arche 3.0.0

An opinionated backend foundation for Axum applications, providing batteries-included integrations for cloud services, databases, authentication, middleware, and logging.
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

use crate::error::AppError;
use crate::llm::{GenerateRequest, GenerateResponse, LlmProvider, LlmStream};
use std::pin::Pin;

use super::config::ResolvedAuth;
use super::providers;

#[derive(Debug, Clone, Copy, PartialEq, Eq)]
pub enum VertexProvider {
    Gemini,
    Anthropic,
}

pub struct VertexClient {
    pub(crate) http: reqwest::Client,
    pub(crate) auth: ResolvedAuth,
    pub(crate) provider: VertexProvider,
}

impl VertexClient {
    pub(crate) fn new(http: reqwest::Client, auth: ResolvedAuth, provider: VertexProvider) -> Self {
        Self {
            http,
            auth,
            provider,
        }
    }

    pub(crate) async fn auth_header(&self) -> Result<Option<String>, AppError> {
        match &self.auth {
            ResolvedAuth::ApiKey { .. } => Ok(None),
            ResolvedAuth::ServiceAccount { authenticator, .. } => {
                let token = authenticator
                    .token(&["https://www.googleapis.com/auth/cloud-platform"])
                    .await
                    .map_err(|e| {
                        AppError::internal_error(
                            format!("Failed to fetch Vertex AI access token: {e}"),
                            None,
                        )
                    })?;

                let bearer = token.token().ok_or_else(|| {
                    AppError::internal_error(
                        "Vertex AI token response contained no access token".into(),
                        None,
                    )
                })?;

                Ok(Some(format!("Bearer {bearer}")))
            }
        }
    }

    pub(crate) async fn send(
        &self,
        req: reqwest::RequestBuilder,
    ) -> Result<reqwest::Response, AppError> {
        let resp = req.send().await.map_err(|e| {
            AppError::dependency_failed("vertex-ai", format!("Request failed: {e}"))
        })?;

        if !resp.status().is_success() {
            let status = resp.status();
            let body = resp.text().await.unwrap_or_default();
            return Err(AppError::dependency_failed(
                "vertex-ai",
                format!("API error ({status}): {body}"),
            ));
        }

        Ok(resp)
    }
}

impl LlmProvider for VertexClient {
    fn generate<'a>(
        &'a self,
        request: &'a GenerateRequest,
    ) -> Pin<Box<dyn std::future::Future<Output = Result<GenerateResponse, AppError>> + Send + 'a>>
    {
        Box::pin(async move {
            match self.provider {
                VertexProvider::Gemini => providers::gemini::generate(self, request).await,
                VertexProvider::Anthropic => providers::anthropic::generate(self, request).await,
            }
        })
    }

    fn stream_generate<'a>(
        &'a self,
        request: &'a GenerateRequest,
    ) -> Pin<Box<dyn std::future::Future<Output = Result<LlmStream, AppError>> + Send + 'a>> {
        Box::pin(async move {
            match self.provider {
                VertexProvider::Gemini => providers::gemini::stream_generate(self, request).await,
                VertexProvider::Anthropic => {
                    providers::anthropic::stream_generate(self, request).await
                }
            }
        })
    }
}