searchengines-web/MeiliSearch
searchengines-web/MeiliSearch
1
0
Fork 0
mirror of https://github.com/meilisearch/MeiliSearch synced 2024-12-23 21:20:24 +01:00
Code Issues Releases Wiki Activity

Introduce a WordDerivationsCache struct

Browse Source
This commit is contained in:
Clément Renault 2021-03-05 11:02:24 +01:00 committed by Kerollmops
parent 2606c92ef9
commit 5fcaedb880
No known key found for this signature in database
GPG Key ID: 92ADA4E935E71FA4
7 changed files with 169 additions and 100 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

7
milli/src/search/criteria/asc_desc.rs
View File

@ -15,6 +15,7 @@ use crate::heed_codec::facet::{FieldDocIdFacetI64Codec, FieldDocIdFacetF64Codec}
use crate::search::criteria::{resolve_query_tree, CriteriaBuilder};
use crate::search::facet::FacetIter;
use crate::search::query_tree::Operation;
use crate::search::WordDerivationsCache;
use crate::{FieldsIdsMap, FieldId, Index};
use super::{Criterion, CriterionResult};
@ -92,7 +93,7 @@ impl<'t> AscDesc<'t> {
let candidates = match &query_tree {
Some(qt) => {
let context = CriteriaBuilder::new(rtxn, index)?;
let mut qt_candidates = resolve_query_tree(&context, qt, &mut HashMap::new())?;
let mut qt_candidates = resolve_query_tree(&context, qt, &mut HashMap::new(), &mut WordDerivationsCache::new())?;
if let Some(candidates) = candidates {
qt_candidates.intersect_with(&candidates);
}
@ -145,7 +146,7 @@ impl<'t> AscDesc<'t> {
}
impl<'t> Criterion for AscDesc<'t> {
fn next(&mut self) -> anyhow::Result<Option<CriterionResult>> {
fn next(&mut self, wdcache: &mut WordDerivationsCache) -> anyhow::Result<Option<CriterionResult>> {
loop {
debug!("Facet {}({}) iteration",
if self.ascending { "Asc" } else { "Desc" }, self.field_name
@ -157,7 +158,7 @@ impl<'t> Criterion for AscDesc<'t> {
let bucket_candidates = take(&mut self.bucket_candidates);
match self.parent.as_mut() {
Some(parent) => {
match parent.next()? {
match parent.next(wdcache)? {
Some(CriterionResult { query_tree, mut candidates, bucket_candidates }) => {
self.query_tree = query_tree;
candidates.intersect_with(&self.faceted_candidates);

11
milli/src/search/criteria/fetcher.rs
View File

@ -5,6 +5,7 @@ use log::debug;
use roaring::RoaringBitmap;
use crate::search::query_tree::Operation;
use crate::search::WordDerivationsCache;
use super::{resolve_query_tree, Candidates, Criterion, CriterionResult, Context};
pub struct Fetcher<'t> {
@ -47,7 +48,7 @@ impl<'t> Fetcher<'t> {
}
impl<'t> Criterion for Fetcher<'t> {
fn next(&mut self) -> anyhow::Result<Option<CriterionResult>> {
fn next(&mut self, wdcache: &mut WordDerivationsCache) -> anyhow::Result<Option<CriterionResult>> {
use Candidates::{Allowed, Forbidden};
loop {
debug!("Fetcher iteration (should_get_documents_ids: {}) ({:?})",
@ -60,7 +61,7 @@ impl<'t> Criterion for Fetcher<'t> {
let candidates = take(&mut self.candidates).into_inner();
let candidates = match &self.query_tree {
Some(qt) if should_get_documents_ids => {
let mut docids = resolve_query_tree(self.ctx, &qt, &mut HashMap::new())?;
let mut docids = resolve_query_tree(self.ctx, &qt, &mut HashMap::new(), wdcache)?;
docids.intersect_with(&candidates);
docids
},
@ -76,11 +77,11 @@ impl<'t> Criterion for Fetcher<'t> {
Forbidden(_) => {
match self.parent.as_mut() {
Some(parent) => {
match parent.next()? {
match parent.next(wdcache)? {
Some(result) => return Ok(Some(result)),
None => if should_get_documents_ids {
let candidates = match &self.query_tree {
Some(qt) => resolve_query_tree(self.ctx, &qt, &mut HashMap::new())?,
Some(qt) => resolve_query_tree(self.ctx, &qt, &mut HashMap::new(), wdcache)?,
None => self.ctx.documents_ids()?,
};
@ -94,7 +95,7 @@ impl<'t> Criterion for Fetcher<'t> {
},
None => if should_get_documents_ids {
let candidates = match &self.query_tree {
Some(qt) => resolve_query_tree(self.ctx, &qt, &mut HashMap::new())?,
Some(qt) => resolve_query_tree(self.ctx, &qt, &mut HashMap::new(), wdcache)?,
None => self.ctx.documents_ids()?,
};

