llmshim 0.4.0

Blazing fast LLM API translation layer in pure Rust
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
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
202
203
204
205
206
207

use llmshim::provider::Provider;
use llmshim::providers::openai_compat::OpenAiCompatible;
use serde_json::{json, Value};

fn vllm() -> OpenAiCompatible {
    OpenAiCompatible::new("vllm", "http://localhost:8000/v1", None)
}

fn sglang_with_key() -> OpenAiCompatible {
    OpenAiCompatible::new(
        "sglang",
        "http://localhost:30000/v1",
        Some("secret".to_string()),
    )
}

// ============================================================
// transform_request
// ============================================================

#[test]
fn request_url_from_base_and_no_auth_when_keyless() {
    let p = vllm();
    let req = json!({"model": "m", "messages": [{"role": "user", "content": "hi"}]});
    let result = p.transform_request("m", &req).unwrap();
    assert_eq!(result.url, "http://localhost:8000/v1/chat/completions");
    // Self-hosted with no key: no Authorization header.
    assert!(result.headers.iter().all(|(k, _)| k != "Authorization"));
}

#[test]
fn request_sends_bearer_when_key_set() {
    let p = sglang_with_key();
    let req = json!({"model": "m", "messages": [{"role": "user", "content": "hi"}]});
    let result = p.transform_request("m", &req).unwrap();
    let auth = result.headers.iter().find(|(k, _)| k == "Authorization");
    assert_eq!(auth.unwrap().1, "Bearer secret");
}

#[test]
fn request_trailing_slash_base_url_is_normalized() {
    let p = OpenAiCompatible::new("vllm", "http://localhost:8000/v1/", None);
    let req = json!({"model": "m", "messages": [{"role": "user", "content": "hi"}]});
    let result = p.transform_request("m", &req).unwrap();
    assert_eq!(result.url, "http://localhost:8000/v1/chat/completions");
}

#[test]
fn request_preserves_hf_slug_model() {
    // The served model name (an HF path with a slash) passes through as `model`.
    let p = vllm();
    let req = json!({"model": "meta-llama/Llama-3.1-8B-Instruct", "messages": [{"role": "user", "content": "hi"}]});
    let result = p
        .transform_request("meta-llama/Llama-3.1-8B-Instruct", &req)
        .unwrap();
    assert_eq!(result.body["model"], "meta-llama/Llama-3.1-8B-Instruct");
}

#[test]
fn request_forwards_standard_params_and_reasoning_effort() {
    let p = vllm();
    let tools = json!([{"type": "function", "function": {"name": "f", "parameters": {}}}]);
    let req = json!({
        "model": "m",
        "messages": [{"role": "user", "content": "hi"}],
        "tools": tools,
        "response_format": {"type": "json_object"},
        "reasoning_effort": "high",
        "temperature": 0.3,
    });
    let result = p.transform_request("m", &req).unwrap();
    assert_eq!(result.body["tools"], tools);
    assert_eq!(result.body["response_format"]["type"], "json_object");
    assert_eq!(result.body["reasoning_effort"], "high");
    assert_eq!(result.body["temperature"], 0.3);
}

#[test]
fn request_x_namespace_params_copied_to_body() {
    let p = vllm();
    let req = json!({
        "model": "m",
        "messages": [{"role": "user", "content": "hi"}],
        "x-vllm": {
            "top_k": 20,
            "guided_json": {"type": "object"},
            "chat_template_kwargs": {"enable_thinking": true}
        },
    });
    let result = p.transform_request("m", &req).unwrap();
    assert_eq!(result.body["top_k"], 20);
    assert_eq!(result.body["guided_json"]["type"], "object");
    assert_eq!(result.body["chat_template_kwargs"]["enable_thinking"], true);
}

