feat(ai-proxy): add batches & files support #2355
Conversation
Signed-off-by: Xijun Dai <[email protected]>
Signed-off-by: Xijun Dai <[email protected]>
…/v1/batches/{batche_id}
Signed-off-by: Xijun Dai <[email protected]>
Signed-off-by: Xijun Dai <[email protected]>
feat(ai-proxy): 添加对/v1/files和/v1/batches接口的支持变更文件
时序图Sequence diagram for API routing logic
participant Client
participant Router as AIProxyRouter
participant Provider as OpenAI/QwenProvider
Client->>Router: 发送/v1/files或/v1/batches请求
Router->>Router: 通过getApiName解析路径
Router->>Provider: 调用对应接口处理逻辑
Provider-->>Router: 返回处理结果
Router-->>Client: 响应客户端请求
💡 小贴士与 lingma-agents 交流的方式📜 直接回复评论
📜 在代码行处标记
📜 在讨论中提问
|
![lingma-agents[bot]](https://avatars.githubusercontent.com/in/1198481?s=60&v=4){kind=small}
There was a problem hiding this comment.
🔍 代码评审报告
📋 评审意见详情
💡 单文件建议
✅ 未发现需要特别关注的代码问题。
🚀 跨文件建议
以下是对代码架构和设计的综合分析，聚焦于跨文件交互、系统一致性和潜在优化空间。
🔍 1. 路径匹配逻辑可能导致API路由错误
在main.go的getApiName函数中，新增的/v1/batches和/v1/files路径检查使用strings.Contains代替原来的strings.HasSuffix。这种模糊匹配可能导致误判，例如路径/v1/batchses/example会被错误识别为ApiNameBatches。建议改为精确匹配如strings.HasSuffix或使用路由框架的精确路径匹配，以避免路由逻辑错误。
📌 关键代码：
+if strings.Contains(path, "/v1/batches") {
+ return provider.ApiNameBatches
+}
+if strings.Contains(path, "/v1/files") {
+ return provider.ApiNameFiles- 未指定文件
🔍 2. 兼容模式路径映射不完整风险
在qwen.go的DefaultCapabilities方法中，新增的qwenCompatibleFilesPath和qwenCompatibleBatchesPath路径未在非兼容模式下映射，可能导致非兼容模式下文件/批次API不可用。需检查所有新增功能是否在两种模式下均有对应路径定义。
📌 关键代码：
+string(ApiNameFiles): qwenCompatibleFilesPath,
+string(ApiNameBatches): qwenCompatibleBatchesPath,🔍 3. API路由逻辑架构不一致
新增的/v1/batches和/v1/files端点采用路径包含匹配，而原有/v1/models等端点使用后缀匹配。建议统一整个项目的API路径匹配策略（如全部使用精确后缀匹配或路由框架），避免混合模式导致的维护复杂度增加。
📌 关键代码：
+if strings.Contains(path, "/v1/batches") {🔍 4. Qwen兼容模式路径覆盖不完整
在qwen.go的GetApiName函数中，新增的兼容路径如qwenCompatibleChatCompletionPath被正确识别，但未检查原有非兼容路径的覆盖情况。需确保所有新增功能在两种模式下的路径均被路由逻辑覆盖，避免功能缺失。
📌 关键代码：
+strings.Contains(path, qwenCompatibleChatCompletionPath):
+return ApiNameChatCompletion💡 小贴士
与 lingma-agents 交流的方式
📜 直接回复评论
直接回复本条评论，lingma-agents 将自动处理您的请求。例如：
-
在当前代码中添加详细的注释说明。
-
请详细介绍一下你说的 LRU 改造方案，并使用伪代码加以说明。
📜 在代码行处标记
在文件的特定位置创建评论并 @lingma-agents。例如：
-
@Lingma-Agent 分析这个方法的性能瓶颈并提供优化建议。
-
@Lingma-Agent 对这个方法生成优化代码。
📜 在讨论中提问
在任何讨论中 @lingma-agents 来获取帮助。例如：
-
@Lingma-Agent 请总结上述讨论并提出解决方案。
-
@Lingma-Agent 请根据讨论内容生成优化代码。
Codecov ReportAll modified and coverable lines are covered by tests ✅
Additional details and impacted files@@ Coverage Diff @@
## main #2355 +/- ##
===========================================
+ Coverage 35.91% 46.06% +10.15%
===========================================
Files 69 81 +12
Lines 11576 13010 +1434
===========================================
+ Hits 4157 5993 +1836
+ Misses 7104 6671 -433
- Partials 315 346 +31 🚀 New features to boost your workflow:
|
| if strings.HasSuffix(path, "/v1/batches") { | ||
| return provider.ApiNameBatches | ||
| } |
Signed-off-by: Xijun Dai <[email protected]>
Ⅰ. Describe what this PR did
添加 openai 、 Qwen 对 /v1/files 与 /v1/batches 接口的支持
Ⅱ. Does this pull request fix one issue?