57
milli/src/search/criteria/mod.rs
View File

@ -4,8 +4,8 @@ use std::borrow::Cow;
use anyhow::bail;
use roaring::RoaringBitmap;
use crate::search::word_derivations;
use crate::{DocumentId, Index};
use crate::search::{word_derivations, WordDerivationsCache};
use crate::{Index, DocumentId};
use super::query_tree::{Operation, Query, QueryKind};
use self::typo::Typo;
@ -21,7 +21,7 @@ pub mod proximity;
pub mod fetcher;
pub trait Criterion {
fn next(&mut self) -> anyhow::Result<Option<CriterionResult>>;
fn next(&mut self, wdcache: &mut WordDerivationsCache) -> anyhow::Result<Option<CriterionResult>>;
}
/// The result of a call to the parent criterion.
@ -164,12 +164,14 @@ pub fn resolve_query_tree<'t>(
ctx: &'t dyn Context,
query_tree: &Operation,
cache: &mut HashMap<(Operation, u8), RoaringBitmap>,
wdcache: &mut WordDerivationsCache,
) -> anyhow::Result<RoaringBitmap>
{
fn resolve_operation<'t>(
ctx: &'t dyn Context,
query_tree: &Operation,
cache: &mut HashMap<(Operation, u8), RoaringBitmap>,
wdcache: &mut WordDerivationsCache,
) -> anyhow::Result<RoaringBitmap>
{
use Operation::{And, Consecutive, Or, Query};
@ -177,7 +179,7 @@ pub fn resolve_query_tree<'t>(
match query_tree {
And(ops) => {
let mut ops = ops.iter().map(|op| {
resolve_operation(ctx, op, cache)
resolve_operation(ctx, op, cache, wdcache)
}).collect::<anyhow::Result<Vec<_>>>()?;
ops.sort_unstable_by_key(|cds| cds.len());
@ -200,7 +202,7 @@ pub fn resolve_query_tree<'t>(
for slice in ops.windows(2) {
match (&slice[0], &slice[1]) {
(Operation::Query(left), Operation::Query(right)) => {
match query_pair_proximity_docids(ctx, left, right, 1)? {
match query_pair_proximity_docids(ctx, left, right, 1, wdcache)? {
pair_docids if pair_docids.is_empty() => {
return Ok(RoaringBitmap::new())
},
@ -221,16 +223,16 @@ pub fn resolve_query_tree<'t>(
Or(_, ops) => {
let mut candidates = RoaringBitmap::new();
for op in ops {
let docids = resolve_operation(ctx, op, cache)?;
let docids = resolve_operation(ctx, op, cache, wdcache)?;
candidates.union_with(&docids);
}
Ok(candidates)
},
Query(q) => Ok(query_docids(ctx, q)?),
Query(q) => Ok(query_docids(ctx, q, wdcache)?),
}
}
resolve_operation(ctx, query_tree, cache)
resolve_operation(ctx, query_tree, cache, wdcache)
}
@ -239,7 +241,8 @@ fn all_word_pair_proximity_docids<T: AsRef<str>, U: AsRef<str>>(
left_words: &[(T, u8)],
right_words: &[(U, u8)],
proximity: u8
) -> anyhow::Result<RoaringBitmap> {
) -> anyhow::Result<RoaringBitmap>
{
let mut docids = RoaringBitmap::new();
for (left, _l_typo) in left_words {
for (right, _r_typo) in right_words {
@ -250,13 +253,18 @@ fn all_word_pair_proximity_docids<T: AsRef<str>, U: AsRef<str>>(
Ok(docids)
}
fn query_docids(ctx: &dyn Context, query: &Query) -> anyhow::Result<RoaringBitmap> {
fn query_docids(
ctx: &dyn Context,
query: &Query,
wdcache: &mut WordDerivationsCache,
) -> anyhow::Result<RoaringBitmap>
{
match &query.kind {
QueryKind::Exact { word, .. } => {
if query.prefix && ctx.in_prefix_cache(&word) {
Ok(ctx.word_prefix_docids(&word)?.unwrap_or_default())
} else if query.prefix {
let words = word_derivations(&word, true, 0, ctx.words_fst())?;
let words = word_derivations(&word, true, 0, ctx.words_fst(), wdcache)?;
let mut docids = RoaringBitmap::new();
for (word, _typo) in words {
let current_docids = ctx.word_docids(&word)?.unwrap_or_default();
