searchengines-web/MeiliSearch
searchengines-web/MeiliSearch
1
0
Fork 0
mirror of https://github.com/meilisearch/MeiliSearch synced 2024-12-23 13:10:06 +01:00
Code Issues Releases Wiki Activity

Keep the caches in the AppendOnlyVec

Browse Source
This commit is contained in:
Clément Renault 2024-10-01 11:07:56 +02:00
parent 0a8cb471df
commit dead7a56a3
No known key found for this signature in database
GPG Key ID: F250A4C4E3AE5F5F
4 changed files with 95 additions and 103 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

12
milli/src/update/new/append_only_vec.rs
View File

@ -99,12 +99,6 @@ fn test_indices() {
}
}
impl<T> Default for AppendOnlyVec<T> {
fn default() -> Self {
Self::new()
}
}
impl<T> AppendOnlyVec<T> {
const EMPTY: UnsafeCell<*mut T> = UnsafeCell::new(ptr::null_mut());
@ -220,6 +214,12 @@ impl<T> AppendOnlyVec<T> {
}
}
impl<T> Default for AppendOnlyVec<T> {
fn default() -> Self {
Self::new()
}
}
impl<T> Debug for AppendOnlyVec<T> {
fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
f.debug_struct("AppendOnlyVec").field("len", &self.len()).finish()

69
milli/src/update/new/extract/faceted/extract_facets.rs
View File

@ -12,6 +12,7 @@ use super::super::cache::CboCachedSorter;
use super::facet_document::extract_document_facets;
use super::FacetKind;
use crate::facet::value_encoding::f64_into_bytes;
use crate::update::new::append_only_vec::AppendOnlyVec;
use crate::update::new::extract::DocidsExtractor;
use crate::update::new::items_pool::ParallelIteratorExt;
use crate::update::new::{DocumentChange, ItemsPool};
@ -209,45 +210,40 @@ impl DocidsExtractor for FacetedDocidsExtractor {
let attributes_to_extract = Self::attributes_to_extract(&rtxn, index)?;
let attributes_to_extract: Vec<_> =
attributes_to_extract.iter().map(|s| s.as_ref()).collect();
let context_pool = ItemsPool::new(|| {
Ok((
fields_ids_map.clone(),
Vec::new(),
CboCachedSorter::new(
// TODO use a better value
100.try_into().unwrap(),
create_sorter(
grenad::SortAlgorithm::Stable,
MergeDeladdCboRoaringBitmaps,
indexer.chunk_compression_type,
indexer.chunk_compression_level,
indexer.max_nb_chunks,
max_memory,
),
),
))
});
let caches = AppendOnlyVec::new();
{
let span =
tracing::trace_span!(target: "indexing::documents::extract", "docids_extraction");
let _entered = span.enter();
document_changes.into_par_iter().try_arc_for_each_try_init(
|| index.read_txn().map_err(Error::from),
|rtxn, document_change| {
context_pool.with(|(fields_ids_map, buffer, cached_sorter)| {
Self::extract_document_change(
rtxn,
index,
buffer,
fields_ids_map,
&attributes_to_extract,
cached_sorter,
document_change?,
)
.map_err(Arc::new)
})
|| {
let rtxn = index.read_txn().map_err(Error::from)?;
let cache = caches.push(CboCachedSorter::new(
// TODO use a better value
100.try_into().unwrap(),
create_sorter(
grenad::SortAlgorithm::Stable,
MergeDeladdCboRoaringBitmaps,
indexer.chunk_compression_type,
indexer.chunk_compression_level,
indexer.max_nb_chunks,
max_memory,
),
));
Ok((rtxn, fields_ids_map.clone(), Vec::new(), cache))
},
|(rtxn, fields_ids_map, buffer, cached_sorter), document_change| {
Self::extract_document_change(
rtxn,
index,
buffer,
fields_ids_map,
&attributes_to_extract,
cached_sorter,
document_change?,
)
.map_err(Arc::new)
},
)?;
}
@ -257,14 +253,15 @@ impl DocidsExtractor for FacetedDocidsExtractor {
tracing::trace_span!(target: "indexing::documents::extract", "merger_building");
let _entered = span.enter();
let readers: Vec<_> = context_pool
.into_items()
let readers: Vec<_> = caches
.into_iter()
.par_bridge()
.map(|(_tokenizer, _fields_ids_map, cached_sorter)| {
.map(|cached_sorter| {
let sorter = cached_sorter.into_sorter()?;
sorter.into_reader_cursors()
})
.collect();
for reader in readers {
builder.extend(reader?);
}