#[test]
fn request_namespace_is_provider_specific() {
    // An x-sglang block is ignored by the vllm provider (wrong namespace).
    let p = vllm();
    let req = json!({
        "model": "m",
        "messages": [{"role": "user", "content": "hi"}],
        "x-sglang": {"separate_reasoning": false},
    });
    let result = p.transform_request("m", &req).unwrap();
    assert!(result.body.get("separate_reasoning").is_none());
}

#[test]
fn request_sanitizes_foreign_reasoning_fields() {
    let p = vllm();
    let req = json!({
        "model": "m",
        "messages": [{
            "role": "assistant", "content": "hi",
            "reasoning_signature": "anthropic-sig-should-not-leak",
            "redacted_reasoning_content": "redacted-should-not-leak"
        }]
    });
    let result = p.transform_request("m", &req).unwrap();
    let body = serde_json::to_string(&result.body).unwrap();
    assert!(!body.contains("anthropic-sig-should-not-leak"));
    assert!(!body.contains("redacted-should-not-leak"));
}

#[test]
fn request_preserves_image_url_vision() {
    let p = vllm();
    let req = json!({
        "model": "m",
        "messages": [{"role": "user", "content": [
            {"type": "text", "text": "what is this?"},
            {"type": "image_url", "image_url": {"url": "data:image/png;base64,AAAA"}}
        ]}]
    });
    let result = p.transform_request("m", &req).unwrap();
    let content = &result.body["messages"][0]["content"];
    assert_eq!(content[0]["type"], "text");
    assert_eq!(content[1]["type"], "image_url");
}

// ============================================================
// transform_response / stream
// ============================================================

#[test]
fn response_normalizes_reasoning_field_to_reasoning_content() {
    // vLLM's newer `reasoning` field -> llmshim's reasoning_content.
    let p = vllm();
    let resp = json!({
        "choices": [{"index": 0, "message": {"role": "assistant", "content": "42", "reasoning": "think"}, "finish_reason": "stop"}]
    });
    let result = p.transform_response("m", resp).unwrap();
    assert_eq!(
        result["choices"][0]["message"]["reasoning"][0]["text"],
        "think"
    );
}

#[test]
fn response_keeps_native_reasoning_content() {
    // SGLang already returns reasoning_content — leave it untouched.
    let p = sglang_with_key();
    let resp = json!({
        "choices": [{"index": 0, "message": {"role": "assistant", "content": "42", "reasoning_content": "native"}, "finish_reason": "stop"}]
    });
    let result = p.transform_response("m", resp).unwrap();
    assert_eq!(
        result["choices"][0]["message"]["reasoning"][0]["text"],
        "native"
    );
}

#[test]
fn response_passes_tool_calls_and_errors() {
    let p = vllm();
    let resp = json!({
        "choices": [{"index": 0, "message": {"role": "assistant", "content": null, "tool_calls": [{"id": "c1", "type": "function", "function": {"name": "f", "arguments": "{}"}}]}, "finish_reason": "tool_calls"}]
    });
    let result = p.transform_response("m", resp).unwrap();
    assert_eq!(
        result["choices"][0]["message"]["tool_calls"][0]["wire_ids"][0]["id"],
        "c1"
    );

    let err_resp = json!({"error": {"code": 404, "message": "model not found"}});
    assert!(p.transform_response("m", err_resp).is_err());
}

#[test]
fn stream_normalizes_reasoning_delta() {
    let p = vllm();
    let chunk =
        json!({"choices": [{"index": 0, "delta": {"reasoning": "th"}, "finish_reason": null}]});
    let out = p
        .transform_stream_chunk("m", &serde_json::to_string(&chunk).unwrap())
        .unwrap()
        .unwrap();
    let parsed: Value = serde_json::from_str(&out).unwrap();
    assert_eq!(parsed["choices"][0]["delta"]["reasoning"][0]["text"], "th");
}

#[test]
fn name_reflects_configured_provider() {
    assert_eq!(vllm().name(), "vllm");
    assert_eq!(sglang_with_key().name(), "sglang");
}