hey 0.1.0

Minimal terminal AI coding agent: kernel loop + MCP/Skills self-evolution
Documentation 
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

//! OpenAI Responses API provider 集成测试：MockServer 模拟 wire。
//!
//! 场景：SSE 事件流（正文增量、推理增量、工具调用分片）、completed 收尾、
//! 请求 path 与 body 形状断言（/responses + instructions/input/reasoning）。

mod common;

use std::sync::Arc;

use common::{MockResponse, MockServer};
use hey::config::ProviderConfig;
use hey::llm::build_provider;
use hey::llm::ir::{ChatRequest, EffortLevel, Message, Role};
use hey::llm::{Delta, LlmError};

/// 构造一条 responses SSE 事件行。
fn ev(line: &str) -> String {
    format!("data: {line}\n\n")
}

fn request() -> ChatRequest {
    ChatRequest {
        messages: vec![Message::text(Role::User, "hi")],
        tools: vec![],
        effort: EffortLevel::Medium,
    }
}

fn provider_at(server: &MockServer) -> Arc<dyn hey::llm::Provider> {
    let cfg = ProviderConfig {
        protocol: Some("responses".into()),
        base_url: format!("http://{}/v1", server.addr),
        api_key: Some("test-key".into()),
        models: vec!["o3".into()],
        effort: Some(EffortLevel::High),
        ..Default::default()
    };
    let retry = hey::config::RetryConfig {
        enabled: false,
        ..Default::default()
    };
    build_provider(&cfg, reqwest::Client::new(), &retry).unwrap()
}

#[tokio::test]
async fn streams_text_thinking_and_tool_call() {
    let body = ev(r#"{"type":"response.created","response":{}}"#)
        + &ev(
            r#"{"type":"response.output_item.added","output_index":0,"item":{"type":"function_call","call_id":"fc_1","name":"bash","arguments":""}}"#,
        )
        + &ev(
            r#"{"type":"response.function_call_arguments.delta","output_index":0,"delta":"{\"cmd\""}"#,
        )
        + &ev(
            r#"{"type":"response.function_call_arguments.delta","output_index":0,"delta":":\"ls\"}"}"#,
        )
        + &ev(
            r#"{"type":"response.reasoning_summary_text.delta","item_id":"r1","output_index":0,"delta":"thinking-out"}"#,
        )
        + &ev(
            r#"{"type":"response.output_text.delta","item_id":"m1","output_index":1,"delta":"answer"}"#,
        )
        + &ev(r#"{"type":"response.completed","response":{"id":"r1"}}"#);
    let server = MockServer::start(vec![MockResponse::sse(&body)]).await;
    let provider = provider_at(&server);

    let mut deltas = Vec::new();
    let completion = provider
        .stream(&request(), &mut |d| deltas.push(d))
        .await
        .unwrap();

    assert_eq!(completion.text, "answer");
    assert_eq!(completion.thinking, "thinking-out");
    assert_eq!(completion.tool_calls.len(), 1);
    let tc = &completion.tool_calls[0];
    assert_eq!(tc.id, "fc_1");
    assert_eq!(tc.name, "bash");
    assert_eq!(tc.arguments, r#"{"cmd":"ls"}"#);
    // 增量顺序：thinking → text
    assert_eq!(
        deltas,
        vec![
            Delta::Thinking("thinking-out".into()),
            Delta::Text("answer".into()),
        ]
    );
    // 请求路径 + body 形状
    let reqs = server.requests.lock().unwrap();
    assert_eq!(reqs[0].path, "/v1/responses");
    let body: serde_json::Value = serde_json::from_str(&reqs[0].body).unwrap();
    assert_eq!(body["model"], "o3");
    assert_eq!(
        body["reasoning"]["effort"], "medium",
        "effort 直通（请求级）"
    );
    assert_eq!(body["input"][0]["role"], "user");
    assert_eq!(body["input"][0]["content"][0]["type"], "input_text");
}

#[tokio::test]
async fn streams_error_on_missing_completed() {
    // 有输出但没收到 response.completed → 断流报错（防模型拿到半截输出）
    let body = ev(r#"{"type":"response.output_text.delta","delta":"half"}"#);
    let server = MockServer::start(vec![MockResponse::sse(&body)]).await;
    let provider = provider_at(&server);

    let e = provider
        .stream(&request(), &mut |_| {})
        .await
        .expect_err("断流应报错");
    assert!(matches!(e, LlmError::StreamInterrupted(_)), "{e}");
}

#[tokio::test]
async fn streams_error_on_non_sse_response() {
    // 200 + 非 SSE 内容 → 明确报错（此前静默空回复）
    let server = MockServer::start(vec![MockResponse::status(
        200,
        "<html>gateway error</html>",
    )])
    .await;
    let provider = provider_at(&server);

    let e = provider
        .stream(&request(), &mut |_| {})
        .await
        .expect_err("非 SSE 应报错");
    assert!(matches!(e, LlmError::StreamInterrupted(_)), "{e}");
}

#[tokio::test]
async fn streams_error_on_http_429() {
    let server = MockServer::start(vec![MockResponse::status(
        429,
        r#"{"error":{"message":"rate limited"}}"#,
    )])
    .await;
    let provider = provider_at(&server);

    let e = provider
        .stream(&request(), &mut |_| {})
        .await
        .expect_err("429 应报 RateLimit");
    assert!(matches!(e, LlmError::RateLimit { .. }), "{e}");
}