mirror of
https://github.com/meilisearch/MeiliSearch
synced 2024-12-24 13:40:31 +01:00
Factorize phrase computation
This commit is contained in:
parent
cbb3b25459
commit
b389be48a0
@ -326,43 +326,7 @@ pub fn resolve_query_tree<'t>(
|
||||
}
|
||||
Ok(candidates)
|
||||
}
|
||||
Phrase(words) => {
|
||||
let mut candidates = RoaringBitmap::new();
|
||||
let mut first_iter = true;
|
||||
let winsize = words.len().min(7);
|
||||
|
||||
for win in words.windows(winsize) {
|
||||
// Get all the documents with the matching distance for each word pairs.
|
||||
let mut bitmaps = Vec::with_capacity(winsize.pow(2));
|
||||
for (offset, s1) in win.iter().enumerate() {
|
||||
for (dist, s2) in win.iter().skip(offset + 1).enumerate() {
|
||||
match ctx.word_pair_proximity_docids(s1, s2, dist as u8 + 1)? {
|
||||
Some(m) => bitmaps.push(m),
|
||||
// If there are no document for this distance, there will be no
|
||||
// results for the phrase query.
|
||||
None => return Ok(RoaringBitmap::new()),
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
// We sort the bitmaps so that we perform the small intersections first, which is faster.
|
||||
bitmaps.sort_unstable_by(|a, b| a.len().cmp(&b.len()));
|
||||
|
||||
for bitmap in bitmaps {
|
||||
if first_iter {
|
||||
candidates = bitmap;
|
||||
first_iter = false;
|
||||
} else {
|
||||
candidates &= bitmap;
|
||||
}
|
||||
// There will be no match, return early
|
||||
if candidates.is_empty() {
|
||||
break;
|
||||
}
|
||||
}
|
||||
}
|
||||
Ok(candidates)
|
||||
}
|
||||
Phrase(words) => resolve_phrase(ctx, &words),
|
||||
Or(_, ops) => {
|
||||
let mut candidates = RoaringBitmap::new();
|
||||
for op in ops {
|
||||
@ -378,6 +342,44 @@ pub fn resolve_query_tree<'t>(
|
||||
resolve_operation(ctx, query_tree, wdcache)
|
||||
}
|
||||
|
||||
pub fn resolve_phrase<'t>(ctx: &'t dyn Context, phrase: &[String]) -> Result<RoaringBitmap> {
|
||||
let mut candidates = RoaringBitmap::new();
|
||||
let mut first_iter = true;
|
||||
let winsize = phrase.len().min(7);
|
||||
|
||||
for win in phrase.windows(winsize) {
|
||||
// Get all the documents with the matching distance for each word pairs.
|
||||
let mut bitmaps = Vec::with_capacity(winsize.pow(2));
|
||||
for (offset, s1) in win.iter().enumerate() {
|
||||
for (dist, s2) in win.iter().skip(offset + 1).enumerate() {
|
||||
match ctx.word_pair_proximity_docids(s1, s2, dist as u8 + 1)? {
|
||||
Some(m) => bitmaps.push(m),
|
||||
// If there are no document for this distance, there will be no
|
||||
// results for the phrase query.
|
||||
None => return Ok(RoaringBitmap::new()),
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
// We sort the bitmaps so that we perform the small intersections first, which is faster.
|
||||
bitmaps.sort_unstable_by(|a, b| a.len().cmp(&b.len()));
|
||||
|
||||
for bitmap in bitmaps {
|
||||
if first_iter {
|
||||
candidates = bitmap;
|
||||
first_iter = false;
|
||||
} else {
|
||||
candidates &= bitmap;
|
||||
}
|
||||
// There will be no match, return early
|
||||
if candidates.is_empty() {
|
||||
break;
|
||||
}
|
||||
}
|
||||
}
|
||||
Ok(candidates)
|
||||
}
|
||||
|
||||
fn all_word_pair_proximity_docids<T: AsRef<str>, U: AsRef<str>>(
|
||||
ctx: &dyn Context,
|
||||
left_words: &[(T, u8)],
|
||||
|
@ -6,8 +6,8 @@ use log::debug;
|
||||
use roaring::RoaringBitmap;
|
||||
|
||||
use super::{
|
||||
query_docids, query_pair_proximity_docids, resolve_query_tree, Context, Criterion,
|
||||
CriterionParameters, CriterionResult,
|
||||
query_docids, query_pair_proximity_docids, resolve_phrase, resolve_query_tree, Context,
|
||||
Criterion, CriterionParameters, CriterionResult,
|
||||
};
|
||||
use crate::search::query_tree::{maximum_proximity, Operation, Query, QueryKind};
|
||||
use crate::search::{build_dfa, WordDerivationsCache};
|
||||
@ -192,42 +192,9 @@ fn resolve_candidates<'t>(
|
||||
let most_right = words
|
||||
.last()
|
||||
.map(|w| Query { prefix: false, kind: QueryKind::exact(w.clone()) });
|
||||
let mut candidates = RoaringBitmap::new();
|
||||
let mut first_iter = true;
|
||||
let winsize = words.len().min(7);
|
||||
|
||||
for win in words.windows(winsize) {
|
||||
// Get all the documents with the matching distance for each word pairs.
|
||||
let mut bitmaps = Vec::with_capacity(winsize.pow(2));
|
||||
for (offset, s1) in win.iter().enumerate() {
|
||||
for (dist, s2) in win.iter().skip(offset + 1).enumerate() {
|
||||
match ctx.word_pair_proximity_docids(s1, s2, dist as u8 + 1)? {
|
||||
Some(m) => bitmaps.push(m),
|
||||
// If there are no document for this distance, there will be no
|
||||
// results for the phrase query.
|
||||
None => return Ok(Default::default()),
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
// We sort the bitmaps so that we perform the small intersections first, which is faster.
|
||||
bitmaps.sort_unstable_by(|a, b| a.len().cmp(&b.len()));
|
||||
|
||||
for bitmap in bitmaps {
|
||||
if first_iter {
|
||||
candidates = bitmap;
|
||||
first_iter = false;
|
||||
} else {
|
||||
candidates &= bitmap;
|
||||
}
|
||||
// There will be no match, return early
|
||||
if candidates.is_empty() {
|
||||
break;
|
||||
}
|
||||
}
|
||||
}
|
||||
match (most_left, most_right) {
|
||||
(Some(l), Some(r)) => vec![(l, r, candidates)],
|
||||
(Some(l), Some(r)) => vec![(l, r, resolve_phrase(ctx, &words)?)],
|
||||
_otherwise => Default::default(),
|
||||
}
|
||||
} else {
|
||||
|
Loading…
x
Reference in New Issue
Block a user