xz_rag/types/
chunk.rs

1use serde::{Deserialize, Serialize};
2use std::collections::HashMap;
3
4/// Document chunk with metadata and optional embedding.
5#[derive(Debug, Clone, Serialize, Deserialize)]
6pub struct Chunk {
7    /// Unique chunk identifier.
8    pub id: String,
9    /// ID of the parent document.
10    pub document_id: String,
11    /// Index of this chunk within the document.
12    pub chunk_index: u32,
13    /// Text content of the chunk.
14    pub content: String,
15    /// Optional summary of the chunk content.
16    pub summary: Option<String>,
17    /// Metadata associated with this chunk.
18    pub metadata: ChunkMetadata,
19    /// Optional embedding vector for vector search.
20    pub embedding: Option<Vec<f32>>,
21    /// Timestamp when the chunk was created (milliseconds since epoch).
22    pub created_at: u64,
23    /// Optional expiry timestamp (milliseconds since epoch).
24    pub expires_at: Option<u64>,
25}
26
27/// Metadata attached to each chunk.
28#[derive(Debug, Clone, Serialize, Deserialize, Default)]
29pub struct ChunkMetadata {
30    /// Source of the document (e.g. file path, URL).
31    pub source: Option<String>,
32    /// Title of the parent document.
33    pub document_title: Option<String>,
34    /// Author of the document.
35    pub author: Option<String>,
36    /// Timestamp when the chunk was created (milliseconds since epoch).
37    pub created_at: Option<u64>,
38    /// List of tags associated with the chunk.
39    pub tags: Vec<String>,
40    /// Namespace for scoping the chunk.
41    pub namespace: Option<String>,
42    /// Extra key-value metadata.
43    pub extra: HashMap<String, String>,
44}
45
46impl Chunk {
47    /// Create a new chunk with the given id, document_id, content, and index.
48    pub fn new(id: String, document_id: String, content: String, chunk_index: u32) -> Self {
49        Self {
50            id,
51            document_id,
52            chunk_index,
53            content,
54            summary: None,
55            metadata: ChunkMetadata::default(),
56            embedding: None,
57            created_at: std::time::SystemTime::now()
58                .duration_since(std::time::UNIX_EPOCH)
59                .unwrap_or_default()
60                .as_millis() as u64,
61            expires_at: None,
62        }
63    }
64}
xz_rag/types/chunk.rs

xz_rag/types/
chunk.rs
