api_endpoints_test.rs 64.3 KB
Newer Older
1
2
mod common;

3
4
use std::sync::Arc;

5
6
7
8
9
use axum::{
    body::Body,
    extract::Request,
    http::{header::CONTENT_TYPE, StatusCode},
};
10
11
12
use common::mock_worker::{HealthStatus, MockWorker, MockWorkerConfig, WorkerType};
use reqwest::Client;
use serde_json::json;
13
use sglang_router_rs::{
14
15
    config::{CircuitBreakerConfig, PolicyConfig, RetryConfig, RouterConfig, RoutingMode},
    core::{ConnectionMode, Job},
16
17
    routers::{RouterFactory, RouterTrait},
    server::AppContext,
18
};
19
use tower::ServiceExt;
20
21
22
23

/// Test context that manages mock workers
struct TestContext {
    workers: Vec<MockWorker>,
24
    router: Arc<dyn RouterTrait>,
25
26
27
    _client: Client,
    _config: RouterConfig,
    app_context: Arc<AppContext>,
28
29
30
31
32
33
}

impl TestContext {
    async fn new(worker_configs: Vec<MockWorkerConfig>) -> Self {
        // Create default router config
        let config = RouterConfig {
34
            chat_template: None,
35
36
37
38
39
40
41
42
43
44
            mode: RoutingMode::Regular {
                worker_urls: vec![],
            },
            policy: PolicyConfig::Random,
            host: "127.0.0.1".to_string(),
            port: 3002,
            max_payload_size: 256 * 1024 * 1024,
            request_timeout_secs: 600,
            worker_startup_timeout_secs: 1,
            worker_startup_check_interval_secs: 1,
45
            discovery: None,
46
47
            dp_aware: false,
            api_key: None,
48
49
50
            metrics: None,
            log_dir: None,
            log_level: None,
51
            request_id_headers: None,
52
            max_concurrent_requests: 64,
53
54
55
            queue_size: 0,
            queue_timeout_secs: 60,
            rate_limit_tokens_per_second: None,
56
            cors_allowed_origins: vec![],
57
            retry: RetryConfig::default(),
58
            circuit_breaker: CircuitBreakerConfig::default(),
59
60
            disable_retries: false,
            disable_circuit_breaker: false,
61
            health_check: sglang_router_rs::config::HealthCheckConfig::default(),
62
            enable_igw: false,
63
64
65
            connection_mode: ConnectionMode::Http,
            model_path: None,
            tokenizer_path: None,
66
            history_backend: sglang_router_rs::config::HistoryBackend::Memory,
67
            oracle: None,
68
69
            reasoning_parser: None,
            tool_call_parser: None,
70
            tokenizer_cache: sglang_router_rs::config::TokenizerCacheConfig::default(),
71
72
73
74
75
        };

        Self::new_with_config(config, worker_configs).await
    }

76
77
78
79
    async fn new_with_config(
        mut config: RouterConfig,
        worker_configs: Vec<MockWorkerConfig>,
    ) -> Self {
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
        let mut workers = Vec::new();
        let mut worker_urls = Vec::new();

        // Start mock workers if any
        for worker_config in worker_configs {
            let mut worker = MockWorker::new(worker_config);
            let url = worker.start().await.unwrap();
            worker_urls.push(url);
            workers.push(worker);
        }

        if !workers.is_empty() {
            tokio::time::sleep(tokio::time::Duration::from_millis(200)).await;
        }

95
96
97
98
99
100
101
102
103
104
        // Update config with worker URLs if not already set
        if let RoutingMode::Regular {
            worker_urls: ref mut urls,
        } = config.mode
        {
            if urls.is_empty() {
                *urls = worker_urls.clone();
            }
        }

105
106
107
108
109
        let client = Client::builder()
            .timeout(std::time::Duration::from_secs(config.request_timeout_secs))
            .build()
            .unwrap();

110
111
        // Create app context
        let app_context = common::create_test_context(config.clone());
112

113
        // Submit worker initialization job (same as real server does)
114
        if !worker_urls.is_empty() {
115
116
117
118
119
120
121
122
123
            let job_queue = app_context
                .worker_job_queue
                .get()
                .expect("JobQueue should be initialized");
            let job = Job::InitializeWorkersFromConfig {
                router_config: Box::new(config.clone()),
            };
            job_queue
                .submit(job)
124
                .await
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
                .expect("Failed to submit worker initialization job");

            // Poll until all workers are healthy (up to 10 seconds)
            let expected_count = worker_urls.len();
            let start = tokio::time::Instant::now();
            let timeout_duration = tokio::time::Duration::from_secs(10);
            loop {
                let healthy_workers = app_context
                    .worker_registry
                    .get_all()
                    .iter()
                    .filter(|w| w.is_healthy())
                    .count();

                if healthy_workers >= expected_count {
                    break;
                }

                if start.elapsed() > timeout_duration {
                    panic!(
                        "Timeout waiting for {} workers to become healthy (only {} ready)",
                        expected_count, healthy_workers
                    );
                }

                tokio::time::sleep(tokio::time::Duration::from_millis(100)).await;
            }
152
153
        }

154
155
        // Create router
        let router = RouterFactory::create_router(&app_context).await.unwrap();
156
        let router = Arc::from(router);
157

158
159
160
        Self {
            workers,
            router,
161
162
163
            _client: client,
            _config: config,
            app_context,
164
        }
165
166
    }

167
    async fn create_app(&self) -> axum::Router {
168
        common::test_app::create_test_app_with_context(
169
            Arc::clone(&self.router),
170
            Arc::clone(&self.app_context),
171
172
173
174
175
176
177
178
179
180
181
182
183
184
        )
    }

    async fn shutdown(mut self) {
        for worker in &mut self.workers {
            worker.stop().await;
        }
    }
}

#[cfg(test)]
mod health_tests {
    use super::*;

185
186
187
188
    #[tokio::test]
    async fn test_liveness_endpoint() {
        let ctx = TestContext::new(vec![]).await;
        let app = ctx.create_app().await;
189

190
191
192
193
194
        let req = Request::builder()
            .method("GET")
            .uri("/liveness")
            .body(Body::empty())
            .unwrap();
195

196
197
        let resp = app.oneshot(req).await.unwrap();
        assert_eq!(resp.status(), StatusCode::OK);
198

199
        ctx.shutdown().await;
200
201
    }

202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
    #[tokio::test]
    async fn test_readiness_with_healthy_workers() {
        let ctx = TestContext::new(vec![MockWorkerConfig {
            port: 18001,
            worker_type: WorkerType::Regular,
            health_status: HealthStatus::Healthy,
            response_delay_ms: 0,
            fail_rate: 0.0,
        }])
        .await;

        let app = ctx.create_app().await;

        let req = Request::builder()
            .method("GET")
            .uri("/readiness")
            .body(Body::empty())
            .unwrap();
220

221
222
        let resp = app.oneshot(req).await.unwrap();
        assert_eq!(resp.status(), StatusCode::OK);
223

224
        ctx.shutdown().await;
225
226
    }

227
228
229
    #[tokio::test]
    async fn test_readiness_with_unhealthy_workers() {
        let ctx = TestContext::new(vec![]).await;
230

231
        let app = ctx.create_app().await;
232

233
234
235
236
237
        let req = Request::builder()
            .method("GET")
            .uri("/readiness")
            .body(Body::empty())
            .unwrap();
238

239
240
241
        let resp = app.oneshot(req).await.unwrap();
        // With no workers, readiness should return SERVICE_UNAVAILABLE
        assert_eq!(resp.status(), StatusCode::SERVICE_UNAVAILABLE);
242

243
        ctx.shutdown().await;
244
245
    }

246
247
248
249
250
251
252
253
254
255
256
257
    #[tokio::test]
    async fn test_health_endpoint_details() {
        let ctx = TestContext::new(vec![
            MockWorkerConfig {
                port: 18003,
                worker_type: WorkerType::Regular,
                health_status: HealthStatus::Healthy,
                response_delay_ms: 0,
                fail_rate: 0.0,
            },
            MockWorkerConfig {
                port: 18004,
258
259
260
261
                worker_type: WorkerType::Regular,
                health_status: HealthStatus::Healthy,
                response_delay_ms: 0,
                fail_rate: 0.0,
262
263
264
            },
        ])
        .await;
265

266
        let app = ctx.create_app().await;
267

268
269
270
271
272
        let req = Request::builder()
            .method("GET")
            .uri("/health")
            .body(Body::empty())
            .unwrap();
273

274
275
        let resp = app.oneshot(req).await.unwrap();
        assert_eq!(resp.status(), StatusCode::OK);
276

277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
        ctx.shutdown().await;
    }

