mirror of
https://github.com/meilisearch/meilisearch.git
synced 2025-07-19 21:10:34 +00:00
Compare commits
10 Commits
reduce-pre
...
incrementa
Author | SHA1 | Date | |
---|---|---|---|
8a7a465c19 | |||
1079c55ea5 | |||
77c97960e2 | |||
7813bc748a | |||
46104a2836 | |||
50b41509cb | |||
eb5ebdb6d7 | |||
025be5147f | |||
0780b04f6f | |||
b7ff74a090 |
@ -79,22 +79,29 @@ pub const FACET_MIN_LEVEL_SIZE: u8 = 5;
|
||||
use std::collections::BTreeSet;
|
||||
use std::fs::File;
|
||||
use std::io::BufReader;
|
||||
use std::ops::Bound;
|
||||
|
||||
use grenad::Merger;
|
||||
use heed::types::{Bytes, DecodeIgnore};
|
||||
use heed::BytesDecode as _;
|
||||
use roaring::RoaringBitmap;
|
||||
use time::OffsetDateTime;
|
||||
use tracing::debug;
|
||||
|
||||
use self::incremental::FacetsUpdateIncremental;
|
||||
use super::{FacetsUpdateBulk, MergeDeladdBtreesetString, MergeDeladdCboRoaringBitmaps};
|
||||
use crate::facet::FacetType;
|
||||
use crate::heed_codec::facet::{FacetGroupKey, FacetGroupKeyCodec, FacetGroupValueCodec};
|
||||
use crate::heed_codec::facet::{
|
||||
FacetGroupKey, FacetGroupKeyCodec, FacetGroupValueCodec, OrderedF64Codec,
|
||||
};
|
||||
use crate::heed_codec::BytesRefCodec;
|
||||
use crate::search::facet::get_highest_level;
|
||||
use crate::update::del_add::{DelAdd, KvReaderDelAdd};
|
||||
use crate::{try_split_array_at, FieldId, Index, Result};
|
||||
|
||||
pub mod bulk;
|
||||
pub mod incremental;
|
||||
pub mod new_incremental;
|
||||
|
||||
/// A builder used to add new elements to the `facet_id_string_docids` or `facet_id_f64_docids` databases.
|
||||
///
|
||||
@ -646,3 +653,191 @@ mod comparison_bench {
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
/// Run sanity checks on the specified fid tree
|
||||
///
|
||||
/// 1. No "orphan" child value, any child value has a parent
|
||||
/// 2. Any docid in the child appears in the parent
|
||||
/// 3. No docid in the parent is missing from all its children
|
||||
/// 4. no group is bigger than max_group_size
|
||||
/// 5. Less than 50% of groups are bigger than group_size
|
||||
/// 6. group size matches the number of children
|
||||
/// 7. max_level is < 255
|
||||
pub(crate) fn sanity_checks(
|
||||
index: &Index,
|
||||
rtxn: &heed::RoTxn,
|
||||
field_id: FieldId,
|
||||
facet_type: FacetType,
|
||||
group_size: usize,
|
||||
_min_level_size: usize, // might add a check on level size later
|
||||
max_group_size: usize,
|
||||
) -> Result<()> {
|
||||
tracing::info!(%field_id, ?facet_type, "performing sanity checks");
|
||||
let database = match facet_type {
|
||||
FacetType::String => {
|
||||
index.facet_id_string_docids.remap_key_type::<FacetGroupKeyCodec<BytesRefCodec>>()
|
||||
}
|
||||
FacetType::Number => {
|
||||
index.facet_id_f64_docids.remap_key_type::<FacetGroupKeyCodec<BytesRefCodec>>()
|
||||
}
|
||||
};
|
||||
|
||||
let leaf_prefix: FacetGroupKey<&[u8]> = FacetGroupKey { field_id, level: 0, left_bound: &[] };
|
||||
|
||||
let leaf_it = database.prefix_iter(rtxn, &leaf_prefix)?;
|
||||
|
||||
let max_level = get_highest_level(rtxn, database, field_id)?;
|
||||
if max_level == u8::MAX {
|
||||
panic!("max_level == 255");
|
||||
}
|
||||
|
||||
for leaf in leaf_it {
|
||||
let (leaf_facet_value, leaf_docids) = leaf?;
|
||||
let mut current_level = 0;
|
||||
|
||||
let mut current_parent_facet_value: Option<FacetGroupKey<&[u8]>> = None;
|
||||
let mut current_parent_docids: Option<crate::heed_codec::facet::FacetGroupValue> = None;
|
||||
loop {
|
||||
current_level += 1;
|
||||
if current_level == max_level {
|
||||
break;
|
||||
}
|
||||
let parent_key_right_bound = FacetGroupKey {
|
||||
field_id,
|
||||
level: current_level,
|
||||
left_bound: leaf_facet_value.left_bound,
|
||||
};
|
||||
let (parent_facet_value, parent_docids) = database
|
||||
.get_lower_than_or_equal_to(rtxn, &parent_key_right_bound)?
