显式发送 enable_thinking:false（此前无法关闭服务端默认思考）', async () => {
|
||
const adapter = new AgnesAdapter({
|
||
provider: 'agnes',
|
||
baseURL: 'http://g.test/v1',
|
||
apiKey: 'k',
|
||
defaultModel: 'agnes-2.0-flash',
|
||
});
|
||
const { bodies } = captureFetch();
|
||
|
||
await adapter.send(
|
||
makeRequest({
|
||
params: {
|
||
maxTokens: 4096,
|
||
temperature: 0,
|
||
stream: false,
|
||
thinkingEnabled: true,
|
||
thinkingEffort: 'high',
|
||
},
|
||
}),
|
||
);
|
||
expect(
|
||
((bodies[0].chat_template_kwargs as Record<string, unknown>) ?? {}).enable_thinking,
|
||
).toBe(true);
|
||
|
||
await adapter.send(
|
||
makeRequest({
|
||
params: { maxTokens: 4096, temperature: 0, stream: false, thinkingEnabled: false },
|
||
}),
|
||
);
|
||
expect(
|
||
((bodies[1].chat_template_kwargs as Record<string, unknown>) ?? {}).enable_thinking,
|
||
).toBe(false);
|
||
|
||
// 未配置 thinkingEnabled 同样视为关闭（显式 disabled 保持与服务端默认的确定性）
|
||
await adapter.send(makeRequest({ params: { maxTokens: 4096, temperature: 0, stream: false } }));
|
||
expect(
|
||
((bodies[2].chat_template_kwargs as Record<string, unknown>) ?? {}).enable_thinking,
|
||
).toBe(false);
|
||
});
|
||
});
|
||
|
||
// ===== Anthropic 追加：system 四态 / thinking budget 矩阵 / maxTokens 钳制 =====
|
||
|
||
describe('AnthropicAdapter — system 块数组与 cache_control 四态', () => {
|
||
function makeAdapter(model = 'claude-sonnet-4-5'): AnthropicAdapter {
|
||
return new AnthropicAdapter({
|
||
provider: 'anthropic',
|
||
baseURL: 'http://a.test',
|
||
apiKey: 'k',
|
||
defaultModel: model,
|
||
});
|
||
}
|
||
|
||
it('system 四段全部填充 → 单一 text 块 + cache_control ephemeral（稳定前缀提示缓存）', async () => {
|
||
const adapter = makeAdapter();
|
||
const { bodies } = captureFetch();
|
||
await adapter.send(makeRequest());
|
||
const system = bodies[0].system as Array<Record<string, unknown>>;
|
||
expect(system).toHaveLength(1);
|
||
expect(system[0].type).toBe('text');
|
||
expect(system[0].text).toContain('You are Metona.');
|
||
expect(system[0].text).toContain('Be concise.');
|
||
expect(system[0].text).toContain('Stay safe.');
|
||
expect(system[0].cache_control).toEqual({ type: 'ephemeral' });
|
||
});
|
||
|
||
it('system 部分段为空 → 过滤后拼接，仍打 cache_control', async () => {
|
||
const adapter = makeAdapter();
|
||
const { bodies } = captureFetch();
|
||
await adapter.send(
|
||
makeRequest({
|
||
systemPrompt: {
|
||
roleDefinition: 'Only role',
|
||
outputConstraints: '',
|
||
safetyGuidelines: '',
|
||
dynamicReminders: '',
|
||
},
|
||
}),
|
||
);
|
||
const system = bodies[0].system as Array<Record<string, unknown>>;
|
||
expect(system).toHaveLength(1);
|
||
expect(system[0].text).toBe('Only role');
|
||
expect(system[0].cache_control).toEqual({ type: 'ephemeral' });
|
||
});
|
||
|
||
it('system 全部为空 → 不发块数组，透传空字符串（无 cache_control）', async () => {
|
||
const adapter = makeAdapter();
|
||
const { bodies } = captureFetch();
|
||
await adapter.send(
|
||
makeRequest({
|
||
systemPrompt: { roleDefinition: '', outputConstraints: '', safetyGuidelines: '' },
|
||
}),
|
||
);
|
||
expect(bodies[0].system).toBe('');
|
||
});
|
||
|
||
it('动态提醒 dynamicReminders 被拼入 system 块', async () => {
|
||
const adapter = makeAdapter();
|
||
const { bodies } = captureFetch();
|
||
await adapter.send(
|
||
makeRequest({
|
||
systemPrompt: {
|
||
roleDefinition: 'rd',
|
||
outputConstraints: '',
|
||
safetyGuidelines: '',
|
||
dynamicReminders: 'Remember X',
|
||
},
|
||
}),
|
||
);
|
||
const system = bodies[0].system as Array<Record<string, unknown>>;
|
||
expect(system[0].text).toContain('Remember X');
|
||
});
|
||
});
|
||
|
||
describe('AnthropicAdapter — thinking budget 按 effort 映射矩阵', () => {
|
||