    #[tokio::test]
    async fn test_health_generate_endpoint() {
        let ctx = TestContext::new(vec![MockWorkerConfig {
            port: 18005,
            worker_type: WorkerType::Regular,
            health_status: HealthStatus::Healthy,
            response_delay_ms: 0,
            fail_rate: 0.0,
        }])
        .await;

        let app = ctx.create_app().await;

        let req = Request::builder()
            .method("GET")
            .uri("/health_generate")
            .body(Body::empty())
            .unwrap();

        let resp = app.oneshot(req).await.unwrap();
        assert_eq!(resp.status(), StatusCode::OK);

        let body = axum::body::to_bytes(resp.into_body(), usize::MAX)
            .await
            .unwrap();
        let body_json: serde_json::Value = serde_json::from_slice(&body).unwrap();
        assert!(body_json.is_object());

        ctx.shutdown().await;
309
310
311
312
313
314
315
    }
}

#[cfg(test)]
mod generation_tests {
    use super::*;

316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
    #[tokio::test]
    async fn test_generate_success() {
        let ctx = TestContext::new(vec![MockWorkerConfig {
            port: 18101,
            worker_type: WorkerType::Regular,
            health_status: HealthStatus::Healthy,
            response_delay_ms: 0,
            fail_rate: 0.0,
        }])
        .await;

        let app = ctx.create_app().await;

        let payload = json!({
            "text": "Hello, world!",
            "stream": false
        });
333

334
335
336
337
338
339
        let req = Request::builder()
            .method("POST")
            .uri("/generate")
            .header(CONTENT_TYPE, "application/json")
            .body(Body::from(serde_json::to_string(&payload).unwrap()))
            .unwrap();
340

341
342
        let resp = app.oneshot(req).await.unwrap();
        assert_eq!(resp.status(), StatusCode::OK);
343

344
345
346
347
348
349
350
351
352
353
354
355
        let body = axum::body::to_bytes(resp.into_body(), usize::MAX)
            .await
            .unwrap();
        let body_json: serde_json::Value = serde_json::from_slice(&body).unwrap();
        assert!(body_json.get("text").is_some());
        assert!(body_json.get("meta_info").is_some());
        let meta_info = &body_json["meta_info"];
        assert!(meta_info.get("finish_reason").is_some());
        assert_eq!(meta_info["finish_reason"]["type"], "stop");

        ctx.shutdown().await;
    }
356

357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
    #[tokio::test]
    async fn test_generate_streaming() {
        let ctx = TestContext::new(vec![MockWorkerConfig {
            port: 18102,
            worker_type: WorkerType::Regular,
            health_status: HealthStatus::Healthy,
            response_delay_ms: 0,
            fail_rate: 0.0,
        }])
        .await;

        let app = ctx.create_app().await;

        let payload = json!({
            "text": "Stream test",
            "stream": true
373
374
        });

375
376
377
378
379
380
        let req = Request::builder()
            .method("POST")
            .uri("/generate")
            .header(CONTENT_TYPE, "application/json")
            .body(Body::from(serde_json::to_string(&payload).unwrap()))
            .unwrap();
381

382
383
        let resp = app.oneshot(req).await.unwrap();
        assert_eq!(resp.status(), StatusCode::OK);
384

385
386
387
388
        // For streaming responses, the router might use chunked encoding or other streaming mechanisms
        // The exact content-type can vary based on the router implementation
        // Just verify we got a successful response
        // Note: In a real implementation, we'd check for text/event-stream or appropriate streaming headers
389

390
391
        ctx.shutdown().await;
    }
392

393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
    #[tokio::test]
    async fn test_generate_with_worker_failure() {
        let ctx = TestContext::new(vec![MockWorkerConfig {
            port: 18103,
            worker_type: WorkerType::Regular,
            health_status: HealthStatus::Healthy,
            response_delay_ms: 0,
            fail_rate: 1.0, // Always fail
        }])
        .await;

        let app = ctx.create_app().await;

        let payload = json!({
            "text": "This should fail",
            "stream": false
        });
410

411
412
413
414
415
416
        let req = Request::builder()
            .method("POST")
            .uri("/generate")
            .header(CONTENT_TYPE, "application/json")
            .body(Body::from(serde_json::to_string(&payload).unwrap()))
            .unwrap();
417

418
419
        let resp = app.oneshot(req).await.unwrap();
        assert_eq!(resp.status(), StatusCode::INTERNAL_SERVER_ERROR);
420

421
422
        ctx.shutdown().await;
    }
423

424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
    #[tokio::test]
    async fn test_v1_chat_completions_success() {
        let ctx = TestContext::new(vec![MockWorkerConfig {
            port: 18104,
            worker_type: WorkerType::Regular,
            health_status: HealthStatus::Healthy,
            response_delay_ms: 0,
            fail_rate: 0.0,
        }])
        .await;

        let app = ctx.create_app().await;

        let payload = json!({
            "model": "test-model",
            "messages": [
                {"role": "user", "content": "Hello!"}
            ],
            "stream": false
        });
444

445
446
447
448
449
450
        let req = Request::builder()
            .method("POST")
            .uri("/v1/chat/completions")
            .header(CONTENT_TYPE, "application/json")
            .body(Body::from(serde_json::to_string(&payload).unwrap()))
            .unwrap();
451

452
453
        let resp = app.oneshot(req).await.unwrap();
        assert_eq!(resp.status(), StatusCode::OK);
454

455
456
457
458
459
        let body = axum::body::to_bytes(resp.into_body(), usize::MAX)
            .await
            .unwrap();
        let body_json: serde_json::Value = serde_json::from_slice(&body).unwrap();
        assert!(body_json.get("choices").is_some());
460

461
        ctx.shutdown().await;
462
    }
463
}
464

465
466
467
#[cfg(test)]
mod model_info_tests {
    use super::*;
468

469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
    #[tokio::test]
    async fn test_get_server_info() {
        let ctx = TestContext::new(vec![MockWorkerConfig {
            port: 18201,
            worker_type: WorkerType::Regular,
            health_status: HealthStatus::Healthy,
            response_delay_ms: 0,
            fail_rate: 0.0,
        }])
        .await;

        let app = ctx.create_app().await;

        let req = Request::builder()
            .method("GET")
            .uri("/get_server_info")
            .body(Body::empty())
            .unwrap();
487

488
489
        let resp = app.oneshot(req).await.unwrap();
        assert_eq!(resp.status(), StatusCode::OK);
490

491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
        let body = axum::body::to_bytes(resp.into_body(), usize::MAX)
            .await
            .unwrap();
        let body_json: serde_json::Value = serde_json::from_slice(&body).unwrap();
        assert!(body_json.is_object());
        // Check for actual sglang server fields
        assert!(body_json.get("version").is_some());
        assert!(body_json.get("model_path").is_some());
        assert!(body_json.get("tokenizer_path").is_some());
        assert!(body_json.get("port").is_some());
        assert!(body_json.get("max_num_batched_tokens").is_some());
        assert!(body_json.get("schedule_policy").is_some());

        ctx.shutdown().await;
    }
506

507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
    #[tokio::test]
    async fn test_get_model_info() {
        let ctx = TestContext::new(vec![MockWorkerConfig {
            port: 18202,
            worker_type: WorkerType::Regular,
            health_status: HealthStatus::Healthy,
            response_delay_ms: 0,
            fail_rate: 0.0,
        }])
        .await;

        let app = ctx.create_app().await;

        let req = Request::builder()
            .method("GET")
            .uri("/get_model_info")
            .body(Body::empty())
            .unwrap();
525

526
527
        let resp = app.oneshot(req).await.unwrap();
        assert_eq!(resp.status(), StatusCode::OK);
528

529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
        let body = axum::body::to_bytes(resp.into_body(), usize::MAX)
            .await
            .unwrap();
        let body_json: serde_json::Value = serde_json::from_slice(&body).unwrap();
        assert!(body_json.is_object());
        // Check for actual sglang model info fields
        assert_eq!(
            body_json.get("model_path").and_then(|v| v.as_str()),
            Some("mock-model-path")
        );
        assert_eq!(
            body_json.get("tokenizer_path").and_then(|v| v.as_str()),
            Some("mock-tokenizer-path")
        );
        assert_eq!(
            body_json.get("is_generation").and_then(|v| v.as_bool()),
            Some(true)
        );
        assert!(body_json.get("preferred_sampling_params").is_some());

        ctx.shutdown().await;
550
551
    }

552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
    #[tokio::test]
    async fn test_v1_models() {
        let ctx = TestContext::new(vec![MockWorkerConfig {
            port: 18203,
            worker_type: WorkerType::Regular,
            health_status: HealthStatus::Healthy,
            response_delay_ms: 0,
            fail_rate: 0.0,
        }])
        .await;

        let app = ctx.create_app().await;

        let req = Request::builder()
            .method("GET")
            .uri("/v1/models")
            .body(Body::empty())
            .unwrap();
570

