ailake_core/
schema.rs

1// SPDX-License-Identifier: MIT OR Apache-2.0
2use crate::types::{VectorMetric, VectorPrecision};
3use serde::{Deserialize, Serialize};
4
5/// Canonical column names for LLM-context tables.
6/// ContextAssembler reads columns by these names.
7pub mod llm_columns {
8    pub const CHUNK_ID: &str = "chunk_id";
9    pub const DOCUMENT_ID: &str = "document_id";
10    pub const CHUNK_INDEX: &str = "chunk_index";
11    pub const TOTAL_CHUNKS: &str = "total_chunks";
12    pub const CHUNK_TEXT: &str = "chunk_text";
13    pub const DOCUMENT_TITLE: &str = "document_title";
14    pub const SECTION_PATH: &str = "section_path";
15    pub const PRECEDING_CONTEXT: &str = "preceding_context";
16    pub const FOLLOWING_CONTEXT: &str = "following_context";
17    pub const DOCUMENT_SUMMARY: &str = "document_summary";
18    pub const CHUNK_SUMMARY: &str = "chunk_summary";
19    pub const SOURCE_URI: &str = "source_uri";
20    pub const PAGE_NUMBER: &str = "page_number";
21    pub const CREATED_AT: &str = "created_at";
22    pub const DOCUMENT_DATE: &str = "document_date";
23    pub const EMBEDDING: &str = "embedding";
24    pub const CONTEXT_EMBEDDING: &str = "context_embedding";
25}
26
27/// Vector storage configuration applied at table creation time.
28/// Stored in Iceberg metadata.json properties.
29#[derive(Debug, Clone, Serialize, Deserialize)]
30pub struct VectorStoragePolicy {
31    pub column_name: String,
32    pub dim: u32,
33    pub metric: VectorMetric,
34    pub precision: VectorPrecision,
35    pub pq: Option<PQConfig>,
36    pub keep_raw_for_reranking: bool,
37}
38
39impl VectorStoragePolicy {
40    pub fn default_f16(column: &str, dim: u32, metric: VectorMetric) -> Self {
41        Self {
42            column_name: column.to_string(),
43            dim,
44            metric,
45            precision: VectorPrecision::F16,
46            pq: None,
47            keep_raw_for_reranking: true,
48        }
49    }
50}
51
52/// Product Quantization configuration
53#[derive(Debug, Clone, Serialize, Deserialize)]
54pub struct PQConfig {
55    /// Number of sub-vectors M (dim must be divisible by M)
56    pub num_subvectors: usize,
57    /// Bits per code (8 = 256 centroids per sub-vector)
58    pub bits_per_code: u8,
59    /// Number of training samples for codebook
60    pub train_sample_size: usize,
61}
62
63/// Marker struct for documentation purposes — actual schema is enforced by
64/// column names in llm_columns module.
65pub struct LlmContextSchema;
ailake_core/schema.rs

ailake_core/
schema.rs