@ -268,7 +276,7 @@ fn query_docids(ctx: &dyn Context, query: &Query) -> anyhow::Result<RoaringBitma
}
},
QueryKind::Tolerant { typo, word } => {
let words = word_derivations(&word, query.prefix, *typo, ctx.words_fst())?;
let words = word_derivations(&word, query.prefix, *typo, ctx.words_fst(), wdcache)?;
let mut docids = RoaringBitmap::new();
for (word, _typo) in words {
let current_docids = ctx.word_docids(&word)?.unwrap_or_default();
@ -279,10 +287,17 @@ fn query_docids(ctx: &dyn Context, query: &Query) -> anyhow::Result<RoaringBitma
}
}
fn query_pair_proximity_docids(ctx: &dyn Context, left: &Query, right: &Query, proximity: u8) -> anyhow::Result<RoaringBitmap> {
fn query_pair_proximity_docids(
ctx: &dyn Context,
left: &Query,
right: &Query,
proximity: u8,
wdcache: &mut WordDerivationsCache,
) -> anyhow::Result<RoaringBitmap>
{
if proximity >= 8 {
let mut candidates = query_docids(ctx, left)?;
let right_candidates = query_docids(ctx, right)?;
let mut candidates = query_docids(ctx, left, wdcache)?;
let right_candidates = query_docids(ctx, right, wdcache)?;
candidates.intersect_with(&right_candidates);
return Ok(candidates);
}
@ -293,14 +308,14 @@ fn query_pair_proximity_docids(ctx: &dyn Context, left: &Query, right: &Query, p
if prefix && ctx.in_prefix_cache(&right) {
Ok(ctx.word_prefix_pair_proximity_docids(left.as_str(), right.as_str(), proximity)?.unwrap_or_default())
} else if prefix {
let r_words = word_derivations(&right, true, 0, ctx.words_fst())?;
let r_words = word_derivations(&right, true, 0, ctx.words_fst(), wdcache)?;
all_word_pair_proximity_docids(ctx, &[(left, 0)], &r_words, proximity)
} else {
Ok(ctx.word_pair_proximity_docids(left.as_str(), right.as_str(), proximity)?.unwrap_or_default())
}
},
(QueryKind::Tolerant { typo, word: left }, QueryKind::Exact { word: right, .. }) => {
let l_words = word_derivations(&left, false, *typo, ctx.words_fst())?;
let l_words = word_derivations(&left, false, *typo, ctx.words_fst(), wdcache)?.to_owned();
if prefix && ctx.in_prefix_cache(&right) {
let mut docids = RoaringBitmap::new();
for (left, _) in l_words {
@ -309,19 +324,19 @@ fn query_pair_proximity_docids(ctx: &dyn Context, left: &Query, right: &Query, p
}
Ok(docids)
} else if prefix {
let r_words = word_derivations(&right, true, 0, ctx.words_fst())?;
let r_words = word_derivations(&right, true, 0, ctx.words_fst(), wdcache)?;
all_word_pair_proximity_docids(ctx, &l_words, &r_words, proximity)
} else {
all_word_pair_proximity_docids(ctx, &l_words, &[(right, 0)], proximity)
}
},
(QueryKind::Exact { word: left, .. }, QueryKind::Tolerant { typo, word: right }) => {
let r_words = word_derivations(&right, prefix, *typo, ctx.words_fst())?;
let r_words = word_derivations(&right, prefix, *typo, ctx.words_fst(), wdcache)?;
all_word_pair_proximity_docids(ctx, &[(left, 0)], &r_words, proximity)
},
(QueryKind::Tolerant { typo: l_typo, word: left }, QueryKind::Tolerant { typo: r_typo, word: right }) => {
let l_words = word_derivations(&left, false, *l_typo, ctx.words_fst())?;
let r_words = word_derivations(&right, prefix, *r_typo, ctx.words_fst())?;
let l_words = word_derivations(&left, false, *l_typo, ctx.words_fst(), wdcache)?.to_owned();
let r_words = word_derivations(&right, prefix, *r_typo, ctx.words_fst(), wdcache)?;
all_word_pair_proximity_docids(ctx, &l_words, &r_words, proximity)
},
}

