langchain-rust 7.0.0

LangChain for Rust, the easiest way to write LLM-based programs in 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

use std::sync::Arc;

use crate::embedding::{embedder_trait::Embedder, EmbedderError};
use crate::observability::{traced_embed_documents, traced_embed_query};
use async_trait::async_trait;
use mistralai_client::v1::{client::Client, constants::EmbedModel};

/// `provider` field of the `embed.*` spans.
const PROVIDER: &str = "mistralai";

pub struct MistralAIEmbedder {
    client: Arc<Client>,
    model: EmbedModel,
}

impl MistralAIEmbedder {
    /// Wire name of the model ("mistral-embed"), for span fields.
    fn model_name(&self) -> String {
        serde_json::to_value(&self.model)
            .ok()
            .and_then(|v| v.as_str().map(str::to_string))
            .unwrap_or_else(|| format!("{:?}", self.model))
    }

    pub fn try_new() -> Result<Self, EmbedderError> {
        Ok(Self {
            client: Arc::new(
                Client::new(None, None, None, None).map_err(EmbedderError::MistralAIClientError)?,
            ),
            model: EmbedModel::MistralEmbed,
        })
    }
}

#[async_trait]
impl Embedder for MistralAIEmbedder {
    async fn embed_documents(&self, documents: &[String]) -> Result<Vec<Vec<f64>>, EmbedderError> {
        traced_embed_documents(PROVIDER, &self.model_name(), documents.len(), async {
            tracing::debug!("Embedding documents: {:?}", documents);

            let response = self
                .client
                .embeddings_async(self.model.clone(), documents.into(), None)
                .await
                .map_err(EmbedderError::MistralAIApiError)?;

            Ok(response
                .data
                .into_iter()
                .map(|item| item.embedding.into_iter().map(|x| x as f64).collect())
                .collect::<Vec<Vec<f64>>>())
        })
        .await
    }

    async fn embed_query(&self, text: &str) -> Result<Vec<f64>, EmbedderError> {
        traced_embed_query(PROVIDER, &self.model_name(), async {
            tracing::debug!("Embedding query: {:?}", text);

            let response = self
                .client
                .embeddings_async(self.model.clone(), vec![text.to_string()], None)
                .await
                .map_err(EmbedderError::MistralAIApiError)?;

            Ok(response.data[0]
                .embedding
                .iter()
                .map(|x| *x as f64)
                .collect())
        })
        .await
    }
}

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

    #[tokio::test]
    #[ignore]
    async fn test_mistralai_embed_query() {
        let mistralai = MistralAIEmbedder::try_new().unwrap();
        let embeddings = mistralai.embed_query("Why is the sky blue?").await.unwrap();
        assert_eq!(embeddings.len(), 1024);
    }

    #[tokio::test]
    #[ignore]
    async fn test_mistralai_embed_documents() {
        let mistralai = MistralAIEmbedder::try_new().unwrap();
        let embeddings = mistralai
            .embed_documents(&["hello world".to_string(), "foo bar".to_string()])
            .await
            .unwrap();
        assert_eq!(embeddings.len(), 2);
    }
}