MeiliSearch/src/rank/mod.rs

151 lines
3.5 KiB
Rust
Raw Normal View History

2018-10-10 16:57:21 +02:00
pub mod criterion;
2018-12-07 11:53:17 +01:00
mod query_builder;
mod distinct_map;
use std::iter::FusedIterator;
use std::slice::Windows;
use sdset::SetBuf;
use group_by::GroupBy;
use crate::{Match, DocumentId};
2018-05-27 15:23:43 +02:00
pub use self::query_builder::{FilterFunc, QueryBuilder, DistinctQueryBuilder};
2018-05-27 15:23:43 +02:00
#[inline]
fn match_query_index(a: &Match, b: &Match) -> bool {
a.query_index == b.query_index
}
#[derive(Debug, Clone)]
2018-05-27 15:23:43 +02:00
pub struct Document {
2018-10-10 16:57:21 +02:00
pub id: DocumentId,
pub matches: Matches,
2018-05-27 15:23:43 +02:00
}
impl Document {
2019-01-06 11:23:21 +01:00
pub fn new(id: DocumentId, match_: Match) -> Self {
let matches = SetBuf::new_unchecked(vec![match_]);
2019-01-06 11:23:21 +01:00
Self::from_matches(id, matches)
2018-05-27 15:23:43 +02:00
}
pub fn from_matches(id: DocumentId, matches: SetBuf<Match>) -> Self {
2019-01-06 11:23:21 +01:00
let matches = Matches::new(matches);
Self { id, matches }
}
pub fn from_unsorted_matches(id: DocumentId, matches: Vec<Match>) -> Self {
let matches = Matches::from_unsorted(matches);
Self { id, matches }
}
}
#[derive(Debug, Clone)]
pub struct Matches {
matches: SetBuf<Match>,
slices: Vec<usize>,
2019-01-06 11:23:21 +01:00
}
impl Matches {
pub fn new(matches: SetBuf<Match>) -> Matches {
let mut last = 0;
let mut slices = vec![0];
2019-01-06 11:23:21 +01:00
for group in GroupBy::new(&matches, match_query_index) {
let index = last + group.len();
slices.push(index);
last = index;
}
2019-01-06 11:23:21 +01:00
Matches { matches, slices }
}
2019-01-06 11:23:21 +01:00
pub fn from_unsorted(mut matches: Vec<Match>) -> Matches {
matches.sort_unstable();
let matches = SetBuf::new_unchecked(matches);
2019-01-06 11:23:21 +01:00
Matches::new(matches)
}
pub fn query_index_groups(&self) -> QueryIndexGroups {
QueryIndexGroups {
matches: &self.matches,
windows: self.slices.windows(2),
}
2018-05-27 15:23:43 +02:00
}
pub fn as_matches(&self) -> &[Match] {
&self.matches
}
2018-05-27 15:23:43 +02:00
}
pub struct QueryIndexGroups<'a, 'b> {
matches: &'a [Match],
windows: Windows<'b, usize>,
}
impl<'a> Iterator for QueryIndexGroups<'a, '_> {
type Item = &'a [Match];
#[inline]
fn next(&mut self) -> Option<Self::Item> {
self.windows.next().map(|range| {
match *range {
[left, right] => &self.matches[left..right],
_ => unreachable!(),
}
})
}
#[inline]
fn size_hint(&self) -> (usize, Option<usize>) {
self.windows.size_hint()
}
#[inline]
fn count(self) -> usize {
self.len()
}
#[inline]
fn nth(&mut self, n: usize) -> Option<Self::Item> {
self.windows.nth(n).map(|range| {
match *range {
[left, right] => &self.matches[left..right],
_ => unreachable!(),
}
})
}
#[inline]
fn last(self) -> Option<Self::Item> {
let (matches, windows) = (self.matches, self.windows);
windows.last().map(|range| {
match *range {
[left, right] => &matches[left..right],
_ => unreachable!(),
}
})
}
}
impl ExactSizeIterator for QueryIndexGroups<'_, '_> {
#[inline]
fn len(&self) -> usize {
self.windows.len()
}
}
impl FusedIterator for QueryIndexGroups<'_, '_> { }
impl DoubleEndedIterator for QueryIndexGroups<'_, '_> {
#[inline]
fn next_back(&mut self) -> Option<Self::Item> {
self.windows.next_back().map(|range| {
match *range {
[left, right] => &self.matches[left..right],
_ => unreachable!(),
}
})
}
}