75
milli/src/search/criteria/proximity.rs
View File

@ -1,3 +1,4 @@
use std::borrow::Cow;
use std::collections::{BTreeMap, HashMap, btree_map};
use std::mem::take;
@ -6,6 +7,7 @@ use log::debug;
use crate::{DocumentId, Position, search::{query_tree::QueryKind, word_derivations}};
use crate::search::query_tree::{maximum_proximity, Operation, Query};
use crate::search::WordDerivationsCache;
use super::{Candidates, Criterion, CriterionResult, Context, query_docids, query_pair_proximity_docids};
pub struct Proximity<'t> {
@ -53,7 +55,7 @@ impl<'t> Proximity<'t> {
}
impl<'t> Criterion for Proximity<'t> {
fn next(&mut self) -> anyhow::Result<Option<CriterionResult>> {
fn next(&mut self, wdcache: &mut WordDerivationsCache) -> anyhow::Result<Option<CriterionResult>> {
use Candidates::{Allowed, Forbidden};
loop {
debug!("Proximity at iteration {} (max {:?}) ({:?})",
@ -94,7 +96,8 @@ impl<'t> Criterion for Proximity<'t> {
let cache = resolve_plane_sweep_candidates(
self.ctx,
query_tree,
candidates
candidates,
wdcache,
)?;
self.plane_sweep_cache = Some(cache.into_iter());
@ -106,6 +109,7 @@ impl<'t> Criterion for Proximity<'t> {
&query_tree,
self.proximity,
&mut self.candidates_cache,
wdcache,
)?
};
@ -135,6 +139,7 @@ impl<'t> Criterion for Proximity<'t> {
&query_tree,
self.proximity,
&mut self.candidates_cache,
wdcache,
)?;
new_candidates.difference_with(&candidates);
@ -164,7 +169,7 @@ impl<'t> Criterion for Proximity<'t> {
(None, Forbidden(_)) => {
match self.parent.as_mut() {
Some(parent) => {
match parent.next()? {
match parent.next(wdcache)? {
Some(CriterionResult { query_tree, candidates, bucket_candidates }) => {
self.query_tree = query_tree.map(|op| (maximum_proximity(&op), op));
self.proximity = 0;
@ -188,6 +193,7 @@ fn resolve_candidates<'t>(
query_tree: &Operation,
proximity: u8,
cache: &mut HashMap<(Operation, u8), Vec<(Query, Query, RoaringBitmap)>>,
wdcache: &mut WordDerivationsCache,
) -> anyhow::Result<RoaringBitmap>
{
fn resolve_operation<'t>(
@ -195,27 +201,28 @@ fn resolve_candidates<'t>(
query_tree: &Operation,
proximity: u8,
cache: &mut HashMap<(Operation, u8), Vec<(Query, Query, RoaringBitmap)>>,
wdcache: &mut WordDerivationsCache,
) -> anyhow::Result<Vec<(Query, Query, RoaringBitmap)>>
{
use Operation::{And, Consecutive, Or, Query};
let result = match query_tree {
And(ops) => mdfs(ctx, ops, proximity, cache)?,
And(ops) => mdfs(ctx, ops, proximity, cache, wdcache)?,
Consecutive(ops) => if proximity == 0 {
mdfs(ctx, ops, 0, cache)?
mdfs(ctx, ops, 0, cache, wdcache)?
} else {
Default::default()
},
Or(_, ops) => {
let mut output = Vec::new();
for op in ops {
let result = resolve_operation(ctx, op, proximity, cache)?;
let result = resolve_operation(ctx, op, proximity, cache, wdcache)?;
output.extend(result);
}
output
},
Query(q) => if proximity == 0 {
let candidates = query_docids(ctx, q)?;
let candidates = query_docids(ctx, q, wdcache)?;
vec![(q.clone(), q.clone(), candidates)]
} else {
Default::default()
@ -231,6 +238,7 @@ fn resolve_candidates<'t>(
right: &Operation,
proximity: u8,
cache: &mut HashMap<(Operation, u8), Vec<(Query, Query, RoaringBitmap)>>,
wdcache: &mut WordDerivationsCache,
) -> anyhow::Result<Vec<(Query, Query, RoaringBitmap)>>
{
fn pair_combinations(mana: u8, left_max: u8) -> impl Iterator<Item = (u8, u8)> {
@ -245,13 +253,13 @@ fn resolve_candidates<'t>(
for (left_p, right_p) in pair_combinations(left_right_p, left_right_p) {
let left_key = (left.clone(), left_p);
if !cache.contains_key(&left_key) {
let candidates = resolve_operation(ctx, left, left_p, cache)?;
let candidates = resolve_operation(ctx, left, left_p, cache, wdcache)?;
cache.insert(left_key.clone(), candidates);
}
let right_key = (right.clone(), right_p);
if !cache.contains_key(&right_key) {
let candidates = resolve_operation(ctx, right, right_p, cache)?;
let candidates = resolve_operation(ctx, right, right_p, cache, wdcache)?;
cache.insert(right_key.clone(), candidates);
}
@ -260,7 +268,7 @@ fn resolve_candidates<'t>(
for (ll, lr, lcandidates) in lefts {
for (rl, rr, rcandidates) in rights {
let mut candidates = query_pair_proximity_docids(ctx, lr, rl, pair_p + 1)?;
let mut candidates = query_pair_proximity_docids(ctx, lr, rl, pair_p + 1, wdcache)?;
if lcandidates.len() < rcandidates.len() {
candidates.intersect_with(lcandidates);
candidates.intersect_with(rcandidates);
@ -284,6 +292,7 @@ fn resolve_candidates<'t>(
branches: &[Operation],
proximity: u8,
cache: &mut HashMap<(Operation, u8), Vec<(Query, Query, RoaringBitmap)>>,
wdcache: &mut WordDerivationsCache,
) -> anyhow::Result<Vec<(Query, Query, RoaringBitmap)>>
{
// Extract the first two elements but gives the tail
@ -293,13 +302,13 @@ fn resolve_candidates<'t>(
});
match next {