|
||||
.expect("no parent found");
|
||||
if parent_facet_value.level != current_level {
|
||||
panic!("wrong parent level");
|
||||
}
|
||||
if parent_facet_value.field_id != field_id {
|
||||
panic!("wrong parent fid");
|
||||
}
|
||||
if parent_facet_value.left_bound > leaf_facet_value.left_bound {
|
||||
panic!("wrong parent left bound");
|
||||
}
|
||||
|
||||
if !leaf_docids.bitmap.is_subset(&parent_docids.bitmap) {
|
||||
panic!(
|
||||
"missing docids from leaf in parent, current_level={}, parent={}, child={}, missing={missing:?}, child_len={}, child={:?}",
|
||||
current_level,
|
||||
facet_to_string(parent_facet_value.left_bound, facet_type),
|
||||
facet_to_string(leaf_facet_value.left_bound, facet_type),
|
||||
leaf_docids.bitmap.len(),
|
||||
leaf_docids.bitmap.clone(),
|
||||
missing=leaf_docids.bitmap - parent_docids.bitmap,
|
||||
)
|
||||
}
|
||||
|
||||
if let Some(current_parent_facet_value) = current_parent_facet_value {
|
||||
if current_parent_facet_value.field_id != parent_facet_value.field_id {
|
||||
panic!("wrong parent parent fid");
|
||||
}
|
||||
if current_parent_facet_value.level + 1 != parent_facet_value.level {
|
||||
panic!("wrong parent parent level");
|
||||
}
|
||||
if current_parent_facet_value.left_bound < parent_facet_value.left_bound {
|
||||
panic!("wrong parent parent left bound");
|
||||
}
|
||||
}
|
||||
|
||||
if let Some(current_parent_docids) = current_parent_docids {
|
||||
if !current_parent_docids.bitmap.is_subset(&parent_docids.bitmap) {
|
||||
panic!("missing docids from intermediate node in parent, parent_level={}, parent={}, intermediate={}, missing={missing:?}, intermediate={:?}",
|
||||
parent_facet_value.level,
|
||||
facet_to_string(parent_facet_value.left_bound, facet_type),
|
||||
facet_to_string(current_parent_facet_value.unwrap().left_bound, facet_type),
|
||||
current_parent_docids.bitmap.clone(),
|
||||
missing=current_parent_docids.bitmap - parent_docids.bitmap,
|
||||
);
|
||||
}
|
||||
}
|
||||
|
||||
current_parent_facet_value = Some(parent_facet_value);
|
||||
current_parent_docids = Some(parent_docids);
|
||||
}
|
||||
}
|
||||
tracing::info!(%field_id, ?facet_type, "checked all leaves");
|
||||
|
||||
let mut current_level = max_level;
|
||||
let mut greater_than_group = 0usize;
|
||||
let mut total = 0usize;
|
||||
loop {
|
||||
if current_level == 0 {
|
||||
break;
|
||||
}
|
||||
let child_level = current_level - 1;
|
||||
tracing::info!(%field_id, ?facet_type, %current_level, "checked groups for level");
|
||||
let level_groups_prefix: FacetGroupKey<&[u8]> =
|
||||
FacetGroupKey { field_id, level: current_level, left_bound: &[] };
|
||||
let mut level_groups_it = database.prefix_iter(rtxn, &level_groups_prefix)?.peekable();
|
||||
|
||||
'group_it: loop {
|
||||
let Some(group) = level_groups_it.next() else { break 'group_it };
|
||||
|
||||
let (group_facet_value, group_docids) = group?;
|
||||
let child_left_bound = group_facet_value.left_bound.to_owned();
|
||||
let mut expected_docids = RoaringBitmap::new();
|
||||
let mut expected_size = 0usize;
|
||||
let right_bound = level_groups_it
|
||||
.peek()
|
||||
.and_then(|res| res.as_ref().ok())
|
||||
.map(|(key, _)| key.left_bound);
|
||||
let child_left_bound = FacetGroupKey {
|
||||
field_id,
|
||||
level: child_level,
|
||||
left_bound: child_left_bound.as_slice(),
|
||||
};
|
||||
let child_left_bound = Bound::Included(&child_left_bound);
|
||||
let child_right_bound;
|
||||
let child_right_bound = if let Some(right_bound) = right_bound {
|
||||
child_right_bound =
|
||||
FacetGroupKey { field_id, level: child_level, left_bound: right_bound };
|
||||
Bound::Excluded(&child_right_bound)
|
||||
} else {
|
||||
Bound::Unbounded
|
||||
};
|
||||
let children = database.range(rtxn, &(child_left_bound, child_right_bound))?;
|
||||
for child in children {
|
||||