50
milli/src/update/new/extract/searchable/extract_word_docids.rs
View File

@ -8,6 +8,8 @@ use heed::RoTxn;
use rayon::iter::IntoParallelIterator;
use super::tokenize_document::{tokenizer_builder, DocumentTokenizer};
use super::SearchableExtractor;
use crate::update::new::append_only_vec::AppendOnlyVec;
use crate::update::new::extract::cache::CboCachedSorter;
use crate::update::new::extract::perm_json_p::contained_in;
use crate::update::new::items_pool::ParallelIteratorExt;
@ -339,37 +341,33 @@ impl WordDocidsExtractors {
max_positions_per_attributes: MAX_POSITION_PER_ATTRIBUTE,
};
let context_pool = ItemsPool::new(|| {
Ok((
&document_tokenizer,
fields_ids_map.clone(),
WordDocidsCachedSorters::new(
indexer,
max_memory,
// TODO use a better value
200_000.try_into().unwrap(),
),
))
});
let caches = AppendOnlyVec::new();
{
let span =
tracing::trace_span!(target: "indexing::documents::extract", "docids_extraction");
let _entered = span.enter();
document_changes.into_par_iter().try_arc_for_each_try_init(
|| index.read_txn().map_err(Error::from),
|rtxn, document_change| {
context_pool.with(|(document_tokenizer, fields_ids_map, cached_sorter)| {
Self::extract_document_change(
rtxn,
index,
document_tokenizer,
fields_ids_map,
cached_sorter,
document_change?,
)
.map_err(Arc::new)
})
|| {
let rtxn = index.read_txn().map_err(Error::from)?;
let cache = caches.push(WordDocidsCachedSorters::new(
indexer,
max_memory,
// TODO use a better value
200_000.try_into().unwrap(),
));
Ok((rtxn, &document_tokenizer, fields_ids_map.clone(), cache))
},
|(rtxn, document_tokenizer, fields_ids_map, cached_sorter), document_change| {
Self::extract_document_change(
rtxn,
index,
document_tokenizer,
fields_ids_map,
cached_sorter,
document_change?,
)
.map_err(Arc::new)
},
)?;
}
@ -379,7 +377,7 @@ impl WordDocidsExtractors {
tracing::trace_span!(target: "indexing::documents::extract", "merger_building");
let _entered = span.enter();
let mut builder = WordDocidsMergerBuilders::new();
for (_tokenizer, _fields_ids_map, cache) in context_pool.into_items() {
for cache in caches.into_iter() {
builder.add_sorters(cache)?;
}

67
milli/src/update/new/extract/searchable/mod.rs
View File

@ -14,6 +14,7 @@ use tokenize_document::{tokenizer_builder, DocumentTokenizer};
use super::cache::CboCachedSorter;
use super::DocidsExtractor;
use crate::update::new::append_only_vec::AppendOnlyVec;
use crate::update::new::items_pool::ParallelIteratorExt;
use crate::update::new::{DocumentChange, ItemsPool};
use crate::update::{create_sorter, GrenadParameters, MergeDeladdCboRoaringBitmaps};
@ -57,44 +58,39 @@ pub trait SearchableExtractor {
localized_attributes_rules: &localized_attributes_rules,
max_positions_per_attributes: MAX_POSITION_PER_ATTRIBUTE,
};
let context_pool = ItemsPool::new(|| {
Ok((
&document_tokenizer,
fields_ids_map.clone(),
CboCachedSorter::new(
// TODO use a better value
1_000_000.try_into().unwrap(),
create_sorter(
grenad::SortAlgorithm::Stable,
MergeDeladdCboRoaringBitmaps,
indexer.chunk_compression_type,
indexer.chunk_compression_level,
indexer.max_nb_chunks,
max_memory,
),
),
))
});
let caches = AppendOnlyVec::new();
{
let span =
tracing::trace_span!(target: "indexing::documents::extract", "docids_extraction");
let _entered = span.enter();
document_changes.into_par_iter().try_arc_for_each_try_init(
|| index.read_txn().map_err(Error::from),
|rtxn, document_change| {
context_pool.with(|(document_tokenizer, fields_ids_map, cached_sorter)| {
Self::extract_document_change(
rtxn,
index,
document_tokenizer,
fields_ids_map,
cached_sorter,
document_change?,
)
.map_err(Arc::new)
})
|| {
let rtxn = index.read_txn().map_err(Error::from)?;
let cache = caches.push(CboCachedSorter::new(
// TODO use a better value
1_000_000.try_into().unwrap(),
create_sorter(
grenad::SortAlgorithm::Stable,
MergeDeladdCboRoaringBitmaps,
indexer.chunk_compression_type,
indexer.chunk_compression_level,
indexer.max_nb_chunks,
max_memory,
),
));
Ok((rtxn, &document_tokenizer, fields_ids_map.clone(), cache))
},
|(rtxn, document_tokenizer, fields_ids_map, cached_sorter), document_change| {
Self::extract_document_change(
rtxn,
index,
document_tokenizer,
fields_ids_map,
cached_sorter,
document_change?,
)
.map_err(Arc::new)
},
)?;
}
@ -104,14 +100,15 @@ pub trait SearchableExtractor {
tracing::trace_span!(target: "indexing::documents::extract", "merger_building");
let _entered = span.enter();
let readers: Vec<_> = context_pool
.into_items()
let readers: Vec<_> = caches
.into_iter()
.par_bridge()
.map(|(_tokenizer, _fields_ids_map, cached_sorter)| {
.map(|cached_sorter| {
let sorter = cached_sorter.into_sorter()?;
sorter.into_reader_cursors()
})
.collect();
for reader in readers {
builder.extend(reader?);
}