MeiliSearch/meilisearch-core/src/update/synonyms_addition.rs

120 lines
3.5 KiB
Rust
Raw Normal View History

2019-10-08 17:06:56 +02:00
use std::collections::BTreeMap;
2019-10-18 13:05:28 +02:00
use fst::{set::OpBuilder, SetBuilder};
2019-10-08 17:06:56 +02:00
use sdset::SetBuf;
use crate::automaton::normalize_str;
2019-11-06 10:49:13 +01:00
use crate::database::{UpdateEvent, UpdateEventsEmitter};
2019-10-18 13:05:28 +02:00
use crate::update::{next_update_id, Update};
2019-10-08 17:31:07 +02:00
use crate::{store, MResult};
2019-10-08 17:06:56 +02:00
pub struct SynonymsAddition {
updates_store: store::Updates,
updates_results_store: store::UpdatesResults,
2019-11-06 10:49:13 +01:00
updates_notifier: UpdateEventsEmitter,
2019-10-08 17:06:56 +02:00
synonyms: BTreeMap<String, Vec<String>>,
}
impl SynonymsAddition {
pub fn new(
updates_store: store::Updates,
updates_results_store: store::UpdatesResults,
2019-11-06 10:49:13 +01:00
updates_notifier: UpdateEventsEmitter,
2019-10-18 13:05:28 +02:00
) -> SynonymsAddition {
2019-10-08 17:06:56 +02:00
SynonymsAddition {
updates_store,
updates_results_store,
updates_notifier,
synonyms: BTreeMap::new(),
}
}
pub fn add_synonym<S, T, I>(&mut self, synonym: S, alternatives: I)
2019-10-18 13:05:28 +02:00
where
S: AsRef<str>,
T: AsRef<str>,
I: IntoIterator<Item = T>,
2019-10-08 17:06:56 +02:00
{
let synonym = normalize_str(synonym.as_ref());
let alternatives = alternatives.into_iter().map(|s| s.as_ref().to_lowercase());
2019-10-18 13:05:28 +02:00
self.synonyms
.entry(synonym)
.or_insert_with(Vec::new)
.extend(alternatives);
2019-10-08 17:06:56 +02:00
}
2019-10-21 12:05:53 +02:00
pub fn finalize(self, writer: &mut heed::RwTxn) -> MResult<u64> {
2019-11-06 10:49:13 +01:00
let _ = self.updates_notifier.send(UpdateEvent::NewUpdate);
2019-10-08 17:06:56 +02:00
let update_id = push_synonyms_addition(
2019-10-11 11:29:47 +02:00
writer,
2019-10-08 17:06:56 +02:00
self.updates_store,
self.updates_results_store,
self.synonyms,
)?;
Ok(update_id)
}
}
pub fn push_synonyms_addition(
2019-10-21 12:05:53 +02:00
writer: &mut heed::RwTxn,
updates_store: store::Updates,
updates_results_store: store::UpdatesResults,
addition: BTreeMap<String, Vec<String>>,
2019-10-18 13:05:28 +02:00
) -> MResult<u64> {
let last_update_id = next_update_id(writer, updates_store, updates_results_store)?;
2019-11-12 18:00:47 +01:00
let update = Update::synonyms_addition(addition);
2019-10-08 17:31:07 +02:00
updates_store.put_update(writer, last_update_id, &update)?;
Ok(last_update_id)
}
2019-10-08 17:06:56 +02:00
pub fn apply_synonyms_addition(
2019-10-21 12:05:53 +02:00
writer: &mut heed::RwTxn,
2019-10-08 17:06:56 +02:00
main_store: store::Main,
synonyms_store: store::Synonyms,
addition: BTreeMap<String, Vec<String>>,
2019-10-18 13:05:28 +02:00
) -> MResult<()> {
2019-10-08 17:06:56 +02:00
let mut synonyms_builder = SetBuilder::memory();
for (word, alternatives) in addition {
synonyms_builder.insert(&word).unwrap();
let alternatives = {
let alternatives = SetBuf::from_dirty(alternatives);
let mut alternatives_builder = SetBuilder::memory();
alternatives_builder.extend_iter(alternatives).unwrap();
let bytes = alternatives_builder.into_inner().unwrap();
fst::Set::from_bytes(bytes).unwrap()
};
synonyms_store.put_synonyms(writer, word.as_bytes(), &alternatives)?;
}
let delta_synonyms = synonyms_builder
.into_inner()
.and_then(fst::Set::from_bytes)
.unwrap();
let synonyms = match main_store.synonyms_fst(writer)? {
Some(synonyms) => {
let op = OpBuilder::new()
.add(synonyms.stream())
.add(delta_synonyms.stream())
.r#union();
let mut synonyms_builder = SetBuilder::memory();
synonyms_builder.extend_stream(op).unwrap();
synonyms_builder
.into_inner()
.and_then(fst::Set::from_bytes)
.unwrap()
2019-10-18 13:05:28 +02:00
}
2019-10-08 17:06:56 +02:00
None => delta_synonyms,
};
main_store.put_synonyms_fst(writer, &synonyms)?;
Ok(())
}