let (child_facet_value, child_docids) = child?;
|
||||
if child_facet_value.field_id != field_id {
|
||||
break;
|
||||
}
|
||||
if child_facet_value.level != child_level {
|
||||
break;
|
||||
}
|
||||
expected_size += 1;
|
||||
expected_docids |= &child_docids.bitmap;
|
||||
}
|
||||
assert_eq!(expected_size, group_docids.size as usize);
|
||||
assert!(expected_size <= max_group_size);
|
||||
assert_eq!(expected_docids, group_docids.bitmap);
|
||||
total += 1;
|
||||
if expected_size > group_size {
|
||||
greater_than_group += 1;
|
||||
}
|
||||
}
|
||||
|
||||
current_level -= 1;
|
||||
}
|
||||
if greater_than_group * 2 > total {
|
||||
panic!("too many groups have a size > group_size");
|
||||
}
|
||||
|
||||
tracing::info!("sanity checks OK");
|
||||
|
||||
Ok(())
|
||||
}
|
||||
|
||||
fn facet_to_string(facet_value: &[u8], facet_type: FacetType) -> String {
|
||||
match facet_type {
|
||||
FacetType::String => bstr::BStr::new(facet_value).to_string(),
|
||||
FacetType::Number => match OrderedF64Codec::bytes_decode(facet_value) {
|
||||
Ok(value) => value.to_string(),
|
||||
Err(e) => format!("error: {e} (bytes: {facet_value:?}"),
|
||||
},
|
||||
}
|
||||
}
|
||||
|
464
crates/milli/src/update/facet/new_incremental.rs
Normal file
464
crates/milli/src/update/facet/new_incremental.rs
Normal file
@ -0,0 +1,464 @@
|
||||
use std::ops::Bound;
|
||||
|
||||
use heed::types::{Bytes, DecodeIgnore};
|
||||
use heed::{BytesDecode as _, Database, RwTxn};
|
||||
use roaring::RoaringBitmap;
|
||||
|
||||
use crate::facet::FacetType;
|
||||
use crate::heed_codec::facet::{
|
||||
FacetGroupKey, FacetGroupKeyCodec, FacetGroupValue, FacetGroupValueCodec,
|
||||
};
|
||||
use crate::heed_codec::BytesRefCodec;
|
||||
use crate::search::facet::get_highest_level;
|
||||
use crate::{FieldId, Index, Result};
|
||||
|
||||
pub struct FacetsUpdateIncremental {
|
||||
inner: FacetsUpdateIncrementalInner,
|
||||
delta_data: Vec<FacetFieldIdChange>,
|
||||
}
|
||||
|
||||
struct FacetsUpdateIncrementalInner {
|
||||
db: Database<FacetGroupKeyCodec<BytesRefCodec>, FacetGroupValueCodec>,
|
||||
field_id: FieldId,
|
||||
group_size: u8,
|
||||
min_level_size: u8,
|
||||
max_group_size: u8,
|
||||
}
|
||||
|
||||
impl FacetsUpdateIncremental {
|
||||
pub fn new(
|
||||
index: &Index,
|
||||
facet_type: FacetType,
|
||||
field_id: FieldId,
|
||||
delta_data: Vec<FacetFieldIdChange>,
|
||||
group_size: u8,
|
||||
min_level_size: u8,
|
||||
max_group_size: u8,
|
||||
) -> Self {
|
||||
FacetsUpdateIncremental {
|
||||
inner: FacetsUpdateIncrementalInner {
|
||||
db: match facet_type {
|
||||
FacetType::String => index
|
||||
.facet_id_string_docids
|
||||
.remap_key_type::<FacetGroupKeyCodec<BytesRefCodec>>(),
|
||||
FacetType::Number => index
|
||||
.facet_id_f64_docids
|
||||
.remap_key_type::<FacetGroupKeyCodec<BytesRefCodec>>(),
|
||||
},
|
||||
field_id,
|
||||
group_size,
|
||||
min_level_size,
|
||||
max_group_size,
|
||||
},
|
||||
|
||||
delta_data,
|
||||
}
|
||||
}
|
||||
|
||||
#[tracing::instrument(level = "trace", skip_all, target = "indexing::facets::incremental")]
|
||||
pub fn execute(mut self, wtxn: &mut RwTxn) -> Result<()> {
|
||||
if self.delta_data.is_empty() {
|
||||
return Ok(());
|
||||
}
|
||||
self.delta_data.sort_unstable_by(
|
||||
|FacetFieldIdChange { facet_value: left, .. },
|
||||
FacetFieldIdChange { facet_value: right, .. }| left.cmp(right),
|
||||
);
|
||||
|
||||
self.inner.find_touched_parents(
|
||||
wtxn,
|
||||
0,
|
||||
self.delta_data
|
||||
.into_iter()
|
||||
// reverse lexicographic order
|
||||
.rev()
|
||||
.map(|change| change.facet_value.into()),
|
||||
)?;
|
||||
|
||||
self.inner.add_or_delete_level(wtxn)
|
||||
}
|
||||
}
|
||||
|
||||
impl FacetsUpdateIncrementalInner {
|
||||
/// WARNING: `touched_children` must be sorted in **reverse** lexicographic order.