571
572
        let resp = app.oneshot(req).await.unwrap();
        assert_eq!(resp.status(), StatusCode::OK);
573

574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
        let body = axum::body::to_bytes(resp.into_body(), usize::MAX)
            .await
            .unwrap();
        let body_json: serde_json::Value = serde_json::from_slice(&body).unwrap();
        assert!(body_json.get("object").is_some());
        assert_eq!(
            body_json.get("object").and_then(|v| v.as_str()),
            Some("list")
        );

        let data = body_json.get("data").and_then(|v| v.as_array());
        assert!(data.is_some());

        let models = data.unwrap();
        assert!(!models.is_empty());

        let first_model = &models[0];
        assert_eq!(
            first_model.get("id").and_then(|v| v.as_str()),
            Some("mock-model")
        );
        assert_eq!(
            first_model.get("object").and_then(|v| v.as_str()),
            Some("model")
        );
        assert!(first_model.get("created").is_some());
        assert_eq!(
            first_model.get("owned_by").and_then(|v| v.as_str()),
            Some("organization-owner")
        );

        ctx.shutdown().await;
    }
607

608
609
610
611
    #[tokio::test]
    async fn test_model_info_with_no_workers() {
        let ctx = TestContext::new(vec![]).await;
        let app = ctx.create_app().await;
612

613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
        let req = Request::builder()
            .method("GET")
            .uri("/get_server_info")
            .body(Body::empty())
            .unwrap();
        let resp = app.clone().oneshot(req).await.unwrap();
        // Router may return various error codes when no workers
        assert!(
            resp.status() == StatusCode::OK
                || resp.status() == StatusCode::SERVICE_UNAVAILABLE
                || resp.status() == StatusCode::NOT_FOUND
                || resp.status() == StatusCode::INTERNAL_SERVER_ERROR,
            "Unexpected status code: {:?}",
            resp.status()
        );

        let req = Request::builder()
            .method("GET")
            .uri("/get_model_info")
            .body(Body::empty())
            .unwrap();
        let resp = app.clone().oneshot(req).await.unwrap();
        // Router may return various error codes when no workers
        assert!(
            resp.status() == StatusCode::OK
                || resp.status() == StatusCode::SERVICE_UNAVAILABLE
                || resp.status() == StatusCode::NOT_FOUND
                || resp.status() == StatusCode::INTERNAL_SERVER_ERROR,
            "Unexpected status code: {:?}",
            resp.status()
        );

        let req = Request::builder()
            .method("GET")
            .uri("/v1/models")
            .body(Body::empty())
            .unwrap();
        let resp = app.oneshot(req).await.unwrap();
        // Router may return various error codes when no workers
        assert!(
            resp.status() == StatusCode::OK
                || resp.status() == StatusCode::SERVICE_UNAVAILABLE
                || resp.status() == StatusCode::NOT_FOUND
                || resp.status() == StatusCode::INTERNAL_SERVER_ERROR,
            "Unexpected status code: {:?}",
            resp.status()
        );

        ctx.shutdown().await;
662
663
    }

664
665
666
667
668
669
670
671
672
673
674
675
    #[tokio::test]
    async fn test_model_info_with_multiple_workers() {
        let ctx = TestContext::new(vec![
            MockWorkerConfig {
                port: 18204,
                worker_type: WorkerType::Regular,
                health_status: HealthStatus::Healthy,
                response_delay_ms: 0,
                fail_rate: 0.0,
            },
            MockWorkerConfig {
                port: 18205,
676
677
678
679
                worker_type: WorkerType::Regular,
                health_status: HealthStatus::Healthy,
                response_delay_ms: 0,
                fail_rate: 0.0,
680
681
682
            },
        ])
        .await;
683

684
        let app = ctx.create_app().await;
685

686
687
688
        for _ in 0..5 {
            let req = Request::builder()
                .method("GET")
689
                .uri("/get_model_info")
690
691
                .body(Body::empty())
                .unwrap();
692

693
            let resp = app.clone().oneshot(req).await.unwrap();
694
695
            assert_eq!(resp.status(), StatusCode::OK);

696
697
698
699
            let body = axum::body::to_bytes(resp.into_body(), usize::MAX)
                .await
                .unwrap();
            let body_json: serde_json::Value = serde_json::from_slice(&body).unwrap();
700
            assert_eq!(
701
                body_json.get("model_path").and_then(|v| v.as_str()),
702
703
                Some("mock-model-path")
            );
704
        }
705

706
        ctx.shutdown().await;
707
708
    }

709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
    #[tokio::test]
    async fn test_model_info_with_unhealthy_worker() {
        let ctx = TestContext::new(vec![MockWorkerConfig {
            port: 18206,
            worker_type: WorkerType::Regular,
            health_status: HealthStatus::Healthy,
            response_delay_ms: 0,
            fail_rate: 1.0, // Always fail
        }])
        .await;

        let app = ctx.create_app().await;

        let req = Request::builder()
            .method("GET")
            .uri("/get_model_info")
            .body(Body::empty())
            .unwrap();
727

728
729
730
731
732
733
734
735
        let resp = app.oneshot(req).await.unwrap();
        // Worker with fail_rate: 1.0 should always return an error status
        assert!(
            resp.status() == StatusCode::INTERNAL_SERVER_ERROR
                || resp.status() == StatusCode::SERVICE_UNAVAILABLE,
            "Expected error status for always-failing worker, got: {:?}",
            resp.status()
        );
736

737
738
739
        ctx.shutdown().await;
    }
}
740

741
742
743
#[cfg(test)]
mod router_policy_tests {
    use super::*;
744

745
746
747
748
749
    #[tokio::test]
    async fn test_random_policy() {
        let ctx = TestContext::new(vec![
            MockWorkerConfig {
                port: 18801,
750
751
752
753
                worker_type: WorkerType::Regular,
                health_status: HealthStatus::Healthy,
                response_delay_ms: 0,
                fail_rate: 0.0,
754
755
756
            },
            MockWorkerConfig {
                port: 18802,
757
                worker_type: WorkerType::Regular,
758
                health_status: HealthStatus::Healthy,
759
760
                response_delay_ms: 0,
                fail_rate: 0.0,
761
762
763
764
765
766
767
768
769
770
771
            },
        ])
        .await;

        // Send multiple requests and verify they succeed
        let app = ctx.create_app().await;

        for i in 0..10 {
            let payload = json!({
                "text": format!("Request {}", i),
                "stream": false
772
773
            });

774
775
776
777
778
779
            let req = Request::builder()
                .method("POST")
                .uri("/generate")
                .header(CONTENT_TYPE, "application/json")
                .body(Body::from(serde_json::to_string(&payload).unwrap()))
                .unwrap();
780

781
782
783
            let resp = app.clone().oneshot(req).await.unwrap();
            assert_eq!(resp.status(), StatusCode::OK);
        }
784

785
786
        ctx.shutdown().await;
    }
787

788
789
790
    #[tokio::test]
    async fn test_worker_selection() {
        let ctx = TestContext::new(vec![MockWorkerConfig {
791
            port: 18207,
792
793
794
795
796
797
798
799
800
801
            worker_type: WorkerType::Regular,
            health_status: HealthStatus::Healthy,
            response_delay_ms: 0,
            fail_rate: 0.0,
        }])
        .await;

        let _payload = json!({
            "text": "Test selection",
            "stream": false
802
        });
803
804

        // Check that router has the worker
805
806
        // TODO: Update test after worker management refactoring
        // For now, skip this check
807
808

        ctx.shutdown().await;
809
810
811
    }
}

812
813
#[cfg(test)]
mod responses_endpoint_tests {
814
    use reqwest::Client as HttpClient;
815

816
817
    use super::*;

818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
    #[tokio::test]
    async fn test_v1_responses_non_streaming() {
        let ctx = TestContext::new(vec![MockWorkerConfig {
            port: 18950,
            worker_type: WorkerType::Regular,
            health_status: HealthStatus::Healthy,
            response_delay_ms: 0,
            fail_rate: 0.0,
        }])
        .await;

        let app = ctx.create_app().await;

        let payload = json!({
            "input": "Hello Responses API",
            "model": "mock-model",
            "stream": false
        });

        let req = Request::builder()
            .method("POST")
            .uri("/v1/responses")
            .header(CONTENT_TYPE, "application/json")
            .body(Body::from(serde_json::to_string(&payload).unwrap()))
            .unwrap();

        let resp = app.clone().oneshot(req).await.unwrap();
        assert_eq!(resp.status(), StatusCode::OK);

        let body = axum::body::to_bytes(resp.into_body(), usize::MAX)
            .await
            .unwrap();
        let body_json: serde_json::Value = serde_json::from_slice(&body).unwrap();
        assert_eq!(body_json["object"], "response");
        assert_eq!(body_json["status"], "completed");

        ctx.shutdown().await;
    }

    #[tokio::test]
    async fn test_v1_responses_streaming() {
        let ctx = TestContext::new(vec![MockWorkerConfig {
            port: 18951,
            worker_type: WorkerType::Regular,
            health_status: HealthStatus::Healthy,
            response_delay_ms: 0,
            fail_rate: 0.0,
        }])
        .await;

        let app = ctx.create_app().await;

        let payload = json!({
            "input": "Hello Responses API",
            "model": "mock-model",
            "stream": true
        });

        let req = Request::builder()
            .method("POST")
            .uri("/v1/responses")
            .header(CONTENT_TYPE, "application/json")
            .body(Body::from(serde_json::to_string(&payload).unwrap()))
            .unwrap();

        let resp = app.clone().oneshot(req).await.unwrap();
        assert_eq!(resp.status(), StatusCode::OK);

