snowykami/meilisearch
1
0
Fork 0
mirror of https://github.com/meilisearch/meilisearch.git synced 2025-07-31 10:50:03 +00:00
Code Issues Packages Projects Releases Wiki Activity

Rewrite cheapest path algorithm and empty path cache

Browse Source 
It is now much simpler and has much better performance.
This commit is contained in:
Loïc Lecrenier
2023-03-02 21:27:42 +01:00
parent caa1e1b923
commit c27ea2677f
14 changed files with 782 additions and 530 deletions
Download Patch File Download Diff File Expand all files Collapse all files

14
milli/src/search/new/ranking_rule_graph/proximity/build.rs
View File

@ -16,9 +16,9 @@ pub fn visit_from_node(from_node: &QueryNode) -> Result<Option<(WordDerivations,
QueryNode::Term(LocatedQueryTerm { value: value1, positions: pos1 }) => {
match value1 {
QueryTerm::Word { derivations } => (derivations.clone(), *pos1.end()),
QueryTerm::Phrase(phrase1) => {
QueryTerm::Phrase { phrase: phrase1 } => {
// TODO: remove second unwrap
let original = phrase1.last().unwrap().as_ref().unwrap().clone();
let original = phrase1.words.last().unwrap().as_ref().unwrap().clone();
(
WordDerivations {
original: original.clone(),
@ -26,6 +26,8 @@ pub fn visit_from_node(from_node: &QueryNode) -> Result<Option<(WordDerivations,
one_typo: vec![],
two_typos: vec![],
use_prefix_db: false,
synonyms: vec![],
split_words: None,
},
*pos1.end(),
)
@ -39,6 +41,8 @@ pub fn visit_from_node(from_node: &QueryNode) -> Result<Option<(WordDerivations,
one_typo: vec![],
two_typos: vec![],
use_prefix_db: false,
synonyms: vec![],
split_words: None,
},
-100,
),
@ -63,9 +67,9 @@ pub fn visit_to_node<'transaction, 'from_data>(
let (derivations2, pos2, ngram_len2) = match value2 {
QueryTerm::Word { derivations } => (derivations.clone(), *pos2.start(), pos2.len()),
QueryTerm::Phrase(phrase2) => {
QueryTerm::Phrase { phrase: phrase2 } => {
// TODO: remove second unwrap
let original = phrase2.last().unwrap().as_ref().unwrap().clone();
let original = phrase2.words.last().unwrap().as_ref().unwrap().clone();
(
WordDerivations {
original: original.clone(),
@ -73,6 +77,8 @@ pub fn visit_to_node<'transaction, 'from_data>(
one_typo: vec![],
two_typos: vec![],
use_prefix_db: false,
synonyms: vec![],
split_words: None,
},
*pos2.start(),
1,

26
milli/src/search/new/ranking_rule_graph/proximity/mod.rs
View File

@ -2,18 +2,21 @@ pub mod build;
pub mod compute_docids;
use heed::RoTxn;
use roaring::RoaringBitmap;
use super::empty_paths_cache::EmptyPathsCache;
use super::paths_map::PathsMap;
use super::{EdgeDetails, RankingRuleGraphTrait};
use crate::new::db_cache::DatabaseCache;
use crate::new::logger::SearchLogger;
use crate::new::query_term::WordDerivations;
use crate::new::QueryNode;
use crate::new::{QueryGraph, QueryNode};
use crate::{Index, Result};
// TODO: intern the strings, refer to them by their pointer?
#[derive(Debug, Clone)]
pub enum WordPair {
// TODO: add WordsSwapped and WordPrefixSwapped case
Words { left: String, right: String },
WordsSwapped { left: String, right: String },
WordPrefix { left: String, right_prefix: String },
@ -22,6 +25,7 @@ pub enum WordPair {
#[derive(Clone)]
pub struct ProximityEdge {
// TODO: use a list of pointers to the word pairs instead?
pairs: Vec<WordPair>,
proximity: u8,
}
@ -67,10 +71,20 @@ impl RankingRuleGraphTrait for ProximityGraph {
fn log_state(
graph: &super::RankingRuleGraph<Self>,
paths: &PathsMap<u64>,
paths: &[Vec<u32>],
empty_paths_cache: &EmptyPathsCache,
logger: &mut dyn crate::new::logger::SearchLogger<crate::new::QueryGraph>,
universe: &RoaringBitmap,
distances: &[Vec<u64>],
cost: u64,
logger: &mut dyn SearchLogger<QueryGraph>,
) {
logger.log_proximity_state(graph, paths, empty_paths_cache);
logger.log_proximity_state(
graph,
paths,
empty_paths_cache,
universe,
distances.to_vec(),
cost,
);
}
}