|
||||
fn find_touched_parents(
|
||||
&self,
|
||||
wtxn: &mut RwTxn,
|
||||
child_level: u8,
|
||||
mut touched_children: impl Iterator<Item = Vec<u8>>,
|
||||
) -> Result<()> {
|
||||
let mut touched_parents = vec![];
|
||||
let Some(parent_level) = child_level.checked_add(1) else { return Ok(()) };
|
||||
let parent_level_left_bound: FacetGroupKey<&[u8]> =
|
||||
FacetGroupKey { field_id: self.field_id, level: parent_level, left_bound: &[] };
|
||||
|
||||
let mut last_parent: Option<Vec<u8>> = None;
|
||||
|
||||
for child in &mut touched_children {
|
||||
if let Some(last_parent) = &last_parent {
|
||||
if child.as_slice() >= last_parent.as_slice() {
|
||||
self.compute_parent_group(wtxn, child_level, child)?;
|
||||
continue;
|
||||
}
|
||||
}
|
||||
|
||||
// need to find a new parent
|
||||
let parent_key_prefix = FacetGroupKey {
|
||||
field_id: self.field_id,
|
||||
level: parent_level,
|
||||
left_bound: child.as_slice(),
|
||||
};
|
||||
|
||||
let parent = self
|
||||
.db
|
||||
.remap_data_type::<DecodeIgnore>()
|
||||
.rev_range(
|
||||
wtxn,
|
||||
&(
|
||||
Bound::Excluded(&parent_level_left_bound),
|
||||
Bound::Included(&parent_key_prefix),
|
||||
),
|
||||
)?
|
||||
.next();
|
||||
|
||||
match parent {
|
||||
Some(Ok((parent_key, _parent_value))) => {
|
||||
// found parent, cache it for next keys
|
||||
last_parent = Some(parent_key.left_bound.to_owned());
|
||||
|
||||
// add to modified list for parent level
|
||||
touched_parents.push(parent_key.left_bound.to_owned());
|
||||
self.compute_parent_group(wtxn, child_level, child)?;
|
||||
}
|
||||
Some(Err(err)) => return Err(err.into()),
|
||||
None => {
|
||||
self.compute_parent_group(wtxn, child_level, child)?;
|
||||
break;
|
||||
}
|
||||
}
|
||||
}
|
||||
// do we have children without parents?
|
||||
if let Some(child) = touched_children.next() {
|
||||
// no parent for that key
|
||||
let mut it = self
|
||||
.db
|
||||
.remap_data_type::<DecodeIgnore>()
|
||||
.prefix_iter_mut(wtxn, &parent_level_left_bound)?;
|
||||
match it.next() {
|
||||
// 1. left of the current left bound, or
|
||||
Some(Ok((first_key, _first_value))) => 'change_left_bound: {
|
||||
// make sure we don't spill on the neighboring fid (level also included defensively)
|
||||
if first_key.field_id != self.field_id || first_key.level != parent_level {
|
||||
break 'change_left_bound;
|
||||
}
|
||||
// remove old left bound
|
||||
unsafe { it.del_current()? };
|
||||
drop(it);
|
||||
// pop all elements and order to visit the new left bound
|
||||
touched_parents.push(child.clone());
|
||||
self.compute_parent_group(wtxn, child_level, child)?;
|
||||
for child in touched_children {
|
||||
let new_left_bound = touched_parents.last_mut().unwrap();
|
||||
new_left_bound.clear();
|
||||
new_left_bound.extend_from_slice(&child);
|
||||
self.compute_parent_group(wtxn, child_level, child)?;
|
||||
}
|
||||
}
|
||||
Some(Err(err)) => return Err(err.into()),
|
||||
// 2. max level reached, exit
|
||||
None => {
|
||||
drop(it);
|
||||
self.compute_parent_group(wtxn, child_level, child)?;
|
||||
for child in touched_children {
|
||||
self.compute_parent_group(wtxn, child_level, child)?;
|
||||
}
|
||||
}
|
||||
}
|
||||
}
|
||||
self.find_touched_parents(
|
||||
wtxn,
|
||||
parent_level,
|
||||
touched_parents
|
||||
// no need to `rev` here because the parents were already visited in reverse order
|
||||
.into_iter(),
|
||||
)
|
||||
}
|
||||
|
||||
fn compute_parent_group(
|
||||
&self,
|
||||
wtxn: &mut RwTxn<'_>,
|
||||
parent_level: u8,
|
||||
parent_left_bound: Vec<u8>,
|
||||
) -> Result<()> {
|
||||
let mut range_left_bound = parent_left_bound;
|
||||
if parent_level == 0 {
|
||||
return Ok(());
|
||||
}
|
||||
let child_level = parent_level - 1;
|
||||
|
||||
let parent_key = FacetGroupKey {
|
||||
field_id: self.field_id,
|
||||
level: parent_level,
|
||||
left_bound: range_left_bound.as_slice(),
|
||||
};
|
||||
let child_right_bound = self
|
||||
.db
|
||||
.remap_data_type::<DecodeIgnore>()
|
||||
.get_greater_than(wtxn, &parent_key)?
