import { describe, it, expect } from "vitest"; // 由于 extractUsageMetrics 是内部函数，需要通过 parseUsageFromResponseText 间接测试 // 或者将其导出用于测试 // 这里我们通过构造 JSON 响应来测试 parseUsageFromResponseText import { parseUsageFromResponseText } from "@/app/v1/_lib/proxy/response-handler"; describe("extractUsageMetrics", () => { describe("基本 token 提取", () => { it("应正确提取 input_tokens 和 output_tokens", () => { const response = JSON.stringify({ usage: { input_tokens: 1000, output_tokens: 500, }, }); const result = parseUsageFromResponseText(response, "claude"); expect(result.usageMetrics).not.toBeNull(); expect(result.usageMetrics?.input_tokens).toBe(1000); expect(result.usageMetrics?.output_tokens).toBe(500); }); it("空值或非对象应返回 null", () => { expect(parseUsageFromResponseText("", "claude").usageMetrics).toBeNull(); expect(parseUsageFromResponseText("null", "claude").usageMetrics).toBeNull(); expect(parseUsageFromResponseText('"string"', "claude").usageMetrics).toBeNull(); }); }); describe("Claude 嵌套格式 (cache_creation.ephemeral_*)", () => { it("应从 cache_creation 嵌套对象提取 5m 和 1h token", () => { const response = JSON.stringify({ usage: { input_tokens: 1000, output_tokens: 500, cache_creation_input_tokens: 800, cache_creation: { ephemeral_5m_input_tokens: 300, ephemeral_1h_input_tokens: 500, }, cache_read_input_tokens: 200, }, }); const result = parseUsageFromResponseText(response, "claude"); expect(result.usageMetrics).not.toBeNull(); expect(result.usageMetrics?.cache_creation_input_tokens).toBe(800); expect(result.usageMetrics?.cache_creation_5m_input_tokens).toBe(300); expect(result.usageMetrics?.cache_creation_1h_input_tokens).toBe(500); expect(result.usageMetrics?.cache_read_input_tokens).toBe(200); expect(result.usageMetrics?.cache_ttl).toBe("mixed"); }); it("只有 5m 时应推断 cache_ttl 为 5m", () => { const response = JSON.stringify({ usage: { cache_creation_input_tokens: 300, cache_creation: { ephemeral_5m_input_tokens: 300, }, }, }); const result = parseUsageFromResponseText(response, "claude"); expect(result.usageMetrics?.cache_creation_5m_input_tokens).toBe(300); expect(result.usageMetrics?.cache_creation_1h_input_tokens).toBeUndefined(); expect(result.usageMetrics?.cache_ttl).toBe("5m"); }); it("只有 1h 时应推断 cache_ttl 为 1h", () => { const response = JSON.stringify({ usage: { cache_creation_input_tokens: 500, cache_creation: { ephemeral_1h_input_tokens: 500, }, }, }); const result = parseUsageFromResponseText(response, "claude"); expect(result.usageMetrics?.cache_creation_1h_input_tokens).toBe(500); expect(result.usageMetrics?.cache_creation_5m_input_tokens).toBeUndefined(); expect(result.usageMetrics?.cache_ttl).toBe("1h"); }); }); describe("旧 relay 格式 (claude_cache_creation_*)", () => { it("应从旧 relay 字段提取 5m 和 1h token", () => { const response = JSON.stringify({ usage: { input_tokens: 1000, output_tokens: 500, cache_creation_input_tokens: 800, claude_cache_creation_5_m_tokens: 300, claude_cache_creation_1_h_tokens: 500, cache_read_input_tokens: 200, }, }); const result = parseUsageFromResponseText(response, "claude"); expect(result.usageMetrics?.cache_creation_5m_input_tokens).toBe(300); expect(result.usageMetrics?.cache_creation_1h_input_tokens).toBe(500); expect(result.usageMetrics?.cache_ttl).toBe("mixed"); }); it("嵌套格式应优先于旧 relay 格式", () => { const response = JSON.stringify({ usage: { cache_creation: { ephemeral_5m_input_tokens: 100, ephemeral_1h_input_tokens: 200, }, claude_cache_creation_5_m_tokens: 999, claude_cache_creation_1_h_tokens: 