Some((head1, Some((head2, [_])))) => mdfs_pair(ctx, head1, head2, proximity, cache),
Some((head1, Some((head2, [_])))) => mdfs_pair(ctx, head1, head2, proximity, cache, wdcache),
Some((head1, Some((head2, tail)))) => {
let mut output = Vec::new();
for p in 0..=proximity {
for (lhead, _, head_candidates) in mdfs_pair(ctx, head1, head2, p, cache)? {
for (lhead, _, head_candidates) in mdfs_pair(ctx, head1, head2, p, cache, wdcache)? {
if !head_candidates.is_empty() {
for (_, rtail, mut candidates) in mdfs(ctx, tail, proximity - p, cache)? {
for (_, rtail, mut candidates) in mdfs(ctx, tail, proximity - p, cache, wdcache)? {
candidates.intersect_with(&head_candidates);
if !candidates.is_empty() {
output.push((lhead.clone(), rtail, candidates));
@ -310,13 +319,13 @@ fn resolve_candidates<'t>(
}
Ok(output)
},
Some((head1, None)) => resolve_operation(ctx, head1, proximity, cache),
Some((head1, None)) => resolve_operation(ctx, head1, proximity, cache, wdcache),
None => return Ok(Default::default()),
}
}
let mut candidates = RoaringBitmap::new();
for (_, _, cds) in resolve_operation(ctx, query_tree, proximity, cache)? {
for (_, _, cds) in resolve_operation(ctx, query_tree, proximity, cache, wdcache)? {
candidates.union_with(&cds);
}
Ok(candidates)
@ -326,6 +335,7 @@ fn resolve_plane_sweep_candidates<'t>(
ctx: &'t dyn Context,
query_tree: &Operation,
allowed_candidates: &RoaringBitmap,
wdcache: &mut WordDerivationsCache,
) -> anyhow::Result<BTreeMap<u8, RoaringBitmap>>
{
/// FIXME may be buggy with query like "new new york"
@ -334,8 +344,14 @@ fn resolve_plane_sweep_candidates<'t>(
operations: &[Operation],
docid: DocumentId,
consecutive: bool,
) -> anyhow::Result<Vec<(Position, u8, Position)>> {
fn compute_groups_proximity(groups: &Vec<(usize, (Position, u8, Position))>, consecutive: bool) -> Option<(Position, u8, Position)> {
wdcache: &mut WordDerivationsCache,
) -> anyhow::Result<Vec<(Position, u8, Position)>>
{
fn compute_groups_proximity(
groups: &[(usize, (Position, u8, Position))],
consecutive: bool,
) -> Option<(Position, u8, Position)>
{
// take the inner proximity of the first group as initial
let mut proximity = groups.first()?.1.1;
let left_most_pos = groups.first()?.1.0;
@ -360,14 +376,16 @@ fn resolve_plane_sweep_candidates<'t>(
// if groups should be consecutives, we will only accept groups with a proximity of 0
if !consecutive || proximity == 0 {
Some((left_most_pos, proximity, right_most_pos))
} else { None }
} else {
None
}
}
let groups_len = operations.len();
let mut groups_positions = Vec::with_capacity(groups_len);
for operation in operations {
let positions = resolve_operation(ctx, operation, docid)?;
let positions = resolve_operation(ctx, operation, docid, wdcache)?;
groups_positions.push(positions.into_iter());
}
@ -442,16 +460,17 @@ fn resolve_plane_sweep_candidates<'t>(
ctx: &'t dyn Context,
query_tree: &Operation,
docid: DocumentId,
wdcache: &mut WordDerivationsCache,
) -> anyhow::Result<Vec<(Position, u8, Position)>> {
use Operation::{And, Consecutive, Or};
match query_tree {
And(ops) => plane_sweep(ctx, ops, docid, false),
Consecutive(ops) => plane_sweep(ctx, ops, docid, true),
And(ops) => plane_sweep(ctx, ops, docid, false, wdcache),
Consecutive(ops) => plane_sweep(ctx, ops, docid, true, wdcache),
Or(_, ops) => {
let mut result = Vec::new();
for op in ops {
result.extend(resolve_operation(ctx, op, docid)?)
result.extend(resolve_operation(ctx, op, docid, wdcache)?)
}
result.sort_unstable();
@ -462,19 +481,19 @@ fn resolve_plane_sweep_candidates<'t>(
let words = match kind {
QueryKind::Exact { word, .. } => {
if *prefix {
word_derivations(word, true, 0, fst)?
Cow::Borrowed(word_derivations(word, true, 0, fst, wdcache)?)
} else {
vec![(word.to_string(), 0)]
Cow::Owned(vec![(word.to_string(), 0)])
}
},
QueryKind::Tolerant { typo, word } => {
word_derivations(word, *prefix, *typo, fst)?
Cow::Borrowed(word_derivations(word, *prefix, *typo, fst, wdcache)?)
}
};
let mut result = Vec::new();
for (word, _) in words {
if let Some(positions) = ctx.docid_word_positions(docid, &word)? {
for (word, _) in words.as_ref() {
if let Some(positions) = ctx.docid_word_positions(docid, word)? {
let iter = positions.iter().map(|p| (p, 0, p));
result.extend(iter);
}
@ -488,7 +507,7 @@ fn resolve_plane_sweep_candidates<'t>(
let mut candidates = BTreeMap::new();
for docid in allowed_candidates {
let positions = resolve_operation(ctx, query_tree, docid)?;
let positions = resolve_operation(ctx, query_tree, docid, wdcache)?;
let best_proximity = positions.into_iter().min_by_key(|(_, proximity, _)| *proximity);
let best_proximity = best_proximity.map(|(_, proximity, _)| proximity).unwrap_or(7);
candidates.entry(best_proximity).or_insert_with(RoaringBitmap::new).insert(docid);