|
||||
.and_then(
|
||||
|(
|
||||
FacetGroupKey {
|
||||
level: right_level,
|
||||
field_id: right_fid,
|
||||
left_bound: right_bound,
|
||||
},
|
||||
_,
|
||||
)| {
|
||||
if parent_level != right_level || self.field_id != right_fid {
|
||||
// there was a greater key, but with a greater level or fid, so not a sibling to the parent: ignore
|
||||
return None;
|
||||
}
|
||||
Some(right_bound.to_owned())
|
||||
},
|
||||
);
|
||||
let child_right_bound = match &child_right_bound {
|
||||
Some(right_bound) => Bound::Excluded(FacetGroupKey {
|
||||
left_bound: right_bound.as_slice(),
|
||||
field_id: self.field_id,
|
||||
level: child_level,
|
||||
}),
|
||||
None => Bound::Unbounded,
|
||||
};
|
||||
|
||||
let child_left_key = FacetGroupKey {
|
||||
field_id: self.field_id,
|
||||
level: child_level,
|
||||
left_bound: range_left_bound.as_slice(),
|
||||
};
|
||||
let mut child_left_bound = Bound::Included(child_left_key);
|
||||
|
||||
loop {
|
||||
let mut child_it = self.db.range(wtxn, &(child_left_bound, child_right_bound))?;
|
||||
let res: Result<_> = child_it
|
||||
.by_ref()
|
||||
.take(self.max_group_size as usize)
|
||||
// stop if we go to the next level or field id
|
||||
.take_while(|res| match res {
|
||||
Ok((child_key, _)) => {
|
||||
child_key.field_id == self.field_id && child_key.level == child_level
|
||||
}
|
||||
Err(_) => true,
|
||||
})
|
||||
.try_fold(
|
||||
(None, FacetGroupValue { size: 0, bitmap: Default::default() }),
|
||||
|(bounds, mut group_value), child_res| {
|
||||
let (child_key, child_value) = child_res?;
|
||||
let bounds = match bounds {
|
||||
Some((left_bound, _)) => Some((left_bound, child_key.left_bound)),
|
||||
None => Some((child_key.left_bound, child_key.left_bound)),
|
||||
};
|
||||
// max_group_size <= u8::MAX
|
||||
group_value.size += 1;
|
||||
group_value.bitmap |= &child_value.bitmap;
|
||||
Ok((bounds, group_value))
|
||||
},
|
||||
);
|
||||
|
||||
let (bounds, group_value) = res?;
|
||||
|
||||
let Some((group_left_bound, right_bound)) = bounds else {
|
||||
let update_key = FacetGroupKey {
|
||||
field_id: self.field_id,
|
||||
level: parent_level,
|
||||
left_bound: range_left_bound.as_slice(),
|
||||
};
|
||||
drop(child_it);
|
||||
if let Bound::Included(_) = child_left_bound {
|
||||
self.db.delete(wtxn, &update_key)?;
|
||||
}
|
||||
|
||||
break;
|
||||
};
|
||||
|
||||
drop(child_it);
|
||||
let current_left_bound = group_left_bound.to_owned();
|
||||
|
||||
let delete_old_bound = match child_left_bound {
|
||||
Bound::Included(bound) => {
|
||||
if bound.left_bound != current_left_bound {
|
||||
Some(range_left_bound.clone())
|
||||
} else {
|
||||
None
|
||||
}
|
||||
}
|
||||
_ => None,
|
||||
};
|
||||
|
||||
range_left_bound.clear();
|
||||
range_left_bound.extend_from_slice(right_bound);
|
||||
let child_left_key = FacetGroupKey {
|
||||
field_id: self.field_id,
|
||||
level: child_level,
|
||||
left_bound: range_left_bound.as_slice(),
|
||||
};
|
||||
child_left_bound = Bound::Excluded(child_left_key);
|
||||
|
||||
if let Some(old_bound) = delete_old_bound {
|
||||
let update_key = FacetGroupKey {
|
||||
field_id: self.field_id,
|
||||
level: parent_level,
|
||||
left_bound: old_bound.as_slice(),
|
||||
};
|
||||
self.db.delete(wtxn, &update_key)?;
|
||||
}
|
||||
|
||||
let update_key = FacetGroupKey {
|
||||
field_id: self.field_id,
|
||||
level: parent_level,
|
||||
left_bound: current_left_bound.as_slice(),
|
||||
};
|
||||
if group_value.bitmap.is_empty() {
|
||||
self.db.delete(wtxn, &update_key)?;
|
||||
} else {
|
||||
self.db.put(wtxn, &update_key, &group_value)?;
|
||||
}
|
||||
}
|
||||
|
||||
Ok(())
|
||||
}
|
||||
|
||||
/// Check whether the highest level has exceeded `min_level_size` * `self.group_size`.