        // Check that content-type indicates SSE
        let headers = resp.headers().clone();
        let ct = headers
            .get("content-type")
            .and_then(|v| v.to_str().ok())
            .unwrap_or("");
        assert!(ct.contains("text/event-stream"));

        // We don't fully consume the stream in this test harness.
        ctx.shutdown().await;
    }
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911

    #[tokio::test]
    async fn test_v1_responses_get() {
        let ctx = TestContext::new(vec![MockWorkerConfig {
            port: 18952,
            worker_type: WorkerType::Regular,
            health_status: HealthStatus::Healthy,
            response_delay_ms: 0,
            fail_rate: 0.0,
        }])
        .await;

        let app = ctx.create_app().await;

        // First create a response to obtain an id
912
        let resp_id = "test-get-resp-id-123";
913
914
915
        let payload = json!({
            "input": "Hello Responses API",
            "model": "mock-model",
916
917
918
919
            "stream": false,
            "store": true,
            "background": true,
            "request_id": resp_id
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
        });
        let req = Request::builder()
            .method("POST")
            .uri("/v1/responses")
            .header(CONTENT_TYPE, "application/json")
            .body(Body::from(serde_json::to_string(&payload).unwrap()))
            .unwrap();
        let resp = app.clone().oneshot(req).await.unwrap();
        assert_eq!(resp.status(), StatusCode::OK);

        // Retrieve the response
        let req = Request::builder()
            .method("GET")
            .uri(format!("/v1/responses/{}", resp_id))
            .body(Body::empty())
            .unwrap();
        let resp = app.clone().oneshot(req).await.unwrap();
        assert_eq!(resp.status(), StatusCode::OK);
        let body = axum::body::to_bytes(resp.into_body(), usize::MAX)
            .await
            .unwrap();
        let get_json: serde_json::Value = serde_json::from_slice(&body).unwrap();
        assert_eq!(get_json["object"], "response");

        ctx.shutdown().await;
    }

    #[tokio::test]
    async fn test_v1_responses_cancel() {
        let ctx = TestContext::new(vec![MockWorkerConfig {
            port: 18953,
            worker_type: WorkerType::Regular,
            health_status: HealthStatus::Healthy,
            response_delay_ms: 0,
            fail_rate: 0.0,
        }])
        .await;

        let app = ctx.create_app().await;

        // First create a response to obtain an id
961
        let resp_id = "test-cancel-resp-id-456";
962
963
964
        let payload = json!({
            "input": "Hello Responses API",
            "model": "mock-model",
965
966
967
968
            "stream": false,
            "store": true,
            "background": true,
            "request_id": resp_id
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
1001
1002
1003
1004
1005
1006
1007
1008
1009
1010
1011
1012
1013
1014
1015
1016
1017
1018
1019
1020
1021
1022
1023
1024
1025
1026
1027
1028
1029
1030
1031
1032
1033
1034
1035
1036
1037
1038
1039
1040
1041
1042
1043
1044
1045
1046
1047
1048
1049
1050
1051
1052
1053
1054
1055
1056
1057
1058
1059
1060
1061
1062
1063
1064
1065
1066
1067
1068
1069
1070
1071
1072
1073
1074
1075
1076
1077
1078
1079
1080
1081
1082
1083
1084
        });
        let req = Request::builder()
            .method("POST")
            .uri("/v1/responses")
            .header(CONTENT_TYPE, "application/json")
            .body(Body::from(serde_json::to_string(&payload).unwrap()))
            .unwrap();
        let resp = app.clone().oneshot(req).await.unwrap();
        assert_eq!(resp.status(), StatusCode::OK);

        // Cancel the response
        let req = Request::builder()
            .method("POST")
            .uri(format!("/v1/responses/{}/cancel", resp_id))
            .body(Body::empty())
            .unwrap();
        let resp = app.clone().oneshot(req).await.unwrap();
        assert_eq!(resp.status(), StatusCode::OK);
        let body = axum::body::to_bytes(resp.into_body(), usize::MAX)
            .await
            .unwrap();
        let cancel_json: serde_json::Value = serde_json::from_slice(&body).unwrap();
        assert_eq!(cancel_json["status"], "cancelled");

        ctx.shutdown().await;
    }

    #[tokio::test]
    async fn test_v1_responses_delete_and_list_not_implemented() {
        let ctx = TestContext::new(vec![MockWorkerConfig {
            port: 18954,
            worker_type: WorkerType::Regular,
            health_status: HealthStatus::Healthy,
            response_delay_ms: 0,
            fail_rate: 0.0,
        }])
        .await;

        let app = ctx.create_app().await;

        // Use an arbitrary id for delete/list
        let resp_id = "resp-test-123";

        let req = Request::builder()
            .method("DELETE")
            .uri(format!("/v1/responses/{}", resp_id))
            .body(Body::empty())
            .unwrap();
        let resp = app.clone().oneshot(req).await.unwrap();
        assert_eq!(resp.status(), StatusCode::NOT_IMPLEMENTED);

        let req = Request::builder()
            .method("GET")
            .uri(format!("/v1/responses/{}/input", resp_id))
            .body(Body::empty())
            .unwrap();
        let resp = app.clone().oneshot(req).await.unwrap();
        assert_eq!(resp.status(), StatusCode::NOT_IMPLEMENTED);

        ctx.shutdown().await;
    }

    #[tokio::test]
    async fn test_v1_responses_get_multi_worker_fanout() {
        // Start two mock workers
        let ctx = TestContext::new(vec![
            MockWorkerConfig {
                port: 18960,
                worker_type: WorkerType::Regular,
                health_status: HealthStatus::Healthy,
                response_delay_ms: 0,
                fail_rate: 0.0,
            },
            MockWorkerConfig {
                port: 18961,
                worker_type: WorkerType::Regular,
                health_status: HealthStatus::Healthy,
                response_delay_ms: 0,
                fail_rate: 0.0,
            },
        ])
        .await;

        let app = ctx.create_app().await;

        // Create a background response with a known id
        let rid = format!("resp_{}", 18960); // arbitrary unique id
        let payload = json!({
            "input": "Hello Responses API",
            "model": "mock-model",
            "background": true,
            "store": true,
            "request_id": rid,
        });

        let req = Request::builder()
            .method("POST")
            .uri("/v1/responses")
            .header(CONTENT_TYPE, "application/json")
            .body(Body::from(serde_json::to_string(&payload).unwrap()))
            .unwrap();
        let resp = app.clone().oneshot(req).await.unwrap();
        assert_eq!(resp.status(), StatusCode::OK);

        // Using the router, GET should succeed by fanning out across workers
        let req = Request::builder()
            .method("GET")
            .uri(format!("/v1/responses/{}", rid))
            .body(Body::empty())
            .unwrap();
        let resp = app.clone().oneshot(req).await.unwrap();
        assert_eq!(resp.status(), StatusCode::OK);

        // Validate only one worker holds the metadata: direct calls
        let client = HttpClient::new();
        let mut ok_count = 0usize;
1085
1086
1087
1088
1089
1090
        // Get the actual worker URLs from the context
        let worker_urls: Vec<String> = vec![
            "http://127.0.0.1:18960".to_string(),
            "http://127.0.0.1:18961".to_string(),
        ];
        for url in worker_urls {
1091
1092
1093
1094
1095
1096
1097
1098
1099
1100
            let get_url = format!("{}/v1/responses/{}", url, rid);
            let res = client.get(get_url).send().await.unwrap();
            if res.status() == StatusCode::OK {
                ok_count += 1;
            }
        }
        assert_eq!(ok_count, 1, "exactly one worker should store the response");

        ctx.shutdown().await;
    }
1101
1102
}

1103
1104
1105
1106
#[cfg(test)]
mod error_tests {
    use super::*;