888, }, }); const result = parseUsageFromResponseText(response, "claude"); // 嵌套格式优先 expect(result.usageMetrics?.cache_creation_5m_input_tokens).toBe(100); expect(result.usageMetrics?.cache_creation_1h_input_tokens).toBe(200); }); }); describe("顶层扁平格式 (cache_creation_5m_input_tokens)", () => { it("应从顶层扁平字段提取 5m 和 1h token", () => { const response = JSON.stringify({ usage: { input_tokens: 1000, output_tokens: 500, cache_creation_input_tokens: 800, cache_creation_5m_input_tokens: 300, cache_creation_1h_input_tokens: 500, cache_read_input_tokens: 200, }, }); const result = parseUsageFromResponseText(response, "claude"); expect(result.usageMetrics?.cache_creation_input_tokens).toBe(800); expect(result.usageMetrics?.cache_creation_5m_input_tokens).toBe(300); expect(result.usageMetrics?.cache_creation_1h_input_tokens).toBe(500); expect(result.usageMetrics?.cache_read_input_tokens).toBe(200); expect(result.usageMetrics?.cache_ttl).toBe("mixed"); }); it("只有顶层 5m 时应正确提取并推断 TTL", () => { const response = JSON.stringify({ usage: { cache_creation_input_tokens: 300, cache_creation_5m_input_tokens: 300, }, }); const result = parseUsageFromResponseText(response, "claude"); expect(result.usageMetrics?.cache_creation_5m_input_tokens).toBe(300); expect(result.usageMetrics?.cache_ttl).toBe("5m"); }); it("只有顶层 1h 时应正确提取并推断 TTL", () => { const response = JSON.stringify({ usage: { cache_creation_input_tokens: 500, cache_creation_1h_input_tokens: 500, }, }); const result = parseUsageFromResponseText(response, "claude"); expect(result.usageMetrics?.cache_creation_1h_input_tokens).toBe(500); expect(result.usageMetrics?.cache_ttl).toBe("1h"); }); it("嵌套格式应优先于顶层扁平格式", () => { const response = JSON.stringify({ usage: { cache_creation: { ephemeral_5m_input_tokens: 100, ephemeral_1h_input_tokens: 200, }, cache_creation_5m_input_tokens: 999, cache_creation_1h_input_tokens: 888, }, }); const result = parseUsageFromResponseText(response, "claude"); // 嵌套格式优先 expect(result.usageMetrics?.cache_creation_5m_input_tokens).toBe(100); expect(result.usageMetrics?.cache_creation_1h_input_tokens).toBe(200); }); it("顶层扁平格式应优先于旧 relay 格式", () => { const response = JSON.stringify({ usage: { cache_creation_5m_input_tokens: 300, cache_creation_1h_input_tokens: 500, claude_cache_creation_5_m_tokens: 999, claude_cache_creation_1_h_tokens: 888, }, }); const result = parseUsageFromResponseText(response, "claude"); // 顶层扁平格式优先于旧 relay 格式 expect(result.usageMetrics?.cache_creation_5m_input_tokens).toBe(300); expect(result.usageMetrics?.cache_creation_1h_input_tokens).toBe(500); }); it("三种格式同时存在时应按优先级提取", () => { const response = JSON.stringify({ usage: { cache_creation: { ephemeral_5m_input_tokens: 100, ephemeral_1h_input_tokens: 200, }, cache_creation_5m_input_tokens: 300, cache_creation_1h_input_tokens: 400, claude_cache_creation_5_m_tokens: 500, claude_cache_creation_1_h_tokens: 600, }, }); const result = parseUsageFromResponseText(response, "claude"); // 嵌套格式最优先 expect(result.usageMetrics?.cache_creation_5m_input_tokens).toBe(100); expect(result.usageMetrics?.cache_creation_1h_input_tokens).toBe(200); expect(result.usageMetrics?.cache_ttl).toBe("mixed"); }); }); describe("cache_creation_input_tokens 自动计算", () => { it("当 cache_creation_input_tokens 缺失时应自动计算总量", () => { const response = JSON.stringify({ usage: { cache_creation: { ephemeral_5m_input_tokens: 300, ephemeral_1h_input_tokens: 500, }, }, }); const result = parseUsageFromResponseText(response, "claude"); expect(result.usageMetrics?.cache_creation_input_tokens).toBe(800); }); it("顶层扁平格式缺失 cache_creation_input_tokens 时应自动计算总量", () => { const response = JSON.stringify({ usage: { cache_creation_5m_input_tokens: 400, cache_creation_1h_input_tokens: 600, }, }); const result = parseUsageFromResponseText(response, "claude"); expect(result.usageMetrics?.cache_creation_input_tokens).toBe(1000); expect(result.usageMetrics?.cache_creation_5m_input_tokens).toBe(400); expect(result.usageMetrics?.cache_creation_1h_input_tokens).toBe(600); }); it("混合回退：嵌套缺失某字段时顶层扁平补齐", () => { const response = JSON.stringify({ usage: { cache_creation: { ephemeral_5m_input_tokens: 200, // 缺失 ephemeral_1h_input_tokens }, cache_creation_1h_input_tokens: 300, // 顶层扁平补齐 }, }); const result = parseUsageFromResponseText(response, "claude"); // 5m 来自嵌套，1h 来自顶层扁平 expect(result.usageMetrics?.cache_creation_5m_input_tokens).toBe(200); expect(result.usageMetrics?.cache_creation_1h_input_tokens).toBe(300); expect(result.usageMetrics?.cache_creation_input_tokens).toBe(500); expect(result.usageMetrics?.cache_ttl).toBe("mixed"); }); it("当 cache_creation_input_tokens 存在时不应覆盖", () => { const response = JSON.stringify({ usage: { cache_creation_input_tokens: 1000, cache_creation: { ephemeral_5m_input_tokens: 300, ephemeral_1h_input_tokens: 500, }, }, }); const result = parseUsageFromResponseText(response, "claude"); // 保留原值 expect(result.usageMetrics?.cache_creation_input_tokens).toBe(1000); }); }); describe("Gemini 格式支持", () => { it("应正确提取 Gemini usage 字段", () => { const response = JSON.stringify({ usageMetadata: { promptTokenCount: 1000, candidatesTokenCount: 500, cachedContentTokenCount: 200, }, }); const result = parseUsageFromResponseText(response, "gemini"); expect(result.usageMetrics).not.toBeNull(); // input_tokens = promptTokenCount - cachedContentTokenCount expect(result.usageMetrics?.input_tokens).toBe(800); expect(result.usageMetrics?.output_tokens).toBe(500); expect(result.usageMetrics?.cache_read_input_tokens).toBe(200); }); it("应正确处理 Gemini thoughtsTokenCount", () => { const response = JSON.stringify({ usageMetadata: { promptTokenCount: 1000, candidatesTokenCount: 500, thoughtsTokenCount: 100, }, }); const result = parseUsageFromResponseText(response, "gemini"); // output_tokens = candidatesTokenCount + thoughtsTokenCount expect(result.usageMetrics?.output_tokens).toBe(600); }); it("应从 candidatesTokensDetails 提取 IMAGE modality tokens", () => { const response = JSON.stringify({ usageMetadata: { promptTokenCount: 326, candidatesTokenCount: 2340, candidatesTokensDetails: [ { modality: "IMAGE", tokenCount: 2000 }, { modality: "TEXT", tokenCount: 340 }, ], }, }); const result = parseUsageFromResponseText(response, "gemini"); expect(result.usageMetrics?.output_image_tokens).toBe(2000); expect(result.usageMetrics?.output_tokens).toBe(340); }); it("应从 promptTokensDetails 提取 IMAGE modality tokens", () => { const response = JSON.stringify({ usageMetadata: { promptTokenCount: 886, candidatesTokenCount: 500, promptTokensDetails: [ { modality: "TEXT", tokenCount: 326 }, { modality: "IMAGE", tokenCount: 560 }, ], }, }); const result = parseUsageFromResponseText(response, "gemini"); expect(result.usageMetrics?.input_image_tokens).toBe(560); expect(result.usageMetrics?.input_tokens).toBe(326); }); it("应正确解析混合输入输出的完整 usage", () => { const response = JSON.stringify({ usageMetadata: { promptTokenCount: 357, candidatesTokenCount: 