|
||||
/// If it has, we must build an addition level above it.
|
||||
/// Then check whether the highest level is under `min_level_size`.
|
||||
/// If it has, we must remove the complete level.
|
||||
pub(crate) fn add_or_delete_level(&self, txn: &mut RwTxn<'_>) -> Result<()> {
|
||||
let highest_level = get_highest_level(txn, self.db, self.field_id)?;
|
||||
let mut highest_level_prefix = vec![];
|
||||
highest_level_prefix.extend_from_slice(&self.field_id.to_be_bytes());
|
||||
highest_level_prefix.push(highest_level);
|
||||
|
||||
let size_highest_level =
|
||||
self.db.remap_types::<Bytes, Bytes>().prefix_iter(txn, &highest_level_prefix)?.count();
|
||||
|
||||
if size_highest_level >= self.group_size as usize * self.min_level_size as usize {
|
||||
self.add_level(txn, highest_level, &highest_level_prefix, size_highest_level)
|
||||
} else if size_highest_level < self.min_level_size as usize && highest_level != 0 {
|
||||
self.delete_level(txn, &highest_level_prefix)
|
||||
} else {
|
||||
Ok(())
|
||||
}
|
||||
}
|
||||
|
||||
/// Delete a level.
|
||||
fn delete_level(&self, txn: &mut RwTxn<'_>, highest_level_prefix: &[u8]) -> Result<()> {
|
||||
let mut to_delete = vec![];
|
||||
let mut iter =
|
||||
self.db.remap_types::<Bytes, Bytes>().prefix_iter(txn, highest_level_prefix)?;
|
||||
for el in iter.by_ref() {
|
||||
let (k, _) = el?;
|
||||
to_delete.push(
|
||||
FacetGroupKeyCodec::<BytesRefCodec>::bytes_decode(k)
|
||||
.map_err(heed::Error::Encoding)?
|
||||
.into_owned(),
|
||||
);
|
||||
}
|
||||
drop(iter);
|
||||
for k in to_delete {
|
||||
self.db.delete(txn, &k.as_ref())?;
|
||||
}
|
||||
Ok(())
|
||||
}
|
||||
|
||||
/// Build an additional level for the field id.
|
||||
fn add_level(
|
||||
&self,
|
||||
txn: &mut RwTxn<'_>,
|
||||
highest_level: u8,
|
||||
highest_level_prefix: &[u8],
|
||||
size_highest_level: usize,
|
||||
) -> Result<()> {
|
||||
let mut groups_iter = self
|
||||
.db
|
||||
.remap_types::<Bytes, FacetGroupValueCodec>()
|
||||
.prefix_iter(txn, highest_level_prefix)?;
|
||||
|
||||
let nbr_new_groups = size_highest_level / self.group_size as usize;
|
||||
let nbr_leftover_elements = size_highest_level % self.group_size as usize;
|
||||
|
||||
let mut to_add = vec![];
|
||||
for _ in 0..nbr_new_groups {
|
||||
let mut first_key = None;
|
||||
let mut values = RoaringBitmap::new();
|
||||
for _ in 0..self.group_size {
|
||||
let (key_bytes, value_i) = groups_iter.next().unwrap()?;
|
||||
let key_i = FacetGroupKeyCodec::<BytesRefCodec>::bytes_decode(key_bytes)
|
||||
.map_err(heed::Error::Encoding)?;
|
||||
|
||||
if first_key.is_none() {
|
||||
first_key = Some(key_i);
|
||||
}
|
||||
values |= value_i.bitmap;
|
||||
}
|
||||
let key = FacetGroupKey {
|
||||
field_id: self.field_id,
|
||||
level: highest_level + 1,
|
||||
left_bound: first_key.unwrap().left_bound,
|
||||
};
|
||||
let value = FacetGroupValue { size: self.group_size, bitmap: values };
|
||||
to_add.push((key.into_owned(), value));
|
||||
}
|
||||
// now we add the rest of the level, in case its size is > group_size * min_level_size
|
||||
// this can indeed happen if the min_level_size parameter changes between two calls to `insert`
|
||||
if nbr_leftover_elements > 0 {
|
||||
let mut first_key = None;
|
||||
let mut values = RoaringBitmap::new();
|
||||
for _ in 0..nbr_leftover_elements {
|
||||
let (key_bytes, value_i) = groups_iter.next().unwrap()?;
|
||||
let key_i = FacetGroupKeyCodec::<BytesRefCodec>::bytes_decode(key_bytes)
|
||||
.map_err(heed::Error::Encoding)?;
|
||||
|
||||
if first_key.is_none() {
|
||||
first_key = Some(key_i);
|
||||
}
|
||||
values |= value_i.bitmap;
|
||||
}
|
||||
let key = FacetGroupKey {
|
||||
field_id: self.field_id,
|
||||
level: highest_level + 1,
|
||||
left_bound: first_key.unwrap().left_bound,
|
||||
};
|
||||
// Note: nbr_leftover_elements can be casted to a u8 since it is bounded by `max_group_size`
|
||||
// when it is created above.
