sapphire_retrieve/
config.rs

1use serde::{Deserialize, Serialize};
2
3use crate::embed::EmbedderConfig;
4
5/// Top-level retrieve configuration (`[retrieve]` section).
6///
7/// Controls which vector database backend to use and, optionally, text
8/// embedding settings for approximate semantic search.
9#[derive(Debug, Clone, Serialize, Deserialize, Default)]
10pub struct RetrieveConfig {
11    /// Vector database backend (default: `none` — vector search disabled).
12    #[serde(default)]
13    pub db: VectorDb,
14    /// Text embedding settings.  When absent, embedding is disabled even
15    /// if `db` is set to a non-`none` value.
16    #[serde(default)]
17    pub embedding: Option<EmbeddingConfig>,
18    /// Hybrid search tuning (FTS + semantic merged via Reciprocal Rank Fusion).
19    #[serde(default)]
20    pub hybrid: HybridConfig,
21}
22
23/// Settings for hybrid (FTS + semantic) search merging via Reciprocal Rank
24/// Fusion (RRF).
25#[derive(Debug, Clone, Serialize, Deserialize)]
26pub struct HybridConfig {
27    /// Weight for FTS results in RRF fusion (0.0–1.0, default 0.5).
28    /// The semantic weight is `1.0 - fts_weight`.
29    #[serde(default = "default_fts_weight")]
30    pub fts_weight: f64,
31    /// Constant *k* in the RRF formula: `score = 1 / (k + rank)`.
32    /// Default 60.
33    #[serde(default = "default_rrf_k")]
34    pub rrf_k: u32,
35}
36
37fn default_fts_weight() -> f64 {
38    0.5
39}
40
41fn default_rrf_k() -> u32 {
42    60
43}
44
45impl Default for HybridConfig {
46    fn default() -> Self {
47        Self {
48            fts_weight: default_fts_weight(),
49            rrf_k: default_rrf_k(),
50        }
51    }
52}
53
54/// Vector database backend for approximate (semantic) text search.
55///
56/// | Variant      | Description                                              |
57/// |--------------|----------------------------------------------------------|
58/// | `none`       | Vector search disabled (default, no extra dependencies)  |
59/// | `sqlite_vec` | sqlite-vec extension, stored inside the SQLite cache DB  |
60/// | `lancedb`    | LanceDB — suitable for larger-scale / multimodal use     |
61#[derive(Debug, Clone, Copy, Serialize, Deserialize, Default, PartialEq, Eq)]
62#[serde(rename_all = "snake_case")]
63pub enum VectorDb {
64    /// Vector search is disabled. No embedding model is required.
65    #[default]
66    None,
67    /// sqlite-vec extension stored in the existing SQLite cache database.
68    SqliteVec,
69    /// LanceDB stored in a separate data directory alongside the cache.
70    #[serde(rename = "lancedb")]
71    LanceDb,
72}
73
74impl VectorDb {
75    /// Human-readable name, matching the TOML serialization.
76    pub fn as_str(self) -> &'static str {
77        match self {
78            VectorDb::None => "none",
79            VectorDb::SqliteVec => "sqlite_vec",
80            VectorDb::LanceDb => "lancedb",
81        }
82    }
83}
84
85/// Text embedding provider configuration (`[retrieve.embedding]` subsection).
86#[derive(Debug, Clone, Serialize, Deserialize, Default)]
87pub struct EmbeddingConfig {
88    /// Enable embedding and vector search.
89    #[serde(default)]
90    pub enabled: bool,
91
92    /// Embedding provider identifier: `"openai"`, `"ollama"`, or `"fastembed"`.
93    #[serde(default)]
94    pub provider: String,
95
96    /// Model name understood by the provider.
97    #[serde(default)]
98    pub model: String,
99
100    /// Name of the environment variable holding the API key.
101    /// Used by OpenAI-compatible providers; defaults to `OPENAI_API_KEY`.
102    #[serde(skip_serializing_if = "Option::is_none")]
103    pub api_key_env: Option<String>,
104
105    /// Base URL of the embedding API endpoint.
106    #[serde(skip_serializing_if = "Option::is_none")]
107    pub base_url: Option<String>,
108
109    /// Output vector dimension of the model.
110    /// Required when `db = "sqlite_vec"`.
111    #[serde(skip_serializing_if = "Option::is_none")]
112    pub dimension: Option<u32>,
113}
114
115impl EmbeddingConfig {
116    /// Convert to the runtime [`EmbedderConfig`] used by [`crate::build_embedder`].
117    /// Convert to the runtime [`EmbedderConfig`].
118    ///
119    /// `cache_dir` is left as `None`; callers should set it to the
120    /// app-provided model cache directory before calling [`crate::build_embedder`].
121    pub fn to_embedder_config(&self) -> EmbedderConfig {
122        EmbedderConfig {
123            provider: self.provider.clone(),
124            model: self.model.clone(),
125            api_key_env: self.api_key_env.clone(),
126            base_url: self.base_url.clone(),
127            cache_dir: None,
128        }
129    }
130}
sapphire_retrieve/config.rs

sapphire_retrieve/
config.rs
