Files
tools/search/src/index.rs
T

133 lines
4.0 KiB
Rust
Raw Normal View History

2026-06-08 19:33:49 -04:00
use crate::error;
use crate::error::SearchError;
use crate::schema::Schema;
use std::path::PathBuf;
use tantivy::collector::TopDocs;
use tantivy::directory::{ManagedDirectory, MmapDirectory};
use tantivy::query::{BooleanQuery, Occur, QueryParser, TermQuery};
use tantivy::schema::{Field, IndexRecordOption, Value};
use tantivy::tokenizer::{NgramTokenizer, TokenizerManager};
use tantivy::{Index, IndexReader, IndexWriter, ReloadPolicy, TantivyDocument, Term, doc};
#[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> {
if let Some(path) = self.path {
let ngram_32 = NgramTokenizer::new(1, 32, false)?;
let tokenizer_manager = TokenizerManager::default();
tokenizer_manager.register("ngram_32", ngram_32);
let mmap_directory = MmapDirectory::open(path)?;
let managed_directory = ManagedDirectory::wrap(Box::new(mmap_directory))?;
let index = Index::builder()
.schema(Schema::schema().clone())
.tokenizers(tokenizer_manager)
.open_or_create(managed_directory)?;
let reader = index
.reader_builder()
.reload_policy(ReloadPolicy::OnCommitWithDelay)
.try_into()?;
let writer = index.writer(50_000_000)?;
Ok(SearchIndex {
index,
reader,
writer,
})
} else {
Err(SearchError::IndexPathNotSpecified)
}
}
}
pub struct SearchIndex {
index: Index,
reader: IndexReader,
writer: IndexWriter,
}
impl SearchIndex {
pub fn builder() -> SearchIndexBuilder {
SearchIndexBuilder::default()
}
pub fn remove_all_annotated_iris(&mut self) {
self.writer
.delete_term(Term::from_field_u64(Schema::type_field(), 0u64));
self.writer.commit().unwrap();
}
pub fn add_annotated_iri(
&self,
iri: &str,
label: Option<&str>,
comment: Option<&str>,
) -> crate::Result<()> {
let mut document = doc!(
Schema::type_field() => 0u64,
Schema::iri_field() => iri,
);
if let Some(label) = label {
document.add_text(Schema::label_field(), label.to_lowercase());
}
if let Some(comment) = comment {
document.add_text(Schema::comment_field(), comment);
}
self.writer.add_document(document)?;
Ok(())
}
pub fn add<'a>(&self, document: TantivyDocument) -> crate::Result<()> {
self.writer.add_document(document)?;
Ok(())
}
pub fn commit(&mut self) -> crate::Result<()> {
self.writer.commit()?;
Ok(())
}
pub fn query(
&self,
type_: u64,
user_query: &str,
default_fields: Vec<Field>,
) -> error::Result<Vec<String>> {
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));
let parser = QueryParser::for_index(&self.index, default_fields);
let (user_query, _) = parser.parse_query_lenient(user_query);
let query = BooleanQuery::new(vec![
(Occur::Must, doc_type_query),
(Occur::Must, user_query),
]);
let searcher = self.reader.searcher();
let results = searcher.search(&query, &TopDocs::with_limit(10000).order_by_score())?;
let mut iris = vec![];
for (_score, address) in results.iter() {
let doc: TantivyDocument = searcher.doc(*address)?;
if let Some(doc_iri) = doc.get_first(Schema::iri_field()) {
let doc_iri_string = doc_iri.as_str().unwrap_or("???").to_string();
iris.push(doc_iri_string);
}
}
Ok(iris)
}
}