|
||||
let value = FacetGroupValue { size: nbr_leftover_elements as u8, bitmap: values };
|
||||
to_add.push((key.into_owned(), value));
|
||||
}
|
||||
|
||||
drop(groups_iter);
|
||||
for (key, value) in to_add {
|
||||
self.db.put(txn, &key.as_ref(), &value)?;
|
||||
}
|
||||
Ok(())
|
||||
}
|
||||
}
|
||||
|
||||
#[derive(Debug)]
|
||||
pub struct FacetFieldIdChange {
|
||||
pub facet_value: Box<[u8]>,
|
||||
pub operation: FacetFieldIdOperation,
|
||||
}
|
||||
|
||||
#[derive(Debug, Clone, Copy)]
|
||||
pub enum FacetFieldIdOperation {
|
||||
/// The docids have been modified for an existing facet value
|
||||
///
|
||||
/// The modification must be propagated to upper levels, without changing the structure of the tree
|
||||
InPlace,
|
||||
/// A new value has been inserted
|
||||
///
|
||||
/// The modification must be propagated to upper levels, splitting nodes and adding new levels as necessary.
|
||||
Insert,
|
||||
/// An existing value has been deleted
|
||||
///
|
||||
/// The modification must be propagated to upper levels, merging nodes and removing levels as necessary.
|
||||
Remove,
|
||||
}
|
@ -36,6 +36,8 @@ use crate::index::main_key::{WORDS_FST_KEY, WORDS_PREFIXES_FST_KEY};
|
||||
use crate::progress::Progress;
|
||||
use crate::proximity::ProximityPrecision;
|
||||
use crate::update::del_add::DelAdd;
|
||||
use crate::update::facet::new_incremental::FacetsUpdateIncremental;
|
||||
use crate::update::facet::{FACET_GROUP_SIZE, FACET_MAX_GROUP_SIZE, FACET_MIN_LEVEL_SIZE};
|
||||
use crate::update::new::extract::EmbeddingExtractor;
|
||||
use crate::update::new::merger::merge_and_send_rtree;
|
||||
use crate::update::new::words_prefix_docids::compute_exact_word_prefix_docids;
|
||||
@ -203,6 +205,7 @@ where
|
||||
caches,
|
||||
FacetDatabases::new(index),
|
||||
index,
|
||||
&rtxn,
|
||||
extractor_sender.facet_docids(),
|
||||
)?;
|
||||
}
|
||||
@ -735,27 +738,67 @@ fn compute_facet_search_database(
|
||||
fn compute_facet_level_database(
|
||||
index: &Index,
|
||||
wtxn: &mut RwTxn,
|
||||
facet_field_ids_delta: FacetFieldIdsDelta,
|
||||
mut facet_field_ids_delta: FacetFieldIdsDelta,
|
||||
) -> Result<()> {
|
||||
if let Some(modified_facet_string_ids) = facet_field_ids_delta.modified_facet_string_ids() {
|
||||
for (fid, delta) in facet_field_ids_delta.consume_facet_string_delta() {
|
||||
let span = tracing::trace_span!(target: "indexing::facet_field_ids", "string");
|
||||
let _entered = span.enter();
|
||||
FacetsUpdateBulk::new_not_updating_level_0(
|
||||
index,
|
||||
modified_facet_string_ids,
|
||||
FacetType::String,
|
||||
)
|
||||
.execute(wtxn)?;
|
||||
match delta {
|
||||
super::merger::FacetFieldIdDelta::Bulk => {
|
||||
tracing::debug!(%fid, "bulk string facet processing");
|
||||
FacetsUpdateBulk::new_not_updating_level_0(index, vec![fid], FacetType::String)
|
||||
.execute(wtxn)?