1107
1108
1109
1110
1111
1112
1113
1114
1115
1116
1117
1118
1119
1120
1121
1122
1123
1124
    #[tokio::test]
    async fn test_404_not_found() {
        let ctx = TestContext::new(vec![MockWorkerConfig {
            port: 18401,
            worker_type: WorkerType::Regular,
            health_status: HealthStatus::Healthy,
            response_delay_ms: 0,
            fail_rate: 0.0,
        }])
        .await;

        let app = ctx.create_app().await;

        let req = Request::builder()
            .method("GET")
            .uri("/unknown_endpoint")
            .body(Body::empty())
            .unwrap();
1125

1126
1127
1128
1129
1130
1131
1132
1133
1134
1135
1136
        let resp = app.clone().oneshot(req).await.unwrap();
        assert_eq!(resp.status(), StatusCode::NOT_FOUND);

        let req = Request::builder()
            .method("POST")
            .uri("/api/v2/generate")
            .header(CONTENT_TYPE, "application/json")
            .body(Body::from(
                serde_json::to_string(&json!({"text": "test"})).unwrap(),
            ))
            .unwrap();
1137

1138
1139
        let resp = app.oneshot(req).await.unwrap();
        assert_eq!(resp.status(), StatusCode::NOT_FOUND);
1140

1141
1142
        ctx.shutdown().await;
    }
1143

1144
1145
1146
1147
1148
1149
1150
1151
1152
1153
1154
1155
1156
1157
1158
1159
1160
1161
1162
    #[tokio::test]
    async fn test_method_not_allowed() {
        let ctx = TestContext::new(vec![MockWorkerConfig {
            port: 18402,
            worker_type: WorkerType::Regular,
            health_status: HealthStatus::Healthy,
            response_delay_ms: 0,
            fail_rate: 0.0,
        }])
        .await;

        let app = ctx.create_app().await;

        // GET request to POST-only endpoint
        let req = Request::builder()
            .method("GET")
            .uri("/generate")
            .body(Body::empty())
            .unwrap();
1163

1164
1165
1166
        let resp = app.clone().oneshot(req).await.unwrap();
        // Note: Axum returns 405 for wrong methods on matched routes
        assert_eq!(resp.status(), StatusCode::METHOD_NOT_ALLOWED);
1167

1168
1169
1170
1171
1172
1173
1174
1175
1176
1177
1178
1179
        // POST request to GET-only endpoint
        let req = Request::builder()
            .method("POST")
            .uri("/health")
            .header(CONTENT_TYPE, "application/json")
            .body(Body::from("{}"))
            .unwrap();

        let resp = app.oneshot(req).await.unwrap();
        assert_eq!(resp.status(), StatusCode::METHOD_NOT_ALLOWED);

        ctx.shutdown().await;
1180
1181
    }

1182
1183
1184
1185
    #[tokio::test]
    async fn test_payload_too_large() {
        // Create context with small payload limit
        let config = RouterConfig {
1186
            chat_template: None,
1187
1188
1189
1190
1191
1192
1193
1194
1195
1196
1197
1198
1199
1200
1201
1202
1203
1204
            mode: RoutingMode::Regular {
                worker_urls: vec![],
            },
            policy: PolicyConfig::Random,
            host: "127.0.0.1".to_string(),
            port: 3010,
            max_payload_size: 1024, // 1KB limit
            request_timeout_secs: 600,
            worker_startup_timeout_secs: 1,
            worker_startup_check_interval_secs: 1,
            dp_aware: false,
            api_key: None,
            discovery: None,
            metrics: None,
            log_dir: None,
            log_level: None,
            request_id_headers: None,
            max_concurrent_requests: 64,
1205
1206
1207
            queue_size: 0,
            queue_timeout_secs: 60,
            rate_limit_tokens_per_second: None,
1208
            cors_allowed_origins: vec![],
1209
            retry: RetryConfig::default(),
1210
            circuit_breaker: CircuitBreakerConfig::default(),
1211
1212
            disable_retries: false,
            disable_circuit_breaker: false,
1213
            health_check: sglang_router_rs::config::HealthCheckConfig::default(),
1214
            enable_igw: false,
1215
1216
1217
            connection_mode: ConnectionMode::Http,
            model_path: None,
            tokenizer_path: None,
1218
            history_backend: sglang_router_rs::config::HistoryBackend::Memory,
1219
            oracle: None,
1220
1221
            reasoning_parser: None,
            tool_call_parser: None,
1222
            tokenizer_cache: sglang_router_rs::config::TokenizerCacheConfig::default(),
1223
1224
1225
1226
1227
1228
        };

        let ctx = TestContext::new_with_config(
            config,
            vec![MockWorkerConfig {
                port: 18403,
1229
1230
1231
1232
                worker_type: WorkerType::Regular,
                health_status: HealthStatus::Healthy,
                response_delay_ms: 0,
                fail_rate: 0.0,
1233
1234
1235
            }],
        )
        .await;
1236

1237
1238
1239
        // Note: The server would have payload size middleware configured
        // but we cannot test it directly through the test app
        // This test is kept for documentation purposes
1240

1241
1242
        ctx.shutdown().await;
    }
1243

1244
1245
1246
1247
1248
1249
1250
1251
1252
1253
1254
1255
1256
1257
1258
1259
1260
1261
1262
1263
    #[tokio::test]
    async fn test_invalid_json_payload() {
        let ctx = TestContext::new(vec![MockWorkerConfig {
            port: 18404,
            worker_type: WorkerType::Regular,
            health_status: HealthStatus::Healthy,
            response_delay_ms: 0,
            fail_rate: 0.0,
        }])
        .await;

        let app = ctx.create_app().await;

        // Send invalid JSON
        let req = Request::builder()
            .method("POST")
            .uri("/generate")
            .header(CONTENT_TYPE, "application/json")
            .body(Body::from("{invalid json}"))
            .unwrap();
1264

1265
1266
        let resp = app.clone().oneshot(req).await.unwrap();
        assert_eq!(resp.status(), StatusCode::BAD_REQUEST);
1267

1268
1269
1270
1271
1272
1273
1274
        // Send empty body
        let req = Request::builder()
            .method("POST")
            .uri("/generate")
            .header(CONTENT_TYPE, "application/json")
            .body(Body::empty())
            .unwrap();
1275

1276
1277
        let resp = app.oneshot(req).await.unwrap();
        assert_eq!(resp.status(), StatusCode::BAD_REQUEST);
1278

1279
1280
        ctx.shutdown().await;
    }
1281

1282
1283
1284
1285
1286
1287
1288
1289
1290
1291
1292
1293
1294
1295
1296
1297
1298
1299
    #[tokio::test]
    async fn test_invalid_model() {
        let ctx = TestContext::new(vec![MockWorkerConfig {
            port: 18406,
            worker_type: WorkerType::Regular,
            health_status: HealthStatus::Healthy,
            response_delay_ms: 0,
            fail_rate: 0.0,
        }])
        .await;

        let app = ctx.create_app().await;

        let payload = json!({
            "model": "invalid-model-name-that-does-not-exist",
            "messages": [{"role": "user", "content": "Hello"}],
            "stream": false
        });
1300

1301
1302
1303
1304
1305
1306
        let req = Request::builder()
            .method("POST")
            .uri("/v1/chat/completions")
            .header(CONTENT_TYPE, "application/json")
            .body(Body::from(serde_json::to_string(&payload).unwrap()))
            .unwrap();
1307

1308
1309
1310
        let resp = app.oneshot(req).await.unwrap();
        // Mock worker accepts any model, but real implementation might return 400
        assert!(resp.status().is_success() || resp.status() == StatusCode::BAD_REQUEST);
1311

1312
1313
1314
        ctx.shutdown().await;
    }
}
1315

1316
1317
1318
#[cfg(test)]
mod cache_tests {
    use super::*;
1319

1320
1321
1322
1323
1324
1325
1326
1327
1328
1329
1330
1331
1332
1333
1334
1335
1336
1337
    #[tokio::test]
    async fn test_flush_cache() {
        let ctx = TestContext::new(vec![MockWorkerConfig {
            port: 18501,
            worker_type: WorkerType::Regular,
            health_status: HealthStatus::Healthy,
            response_delay_ms: 0,
            fail_rate: 0.0,
        }])
        .await;

        let app = ctx.create_app().await;

        let req = Request::builder()
            .method("POST")
            .uri("/flush_cache")
            .body(Body::empty())
            .unwrap();
1338

1339
1340
1341
1342
1343
1344
1345
1346
1347
1348
1349
1350
1351
1352
1353
1354
        let resp = app.oneshot(req).await.unwrap();
        assert_eq!(resp.status(), StatusCode::OK);