function makeAdapter(model = 'claude-sonnet-4-5'): AnthropicAdapter {
|
||
return new AnthropicAdapter({
|
||
provider: 'anthropic',
|
||
baseURL: 'http://a.test',
|
||
apiKey: 'k',
|
||
defaultModel: model,
|
||
});
|
||
}
|
||
|
||
it.each([
|
||
['low', 1024],
|
||
['medium', 4096],
|
||
['high', 16384],
|
||
['xhigh', 24576], // v0.8.3: 介于 high 与 max 之间
|
||
// v0.8.1: max_tokens 不再按模型钳制（100_000 原样透传）→ budget = min(32768, floor(100000/2)) = 32768
|
||
['max', 32768],
|
||
['true', 16384], // v0.8.3: 模型默认档按 high 同档预算
|
||
] as const)('effort=%s → budget 为该档值且 < max_tokens', async (effort, expectBudget) => {
|
||
const adapter = makeAdapter();
|
||
const { bodies } = captureFetch();
|
||
await adapter.send(
|
||
makeRequest({
|
||
params: {
|
||
maxTokens: 100_000,
|
||
temperature: 0,
|
||
stream: false,
|
||
thinkingEnabled: true,
|
||
thinkingEffort: effort,
|
||
},
|
||
}),
|
||
);
|
||
const thinking = bodies[0].thinking as { type: string; budget_tokens: number };
|
||
expect(thinking.type).toBe('enabled');
|
||
expect(thinking.budget_tokens).toBe(expectBudget);
|
||
expect(thinking.budget_tokens).toBeLessThan(bodies[0].max_tokens as number);
|
||
});
|
||
|
||
it('effort 未配置时缺省 high → budget 16384', async () => {
|
||
const adapter = makeAdapter();
|
||
const { bodies } = captureFetch();
|
||
await adapter.send(
|
||
makeRequest({
|
||
params: { maxTokens: 100_000, temperature: 0, stream: false, thinkingEnabled: true },
|
||
}),
|
||
);
|
||
const thinking = bodies[0].thinking as { budget_tokens: number };
|
||
expect(thinking.budget_tokens).toBe(16384);
|
||
});
|
||
|
||
it('小 max_tokens 时 budget 被 max_tokens/2 二次钳制（budget < max_tokens 协议约束）', async () => {
|
||
const adapter = makeAdapter();
|
||
const { bodies } = captureFetch();
|
||
await adapter.send(
|
||
makeRequest({
|
||
params: {
|
||
maxTokens: 2048,
|
||
temperature: 0,
|
||
stream: false,
|
||
thinkingEnabled: true,
|
||
thinkingEffort: 'high',
|
||
},
|
||
}),
|
||
);
|
||
const thinking = bodies[0].thinking as { budget_tokens: number };
|
||
// effort high=16384 但 max_tokens=2048 → budget 钳到 floor(2048/2)=1024
|
||
expect(thinking.budget_tokens).toBe(1024);
|
||
});
|
||
});
|
||
|
||
describe('AnthropicAdapter — max_tokens 透传矩阵（v0.8.1 无钳制）', () => {
|
||
it.each([
|
||
['claude-sonnet-4-5', 63_488, 63_488],
|
||
['claude-sonnet-4-5', 70_000, 70_000], // 超过任何旧元信息上限 → 原样透传
|
||
['claude-opus-4-1', 63_488, 63_488],
|
||
['claude-haiku-4-5', 63_488, 63_488],
|
||
['claude-sonnet-4-5', 500, 500],
|
||
])('%s maxTokens=%d → max_tokens=%d', async (model, requested, expected) => {
|
||
const adapter = new AnthropicAdapter({
|
||
provider: 'anthropic',
|
||
baseURL: 'http://a.test',
|
||
apiKey: 'k',
|
||
defaultModel: model,
|
||
});
|
||
const { bodies } = captureFetch();
|
||
await adapter.send(
|
||
makeRequest({ params: { maxTokens: requested, temperature: 0, stream: false } }),
|
||
);
|
||
expect(bodies[0].max_tokens).toBe(expected);
|
||
});
|
||
|
||
it('thinking 开启时小 maxTokens 被抬升到安全下限 2048', async () => {
|
||
const adapter = new AnthropicAdapter({
|
||
provider: 'anthropic',
|
||
baseURL: 'http://a.test',
|
||
apiKey: 'k',
|
||
defaultModel: 'claude-sonnet-4-5',
|
||
});
|
||
const { bodies } = captureFetch();
|
||
await adapter.send(
|
||
makeRequest({
|
||
params: {
|
||
maxTokens: 800,
|
||
temperature: 0,
|
||
stream: false,
|
||
thinkingEnabled: true,
|
||
thinkingEffort: 'low',
|
||
},
|
||
}),
|
||
);
|
||
expect(bodies[0].max_tokens).toBe(2048);
|
||
});
|
||
});
|
||
|
||
describe('AnthropicAdapter — temperature 传递与停止序列', () => {
|
||
it('thinking 关闭时 temperature 逐值透传', async () => {
|
||
const adapter = new AnthropicAdapter({
|
||
provider: 'anthropic',
|
||
baseURL: 'http://a.test',
