Files
tools/search/src/index.rs
T

112 lines
3.6 KiB
Rust
Raw Normal View History

2026-06-17 20:50:26 -04:00
use crate::{error, SearchDocument};
2026-06-08 19:33:49 -04:00
use crate::schema::Schema;
use std::path::PathBuf;
use tantivy::collector::TopDocs;
use tantivy::directory::{ManagedDirectory, MmapDirectory};
2026-06-09 20:09:16 -04:00
use tantivy::query::{BooleanQuery, Occur, QueryParser, TermQuery};
2026-06-17 20:50:26 -04:00
use tantivy::schema::{Field, IndexRecordOption, NamedFieldDocument, Value};
use tantivy::tokenizer::{LowerCaser, NgramTokenizer, TextAnalyzer, TokenizerManager};
use tantivy::{Document, Index, IndexReader, IndexWriter, ReloadPolicy, Score, TantivyDocument, Term};
2026-06-29 23:10:22 -04:00
use tantivy::indexer::UserOperation;
use tracing::{info, instrument, span, Level};
2026-06-08 19:33:49 -04:00
#[derive(Default)]
pub struct SearchIndexBuilder {
path: Option<PathBuf>,
}
impl SearchIndexBuilder {
pub fn with_path(mut self, path: impl Into<PathBuf>) -> Self {
self.path = Some(path.into());
self
}
pub fn build(self) -> error::Result<SearchIndex> {
2026-06-29 15:20:02 -04:00
let ngram_32 = NgramTokenizer::new(1, 32, false)?;
let ngram_32_lowercase = TextAnalyzer::builder(ngram_32)
.filter(LowerCaser)
.build();
2026-06-17 20:50:26 -04:00
2026-06-29 15:20:02 -04:00
let tokenizer_manager = TokenizerManager::default();
tokenizer_manager.register("ngram_32", ngram_32_lowercase);
2026-06-08 19:33:49 -04:00
2026-06-29 15:20:02 -04:00
let builder = Index::builder()
.schema(Schema::schema().clone())
.tokenizers(tokenizer_manager);
let index = if let Some(path) = self.path {
2026-06-08 19:33:49 -04:00
let mmap_directory = MmapDirectory::open(path)?;
let managed_directory = ManagedDirectory::wrap(Box::new(mmap_directory))?;
2026-06-29 15:20:02 -04:00
builder.open_or_create(managed_directory)?
2026-06-08 19:33:49 -04:00
} else {
2026-06-29 15:20:02 -04:00
builder.create_in_ram()?
};
let reader = index
.reader_builder()
.reload_policy(ReloadPolicy::OnCommitWithDelay)
.try_into()?;
Ok(SearchIndex {
index,
reader,
})
2026-06-08 19:33:49 -04:00
}
}
2026-06-29 23:10:22 -04:00
#[derive(Clone)]
2026-06-08 19:33:49 -04:00
pub struct SearchIndex {
index: Index,
reader: IndexReader,
}
impl SearchIndex {
pub fn builder() -> SearchIndexBuilder {
SearchIndexBuilder::default()
}
2026-06-29 23:10:22 -04:00
pub fn writer(&self) -> crate::Result<IndexWriter> {
Ok(self.index.writer(128_000_000)?)
2026-06-08 19:33:49 -04:00
}
2026-06-29 23:10:22 -04:00
pub fn remove_all_of_type(&mut self, type_: u64) -> crate::Result<()> {
let mut writer: IndexWriter<TantivyDocument> = self.index.writer(64_000_000)?;
writer.delete_term(Term::from_field_u64(Schema::type_field(), type_));
writer.commit()?;
2026-06-08 19:33:49 -04:00
Ok(())
}
pub fn query(
&self,
2026-06-09 20:09:16 -04:00
type_: Option<u64>,
2026-06-08 19:33:49 -04:00
user_query: &str,
default_fields: Vec<Field>,
2026-06-17 20:50:26 -04:00
) -> error::Result<Vec<NamedFieldDocument>> {
2026-06-29 23:10:22 -04:00
let span = span!(Level::INFO, "Search Query");
let _enter = span.enter();
2026-06-08 19:33:49 -04:00
let parser = QueryParser::for_index(&self.index, default_fields);
let (user_query, _) = parser.parse_query_lenient(user_query);
2026-06-09 20:09:16 -04:00
let mut subqueries = vec![
(Occur::Must, user_query)
];
if let Some(type_) = type_ {
let doc_type_term = Term::from_field_u64(Schema::type_field(), type_);
let doc_type_query = Box::new(TermQuery::new(doc_type_term, IndexRecordOption::Basic));
subqueries.push((Occur::Must, doc_type_query));
}
let query = BooleanQuery::new(subqueries);
2026-06-08 19:33:49 -04:00
let searcher = self.reader.searcher();
2026-06-17 20:50:26 -04:00
let results = searcher.search(&query, &TopDocs::with_limit(10000).order_by_score())?
.iter()
.map(|(_, address)| searcher.doc(*address))
.filter_map(Result::ok)
.map(|doc: SearchDocument| doc.to_named_doc(Schema::schema()))
.collect();
Ok(results)
2026-06-08 19:33:49 -04:00
}
}