hermes_core/segment/builder/
config.rs

1//! Configuration and statistics types for segment builder
2
3use std::path::PathBuf;
4
5use crate::compression::CompressionLevel;
6
7/// Statistics about segment builder state
8#[derive(Debug, Clone, Default)]
9pub struct SegmentBuilderStats {
10    /// Number of documents indexed
11    pub num_docs: u32,
12    /// Number of unique terms in the inverted index
13    pub unique_terms: usize,
14    /// Total postings in memory (across all terms)
15    pub postings_in_memory: usize,
16    /// Number of interned strings
17    pub interned_strings: usize,
18    /// Size of doc_field_lengths vector
19    pub doc_field_lengths_size: usize,
20    /// Estimated total memory usage in bytes
21    pub estimated_memory_bytes: usize,
22    /// Memory breakdown by component
23    pub memory_breakdown: MemoryBreakdown,
24}
25
26/// Detailed memory breakdown by component
27#[derive(Debug, Clone, Default)]
28pub struct MemoryBreakdown {
29    /// Postings memory (CompactPosting structs)
30    pub postings_bytes: usize,
31    /// Inverted index HashMap overhead
32    pub index_overhead_bytes: usize,
33    /// Term interner memory
34    pub interner_bytes: usize,
35    /// Document field lengths
36    pub field_lengths_bytes: usize,
37    /// Dense vector storage
38    pub dense_vectors_bytes: usize,
39    /// Number of dense vectors
40    pub dense_vector_count: usize,
41    /// Sparse vector storage
42    pub sparse_vectors_bytes: usize,
43    /// Position index storage
44    pub position_index_bytes: usize,
45}
46
47/// Configuration for segment builder
48#[derive(Clone)]
49pub struct SegmentBuilderConfig {
50    /// Directory for temporary spill files
51    pub temp_dir: PathBuf,
52    /// Compression level for document store
53    pub compression_level: CompressionLevel,
54    /// Width of the document-store compression pool. Concurrent builders with
55    /// the same width share one process-wide executor.
56    pub num_compression_threads: usize,
57    /// Initial capacity for term interner
58    pub interner_capacity: usize,
59    /// Initial capacity for posting lists hashmap
60    pub posting_map_capacity: usize,
61}
62
63impl Default for SegmentBuilderConfig {
64    fn default() -> Self {
65        Self {
66            #[cfg(feature = "native")]
67            temp_dir: std::env::temp_dir(),
68            #[cfg(not(feature = "native"))]
69            temp_dir: PathBuf::from("/tmp"),
70            compression_level: CompressionLevel(3),
71            #[cfg(feature = "native")]
72            num_compression_threads: crate::default_compression_threads(),
73            #[cfg(not(feature = "native"))]
74            num_compression_threads: 1,
75            interner_capacity: 1_000_000,
76            posting_map_capacity: 500_000,
77        }
78    }
79}
hermes_core/segment/builder/config.rs

hermes_core/segment/builder/
config.rs
