openai-compat 0.2.0

Async Rust client for OpenAI-compatible LLM provider APIs
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

//! Embeddings types, mirroring `openai-python/src/openai/types/embedding*.py`.

use serde::{Deserialize, Serialize};

use super::common::Usage;

/// `input` accepts a string, an array of strings, or (arrays of) token ids.
#[derive(Debug, Clone, Serialize, Deserialize)]
#[serde(untagged)]
pub enum EmbeddingInput {
    Text(String),
    Texts(Vec<String>),
    Tokens(Vec<i64>),
    TokenArrays(Vec<Vec<i64>>),
}

impl From<&str> for EmbeddingInput {
    fn from(text: &str) -> Self {
        Self::Text(text.to_string())
    }
}

impl From<String> for EmbeddingInput {
    fn from(text: String) -> Self {
        Self::Text(text)
    }
}

impl From<Vec<String>> for EmbeddingInput {
    fn from(texts: Vec<String>) -> Self {
        Self::Texts(texts)
    }
}

/// Request body for `POST /embeddings` (`resources/embeddings.py::create`).
#[derive(Debug, Clone, Serialize)]
pub struct EmbeddingRequest {
    pub model: String,
    pub input: EmbeddingInput,
    #[serde(skip_serializing_if = "Option::is_none")]
    pub dimensions: Option<u32>,
    /// `"float"` (default) or `"base64"`.
    #[serde(skip_serializing_if = "Option::is_none")]
    pub encoding_format: Option<String>,
    #[serde(skip_serializing_if = "Option::is_none")]
    pub user: Option<String>,
}

impl EmbeddingRequest {
    pub fn new(model: impl Into<String>, input: impl Into<EmbeddingInput>) -> Self {
        Self {
            model: model.into(),
            input: input.into(),
            dimensions: None,
            encoding_format: None,
            user: None,
        }
    }

    pub fn dimensions(mut self, dimensions: u32) -> Self {
        self.dimensions = Some(dimensions);
        self
    }

    pub fn encoding_format(mut self, encoding_format: impl Into<String>) -> Self {
        self.encoding_format = Some(encoding_format.into());
        self
    }
}

/// The embedding vector: floats by default, or a base64 string when
/// `encoding_format: "base64"` was requested.
#[derive(Debug, Clone, Serialize, Deserialize)]
#[serde(untagged)]
pub enum EmbeddingVector {
    Floats(Vec<f32>),
    Base64(String),
}

/// One embedding result.
#[derive(Debug, Clone, Serialize, Deserialize)]
#[non_exhaustive]
pub struct Embedding {
    pub index: u32,
    pub embedding: EmbeddingVector,
    #[serde(default)]
    pub object: String,
}

/// Response from `POST /embeddings`.
#[derive(Debug, Clone, Serialize, Deserialize)]
#[non_exhaustive]
pub struct EmbeddingResponse {
    pub data: Vec<Embedding>,
    pub model: String,
    #[serde(default)]
    pub object: String,
    #[serde(default)]
    pub usage: Option<Usage>,
}

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

    #[test]
    fn input_variants_serialize() {
        assert_eq!(
            serde_json::to_value(EmbeddingInput::from("hi")).unwrap(),
            serde_json::json!("hi")
        );
        assert_eq!(
            serde_json::to_value(EmbeddingInput::Texts(vec!["a".into(), "b".into()])).unwrap(),
            serde_json::json!(["a", "b"])
        );
        assert_eq!(
            serde_json::to_value(EmbeddingInput::Tokens(vec![1, 2])).unwrap(),
            serde_json::json!([1, 2])
        );
    }

    #[test]
    fn response_deserializes_floats() {
        let body = r#"{
            "object": "list",
            "data": [{"object": "embedding", "index": 0, "embedding": [0.1, -0.2]}],
            "model": "text-embedding-3-small",
            "usage": {"prompt_tokens": 5, "total_tokens": 5}
        }"#;
        let response: EmbeddingResponse = serde_json::from_str(body).unwrap();
        let EmbeddingVector::Floats(floats) = &response.data[0].embedding else {
            panic!("expected float vector");
        };
        assert_eq!(floats.len(), 2);
    }
}