75
milli/src/search/criteria/typo.rs
View File

@ -5,7 +5,7 @@ use log::debug;
use roaring::RoaringBitmap;
use crate::search::query_tree::{maximum_typo, Operation, Query, QueryKind};
use crate::search::word_derivations;
use crate::search::{word_derivations, WordDerivationsCache};
use super::{Candidates, Criterion, CriterionResult, Context, query_docids, query_pair_proximity_docids};
pub struct Typo<'t> {
@ -53,7 +53,7 @@ impl<'t> Typo<'t> {
}
impl<'t> Criterion for Typo<'t> {
fn next(&mut self) -> anyhow::Result<Option<CriterionResult>> {
fn next(&mut self, wdcache: &mut WordDerivationsCache) -> anyhow::Result<Option<CriterionResult>> {
use Candidates::{Allowed, Forbidden};
loop {
debug!("Typo at iteration {} ({:?})", self.number_typos, self.candidates);
@ -73,15 +73,21 @@ impl<'t> Criterion for Typo<'t> {
} else {
let fst = self.ctx.words_fst();
let new_query_tree = if self.number_typos < 2 {
alterate_query_tree(&fst, query_tree.clone(), self.number_typos, &mut self.typo_cache)?
alterate_query_tree(&fst, query_tree.clone(), self.number_typos, &mut self.typo_cache, wdcache)?
} else if self.number_typos == 2 {
*query_tree = alterate_query_tree(&fst, query_tree.clone(), self.number_typos, &mut self.typo_cache)?;
*query_tree = alterate_query_tree(&fst, query_tree.clone(), self.number_typos, &mut self.typo_cache, wdcache)?;
query_tree.clone()
} else {
query_tree.clone()
};
let mut new_candidates = resolve_candidates(self.ctx, &new_query_tree, self.number_typos, &mut self.candidates_cache)?;
let mut new_candidates = resolve_candidates(
self.ctx,
&new_query_tree,
self.number_typos,
&mut self.candidates_cache,
wdcache,
)?;
new_candidates.intersect_with(&candidates);
candidates.difference_with(&new_candidates);
self.number_typos += 1;
@ -105,15 +111,21 @@ impl<'t> Criterion for Typo<'t> {
} else {
let fst = self.ctx.words_fst();
let new_query_tree = if self.number_typos < 2 {
alterate_query_tree(&fst, query_tree.clone(), self.number_typos, &mut self.typo_cache)?
alterate_query_tree(&fst, query_tree.clone(), self.number_typos, &mut self.typo_cache, wdcache)?
} else if self.number_typos == 2 {
*query_tree = alterate_query_tree(&fst, query_tree.clone(), self.number_typos, &mut self.typo_cache)?;
*query_tree = alterate_query_tree(&fst, query_tree.clone(), self.number_typos, &mut self.typo_cache, wdcache)?;
query_tree.clone()
} else {
query_tree.clone()
};
let mut new_candidates = resolve_candidates(self.ctx, &new_query_tree, self.number_typos, &mut self.candidates_cache)?;
let mut new_candidates = resolve_candidates(
self.ctx,
&new_query_tree,
self.number_typos,
&mut self.candidates_cache,
wdcache,
)?;
new_candidates.difference_with(&candidates);
candidates.union_with(&new_candidates);
self.number_typos += 1;
@ -141,7 +153,7 @@ impl<'t> Criterion for Typo<'t> {
(None, Forbidden(_)) => {
match self.parent.as_mut() {
Some(parent) => {
match parent.next()? {
match parent.next(wdcache)? {
Some(CriterionResult { query_tree, candidates, bucket_candidates }) => {
self.query_tree = query_tree.map(|op| (maximum_typo(&op), op));
self.number_typos = 0;
@ -167,6 +179,7 @@ fn alterate_query_tree(
mut query_tree: Operation,
number_typos: u8,
typo_cache: &mut HashMap<(String, bool, u8), Vec<(String, u8)>>,
wdcache: &mut WordDerivationsCache,
) -> anyhow::Result<Operation>
{
fn recurse(
@ -174,13 +187,14 @@ fn alterate_query_tree(
operation: &mut Operation,
number_typos: u8,
typo_cache: &mut HashMap<(String, bool, u8), Vec<(String, u8)>>,
wdcache: &mut WordDerivationsCache,
) -> anyhow::Result<()>
{
use Operation::{And, Consecutive, Or};
match operation {
And(ops) | Consecutive(ops) | Or(_, ops) => {
ops.iter_mut().try_for_each(|op| recurse(words_fst, op, number_typos, typo_cache))
ops.iter_mut().try_for_each(|op| recurse(words_fst, op, number_typos, typo_cache, wdcache))
},
Operation::Query(q) => {
// TODO may be optimized when number_typos == 0
@ -198,7 +212,7 @@ fn alterate_query_tree(
let words = if let Some(derivations) = typo_cache.get(&cache_key) {
derivations.clone()
} else {
let derivations = word_derivations(word, q.prefix, typo, words_fst)?;
let derivations = word_derivations(word, q.prefix, typo, words_fst, wdcache)?.to_vec();
typo_cache.insert(cache_key, derivations.clone());
derivations
};
@ -219,7 +233,7 @@ fn alterate_query_tree(
}
}
recurse(words_fst, &mut query_tree, number_typos, typo_cache)?;
recurse(words_fst, &mut query_tree, number_typos, typo_cache, wdcache)?;
Ok(query_tree)
}
@ -228,6 +242,7 @@ fn resolve_candidates<'t>(
query_tree: &Operation,
number_typos: u8,
cache: &mut HashMap<(Operation, u8), RoaringBitmap>,
wdcache: &mut WordDerivationsCache,
) -> anyhow::Result<RoaringBitmap>
{
fn resolve_operation<'t>(
@ -235,13 +250,14 @@ fn resolve_candidates<'t>(
query_tree: &Operation,
number_typos: u8,
cache: &mut HashMap<(Operation, u8), RoaringBitmap>,
wdcache: &mut WordDerivationsCache,