        // The response might be empty or contain a message
        let body_bytes = axum::body::to_bytes(resp.into_body(), usize::MAX)
            .await
            .unwrap();
        if !body_bytes.is_empty() {
            if let Ok(body) = serde_json::from_slice::<serde_json::Value>(&body_bytes) {
                // Check that we got a successful response with expected fields
                assert!(body.is_object());
                assert!(body.get("message").is_some() || body.get("status").is_some());
            }
        }

        ctx.shutdown().await;
1355
1356
    }

1357
1358
1359
1360
1361
1362
1363
1364
1365
1366
1367
1368
    #[tokio::test]
    async fn test_get_loads() {
        let ctx = TestContext::new(vec![
            MockWorkerConfig {
                port: 18502,
                worker_type: WorkerType::Regular,
                health_status: HealthStatus::Healthy,
                response_delay_ms: 0,
                fail_rate: 0.0,
            },
            MockWorkerConfig {
                port: 18503,
1369
1370
1371
1372
                worker_type: WorkerType::Regular,
                health_status: HealthStatus::Healthy,
                response_delay_ms: 0,
                fail_rate: 0.0,
1373
1374
1375
            },
        ])
        .await;
1376

1377
        let app = ctx.create_app().await;
1378

1379
1380
1381
1382
1383
        let req = Request::builder()
            .method("GET")
            .uri("/get_loads")
            .body(Body::empty())
            .unwrap();
1384

1385
1386
        let resp = app.oneshot(req).await.unwrap();
        assert_eq!(resp.status(), StatusCode::OK);
1387

1388
1389
1390
1391
        let body = axum::body::to_bytes(resp.into_body(), usize::MAX)
            .await
            .unwrap();
        let body_json: serde_json::Value = serde_json::from_slice(&body).unwrap();
1392

1393
1394
1395
        assert!(body_json.is_object());
        // The exact structure depends on the implementation
        // but should contain worker load information
1396

1397
1398
        ctx.shutdown().await;
    }
1399

1400
1401
1402
    #[tokio::test]
    async fn test_flush_cache_no_workers() {
        let ctx = TestContext::new(vec![]).await;
1403

1404
        let app = ctx.create_app().await;
1405

1406
1407
1408
1409
1410
        let req = Request::builder()
            .method("POST")
            .uri("/flush_cache")
            .body(Body::empty())
            .unwrap();
1411

1412
1413
1414
1415
1416
        let resp = app.oneshot(req).await.unwrap();
        // Should either succeed (no-op) or return service unavailable
        assert!(
            resp.status() == StatusCode::OK || resp.status() == StatusCode::SERVICE_UNAVAILABLE
        );
1417

1418
        ctx.shutdown().await;
1419
1420
1421
1422
    }
}

#[cfg(test)]
1423
mod load_balancing_tests {
1424
1425
    use super::*;

1426
1427
1428
1429
1430
1431
    #[tokio::test]
    async fn test_request_distribution() {
        // Create multiple workers
        let ctx = TestContext::new(vec![
            MockWorkerConfig {
                port: 18601,
1432
1433
1434
1435
                worker_type: WorkerType::Regular,
                health_status: HealthStatus::Healthy,
                response_delay_ms: 0,
                fail_rate: 0.0,
1436
1437
1438
1439
1440
1441
1442
1443
1444
1445
            },
            MockWorkerConfig {
                port: 18602,
                worker_type: WorkerType::Regular,
                health_status: HealthStatus::Healthy,
                response_delay_ms: 0,
                fail_rate: 0.0,
            },
        ])
        .await;
1446

1447
        let app = ctx.create_app().await;
1448

1449
1450
1451
1452
1453
1454
1455
        // Send multiple requests and track distribution
        let mut request_count = 0;
        for i in 0..10 {
            let payload = json!({
                "text": format!("Request {}", i),
                "stream": false
            });
1456

1457
1458
1459
1460
1461
1462
            let req = Request::builder()
                .method("POST")
                .uri("/generate")
                .header(CONTENT_TYPE, "application/json")
                .body(Body::from(serde_json::to_string(&payload).unwrap()))
                .unwrap();
1463

1464
1465
1466
            let resp = app.clone().oneshot(req).await.unwrap();
            if resp.status() == StatusCode::OK {
                request_count += 1;
1467
            }
1468
        }
1469

1470
1471
        // With random policy, all requests should succeed
        assert_eq!(request_count, 10);
1472

1473
1474
1475
        ctx.shutdown().await;
    }
}
1476

1477
1478
1479
#[cfg(test)]
mod pd_mode_tests {
    use super::*;
1480

1481
1482
1483
1484
1485
1486
1487
1488
1489
1490
    #[tokio::test]
    async fn test_pd_mode_routing() {
        // Create PD mode configuration with prefill and decode workers
        let mut prefill_worker = MockWorker::new(MockWorkerConfig {
            port: 18701,
            worker_type: WorkerType::Prefill,
            health_status: HealthStatus::Healthy,
            response_delay_ms: 0,
            fail_rate: 0.0,
        });
1491

1492
1493
1494
1495
1496
1497
        let mut decode_worker = MockWorker::new(MockWorkerConfig {
            port: 18702,
            worker_type: WorkerType::Decode,
            health_status: HealthStatus::Healthy,
            response_delay_ms: 0,
            fail_rate: 0.0,
1498
1499
        });

1500
1501
        let prefill_url = prefill_worker.start().await.unwrap();
        let decode_url = decode_worker.start().await.unwrap();
1502

1503
1504
1505
1506
1507
        tokio::time::sleep(tokio::time::Duration::from_millis(200)).await;

        // Extract port from prefill URL
        let prefill_port = prefill_url
            .split(':')
1508
            .next_back()
1509
1510
1511
1512
            .and_then(|p| p.trim_end_matches('/').parse::<u16>().ok())
            .unwrap_or(9000);

        let config = RouterConfig {
1513
            chat_template: None,
1514
1515
1516
1517
1518
1519
1520
1521
1522
1523
1524
1525
1526
1527
1528
1529
1530
1531
1532
1533
1534
            mode: RoutingMode::PrefillDecode {
                prefill_urls: vec![(prefill_url, Some(prefill_port))],
                decode_urls: vec![decode_url],
                prefill_policy: None,
                decode_policy: None,
            },
            policy: PolicyConfig::Random,
            host: "127.0.0.1".to_string(),
            port: 3011,
            max_payload_size: 256 * 1024 * 1024,
            request_timeout_secs: 600,
            worker_startup_timeout_secs: 1,
            worker_startup_check_interval_secs: 1,
            discovery: None,
            metrics: None,
            log_dir: None,
            dp_aware: false,
            api_key: None,
            log_level: None,
            request_id_headers: None,
            max_concurrent_requests: 64,
1535
1536
1537
            queue_size: 0,
            queue_timeout_secs: 60,
            rate_limit_tokens_per_second: None,
1538
            cors_allowed_origins: vec![],
1539
            retry: RetryConfig::default(),
1540
            circuit_breaker: CircuitBreakerConfig::default(),
1541
1542
            disable_retries: false,
            disable_circuit_breaker: false,
1543
            health_check: sglang_router_rs::config::HealthCheckConfig::default(),
1544
            enable_igw: false,
1545
1546
1547
            connection_mode: ConnectionMode::Http,
            model_path: None,
            tokenizer_path: None,
1548
            history_backend: sglang_router_rs::config::HistoryBackend::Memory,
1549
            oracle: None,
1550
1551
            reasoning_parser: None,
            tool_call_parser: None,
1552
            tokenizer_cache: sglang_router_rs::config::TokenizerCacheConfig::default(),
1553
1554
        };

1555
1556
1557
        // Create app context
        let app_context = common::create_test_context(config);

1558
        // Create router - this might fail due to health check issues
1559
        let router_result = RouterFactory::create_router(&app_context).await;
1560
1561
1562
1563
1564
1565
1566

        // Clean up workers
        prefill_worker.stop().await;
        decode_worker.stop().await;

        // For now, just verify the configuration was attempted
        assert!(router_result.is_err() || router_result.is_ok());
1567
1568
1569
1570
    }
}

#[cfg(test)]
1571
mod request_id_tests {
1572
1573
    use super::*;

1574
1575
1576
1577
1578
1579
1580
1581
1582
1583
1584
1585
1586
1587
1588
1589
1590
    #[tokio::test]
    async fn test_request_id_generation() {
        let ctx = TestContext::new(vec![MockWorkerConfig {
            port: 18901,
            worker_type: WorkerType::Regular,
            health_status: HealthStatus::Healthy,
            response_delay_ms: 0,
            fail_rate: 0.0,
        }])
        .await;

        let app = ctx.create_app().await;

        let payload = json!({
            "text": "Test request",
            "stream": false
        });
1591