|
||
apiKey: 'k',
|
||
defaultModel: 'claude-sonnet-4-5',
|
||
});
|
||
const { bodies } = captureFetch();
|
||
for (const t of [0, 0.2, 1.0]) {
|
||
await adapter.send(
|
||
makeRequest({ params: { maxTokens: 4096, temperature: t, stream: false } }),
|
||
);
|
||
}
|
||
expect(bodies[0].temperature).toBe(0);
|
||
expect(bodies[1].temperature).toBe(0.2);
|
||
expect(bodies[2].temperature).toBe(1.0);
|
||
});
|
||
|
||
it('stopSequences 映射为 stop_sequences 数组', async () => {
|
||
const adapter = new AnthropicAdapter({
|
||
provider: 'anthropic',
|
||
baseURL: 'http://a.test',
|
||
apiKey: 'k',
|
||
defaultModel: 'claude-sonnet-4-5',
|
||
});
|
||
const { bodies } = captureFetch();
|
||
await adapter.send(
|
||
makeRequest({
|
||
params: { maxTokens: 4096, temperature: 0, stream: false, stopSequences: ['END', 'STOP'] },
|
||
}),
|
||
);
|
||
expect(bodies[0].stop_sequences).toEqual(['END', 'STOP']);
|
||
});
|
||
|
||
it('未配置 stopSequences 时不发送 stop_sequences', async () => {
|
||
const adapter = new AnthropicAdapter({
|
||
provider: 'anthropic',
|
||
baseURL: 'http://a.test',
|
||
apiKey: 'k',
|
||
defaultModel: 'claude-sonnet-4-5',
|
||
});
|
||
const { bodies } = captureFetch();
|
||
await adapter.send(makeRequest());
|
||
expect(bodies[0].stop_sequences).toBeUndefined();
|
||
});
|
||
|
||
it('tools 定义映射为 input_schema 命名空间', async () => {
|
||
const adapter = new AnthropicAdapter({
|
||
provider: 'anthropic',
|
||
baseURL: 'http://a.test',
|
||
apiKey: 'k',
|
||
defaultModel: 'claude-sonnet-4-5',
|
||
});
|
||
const { bodies } = captureFetch();
|
||
await adapter.send(
|
||
makeRequest({
|
||
tools: [
|
||
{
|
||
name: 'read_file',
|
||
description: 'Read a file',
|
||
parameters: {
|
||
type: 'object',
|
||
properties: { path: { type: 'string', description: 'file path' } },
|
||
required: ['path'],
|
||
},
|
||
category: 'filesystem' as never,
|
||
riskLevel: 'low' as never,
|
||
requiresPermission: false,
|
||
timeoutMs: 1000,
|
||
},
|
||
],
|
||
}),
|
||
);
|
||
const tools = bodies[0].tools as Array<Record<string, unknown>>;
|
||
expect(tools[0].name).toBe('read_file');
|
||
expect(tools[0].input_schema).toBeDefined();
|
||
expect((tools[0].input_schema as Record<string, unknown>).required).toEqual(['path']);
|
||
});
|
||
|
||
it('无 tools 时不发送 tools 字段', async () => {
|
||
const adapter = new AnthropicAdapter({
|
||
provider: 'anthropic',
|
||
baseURL: 'http://a.test',
|
||
apiKey: 'k',
|
||
defaultModel: 'claude-sonnet-4-5',
|
||
});
|
||
const { bodies } = captureFetch();
|
||
await adapter.send(makeRequest());
|
||
expect(bodies[0].tools).toBeUndefined();
|
||
});
|
||
});
|
||
|
||
// ===== DeepSeek thinking 映射矩阵（v0.6.4） =====
|
||
|
||
describe('DeepSeekAdapter — thinking 映射矩阵', () => {
|
||
function makeAdapter(model = 'deepseek-v4-pro'): DeepSeekAdapter {
|
||
return new DeepSeekAdapter({
|
||
provider: 'deepseek',
|
||
baseURL: 'https://api.deepseek.com',
|
||
apiKey: 'k',
|
||
defaultModel: model,
|
||
});
|
||
}
|
||
|
||
it.each([
|
||
['low', 'high'],
|
||
['medium', 'high'],
|
||
['high', 'high'],
|
||
['xhigh', 'high'], // v0.8.3: DeepSeek 无 xhigh 档，就近映射 high
|
||
['max', 'max'],
|
||
['true', 'high'], // v0.8.3: 模型默认档映射 high
|
||
] as const)(
|
||
'effort=%s → reasoning_effort=%s（low/medium/xhigh/true 归一 high）',
|
||
async (effort, expected) => {
|
||
const adapter = makeAdapter();
|
||
const { bodies } = captureFetch();
|
||
await adapter.send(
|
||
makeRequest({
|
||
params: {
|
||
maxTokens: 4096,
|
||
temperature: 0,
|
||
stream: false,
|
||
thinkingEnabled: true,
|
||
thinkingEffort: effort,
|
||
},
|
||
}),
|
||
);
|
||
expect(bodies[0].thinking).toEqual({ type: 'enabled' });
|
||
