Crate lucene_rs 

A Rust port of the core of Apache Lucene 10.5: the Lucene104 postings format, BM25 scoring, Lucene’s block-max top-k algorithms, segment-based indexing with deletes and merges, analyzers and the classic query parser.

On the benchmark corpus (469k Wikipedia articles) it returns the same top-10 documents with bit-identical scores as Lucene 10.5.2, about 1.6–2x faster per query.

Quick start

use lucene_rs::{
    DirectoryReader, Document, Field, IndexSearcher, IndexWriter, IndexWriterConfig, QueryParser,
    StandardAnalyzer, Store,
};

let mut writer = IndexWriter::open(dir, IndexWriterConfig::default())?;
writer.add_document(
    &Document::new()
        .with(Field::string("id", "1", Store::Yes))
        .with(Field::text("title", "Rust in Action", Store::Yes))
        .with(Field::text("body", "systems programming in rust", Store::No)),
)?;
writer.commit()?;

let searcher = IndexSearcher::new(DirectoryReader::open(dir)?);
let query = QueryParser::new("body", StandardAnalyzer::new()).parse("\"systems programming\" title:rust")?;
let top = searcher.search(&query, 10)?;
assert_eq!(top.total_hits.value, 1);
let doc = searcher.doc(top.score_docs[0].doc)?;
assert_eq!(doc.get_str("title"), Some("Rust in Action"));

Map

• document: Document, Field, FieldType (text, keyword, stored fields).
• analysis: Analyzer and the standard, whitespace, simple and keyword analyzers.
• index: IndexWriter (add, update, delete, flush, merge, commit) and DirectoryReader (a point-in-time view of a commit).
• search: Query, BooleanQuery, PhraseQuery and IndexSearcher.
• queryparser: Lucene’s classic query syntax.
• sim: BM25 and Lucene’s one-byte norm encoding.

The on-disk format follows Lucene104’s postings layout but is not file-compatible with Lucene: indexes written by one cannot be read by the other.

Re-exports

pub use analysis::Analyzer;

pub use analysis::StandardAnalyzer;

pub use document::Document;

pub use document::Field;

pub use document::FieldType;

pub use document::FieldValue;

pub use document::IndexOptions;

pub use document::Store;

pub use index::doc_values::SortValue;

pub use index::DirectoryReader;

pub use index::IndexWriter;

pub use index::IndexWriterConfig;

pub use index::OpenMode;

pub use index::Term;

pub use queryparser::QueryParser;

pub use search::BooleanQuery;

pub use search::IndexSearcher;

pub use search::Occur;

pub use search::PhraseQuery;

pub use search::Query;

pub use search::ScoreDoc;

pub use search::TopDocs;

pub use search::FieldDoc;

pub use search::MissingValue;

pub use search::Sort;

pub use search::SortField;

pub use search::SortOrder;

Modules

analysis

Text analysis: turning field values into terms with positions.

document

Documents and fields, mirroring Lucene’s Document, Field, FieldType, TextField, StringField and StoredField.

index

Writing and reading indexes: documents are buffered and inverted in memory, flushed as immutable segments, merged in the background of commits, and published atomically.

queryparser

Lucene “classic” query syntax: title:rust +"exact phrase" -draft (a OR b)^2.

search

Searching: queries, scoring (BM25) and top-k collection.

sim

SmallFloat norm encoding and BM25Similarity, ported from Lucene 10.5.

Enums

Error

Errors returned by indexing and searching.

Type Aliases

Result