2100, totalTokenCount: 2580, promptTokensDetails: [ { modality: "TEXT", tokenCount: 99 }, { modality: "IMAGE", tokenCount: 258 }, ], candidatesTokensDetails: [{ modality: "IMAGE", tokenCount: 2000 }], thoughtsTokenCount: 123, }, }); const result = parseUsageFromResponseText(response, "gemini"); expect(result.usageMetrics?.input_tokens).toBe(99); expect(result.usageMetrics?.input_image_tokens).toBe(258); // output_tokens = (candidatesTokenCount - IMAGE详情) + thoughtsTokenCount // = (2100 - 2000) + 123 = 223 expect(result.usageMetrics?.output_tokens).toBe(223); expect(result.usageMetrics?.output_image_tokens).toBe(2000); }); it("应处理只有 IMAGE modality 的 candidatesTokensDetails", () => { const response = JSON.stringify({ usageMetadata: { promptTokenCount: 100, candidatesTokenCount: 2000, candidatesTokensDetails: [{ modality: "IMAGE", tokenCount: 2000 }], }, }); const result = parseUsageFromResponseText(response, "gemini"); expect(result.usageMetrics?.output_image_tokens).toBe(2000); // candidatesTokenCount = 2000, IMAGE = 2000, 未分类 = 0 expect(result.usageMetrics?.output_tokens).toBe(0); }); it("应计算 candidatesTokenCount 与 details 的差值作为未分类 TEXT", () => { const response = JSON.stringify({ usageMetadata: { promptTokenCount: 326, candidatesTokenCount: 2340, candidatesTokensDetails: [{ modality: "IMAGE", tokenCount: 2000 }], thoughtsTokenCount: 337, }, }); const result = parseUsageFromResponseText(response, "gemini"); // 未分类 = 2340 - 2000 = 340 // output_tokens = 340 + 337 (thoughts) = 677 expect(result.usageMetrics?.output_tokens).toBe(677); expect(result.usageMetrics?.output_image_tokens).toBe(2000); }); it("应处理缺失 candidatesTokensDetails 的情况（向后兼容）", () => { const response = JSON.stringify({ usageMetadata: { promptTokenCount: 1000, candidatesTokenCount: 500, }, }); const result = parseUsageFromResponseText(response, "gemini"); expect(result.usageMetrics?.output_tokens).toBe(500); expect(result.usageMetrics?.output_image_tokens).toBeUndefined(); expect(result.usageMetrics?.input_image_tokens).toBeUndefined(); }); it("应处理空的 candidatesTokensDetails 数组", () => { const response = JSON.stringify({ usageMetadata: { promptTokenCount: 1000, candidatesTokenCount: 500, candidatesTokensDetails: [], }, }); const result = parseUsageFromResponseText(response, "gemini"); expect(result.usageMetrics?.output_tokens).toBe(500); expect(result.usageMetrics?.output_image_tokens).toBeUndefined(); }); it("应处理 candidatesTokensDetails 中无效 tokenCount 的情况", () => { const response = JSON.stringify({ usageMetadata: { promptTokenCount: 1000, candidatesTokenCount: 500, candidatesTokensDetails: [ { modality: "TEXT" }, { modality: "IMAGE", tokenCount: null }, { modality: "TEXT", tokenCount: -1 }, ], }, }); const result = parseUsageFromResponseText(response, "gemini"); // 无效数据不应覆盖原始 candidatesTokenCount expect(result.usageMetrics?.output_tokens).toBe(500); expect(result.usageMetrics?.output_image_tokens).toBeUndefined(); }); it("应处理 modality 大小写变体", () => { const response = JSON.stringify({ usageMetadata: { promptTokenCount: 100, candidatesTokenCount: 2340, candidatesTokensDetails: [ { modality: "image", tokenCount: 2000 }, { modality: "Image", tokenCount: 100 }, { modality: "TEXT", tokenCount: 240 }, ], }, }); const result = parseUsageFromResponseText(response, "gemini"); expect(result.usageMetrics?.output_image_tokens).toBe(2100); expect(result.usageMetrics?.output_tokens).toBe(240); }); }); describe("OpenAI Response API 格式", () => { it("应从 