dwctl 8.62.0

The Doubleword Control Layer - A self-hostable observability and analytics platform for LLM applications
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
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
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201

//! End-to-end tests for the Anthropic `/v1/messages` ingress.
//!
//! These exercise the FULL real stack against a wiremock upstream: the
//! `/messages` route (onwards' strict alias), the edge translation middleware
//! (request -> Chat Completions, response/stream -> Anthropic), onwards routing
//! and sanitization, and the DB-driven model/key config. Unlike the unit tests
//! in `inference::translation`, onwards is the real (locally patched) crate here.

use crate::api::models::users::Role;
use crate::test::utils::{add_auth_headers, create_test_admin_user, create_test_config, create_test_user};
use axum_test::TestServer;
use sqlx::PgPool;

/// Seed a routable `gpt-4` deployment pointing at the wiremock upstream plus an
/// inference API key, and return the key secret. Caller must then sync onwards.
async fn seed_model_and_key(server: &TestServer, pool: &PgPool, mock_uri: &str) -> String {
    let admin = create_test_admin_user(pool, Role::PlatformManager).await;
    let h = add_auth_headers(&admin);

    let endpoint: serde_json::Value = server
        .post("/admin/api/v1/endpoints")
        .add_header(&h[0].0, &h[0].1)
        .add_header(&h[1].0, &h[1].1)
        .json(&serde_json::json!({ "name": "mock", "url": mock_uri, "description": "anthropic e2e", "auto_sync_models": false }))
        .await
        .json();
    let endpoint_id = endpoint["id"].as_str().unwrap();

    let model: serde_json::Value = server
        .post("/admin/api/v1/models")
        .add_header(&h[0].0, &h[0].1)
        .add_header(&h[1].0, &h[1].1)
        .json(&serde_json::json!({ "type": "standard", "model_name": "gpt-4", "alias": "gpt-4", "hosted_on": endpoint_id }))
        .await
        .json();
    let deployment_id = model["id"].as_str().unwrap();

    // Public group (all-zeros UUID) makes the model accessible to all users.
    let group_id = "00000000-0000-0000-0000-000000000000";
    let assoc = server
        .post(&format!("/admin/api/v1/groups/{group_id}/models/{deployment_id}"))
        .add_header(&h[0].0, &h[0].1)
        .add_header(&h[1].0, &h[1].1)
        .await;
    assert!(
        assoc.status_code().is_success(),
        "model-group association failed: {} {}",
        assoc.status_code(),
        assoc.text()
    );

    let user = create_test_user(pool, Role::StandardUser).await;
    let grant = server
        .post("/admin/api/v1/transactions")
        .add_header(&h[0].0, &h[0].1)
        .add_header(&h[1].0, &h[1].1)
        .json(&serde_json::json!({ "user_id": user.id, "transaction_type": "admin_grant", "amount": 1000, "source_id": admin.id }))
        .await;
    assert!(
        grant.status_code().is_success(),
        "credits grant failed: {} {}",
        grant.status_code(),
        grant.text()
    );

    let key: serde_json::Value = server
        .post(&format!("/admin/api/v1/users/{}/api-keys", user.id))
        .add_header(&h[0].0, &h[0].1)
        .add_header(&h[1].0, &h[1].1)
        .json(&serde_json::json!({ "purpose": "realtime", "name": "anthropic e2e key" }))
        .await
        .json();
    key["key"].as_str().unwrap().to_string()
}

/// Poll `/ai/v1/models` until onwards has picked up `gpt-4`.
async fn wait_for_model(server: &TestServer, api_key: &str) {
    let start = std::time::Instant::now();
    let mut ok = false;
    while !ok && start.elapsed() < std::time::Duration::from_secs(3) {
        let r = server
            .get("/ai/v1/models")
            .add_header("Authorization", &format!("Bearer {api_key}"))
            .await;
        if r.status_code() == 200 {
            let m: serde_json::Value = r.json();
            if let Some(d) = m["data"].as_array() {
                ok = d.iter().any(|x| x["id"].as_str() == Some("gpt-4"));
            }
        }
        if !ok {
            tokio::time::sleep(std::time::Duration::from_millis(1)).await;
        }
    }
    assert!(ok, "onwards did not pick up gpt-4 within 3s");
}