expect(bodies[0].reasoning_effort).toBe(expected);
|
||
},
|
||
);
|
||
|
||
it('thinkingEnabled=false → 显式 {type:disabled}（服务端默认开启，必须显式关闭）', async () => {
|
||
const adapter = makeAdapter();
|
||
const { bodies } = captureFetch();
|
||
await adapter.send(
|
||
makeRequest({
|
||
params: { maxTokens: 4096, temperature: 0, stream: false, thinkingEnabled: false },
|
||
}),
|
||
);
|
||
expect(bodies[0].thinking).toEqual({ type: 'disabled' });
|
||
expect(bodies[0].reasoning_effort).toBeUndefined();
|
||
});
|
||
|
||
it('thinkingEnabled 未配置 → 显式 disabled（v0.8.0 P0-3 确定性契约）', async () => {
|
||
// v0.8.0 P0-3 契约变更: DeepSeek 服务端默认 thinking.enabled，未配置即发请求
|
||
// 会得到隐式思考 —— 现在未配置一律显式 disabled，行为不依赖服务端隐式默认
|
||
//（与 Agnes/MiMo 的显式口径对齐；引擎路径恒传布尔值，此处为兜底确定性）。
|
||
const adapter = makeAdapter();
|
||
const { bodies } = captureFetch();
|
||
await adapter.send(makeRequest({ params: { maxTokens: 4096, temperature: 0, stream: false } }));
|
||
expect(bodies[0].thinking).toEqual({ type: 'disabled' });
|
||
});
|
||
|
||
it('effort 未配置缺省 high → reasoning_effort=high', async () => {
|
||
const adapter = makeAdapter();
|
||
const { bodies } = captureFetch();
|
||
await adapter.send(
|
||
makeRequest({
|
||
params: { maxTokens: 4096, temperature: 0, stream: false, thinkingEnabled: true },
|
||
}),
|
||
);
|
||
expect(bodies[0].reasoning_effort).toBe('high');
|
||
});
|
||
|
||
it('temperature 与 stop 序列原样传递', async () => {
|
||
const adapter = makeAdapter();
|
||
const { bodies } = captureFetch();
|
||
await adapter.send(
|
||
makeRequest({
|
||
params: {
|
||
maxTokens: 4096,
|
||
temperature: 0.5,
|
||
stream: false,
|
||
stopSequences: ['<END>'],
|
||
},
|
||
}),
|
||
);
|
||
expect(bodies[0].temperature).toBe(0.5);
|
||
expect(bodies[0].stop).toEqual(['<END>']);
|
||
});
|
||
|
||
it('max_tokens 原样透传（v0.8.1：pro/vision 均不钳制）', async () => {
|
||
const pro = makeAdapter('deepseek-v4-pro');
|
||
const vision = makeAdapter('deepseek-v4-flash-vision-exp');
|
||
const { bodies } = captureFetch();
|
||
await pro.send(makeRequest({ params: { maxTokens: 500_000, temperature: 0, stream: false } }));
|
||
await vision.send(
|
||
makeRequest({ params: { maxTokens: 63_488, temperature: 0, stream: false } }),
|
||
);
|
||
expect(bodies[0].max_tokens).toBe(500_000);
|
||
expect(bodies[1].max_tokens).toBe(63_488);
|
||
});
|
||
});
|
||
|
||
// ===== Agnes enable_thinking 对称性扩展 =====
|
||
|
||
describe('AgnesAdapter — enable_thinking 对称性矩阵', () => {
|
||
function makeAdapter(): AgnesAdapter {
|
||
return new AgnesAdapter({
|
||
provider: 'agnes',
|
||
baseURL: 'http://g.test/v1',
|
||
apiKey: 'k',
|
||
defaultModel: 'agnes-2.0-flash',
|
||
});
|
||
}
|
||
|
||
it.each([
|
||
['high', true],
|
||
['medium', true],
|
||
['max', true],
|
||
['low', false],
|
||
] as const)('effort=%s → enable_thinking=%s（low 映射为关闭）', async (effort, expected) => {
|
||
const adapter = makeAdapter();
|
||
const { bodies } = captureFetch();
|
||
await adapter.send(
|
||
makeRequest({
|
||
params: {
|
||
maxTokens: 4096,
|
||
temperature: 0,
|
||
stream: false,
|
||
thinkingEnabled: true,
|
||
thinkingEffort: effort,
|
||
},
|
||
}),
|
||
);
|
||
expect(
|
||
((bodies[0].chat_template_kwargs as Record<string, unknown>) ?? {}).enable_thinking,
|
||
).toBe(expected);
|
||
});
|
||
|
||
it('thinkingEnabled=true 但 effort 未配置 → 缺省 high → enable_thinking true', async () => {
|
||
const adapter = makeAdapter();
|
||
const { bodies } = captureFetch();
|
||
await adapter.send(
|
||
makeRequest({
|
||
params: { maxTokens: 4096, temperature: 0, stream: false, thinkingEnabled: true },
|
||
}),
|
||
);
|
||
expect(
|
||
((bodies[0].chat_template_kwargs as Record<string, unknown>) ?? {}).enable_thinking,
|
||
).toBe(true);
|
||
});