) -> anyhow::Result<RoaringBitmap>
{
use Operation::{And, Consecutive, Or, Query};
match query_tree {
And(ops) => {
mdfs(ctx, ops, number_typos, cache)
mdfs(ctx, ops, number_typos, cache, wdcache)
},
Consecutive(ops) => {
let mut candidates = RoaringBitmap::new();
@ -249,7 +265,7 @@ fn resolve_candidates<'t>(
for slice in ops.windows(2) {
match (&slice[0], &slice[1]) {
(Operation::Query(left), Operation::Query(right)) => {
match query_pair_proximity_docids(ctx, left, right, 1)? {
match query_pair_proximity_docids(ctx, left, right, 1, wdcache)? {
pair_docids if pair_docids.is_empty() => {
return Ok(RoaringBitmap::new())
},
@ -270,13 +286,13 @@ fn resolve_candidates<'t>(
Or(_, ops) => {
let mut candidates = RoaringBitmap::new();
for op in ops {
let docids = resolve_operation(ctx, op, number_typos, cache)?;
let docids = resolve_operation(ctx, op, number_typos, cache, wdcache)?;
candidates.union_with(&docids);
}
Ok(candidates)
},
Query(q) => if q.kind.typo() == number_typos {
Ok(query_docids(ctx, q)?)
Ok(query_docids(ctx, q, wdcache)?)
} else {
Ok(RoaringBitmap::new())
},
@ -288,6 +304,7 @@ fn resolve_candidates<'t>(
branches: &[Operation],
mana: u8,
cache: &mut HashMap<(Operation, u8), RoaringBitmap>,
wdcache: &mut WordDerivationsCache,
) -> anyhow::Result<RoaringBitmap>
{
match branches.split_first() {
@ -296,7 +313,7 @@ fn resolve_candidates<'t>(
if let Some(candidates) = cache.get(&cache_key) {
Ok(candidates.clone())
} else {
let candidates = resolve_operation(ctx, head, mana, cache)?;
let candidates = resolve_operation(ctx, head, mana, cache, wdcache)?;
cache.insert(cache_key, candidates.clone());
Ok(candidates)
}
@ -310,13 +327,13 @@ fn resolve_candidates<'t>(
if let Some(candidates) = cache.get(&cache_key) {
candidates.clone()
} else {
let candidates = resolve_operation(ctx, head, m, cache)?;
let candidates = resolve_operation(ctx, head, m, cache, wdcache)?;
cache.insert(cache_key, candidates.clone());
candidates
}
};
if !head_candidates.is_empty() {
let tail_candidates = mdfs(ctx, tail, mana - m, cache)?;
let tail_candidates = mdfs(ctx, tail, mana - m, cache, wdcache)?;
head_candidates.intersect_with(&tail_candidates);
candidates.union_with(&head_candidates);
}
@ -328,7 +345,7 @@ fn resolve_candidates<'t>(
}
}
resolve_operation(ctx, query_tree, number_typos, cache)
resolve_operation(ctx, query_tree, number_typos, cache, wdcache)
}
#[cfg(test)]
@ -343,9 +360,10 @@ mod test {
let query_tree = None;
let facet_candidates = None;
let mut wdcache = WordDerivationsCache::new();
let mut criteria = Typo::initial(&context, query_tree, facet_candidates);
assert!(criteria.next().unwrap().is_none());
assert!(criteria.next(&mut wdcache).unwrap().is_none());
}
#[test]
@ -361,6 +379,7 @@ mod test {
let facet_candidates = None;
let mut wdcache = WordDerivationsCache::new();
let mut criteria = Typo::initial(&context, Some(query_tree), facet_candidates);
let candidates_1 = context.word_docids("split").unwrap().unwrap()
@ -378,7 +397,7 @@ mod test {
bucket_candidates: candidates_1,
};
assert_eq!(criteria.next().unwrap(), Some(expected_1));
assert_eq!(criteria.next(&mut wdcache).unwrap(), Some(expected_1));
let candidates_2 = (
context.word_docids("split").unwrap().unwrap()
@ -400,7 +419,7 @@ mod test {
bucket_candidates: candidates_2,
};
assert_eq!(criteria.next().unwrap(), Some(expected_2));
assert_eq!(criteria.next(&mut wdcache).unwrap(), Some(expected_2));
}
#[test]
@ -409,6 +428,7 @@ mod test {
let query_tree = None;
let facet_candidates = context.word_docids("earth").unwrap().unwrap();
let mut wdcache = WordDerivationsCache::new();
let mut criteria = Typo::initial(&context, query_tree, Some(facet_candidates.clone()));
let expected = CriterionResult {
@ -418,10 +438,10 @@ mod test {
};
// first iteration, returns the facet candidates
assert_eq!(criteria.next().unwrap(), Some(expected));
assert_eq!(criteria.next(&mut wdcache).unwrap(), Some(expected));
// second iteration, returns None because there is no more things to do
assert!(criteria.next().unwrap().is_none());
assert!(criteria.next(&mut wdcache).unwrap().is_none());
}
#[test]
@ -437,6 +457,7 @@ mod test {
let facet_candidates = context.word_docids("earth").unwrap().unwrap();
let mut wdcache = WordDerivationsCache::new();
let mut criteria = Typo::initial(&context, Some(query_tree), Some(facet_candidates.clone()));
let candidates_1 = context.word_docids("split").unwrap().unwrap()
@ -454,7 +475,7 @@ mod test {
bucket_candidates: candidates_1 & &facet_candidates,
};
assert_eq!(criteria.next().unwrap(), Some(expected_1));
assert_eq!(criteria.next(&mut wdcache).unwrap(), Some(expected_1));
let candidates_2 = (
context.word_docids("split").unwrap().unwrap()
@ -476,7 +497,7 @@ mod test {
bucket_candidates: candidates_2 & &facet_candidates,
};
assert_eq!(criteria.next().unwrap(), Some(expected_2));
assert_eq!(criteria.next(&mut wdcache).unwrap(), Some(expected_2));
}
}