input_tokens_details.cached_tokens 提取缓存读取", () => { const response = JSON.stringify({ usage: { input_tokens: 1000, output_tokens: 500, input_tokens_details: { cached_tokens: 200, }, }, }); const result = parseUsageFromResponseText(response, "openai"); expect(result.usageMetrics?.cache_read_input_tokens).toBe(200); }); it("顶层 cache_read_input_tokens 应优先于嵌套格式", () => { const response = JSON.stringify({ usage: { input_tokens: 1000, cache_read_input_tokens: 300, input_tokens_details: { cached_tokens: 200, }, }, }); const result = parseUsageFromResponseText(response, "openai"); // 顶层优先 expect(result.usageMetrics?.cache_read_input_tokens).toBe(300); }); }); describe("SSE 流式响应解析", () => { it("应正确合并 message_start 和 message_delta 的 usage", () => { // 模拟 Claude SSE 流式响应 const sseResponse = [ "event: message_start", 'data: {"type":"message_start","message":{"usage":{"input_tokens":1000,"cache_creation_input_tokens":500,"cache_creation":{"ephemeral_5m_input_tokens":200,"ephemeral_1h_input_tokens":300},"cache_read_input_tokens":100}}}', "", "event: message_delta", 'data: {"type":"message_delta","usage":{"output_tokens":800}}', "", ].join("\n"); const result = parseUsageFromResponseText(sseResponse, "claude"); expect(result.usageMetrics).not.toBeNull(); expect(result.usageMetrics?.input_tokens).toBe(1000); expect(result.usageMetrics?.output_tokens).toBe(800); expect(result.usageMetrics?.cache_creation_input_tokens).toBe(500); expect(result.usageMetrics?.cache_creation_5m_input_tokens).toBe(200); expect(result.usageMetrics?.cache_creation_1h_input_tokens).toBe(300); expect(result.usageMetrics?.cache_read_input_tokens).toBe(100); }); it("message_delta 的值应优先于 message_start", () => { const sseResponse = [ "event: message_start", 'data: {"type":"message_start","message":{"usage":{"input_tokens":100,"output_tokens":50}}}', "", "event: message_delta", 'data: {"type":"message_delta","usage":{"input_tokens":1000,"output_tokens":500}}', "", ].join("\n"); const result = parseUsageFromResponseText(sseResponse, "claude"); // message_delta 优先 expect(result.usageMetrics?.input_tokens).toBe(1000); expect(result.usageMetrics?.output_tokens).toBe(500); }); it("message_start 的 cache 细分应补充 message_delta 缺失的字段", () => { const sseResponse = [ "event: message_start", 'data: {"type":"message_start","message":{"usage":{"cache_creation":{"ephemeral_5m_input_tokens":200,"ephemeral_1h_input_tokens":300}}}}', "", "event: message_delta", 'data: {"type":"message_delta","usage":{"input_tokens":1000,"output_tokens":500,"cache_creation_input_tokens":500}}', "", ].join("\n"); const result = parseUsageFromResponseText(sseResponse, "claude"); // message_delta 的值 expect(result.usageMetrics?.input_tokens).toBe(1000); expect(result.usageMetrics?.output_tokens).toBe(500); expect(result.usageMetrics?.cache_creation_input_tokens).toBe(500); // message_start 补充的细分字段 expect(result.usageMetrics?.cache_creation_5m_input_tokens).toBe(200); expect(result.usageMetrics?.cache_creation_1h_input_tokens).toBe(300); }); }); describe("Codex provider 特殊处理", () => { it("Codex 应从 input_tokens 中减去 cached_tokens", () => { const response = JSON.stringify({ usage: { input_tokens: 1000, output_tokens: 500, cache_read_input_tokens: 300, }, }); const result = parseUsageFromResponseText(response, "codex"); // adjustUsageForProviderType 会调整 input_tokens expect(result.usageMetrics?.input_tokens).toBe(700); // 1000 - 300 expect(result.usageMetrics?.cache_read_input_tokens).toBe(300); }); }); describe("边界情况", () => { it("应处理所有值为 