|
||
|
||
it('temperature 与 max_tokens 同时传递（Agnes 支持 temperature）', async () => {
|
||
const adapter = makeAdapter();
|
||
const { bodies } = captureFetch();
|
||
await adapter.send(
|
||
makeRequest({ params: { maxTokens: 70_000, temperature: 0.9, stream: false } }),
|
||
);
|
||
expect(bodies[0].temperature).toBe(0.9);
|
||
// v0.8.1: 原样透传，无 65536 钳制
|
||
expect(bodies[0].max_tokens).toBe(70_000);
|
||
});
|
||
});
|
||
|
||
// ===== MiMo thinking 开关与 providerOptions 扩展 =====
|
||
|
||
describe('MimoAdapter — thinking 显式开关', () => {
|
||
function makeAdapter(overrides: Record<string, unknown> = {}): MimoAdapter {
|
||
return new MimoAdapter({
|
||
provider: 'mimo',
|
||
baseURL: 'http://m.test/v1',
|
||
apiKey: 'k',
|
||
defaultModel: 'mimo-v2.5',
|
||
...overrides,
|
||
});
|
||
}
|
||
|
||
it('thinkingEnabled=false → {type:disabled} + temperature/top_p 显式传递（非思考模式有效）', async () => {
|
||
const adapter = makeAdapter();
|
||
const { bodies } = captureFetch();
|
||
await adapter.send(
|
||
makeRequest({
|
||
params: {
|
||
maxTokens: 4096,
|
||
temperature: 0.7,
|
||
topP: 0.8,
|
||
stream: false,
|
||
thinkingEnabled: false,
|
||
},
|
||
}),
|
||
);
|
||
expect(bodies[0].thinking).toEqual({ type: 'disabled' });
|
||
expect(bodies[0].temperature).toBe(0.7);
|
||
expect(bodies[0].top_p).toBe(0.8);
|
||
});
|
||
|
||
// v0.8.2 P2-6 修订：未配置时显式 disabled（与 DeepSeek/Agnes 的"用户意图优先"对齐；
|
||
// 旧行为隐式 enabled 会静默吞掉用户 temperature/top_p —— 服务端思考模式强制覆盖）
|
||
it('thinkingEnabled 未配置 → 显式 {type:disabled} 且 temperature/top_p 透传', async () => {
|
||
const adapter = makeAdapter();
|
||
const { bodies } = captureFetch();
|
||
await adapter.send(
|
||
makeRequest({ params: { maxTokens: 4096, temperature: 0.7, topP: 0.8, stream: false } }),
|
||
);
|
||
expect(bodies[0].thinking).toEqual({ type: 'disabled' });
|
||
expect(bodies[0].temperature).toBe(0.7);
|
||
expect(bodies[0].top_p).toBe(0.8);
|
||
});
|
||
|
||
it('max_completion_tokens 原样透传（v0.8.1：pro/standard 均不钳制）', async () => {
|
||
const pro = makeAdapter({ defaultModel: 'mimo-v2.5-pro' });
|
||
const std = makeAdapter({ defaultModel: 'mimo-v2.5' });
|
||
const { bodies } = captureFetch();
|
||
await pro.send(makeRequest({ params: { maxTokens: 200_000, temperature: 0, stream: false } }));
|
||
await std.send(makeRequest({ params: { maxTokens: 63_488, temperature: 0, stream: false } }));
|
||
expect(bodies[0].max_completion_tokens).toBe(200_000);
|
||
expect(bodies[1].max_completion_tokens).toBe(63_488);
|
||
});
|
||
|
||
it('thinking 未关闭时未配置 maxTokens → 不下发该字段（v0.8.1：无写死兜底值）', async () => {
|
||
const pro = makeAdapter({ defaultModel: 'mimo-v2.5-pro' });
|
||
const { bodies } = captureFetch();
|
||
await pro.send(makeRequest({ params: { temperature: 0, stream: false } }));
|
||
expect(bodies[0].max_completion_tokens).toBeUndefined();
|
||
});
|
||
|
||
it('enableWebSearch 且存在客户端 tools → web_search 服务端工具追加（不覆盖客户端工具）', async () => {
|
||
const adapter = makeAdapter({ providerOptions: { enableWebSearch: true } });
|
||
const { bodies } = captureFetch();
|
||
await adapter.send(
|
||
makeRequest({
|
||
tools: [
|
||
{
|
||
name: 'fs',
|
||
description: 'd',
|
||
parameters: { type: 'object', properties: {} },
|
||
category: 'filesystem' as never,
|
||
riskLevel: 'low' as never,
|
||
requiresPermission: false,
|
||
timeoutMs: 100,
|
||
},
|
||
],
|
||
}),
|
||
);
|
||
const tools = bodies[0].tools as Array<Record<string, unknown>>;
|
||
expect(tools).toHaveLength(2);
|
||
expect(tools[0].type).toBe('function');
|
||
expect(tools[1]).toEqual({ type: 'web_search' });
|
||
expect(bodies[0].tool_choice).toBe('auto');
|
||
});
|
||
|
||
it('responseFormatJson + thinking 默认开启可共存（response_format 独立于 thinking）', async () => {