9
milli/src/search/criteria/words.rs
View File

@ -5,6 +5,7 @@ use log::debug;
use roaring::RoaringBitmap;
use crate::search::query_tree::Operation;
use crate::search::WordDerivationsCache;
use super::{resolve_query_tree, Candidates, Criterion, CriterionResult, Context};
pub struct Words<'t> {
@ -46,7 +47,7 @@ impl<'t> Words<'t> {
}
impl<'t> Criterion for Words<'t> {
fn next(&mut self) -> anyhow::Result<Option<CriterionResult>> {
fn next(&mut self, wdcache: &mut WordDerivationsCache) -> anyhow::Result<Option<CriterionResult>> {
use Candidates::{Allowed, Forbidden};
loop {
debug!("Words at iteration {} ({:?})", self.query_trees.len(), self.candidates);
@ -61,7 +62,7 @@ impl<'t> Criterion for Words<'t> {
}));
},
(Some(qt), Allowed(candidates)) => {
let mut found_candidates = resolve_query_tree(self.ctx, &qt, &mut self.candidates_cache)?;
let mut found_candidates = resolve_query_tree(self.ctx, &qt, &mut self.candidates_cache, wdcache)?;
found_candidates.intersect_with(&candidates);
candidates.difference_with(&found_candidates);
@ -77,7 +78,7 @@ impl<'t> Criterion for Words<'t> {
}));
},
(Some(qt), Forbidden(candidates)) => {
let mut found_candidates = resolve_query_tree(self.ctx, &qt, &mut self.candidates_cache)?;
let mut found_candidates = resolve_query_tree(self.ctx, &qt, &mut self.candidates_cache, wdcache)?;
found_candidates.difference_with(&candidates);
candidates.union_with(&found_candidates);
@ -103,7 +104,7 @@ impl<'t> Criterion for Words<'t> {
(None, Forbidden(_)) => {
match self.parent.as_mut() {
Some(parent) => {
match parent.next()? {
match parent.next(wdcache)? {
Some(CriterionResult { query_tree, candidates, bucket_candidates }) => {
self.query_trees = query_tree.map(explode_query_tree).unwrap_or_default();
self.candidates = Candidates::Allowed(candidates);

35
milli/src/search/mod.rs
View File

@ -1,5 +1,7 @@
use std::borrow::Cow;
use std::collections::hash_map::{HashMap, Entry};
use std::fmt;
use std::str::Utf8Error;
use std::time::Instant;
use fst::{IntoStreamer, Streamer, Set};
@ -97,8 +99,9 @@ impl<'a> Search<'a> {
let mut offset = self.offset;
let mut limit = self.limit;
let mut documents_ids = Vec::new();
let mut words_derivations_cache = WordDerivationsCache::new();
let mut initial_candidates = RoaringBitmap::new();
while let Some(CriterionResult { candidates, bucket_candidates, .. }) = criteria.next()? {
while let Some(CriterionResult { candidates, bucket_candidates, .. }) = criteria.next(&mut words_derivations_cache)? {
debug!("Number of candidates found {}", candidates.len());
@ -145,24 +148,32 @@ pub struct SearchResult {
pub documents_ids: Vec<DocumentId>,
}
pub fn word_derivations(
pub type WordDerivationsCache = HashMap<(String, bool, u8), Vec<(String, u8)>>;
pub fn word_derivations<'c>(
word: &str,
is_prefix: bool,
max_typo: u8,
fst: &fst::Set<Cow<[u8]>>,
) -> anyhow::Result<Vec<(String, u8)>>
cache: &'c mut WordDerivationsCache,
) -> Result<&'c [(String, u8)], Utf8Error>
{
let mut derived_words = Vec::new();
let dfa = build_dfa(word, max_typo, is_prefix);
let mut stream = fst.search_with_state(&dfa).into_stream();
match cache.entry((word.to_string(), is_prefix, max_typo)) {
Entry::Occupied(entry) => Ok(entry.into_mut()),
Entry::Vacant(entry) => {
let mut derived_words = Vec::new();
let dfa = build_dfa(word, max_typo, is_prefix);
let mut stream = fst.search_with_state(&dfa).into_stream();
while let Some((word, state)) = stream.next() {
let word = std::str::from_utf8(word)?;
let distance = dfa.distance(state);
derived_words.push((word.to_string(), distance.to_u8()));
while let Some((word, state)) = stream.next() {
let word = std::str::from_utf8(word)?;
let distance = dfa.distance(state);
derived_words.push((word.to_string(), distance.to_u8()));
}
Ok(entry.insert(derived_words))
},
}
Ok(derived_words)
}
pub fn build_dfa(word: &str, typos: u8, is_prefix: bool) -> DFA {