0 的情况", () => { const response = JSON.stringify({ usage: { input_tokens: 0, output_tokens: 0, cache_creation_input_tokens: 0, cache_read_input_tokens: 0, }, }); const result = parseUsageFromResponseText(response, "claude"); expect(result.usageMetrics).not.toBeNull(); expect(result.usageMetrics?.input_tokens).toBe(0); expect(result.usageMetrics?.output_tokens).toBe(0); }); it("应处理部分字段缺失的情况", () => { const response = JSON.stringify({ usage: { input_tokens: 1000, }, }); const result = parseUsageFromResponseText(response, "claude"); expect(result.usageMetrics?.input_tokens).toBe(1000); expect(result.usageMetrics?.output_tokens).toBeUndefined(); expect(result.usageMetrics?.cache_creation_input_tokens).toBeUndefined(); }); it("应处理无效的 JSON", () => { const result = parseUsageFromResponseText("invalid json", "claude"); expect(result.usageMetrics).toBeNull(); }); it("应处理空的 usage 对象", () => { const response = JSON.stringify({ usage: {}, }); const result = parseUsageFromResponseText(response, "claude"); expect(result.usageMetrics).toBeNull(); }); }); describe("OpenAI chat completion format (prompt_tokens/completion_tokens)", () => { it("should extract prompt_tokens as input_tokens", () => { const response = JSON.stringify({ usage: { prompt_tokens: 100, completion_tokens: 50, total_tokens: 150, }, }); const result = parseUsageFromResponseText(response, "openai"); expect(result.usageMetrics).not.toBeNull(); expect(result.usageMetrics?.input_tokens).toBe(100); expect(result.usageMetrics?.output_tokens).toBe(50); }); it("should extract completion_tokens as output_tokens", () => { const response = JSON.stringify({ usage: { completion_tokens: 200, }, }); const result = parseUsageFromResponseText(response, "openai"); expect(result.usageMetrics).not.toBeNull(); expect(result.usageMetrics?.output_tokens).toBe(200); }); it("should prefer input_tokens over prompt_tokens (Claude format priority)", () => { const response = JSON.stringify({ usage: { input_tokens: 500, output_tokens: 300, prompt_tokens: 100, completion_tokens: 50, }, }); const result = parseUsageFromResponseText(response, "openai"); expect(result.usageMetrics?.input_tokens).toBe(500); expect(result.usageMetrics?.output_tokens).toBe(300); }); it("should handle OpenAI streaming chunk with usage in final event", () => { const sse = [ 'data: {"id":"chatcmpl-1","object":"chat.completion.chunk","created":1234567890,"model":"gpt-4","choices":[{"index":0,"delta":{"role":"assistant","content":"Hi"}}]}', "", 'data: {"id":"chatcmpl-1","object":"chat.completion.chunk","created":1234567890,"model":"gpt-4","choices":[{"index":0,"delta":{},"finish_reason":"stop"}],"usage":{"prompt_tokens":33,"completion_tokens":31,"total_tokens":64}}', "", "data: [DONE]", ].join("\n"); const result = parseUsageFromResponseText(sse, "openai"); expect(result.usageMetrics).not.toBeNull(); expect(result.usageMetrics?.input_tokens).toBe(33); expect(result.usageMetrics?.output_tokens).toBe(31); }); it("should handle OpenAI completion_tokens_details (reasoning_tokens)", () => { const response = JSON.stringify({ usage: { prompt_tokens: 66, completion_tokens: 57, total_tokens: 123, completion_tokens_details: { reasoning_tokens: 0, }, }, }); const result = parseUsageFromResponseText(response, "openai-compatible"); expect(result.usageMetrics).not.toBeNull(); expect(result.usageMetrics?.input_tokens).toBe(66); expect(result.usageMetrics?.output_tokens).toBe(57); }); }); });