#[sqlx::test]
async fn anthropic_messages_blocking_end_to_end(pool: PgPool) {
    let mock = wiremock::MockServer::start().await;
    wiremock::Mock::given(wiremock::matchers::method("POST"))
        .and(wiremock::matchers::path("/chat/completions"))
        .respond_with(wiremock::ResponseTemplate::new(200).set_body_json(serde_json::json!({
            "id": "chatcmpl-1", "object": "chat.completion", "created": 1, "model": "gpt-4",
            "choices": [ { "index": 0, "message": { "role": "assistant", "content": "Hello from the gateway" }, "finish_reason": "stop" } ],
            "usage": { "prompt_tokens": 10, "completion_tokens": 4, "total_tokens": 14 }
        })))
        .mount(&mock)
        .await;

    let mut config = create_test_config();
    config.onwards.strict_mode = true;
    config.background_services.onwards_sync.enabled = true;
    let app = crate::Application::new_with_pool(config, Some(pool.clone()), None)
        .await
        .expect("app");
    let (server, bg) = app.into_test_server();

    let api_key = seed_model_and_key(&server, &pool, &mock.uri()).await;
    bg.sync_onwards_config(&pool).await.unwrap();
    wait_for_model(&server, &api_key).await;

    // Anthropic request: x-api-key (not Bearer) + anthropic-version + system + message.
    let resp = server
        .post("/ai/v1/messages")
        .add_header("x-api-key", &api_key)
        .add_header("anthropic-version", "2023-06-01")
        .json(&serde_json::json!({
            "model": "gpt-4",
            "max_tokens": 64,
            "system": "be terse",
            "messages": [ { "role": "user", "content": "hi" } ]
        }))
        .await;

    let status = resp.status_code();
    let text = resp.text();
    assert_eq!(status, 200, "got {status}: {text}");
    let body: serde_json::Value = serde_json::from_str(&text).unwrap();
    assert_eq!(body["type"], "message");
    assert_eq!(body["role"], "assistant");
    assert_eq!(body["content"][0]["type"], "text");
    assert_eq!(body["content"][0]["text"], "Hello from the gateway");
    assert_eq!(body["stop_reason"], "end_turn");
    assert_eq!(body["usage"]["input_tokens"], 10);
    assert_eq!(body["usage"]["output_tokens"], 4);
}

#[sqlx::test]
async fn anthropic_messages_streaming_end_to_end(pool: PgPool) {
    let sse = concat!(
        "data: {\"id\":\"c1\",\"object\":\"chat.completion.chunk\",\"model\":\"gpt-4\",\"choices\":[{\"index\":0,\"delta\":{\"role\":\"assistant\"},\"finish_reason\":null}]}\n\n",
        "data: {\"id\":\"c1\",\"object\":\"chat.completion.chunk\",\"model\":\"gpt-4\",\"choices\":[{\"index\":0,\"delta\":{\"content\":\"Hello\"},\"finish_reason\":null}]}\n\n",
        "data: {\"id\":\"c1\",\"object\":\"chat.completion.chunk\",\"model\":\"gpt-4\",\"choices\":[{\"index\":0,\"delta\":{},\"finish_reason\":\"stop\"}]}\n\n",
        "data: [DONE]\n\n",
    );
    let mock = wiremock::MockServer::start().await;
    wiremock::Mock::given(wiremock::matchers::method("POST"))
        .and(wiremock::matchers::path("/chat/completions"))
        .respond_with(wiremock::ResponseTemplate::new(200).set_body_raw(sse.as_bytes().to_vec(), "text/event-stream"))
        .mount(&mock)
        .await;

    let mut config = create_test_config();
    config.onwards.strict_mode = true;
    config.background_services.onwards_sync.enabled = true;
    let app = crate::Application::new_with_pool(config, Some(pool.clone()), None)
        .await
        .expect("app");
    let (server, bg) = app.into_test_server();

    let api_key = seed_model_and_key(&server, &pool, &mock.uri()).await;
    bg.sync_onwards_config(&pool).await.unwrap();
    wait_for_model(&server, &api_key).await;

    let resp = server
        .post("/ai/v1/messages")
        .add_header("x-api-key", &api_key)
        .add_header("anthropic-version", "2023-06-01")
        .json(&serde_json::json!({
            "model": "gpt-4",
            "max_tokens": 64,
            "stream": true,
            "messages": [ { "role": "user", "content": "hi" } ]
        }))
        .await;

    let status = resp.status_code();
    let text = resp.text();
    assert_eq!(status, 200, "got {status}: {text}");
    for ev in [
        "event: message_start",
        "event: content_block_start",
        "event: content_block_delta",
        "event: message_stop",
    ] {
        assert!(text.contains(ev), "missing {ev} in:\n{text}");
    }
    assert!(text.contains(r#""text":"Hello""#), "{text}");
    assert!(text.contains(r#""stop_reason":"end_turn""#), "{text}");
}