|
||
const adapter = makeAdapter({ providerOptions: { responseFormatJson: true } });
|
||
const { bodies } = captureFetch();
|
||
await adapter.send(
|
||
makeRequest({
|
||
params: { maxTokens: 4096, temperature: 0, stream: false, thinkingEnabled: true },
|
||
}),
|
||
);
|
||
expect(bodies[0].response_format).toEqual({ type: 'json_object' });
|
||
expect(bodies[0].thinking).toEqual({ type: 'enabled' });
|
||
});
|
||
});
|
||
|
||
// ===== OpenAI reasoning_effort 与 maxTokens 路由 =====
|
||
|
||
describe('OpenAIAdapter — 推理模型字段路由（v0.6.4 P3-1）', () => {
|
||
function makeAdapter(model: string): OpenAIAdapter {
|
||
return new OpenAIAdapter({
|
||
provider: 'openai',
|
||
baseURL: 'https://api.openai.com/v1',
|
||
apiKey: 'k',
|
||
defaultModel: model,
|
||
});
|
||
}
|
||
|
||
it.each([
|
||
['low', 'low'],
|
||
['medium', 'medium'],
|
||
['high', 'high'],
|
||
['max', 'high'],
|
||
] as const)(
|
||
'o3-mini effort=%s → reasoning_effort=%s（max 归一 high）',
|
||
async (effort, expected) => {
|
||
const adapter = makeAdapter('o3-mini');
|
||
const { bodies } = captureFetch();
|
||
await adapter.send(
|
||
makeRequest({
|
||
params: {
|
||
maxTokens: 4096,
|
||
temperature: 0,
|
||
stream: false,
|
||
thinkingEnabled: true,
|
||
thinkingEffort: effort,
|
||
},
|
||
}),
|
||
);
|
||
expect(bodies[0].reasoning_effort).toBe(expected);
|
||
expect(bodies[0].max_completion_tokens).toBe(4096); // o 系列用新字段名
|
||
},
|
||
);
|
||
|
||
it('o3-mini thinking 未开启 → 不传 reasoning_effort 也不传 temperature（o 系列不支持温度）', async () => {
|
||
const adapter = makeAdapter('o3-mini');
|
||
const { bodies } = captureFetch();
|
||
await adapter.send(
|
||
makeRequest({
|
||
params: { maxTokens: 4096, temperature: 0.7, stream: false, thinkingEnabled: false },
|
||
}),
|
||
);
|
||
expect(bodies[0].reasoning_effort).toBeUndefined();
|
||
expect(bodies[0].temperature).toBeUndefined();
|
||
});
|
||
|
||
it('非推理模型 gpt-4o → max_tokens 字段 + temperature 透传', async () => {
|
||
const adapter = makeAdapter('gpt-4o');
|
||
const { bodies } = captureFetch();
|
||
await adapter.send(
|
||
makeRequest({ params: { maxTokens: 4096, temperature: 0.5, stream: false } }),
|
||
);
|
||
expect(bodies[0].max_tokens).toBe(4096);
|
||
expect(bodies[0].max_completion_tokens).toBeUndefined();
|
||
expect(bodies[0].temperature).toBe(0.5);
|
||
expect(bodies[0].reasoning_effort).toBeUndefined();
|
||
});
|
||
|
||
it('gpt-4.1 → getContextWindow 返回设置面板配置值（v0.8.1：元信息不再承载窗口）', () => {
|
||
const adapter = new OpenAIAdapter({
|
||
provider: 'openai',
|
||
baseURL: 'http://o.test',
|
||
apiKey: 'k',
|
||
defaultModel: 'gpt-4.1',
|
||
contextWindow: 1_000_000,
|
||
});
|
||
expect(adapter.getContextWindow()).toBe(1_000_000);
|
||
// 未配置 → 0（引擎跳过压缩判定），无任何写死兜底
|
||
const noCfg = makeAdapter('gpt-4.1');
|
||
expect(noCfg.getContextWindow()).toBe(0);
|
||
});
|
||
|
||
it('o3-mini max_completion_tokens 原样透传（v0.8.1：无 100000 钳制）', async () => {
|
||
const adapter = makeAdapter('o3-mini');
|
||
const { bodies } = captureFetch();
|
||
await adapter.send(
|
||
makeRequest({ params: { maxTokens: 200_000, temperature: 0, stream: false } }),
|
||
);
|
||
expect(bodies[0].max_completion_tokens).toBe(200_000);
|
||
});
|
||
|
||
it('gpt-4o max_tokens 原样透传（v0.8.1：无 16384 钳制）', async () => {
|
||
const adapter = makeAdapter('gpt-4o');
|
||
const { bodies } = captureFetch();
|
||
await adapter.send(
|
||
makeRequest({ params: { maxTokens: 63_488, temperature: 0, stream: false } }),
|
||
);
|
||
expect(bodies[0].max_tokens).toBe(63_488);
|
||
});
|
||
});
|
||
|
||
// ===== Ollama options 缺省与工具映射 =====
|
||
|
||
describe('OllamaAdapter — options 缺省与工具定义', () => {
|
||
function makeOllama(): OllamaAdapter {
|
