embedrs 0.4.0

Unified embedding — cloud APIs (OpenAI, Cohere, Gemini, Voyage, Jina, Mistral) + local inference, one interface
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

use std::sync::Arc;
use std::sync::atomic::{AtomicU32, Ordering};

use wiremock::matchers::{method, path};
use wiremock::{Mock, MockServer, Respond, ResponseTemplate};

fn success_body() -> serde_json::Value {
    serde_json::json!({
        "object": "list",
        "data": [{"object": "embedding", "index": 0, "embedding": [0.1, 0.2, 0.3]}],
        "model": "text-embedding-3-small",
        "usage": {"prompt_tokens": 5, "total_tokens": 5}
    })
}

struct CountingResponder {
    call_count: Arc<AtomicU32>,
    status: u16,
    body: Option<serde_json::Value>,
}

impl Respond for CountingResponder {
    fn respond(&self, _request: &wiremock::Request) -> ResponseTemplate {
        self.call_count.fetch_add(1, Ordering::SeqCst);
        let resp = ResponseTemplate::new(self.status);
        match &self.body {
            Some(body) => resp.set_body_json(body),
            None => resp.set_body_string("internal server error"),
        }
    }
}

#[tokio::test]
async fn fallback_on_primary_failure() {
    let primary = MockServer::start().await;
    let fallback = MockServer::start().await;

    let primary_count = Arc::new(AtomicU32::new(0));
    let fallback_count = Arc::new(AtomicU32::new(0));

    Mock::given(method("POST"))
        .and(path("/embeddings"))
        .respond_with(CountingResponder {
            call_count: primary_count.clone(),
            status: 500,
            body: None,
        })
        .mount(&primary)
        .await;

    Mock::given(method("POST"))
        .and(path("/embeddings"))
        .respond_with(CountingResponder {
            call_count: fallback_count.clone(),
            status: 200,
            body: Some(success_body()),
        })
        .mount(&fallback)
        .await;

    let client = embedrs::Client::openai_compatible("key", &primary.uri())
        .with_fallback(embedrs::Client::openai_compatible("key2", &fallback.uri()));

    let result = client.embed(vec!["test".into()]).await.unwrap();
    assert_eq!(result.embeddings.len(), 1);
    assert_eq!(primary_count.load(Ordering::SeqCst), 1);
    assert_eq!(fallback_count.load(Ordering::SeqCst), 1);
}

#[tokio::test]
async fn fallback_skipped_on_success() {
    let primary = MockServer::start().await;
    let fallback = MockServer::start().await;

    let primary_count = Arc::new(AtomicU32::new(0));
    let fallback_count = Arc::new(AtomicU32::new(0));

    Mock::given(method("POST"))
        .and(path("/embeddings"))
        .respond_with(CountingResponder {
            call_count: primary_count.clone(),
            status: 200,
            body: Some(success_body()),
        })
        .mount(&primary)
        .await;

    Mock::given(method("POST"))
        .and(path("/embeddings"))
        .respond_with(CountingResponder {
            call_count: fallback_count.clone(),
            status: 200,
            body: Some(success_body()),
        })
        .mount(&fallback)
        .await;

    let client = embedrs::Client::openai_compatible("key", &primary.uri())
        .with_fallback(embedrs::Client::openai_compatible("key2", &fallback.uri()));

    let result = client.embed(vec!["test".into()]).await.unwrap();
    assert_eq!(result.embeddings.len(), 1);
    assert_eq!(primary_count.load(Ordering::SeqCst), 1);
    assert_eq!(fallback_count.load(Ordering::SeqCst), 0);
}

#[tokio::test]
async fn fallback_chain_tries_all() {
    let primary = MockServer::start().await;
    let fallback1 = MockServer::start().await;
    let fallback2 = MockServer::start().await;

    let primary_count = Arc::new(AtomicU32::new(0));
    let fallback1_count = Arc::new(AtomicU32::new(0));
    let fallback2_count = Arc::new(AtomicU32::new(0));

    // primary returns 500
    Mock::given(method("POST"))
        .and(path("/embeddings"))
        .respond_with(CountingResponder {
            call_count: primary_count.clone(),
            status: 500,
            body: None,
        })
        .mount(&primary)
        .await;

    // fallback1 returns 500
    Mock::given(method("POST"))
        .and(path("/embeddings"))
        .respond_with(CountingResponder {
            call_count: fallback1_count.clone(),
            status: 500,
            body: None,
        })
        .mount(&fallback1)
        .await;

    // fallback2 succeeds
    Mock::given(method("POST"))
        .and(path("/embeddings"))
        .respond_with(CountingResponder {
            call_count: fallback2_count.clone(),
            status: 200,
            body: Some(success_body()),
        })
        .mount(&fallback2)
        .await;

    let client = embedrs::Client::openai_compatible("key", &primary.uri())
        .with_fallback(embedrs::Client::openai_compatible("key2", &fallback1.uri()))
        .with_fallback(embedrs::Client::openai_compatible("key3", &fallback2.uri()));

    let result = client.embed(vec!["test".into()]).await.unwrap();
    assert_eq!(result.embeddings.len(), 1);
    assert_eq!(primary_count.load(Ordering::SeqCst), 1);
    assert_eq!(fallback1_count.load(Ordering::SeqCst), 1);
    assert_eq!(fallback2_count.load(Ordering::SeqCst), 1);
}