1592
1593
1594
1595
1596
1597
1598
1599
1600
1601
1602
1603
1604
1605
1606
1607
1608
1609
1610
1611
1612
1613
1614
1615
1616
1617
1618
1619
1620
1621
1622
1623
1624
1625
1626
1627
1628
1629
        let req = Request::builder()
            .method("POST")
            .uri("/generate")
            .header(CONTENT_TYPE, "application/json")
            .body(Body::from(serde_json::to_string(&payload).unwrap()))
            .unwrap();

        let resp = app.clone().oneshot(req).await.unwrap();
        assert_eq!(resp.status(), StatusCode::OK);

        // Check that response has x-request-id header
        let request_id = resp.headers().get("x-request-id");
        assert!(
            request_id.is_some(),
            "Response should have x-request-id header"
        );

        let id_value = request_id.unwrap().to_str().unwrap();
        assert!(
            id_value.starts_with("gnt-"),
            "Generate endpoint should have gnt- prefix"
        );
        assert!(
            id_value.len() > 4,
            "Request ID should have content after prefix"
        );

        let custom_id = "custom-request-id-123";
        let req = Request::builder()
            .method("POST")
            .uri("/generate")
            .header(CONTENT_TYPE, "application/json")
            .header("x-request-id", custom_id)
            .body(Body::from(serde_json::to_string(&payload).unwrap()))
            .unwrap();

        let resp = app.clone().oneshot(req).await.unwrap();
        assert_eq!(resp.status(), StatusCode::OK);
1630

1631
1632
1633
1634
1635
1636
1637
        let response_id = resp.headers().get("x-request-id");
        assert!(response_id.is_some());
        assert_eq!(response_id.unwrap(), custom_id);

        let chat_payload = json!({
            "messages": [{"role": "user", "content": "Hello"}],
            "model": "test-model"
1638
        });
1639
1640
1641
1642
1643
1644
1645
1646
1647
1648
1649
1650
1651
1652
1653
1654
1655
1656
1657
1658
1659
1660
1661
1662
1663
1664
1665
1666
1667
1668
1669
1670
1671
1672
1673

        let req = Request::builder()
            .method("POST")
            .uri("/v1/chat/completions")
            .header(CONTENT_TYPE, "application/json")
            .body(Body::from(serde_json::to_string(&chat_payload).unwrap()))
            .unwrap();

        let resp = app.clone().oneshot(req).await.unwrap();
        assert_eq!(resp.status(), StatusCode::OK);

        let request_id = resp.headers().get("x-request-id");
        assert!(request_id.is_some());
        assert!(request_id
            .unwrap()
            .to_str()
            .unwrap()
            .starts_with("chatcmpl-"));

        let req = Request::builder()
            .method("POST")
            .uri("/generate")
            .header(CONTENT_TYPE, "application/json")
            .header("x-correlation-id", "correlation-123")
            .body(Body::from(serde_json::to_string(&payload).unwrap()))
            .unwrap();

        let resp = app.clone().oneshot(req).await.unwrap();
        assert_eq!(resp.status(), StatusCode::OK);

        let response_id = resp.headers().get("x-request-id");
        assert!(response_id.is_some());
        assert_eq!(response_id.unwrap(), "correlation-123");

        ctx.shutdown().await;
1674
1675
    }

1676
1677
1678
1679
    #[tokio::test]
    async fn test_request_id_with_custom_headers() {
        // Create config with custom request ID headers
        let config = RouterConfig {
1680
            chat_template: None,
1681
1682
1683
1684
1685
1686
1687
1688
1689
1690
1691
1692
1693
1694
1695
1696
1697
1698
            mode: RoutingMode::Regular {
                worker_urls: vec![],
            },
            policy: PolicyConfig::Random,
            host: "127.0.0.1".to_string(),
            port: 3002,
            max_payload_size: 256 * 1024 * 1024,
            request_timeout_secs: 600,
            worker_startup_timeout_secs: 1,
            worker_startup_check_interval_secs: 1,
            discovery: None,
            metrics: None,
            dp_aware: false,
            api_key: None,
            log_dir: None,
            log_level: None,
            request_id_headers: Some(vec!["custom-id".to_string(), "trace-id".to_string()]),
            max_concurrent_requests: 64,
1699
1700
1701
            queue_size: 0,
            queue_timeout_secs: 60,
            rate_limit_tokens_per_second: None,
1702
            cors_allowed_origins: vec![],
1703
            retry: RetryConfig::default(),
1704
            circuit_breaker: CircuitBreakerConfig::default(),
1705
1706
            disable_retries: false,
            disable_circuit_breaker: false,
1707
            health_check: sglang_router_rs::config::HealthCheckConfig::default(),
1708
            enable_igw: false,
1709
1710
1711
            connection_mode: ConnectionMode::Http,
            model_path: None,
            tokenizer_path: None,
1712
            history_backend: sglang_router_rs::config::HistoryBackend::Memory,
1713
            oracle: None,
1714
1715
            reasoning_parser: None,
            tool_call_parser: None,
1716
            tokenizer_cache: sglang_router_rs::config::TokenizerCacheConfig::default(),
1717
        };
1718

1719
1720
1721
1722
1723
        let ctx = TestContext::new_with_config(
            config,
            vec![MockWorkerConfig {
                port: 18902,
                worker_type: WorkerType::Regular,
1724
1725
1726
                health_status: HealthStatus::Healthy,
                response_delay_ms: 0,
                fail_rate: 0.0,
1727
1728
1729
            }],
        )
        .await;
1730

1731
        let app = ctx.create_app().await;
1732

1733
1734
1735
1736
        let payload = json!({
            "text": "Test request",
            "stream": false
        });
1737

1738
1739
1740
1741
1742
1743
1744
        let req = Request::builder()
            .method("POST")
            .uri("/generate")
            .header(CONTENT_TYPE, "application/json")
            .header("custom-id", "my-custom-id")
            .body(Body::from(serde_json::to_string(&payload).unwrap()))
            .unwrap();
1745

1746
1747
1748
1749
1750
1751
1752
1753
        let resp = app.clone().oneshot(req).await.unwrap();
        assert_eq!(resp.status(), StatusCode::OK);

        let response_id = resp.headers().get("x-request-id");
        assert!(response_id.is_some());
        assert_eq!(response_id.unwrap(), "my-custom-id");

        ctx.shutdown().await;
1754
1755
    }
}
1756
1757
1758
1759
1760
1761
1762
1763
1764
1765
1766
1767
1768
1769
1770
1771
1772
1773
1774
1775
1776
1777
1778
1779
1780
1781
1782
1783
1784
1785
1786
1787
1788
1789
1790
1791
1792
1793
1794
1795
1796
1797
1798
1799
1800
1801
1802
1803
1804
1805
1806
1807
1808
1809
1810
1811
1812
1813
1814
1815
1816
1817
1818
1819
1820
1821
1822
1823
1824
1825
1826
1827
1828
1829
1830
1831
1832
1833
1834
1835
1836
1837
1838
1839
1840
1841
1842
1843
1844
1845
1846
1847
1848
1849
1850
1851
1852
1853
1854
1855
1856
1857
1858
1859
1860
1861
1862
1863
1864
1865
1866
1867
1868
1869
1870
1871
1872
1873
1874
1875
1876
1877
1878
1879
1880
1881
1882
1883
1884
1885
1886
1887
1888
1889
1890
1891
1892
1893
1894
1895
1896
1897
1898
1899
1900
1901
1902
1903
1904
1905
1906
1907
1908
1909
1910
1911
1912
1913
1914
1915
1916
1917
1918
1919
1920
1921
1922
1923
1924
1925
1926
1927
1928
1929
1930
1931
1932
1933
1934
1935
1936
1937
1938
1939
1940
1941
1942
1943
1944
1945
1946
1947
1948
1949
1950
1951
1952
1953
1954
1955
1956
1957
1958
1959
1960
1961
1962
1963
1964
1965
1966
1967
1968
1969
1970
1971
1972
1973
1974
1975
1976
1977
1978

#[cfg(test)]
mod rerank_tests {
    use super::*;
    // Note: RerankRequest and RerankResult are available for future use