||
return new OllamaAdapter({
|
||
provider: 'ollama',
|
||
baseURL: 'http://localhost:11434',
|
||
defaultModel: 'qwen3',
|
||
});
|
||
}
|
||
|
||
it('未配置 topP/contextLength/stop 时 options 仅含 temperature/num_predict', async () => {
|
||
const adapter = makeOllama();
|
||
const { bodies } = captureFetch();
|
||
await adapter.send(
|
||
makeRequest({ params: { maxTokens: 4096, temperature: 0.2, stream: false } }),
|
||
);
|
||
const options = bodies[0].options as Record<string, unknown>;
|
||
expect(Object.keys(options).sort()).toEqual(['num_predict', 'temperature']);
|
||
expect(options.num_predict).toBe(4096);
|
||
});
|
||
|
||
it('tools 定义为 {type:function,function:{...}} 形态', async () => {
|
||
const adapter = makeOllama();
|
||
const { bodies } = captureFetch();
|
||
await adapter.send(
|
||
makeRequest({
|
||
tools: [
|
||
{
|
||
name: 'calc',
|
||
description: 'calc',
|
||
parameters: { type: 'object', properties: { a: { type: 'number', description: 'a' } } },
|
||
category: 'calculation' as never,
|
||
riskLevel: 'safe' as never,
|
||
requiresPermission: false,
|
||
timeoutMs: 100,
|
||
},
|
||
],
|
||
}),
|
||
);
|
||
const tools = bodies[0].tools as Array<Record<string, unknown>>;
|
||
expect(tools[0]).toMatchObject({ type: 'function' });
|
||
expect((tools[0].function as Record<string, unknown>).name).toBe('calc');
|
||
});
|
||
|
||
it('assistant 工具调用参数序列化为 JSON 字符串（Ollama REST 要求）', async () => {
|
||
const adapter = makeOllama();
|
||
const { bodies } = captureFetch();
|
||
await adapter.send(
|
||
makeRequest({
|
||
messages: [
|
||
{ role: 'user', content: 'hi', timestamp: Date.now() },
|
||
{
|
||
role: 'assistant',
|
||
content: null,
|
||
toolCalls: [
|
||
{
|
||
id: 'tc1',
|
||
name: 'read',
|
||
args: { path: 'a.txt', lines: [1, 2] },
|
||
iteration: 1,
|
||
timestamp: Date.now(),
|
||
},
|
||
],
|
||
timestamp: Date.now(),
|
||
},
|
||
{
|
||
role: 'tool',
|
||
content: null,
|
||
toolResult: {
|
||
toolCallId: 'tc1',
|
||
toolName: 'read',
|
||
result: 'data',
|
||
success: true,
|
||
durationMs: 1,
|
||
timestamp: Date.now(),
|
||
},
|
||
timestamp: Date.now(),
|
||
},
|
||
],
|
||
}),
|
||
);
|
||
const messages = bodies[0].messages as Array<Record<string, unknown>>;
|
||
const assistantMsg = messages.find((m) => m.role === 'assistant') as {
|
||
tool_calls: Array<Record<string, unknown>>;
|
||
};
|
||
const fn = assistantMsg.tool_calls[0].function as Record<string, unknown>;
|
||
expect(fn.arguments).toBe(JSON.stringify({ path: 'a.txt', lines: [1, 2] }));
|
||
// tool 消息映射 tool_call_id + 结果文本
|
||
const toolMsg = messages.find((m) => m.role === 'tool') as {
|
||
tool_call_id: string;
|
||
content: string;
|
||
};
|
||
expect(toolMsg.tool_call_id).toBe('tc1');
|
||
expect(toolMsg.content).toBe('data');
|
||
});
|
||
|
||
it('assistant reasoning_content 回传保持推理链完整', async () => {
|
||
const adapter = makeOllama();
|
||
const { bodies } = captureFetch();
|
||
await adapter.send(
|
||
makeRequest({
|
||
messages: [
|
||
{ role: 'user', content: 'hi', timestamp: Date.now() },
|
||
{
|
||
role: 'assistant',
|
||
content: 'answer',
|
||
reasoningContent: 'thinking trace',
|
||
timestamp: Date.now(),
|
||
},
|
||
],
|
||
}),
|
||
);
|
||
const messages = bodies[0].messages as Array<Record<string, unknown>>;
|
||
const assistantMsg = messages.find((m) => m.role === 'assistant') as {
|
||
reasoning_content?: string;
|
||
};
|
||
expect(assistantMsg.reasoning_content).toBe('thinking trace');
|
||
});
|
||
|
||
it('assistant 无 content 时映射为空字符串（Ollama 不支持 null content）', async () => {
|
||
const adapter = makeOllama();
|
||
const { bodies } = captureFetch();
|
||
await adapter.send(
|
||
makeRequest({
|
||
messages: [
|
||