Ⅲ. Why don't you add test cases (unit test/integration test)?
Ⅳ. Describe how to verify it
以通义千问为例
docker-compose.yaml
envoy.yaml
Upload file
List files
$ curl -s http://127.0.0.1:10000/v1/files { "object": "list", "has_more": false, "data": [ { "id": "file-batch-7845fab470d8418cb07d1b7f", "object": "file", "bytes": 9104, "created_at": 1748659404, "filename": "BatchTaskCreate.jsonl", "purpose": "batch", "status": "processed" }, { "id": "file-batch_output-mKjljWQT4JbHn9GnqdDRuCWS", "object": "file", "bytes": 534, "created_at": 1748621857, "filename": "7fe9e7fe-4f24-45ea-b90b-6642715641b5_1748621856436_error.jsonl", "purpose": "batch_output", "status": "processed" }, { "id": "file-batch-9bffa2634c274e03b425afa4", "object": "file", "bytes": 9104, "created_at": 1748621073, "filename": "BatchTaskCreate.jsonl", "purpose": "batch", "status": "processed" }, { "id": "file-batch-61271ca8a18546ef9ef67706", "object": "file", "bytes": 9104, "created_at": 1748602930, "filename": "BatchTaskCreate.jsonl", "purpose": "batch", "status": "processed" } ] }Retrieve File
$ curl -s http://127.0.0.1:10000/v1/files/file-batch-7845fab470d8418cb07d1b7f { "id": "file-batch-7845fab470d8418cb07d1b7f", "object": "file", "bytes": 9104, "created_at": 1748659403, "filename": "BatchTaskCreate.jsonl", "purpose": "batch", "status": "processed", "meta": {} }Create Batch
List Batches
$ curl -s http://127.0.0.1:10000/v1/batches { "object": "list", "data": [ { "id": "batch_d2334048-816a-4c6f-a318-2aee1c201730", "object": "batch", "endpoint": "/v1/chat/completions", "errors": null, "input_file_id": "file-batch-7845fab470d8418cb07d1b7f", "completion_window": "24h", "status": "completed", "output_file_id": null, "error_file_id": "file-batch_output-kQMMqeOKDss58Ujcvlp7Xtfb", "created_at": 1748659669, "in_progress_at": 1748659669, "expires_at": null, "finalizing_at": 1748659670, "completed_at": 1748659670, "failed_at": null, "expired_at": null, "cancelling_at": null, "cancelled_at": null, "request_counts": { "total": 1, "completed": 0, "failed": 1 }, "metadata": { "ds_name": "任务名称", "ds_description": "任务描述" } } ], "first_id": "batch_d2334048-816a-4c6f-a318-2aee1c201730", "last_id": "batch_d2334048-816a-4c6f-a318-2aee1c201730", "has_more": false }Retrieve Batch
$ curl -s http://127.0.0.1:10000/v1/batches/batch_d2334048-816a-4c6f-a318-2aee1c201730 { "id": "batch_d2334048-816a-4c6f-a318-2aee1c201730", "object": "batch", "endpoint": "/v1/chat/completions", "errors": null, "input_file_id": "file-batch-7845fab470d8418cb07d1b7f", "completion_window": "24h", "status": "completed", "output_file_id": null, "error_file_id": "file-batch_output-kQMMqeOKDss58Ujcvlp7Xtfb", "created_at": 1748659669, "in_progress_at": 1748659669, "expires_at": null, "finalizing_at": 1748659670, "completed_at": 1748659670, "failed_at": null, "expired_at": null, "cancelling_at": null, "cancelled_at": null, "request_counts": { "total": 1, "completed": 0, "failed": 1 }, "metadata": { "ds_name": "任务名称", "ds_description": "任务描述" } }Ⅴ. Special notes for reviews