recall_echo/graph/
embed.rs1use std::path::Path;
4
5use fastembed::{EmbeddingModel, InitOptions, TextEmbedding};
6
7use super::error::GraphError;
8
9pub trait Embedder {
11 fn embed(&self, texts: Vec<&str>) -> Result<Vec<Vec<f32>>, GraphError>;
12 fn embed_single(&self, text: &str) -> Result<Vec<f32>, GraphError>;
13 fn dimensions(&self) -> usize;
14}
15
16pub struct FastEmbedder {
18 model: TextEmbedding,
19}
20
21impl FastEmbedder {
22 pub fn new(cache_dir: &Path) -> Result<Self, GraphError> {
23 let options = InitOptions::new(EmbeddingModel::BGESmallENV15)
24 .with_cache_dir(cache_dir.to_path_buf())
25 .with_show_download_progress(true);
26
27 let model =
28 TextEmbedding::try_new(options).map_err(|e| GraphError::Embed(e.to_string()))?;
29 Ok(Self { model })
30 }
31}
32
33impl Embedder for FastEmbedder {
34 fn embed(&self, texts: Vec<&str>) -> Result<Vec<Vec<f32>>, GraphError> {
35 let docs: Vec<String> = texts.into_iter().map(|t| t.to_string()).collect();
36 let embeddings = self
37 .model
38 .embed(docs, None)
39 .map_err(|e| GraphError::Embed(e.to_string()))?;
40 Ok(embeddings)
41 }
42
43 fn embed_single(&self, text: &str) -> Result<Vec<f32>, GraphError> {
44 let embeddings = self
45 .model
46 .embed(vec![text.to_string()], None)
47 .map_err(|e| GraphError::Embed(e.to_string()))?;
48 embeddings
49 .into_iter()
50 .next()
51 .ok_or_else(|| GraphError::Embed("no embedding returned".into()))
52 }
53
54 fn dimensions(&self) -> usize {
55 384
56 }
57}