{ role: 'user', content: 'hi', timestamp: Date.now() },
|
||
{ role: 'assistant', content: null, timestamp: Date.now() },
|
||
],
|
||
}),
|
||
);
|
||
const messages = bodies[0].messages as Array<Record<string, unknown>>;
|
||
const assistantMsg = messages.find((m) => m.role === 'assistant') as { content: unknown };
|
||
expect(assistantMsg.content).toBe('');
|
||
});
|
||
|
||
it('纯 base64 图片（无 data: 前缀）原样透传', async () => {
|
||
const adapter = makeOllama();
|
||
const { bodies } = captureFetch();
|
||
await adapter.send(
|
||
makeRequest({
|
||
messages: [
|
||
{
|
||
role: 'user',
|
||
content: '看图',
|
||
images: [{ url: 'iVBORw0KGgoAAAANSUhEUg', detail: 'auto' }],
|
||
timestamp: Date.now(),
|
||
},
|
||
],
|
||
}),
|
||
);
|
||
const messages = bodies[0].messages as Array<Record<string, unknown>>;
|
||
const userMsg = messages[messages.length - 1];
|
||
expect(userMsg.images).toEqual(['iVBORw0KGgoAAAANSUhEUg']);
|
||
});
|
||
|
||
it('工具结果失败时 error 字段优先作为 content（CE-2）', async () => {
|
||
const adapter = makeOllama();
|
||
const { bodies } = captureFetch();
|
||
await adapter.send(
|
||
makeRequest({
|
||
messages: [
|
||
{ role: 'user', content: 'hi', timestamp: Date.now() },
|
||
{
|
||
role: 'assistant',
|
||
content: null,
|
||
toolCalls: [
|
||
{
|
||
id: 'tc_e',
|
||
name: 'run',
|
||
args: {},
|
||
iteration: 1,
|
||
timestamp: Date.now(),
|
||
},
|
||
],
|
||
timestamp: Date.now(),
|
||
},
|
||
{
|
||
role: 'tool',
|
||
content: null,
|
||
toolResult: {
|
||
toolCallId: 'tc_e',
|
||
toolName: 'run',
|
||
result: null,
|
||
success: false,
|
||
error: 'exit code 2',
|
||
durationMs: 1,
|
||
timestamp: Date.now(),
|
||
},
|
||
timestamp: Date.now(),
|
||
},
|
||
],
|
||
}),
|
||
);
|
||
const messages = bodies[0].messages as Array<Record<string, unknown>>;
|
||
const toolMsg = messages.find((m) => m.role === 'tool') as { content: string };
|
||
expect(toolMsg.content).toBe('exit code 2');
|
||
});
|
||
});
|
||
|
||
// ===== 跨 Provider maxTokens 钳制矩阵 =====
|
||
|
||
describe('跨 Provider — maxTokens 透传矩阵汇总（v0.8.1 无钳制）', () => {
|
||
it.each([
|
||
['anthropic', 'claude-opus-4-1', 100_000, 100_000],
|
||
['anthropic', 'claude-sonnet-4-5', 100_000, 100_000],
|
||
['deepseek', 'deepseek-v4-flash-vision-exp', 100_000, 100_000],
|
||
['agnes', 'agnes-2.0-flash', 100_000, 100_000],
|
||
['mimo', 'mimo-v2.5', 100_000, 100_000],
|
||
['openai', 'gpt-4o', 100_000, 100_000],
|
||
] as const)(
|
||
'%s %s maxTokens=100000 → 原样透传 %d',
|
||
async (provider, model, requested, expected) => {
|
||
const adapterMap: Record<string, unknown> = {
|
||
anthropic: new AnthropicAdapter({
|
||
provider: 'anthropic',
|
||
baseURL: 'http://a',
|
||
apiKey: 'k',
|
||
defaultModel: model,
|
||
}),
|
||
deepseek: new DeepSeekAdapter({
|
||
provider: 'deepseek',
|
||
baseURL: 'http://d',
|
||
apiKey: 'k',
|
||
defaultModel: model,
|
||
}),
|
||
agnes: new AgnesAdapter({
|
||
provider: 'agnes',
|
||
baseURL: 'http://g',
|
||
apiKey: 'k',
|
||
defaultModel: model,
|
||
}),
|
||
mimo: new MimoAdapter({
|
||
provider: 'mimo',
|
||
baseURL: 'http://m',
|
||
apiKey: 'k',
|
||
defaultModel: model,
|
||
}),
|
||
openai: new OpenAIAdapter({
|
||
provider: 'openai',
|
||
baseURL: 'http://o',
|
||
apiKey: 'k',
|
||
defaultModel: model,
|
||
}),
|
||
};
|
||
const adapter = adapterMap[provider] as { send: (r: MetonaRequest) => Promise<unknown> };
|
||
const { bodies } = captureFetch();
|
||
await adapter.send(
|
||
makeRequest({ params: { maxTokens: requested, temperature: 0, stream: false } }),
|
||
);
|
||
const body = bodies[bodies.length - 1] as Record<string, unknown>;
|
||
expect(body.max_tokens ?? body.max_completion_tokens).toBe(expected);
|
||
},
|
||
);
|
||
});
|