    #[tokio::test]
    async fn test_rerank_success() {
        let ctx = TestContext::new(vec![MockWorkerConfig {
            port: 18105,
            worker_type: WorkerType::Regular,
            health_status: HealthStatus::Healthy,
            response_delay_ms: 0,
            fail_rate: 0.0,
        }])
        .await;

        let app = ctx.create_app().await;

        let payload = json!({
            "query": "machine learning algorithms",
            "documents": [
                "Introduction to machine learning concepts",
                "Deep learning neural networks tutorial"
            ],
            "model": "test-rerank-model",
            "top_k": 2,
            "return_documents": true,
            "rid": "test-request-123"
        });

        let req = Request::builder()
            .method("POST")
            .uri("/rerank")
            .header(CONTENT_TYPE, "application/json")
            .body(Body::from(serde_json::to_string(&payload).unwrap()))
            .unwrap();

        let resp = app.oneshot(req).await.unwrap();
        assert_eq!(resp.status(), StatusCode::OK);

        let body = axum::body::to_bytes(resp.into_body(), usize::MAX)
            .await
            .unwrap();
        let body_json: serde_json::Value = serde_json::from_slice(&body).unwrap();

        assert!(body_json.get("results").is_some());
        assert!(body_json.get("model").is_some());
        assert_eq!(body_json["model"], "test-rerank-model");

        let results = body_json["results"].as_array().unwrap();
        assert_eq!(results.len(), 2);

        assert!(results[0]["score"].as_f64().unwrap() >= results[1]["score"].as_f64().unwrap());

        ctx.shutdown().await;
    }

    #[tokio::test]
    async fn test_rerank_with_top_k() {
        let ctx = TestContext::new(vec![MockWorkerConfig {
            port: 18106,
            worker_type: WorkerType::Regular,
            health_status: HealthStatus::Healthy,
            response_delay_ms: 0,
            fail_rate: 0.0,
        }])
        .await;

        let app = ctx.create_app().await;

        let payload = json!({
            "query": "test query",
            "documents": [
                "Document 1",
                "Document 2",
                "Document 3"
            ],
            "model": "test-model",
            "top_k": 1,
            "return_documents": true
        });

        let req = Request::builder()
            .method("POST")
            .uri("/rerank")
            .header(CONTENT_TYPE, "application/json")
            .body(Body::from(serde_json::to_string(&payload).unwrap()))
            .unwrap();

        let resp = app.oneshot(req).await.unwrap();
        assert_eq!(resp.status(), StatusCode::OK);

        let body = axum::body::to_bytes(resp.into_body(), usize::MAX)
            .await
            .unwrap();
        let body_json: serde_json::Value = serde_json::from_slice(&body).unwrap();

        // Should only return top_k results
        let results = body_json["results"].as_array().unwrap();
        assert_eq!(results.len(), 1);

        ctx.shutdown().await;
    }

    #[tokio::test]
    async fn test_rerank_without_documents() {
        let ctx = TestContext::new(vec![MockWorkerConfig {
            port: 18107,
            worker_type: WorkerType::Regular,
            health_status: HealthStatus::Healthy,
            response_delay_ms: 0,
            fail_rate: 0.0,
        }])
        .await;

        let app = ctx.create_app().await;

        let payload = json!({
            "query": "test query",
            "documents": ["Document 1", "Document 2"],
            "model": "test-model",
            "return_documents": false
        });

        let req = Request::builder()
            .method("POST")
            .uri("/rerank")
            .header(CONTENT_TYPE, "application/json")
            .body(Body::from(serde_json::to_string(&payload).unwrap()))
            .unwrap();

        let resp = app.oneshot(req).await.unwrap();
        assert_eq!(resp.status(), StatusCode::OK);

        let body = axum::body::to_bytes(resp.into_body(), usize::MAX)
            .await
            .unwrap();
        let body_json: serde_json::Value = serde_json::from_slice(&body).unwrap();

        // Documents should be null when return_documents is false
        let results = body_json["results"].as_array().unwrap();
        for result in results {
            assert!(result.get("document").is_none());
        }

        ctx.shutdown().await;
    }

    #[tokio::test]
    async fn test_rerank_worker_failure() {
        let ctx = TestContext::new(vec![MockWorkerConfig {
            port: 18108,
            worker_type: WorkerType::Regular,
            health_status: HealthStatus::Healthy,
            response_delay_ms: 0,
            fail_rate: 1.0, // Always fail
        }])
        .await;

        let app = ctx.create_app().await;

        let payload = json!({
            "query": "test query",
            "documents": ["Document 1"],
            "model": "test-model"
        });

        let req = Request::builder()
            .method("POST")
            .uri("/rerank")
            .header(CONTENT_TYPE, "application/json")
            .body(Body::from(serde_json::to_string(&payload).unwrap()))
            .unwrap();

        let resp = app.oneshot(req).await.unwrap();
        // Should return the worker's error response
        assert_eq!(resp.status(), StatusCode::INTERNAL_SERVER_ERROR);

        ctx.shutdown().await;
    }

    #[tokio::test]
    async fn test_v1_rerank_compatibility() {
        let ctx = TestContext::new(vec![MockWorkerConfig {
            port: 18110,
            worker_type: WorkerType::Regular,
            health_status: HealthStatus::Healthy,
            response_delay_ms: 0,
            fail_rate: 0.0,
        }])
        .await;

        let app = ctx.create_app().await;

        let payload = json!({
            "query": "machine learning algorithms",
            "documents": [
                "Introduction to machine learning concepts",
                "Deep learning neural networks tutorial",
                "Statistical learning theory basics"
            ]
        });

        let req = Request::builder()
            .method("POST")
            .uri("/v1/rerank")
            .header(CONTENT_TYPE, "application/json")
            .body(Body::from(serde_json::to_string(&payload).unwrap()))
            .unwrap();

        let resp = app.oneshot(req).await.unwrap();
        assert_eq!(resp.status(), StatusCode::OK);

        let body = axum::body::to_bytes(resp.into_body(), usize::MAX)
            .await
            .unwrap();
        let body_json: serde_json::Value = serde_json::from_slice(&body).unwrap();

        assert!(body_json.get("results").is_some());
        assert!(body_json.get("model").is_some());

        // V1 API should use default model name
1979
        assert_eq!(body_json["model"], "unknown");
1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
2018
2019
2020
2021
2022
2023
2024
2025
2026
2027
2028
2029
2030
2031
2032
2033
2034
2035
2036
2037
2038
2039
2040
2041
2042
2043
2044
2045
2046
2047
2048
2049
2050
2051
2052
2053
2054
2055
2056
2057
2058
2059
2060
2061
2062
2063
2064
2065
2066
2067
2068
2069
2070
2071
2072
2073
2074
2075

        let results = body_json["results"].as_array().unwrap();
        assert_eq!(results.len(), 3); // All documents should be returned

        assert!(results[0]["score"].as_f64().unwrap() >= results[1]["score"].as_f64().unwrap());
        assert!(results[1]["score"].as_f64().unwrap() >= results[2]["score"].as_f64().unwrap());

        // V1 API should return documents by default
        for result in results {
            assert!(result.get("document").is_some());
        }

        ctx.shutdown().await;
    }

    #[tokio::test]
    async fn test_rerank_invalid_request() {
        let ctx = TestContext::new(vec![MockWorkerConfig {
            port: 18111,
            worker_type: WorkerType::Regular,
            health_status: HealthStatus::Healthy,
            response_delay_ms: 0,
            fail_rate: 0.0,
        }])
        .await;

        let app = ctx.create_app().await;

        let payload = json!({
            "query": "",
            "documents": ["Document 1", "Document 2"],
            "model": "test-model"
        });

        let req = Request::builder()
            .method("POST")
            .uri("/rerank")
            .header(CONTENT_TYPE, "application/json")
            .body(Body::from(serde_json::to_string(&payload).unwrap()))
            .unwrap();

        let resp = app.clone().oneshot(req).await.unwrap();
        assert_eq!(resp.status(), StatusCode::BAD_REQUEST);

        let payload = json!({
            "query": "   ",
            "documents": ["Document 1", "Document 2"],
            "model": "test-model"
        });

        let req = Request::builder()
            .method("POST")
            .uri("/rerank")
            .header(CONTENT_TYPE, "application/json")
            .body(Body::from(serde_json::to_string(&payload).unwrap()))
            .unwrap();

        let resp = app.clone().oneshot(req).await.unwrap();
        assert_eq!(resp.status(), StatusCode::BAD_REQUEST);

        let payload = json!({
            "query": "test query",
            "documents": [],
            "model": "test-model"
        });

        let req = Request::builder()
            .method("POST")
            .uri("/rerank")
            .header(CONTENT_TYPE, "application/json")
            .body(Body::from(serde_json::to_string(&payload).unwrap()))
            .unwrap();

        let resp = app.clone().oneshot(req).await.unwrap();
        assert_eq!(resp.status(), StatusCode::BAD_REQUEST);

        let payload = json!({
            "query": "test query",
            "documents": ["Document 1", "Document 2"],
            "model": "test-model",
            "top_k": 0
        });

        let req = Request::builder()
            .method("POST")
            .uri("/rerank")
            .header(CONTENT_TYPE, "application/json")
            .body(Body::from(serde_json::to_string(&payload).unwrap()))
            .unwrap();

        let resp = app.oneshot(req).await.unwrap();
        assert_eq!(resp.status(), StatusCode::BAD_REQUEST);

        ctx.shutdown().await;
    }
}