|
||||
}
|
||||
super::merger::FacetFieldIdDelta::Incremental(delta_data) => {
|
||||
tracing::debug!(%fid, len=%delta_data.len(), "incremental string facet processing");
|
||||
FacetsUpdateIncremental::new(
|
||||
index,
|
||||
FacetType::String,
|
||||
fid,
|
||||
delta_data,
|
||||
FACET_GROUP_SIZE,
|
||||
FACET_MIN_LEVEL_SIZE,
|
||||
FACET_MAX_GROUP_SIZE,
|
||||
)
|
||||
.execute(wtxn)?
|
||||
}
|
||||
}
|
||||
}
|
||||
if let Some(modified_facet_number_ids) = facet_field_ids_delta.modified_facet_number_ids() {
|
||||
|
||||
for (fid, delta) in facet_field_ids_delta.consume_facet_number_delta() {
|
||||
let span = tracing::trace_span!(target: "indexing::facet_field_ids", "number");
|
||||
let _entered = span.enter();
|
||||
FacetsUpdateBulk::new_not_updating_level_0(
|
||||
match delta {
|
||||
super::merger::FacetFieldIdDelta::Bulk => {
|
||||
tracing::debug!(%fid, "bulk number facet processing");
|
||||
FacetsUpdateBulk::new_not_updating_level_0(index, vec![fid], FacetType::Number)
|
||||
.execute(wtxn)?
|
||||
}
|
||||
super::merger::FacetFieldIdDelta::Incremental(delta_data) => {
|
||||
tracing::debug!(%fid, len=%delta_data.len(), "incremental number facet processing");
|
||||
/// TODO: check is_valid lmdb key
|
||||
FacetsUpdateIncremental::new(
|
||||
index,
|
||||
FacetType::Number,
|
||||
fid,
|
||||
delta_data,
|
||||
FACET_GROUP_SIZE,
|
||||
FACET_MIN_LEVEL_SIZE,
|
||||
FACET_MAX_GROUP_SIZE,
|
||||
)
|
||||
.execute(wtxn)?
|
||||
}
|
||||
}
|
||||
debug_assert!(crate::update::facet::sanity_checks(
|
||||
index,
|
||||
modified_facet_number_ids,
|
||||
wtxn,
|
||||
fid,
|
||||
FacetType::Number,
|
||||
FACET_GROUP_SIZE as usize,
|
||||
FACET_MIN_LEVEL_SIZE as usize,
|
||||
FACET_MAX_GROUP_SIZE as usize,
|
||||
)
|
||||
.execute(wtxn)?;
|
||||
.is_ok());
|
||||
}
|
||||
|
||||
Ok(())
|
||||
|
@ -1,6 +1,6 @@
|
||||
use std::cell::RefCell;
|
||||
|
||||
use hashbrown::HashSet;
|
||||
use hashbrown::HashMap;
|
||||
use heed::types::Bytes;
|
||||
use heed::{Database, RoTxn};
|
||||
use memmap2::Mmap;
|
||||
@ -12,6 +12,7 @@ use super::extract::{
|
||||
merge_caches_sorted, transpose_and_freeze_caches, BalancedCaches, DelAddRoaringBitmap,
|
||||
FacetKind, GeoExtractorData,
|
||||
};
|
||||
use crate::update::facet::new_incremental::{FacetFieldIdChange, FacetFieldIdOperation};
|
||||
use crate::{CboRoaringBitmapCodec, FieldId, GeoPoint, Index, InternalError, Result};
|
||||
|
||||
#[tracing::instrument(level = "trace", skip_all, target = "indexing::merge")]
|
||||
@ -100,23 +101,32 @@ pub fn merge_and_send_facet_docids<'extractor>(
|
||||
mut caches: Vec<BalancedCaches<'extractor>>,
|
||||
database: FacetDatabases,
|
||||
index: &Index,
|
||||
rtxn: &RoTxn,
|
||||
docids_sender: FacetDocidsSender,
|
||||
) -> Result<FacetFieldIdsDelta> {
|
||||
let max_string_count = (index.facet_id_string_docids.len(rtxn)? * 500) as usize;
|
||||
let max_number_count = (index.facet_id_f64_docids.len(rtxn)? * 500) as usize;
|
||||
transpose_and_freeze_caches(&mut caches)?
|
||||
.into_par_iter()
|
||||
.map(|frozen| {
|
||||
let mut facet_field_ids_delta = FacetFieldIdsDelta::default();
|
||||
let mut facet_field_ids_delta =
|
||||
FacetFieldIdsDelta::new(max_string_count, max_number_count);
|
||||
let rtxn = index.read_txn()?;
|
||||
merge_caches_sorted(frozen, |key, DelAddRoaringBitmap { del, add }| {
|
||||
let current = database.get_cbo_roaring_bytes_value(&rtxn, key)?;
|
||||
match merge_cbo_bitmaps(current, del, add)? {
|
||||
Operation::Write(bitmap) => {
|
||||
facet_field_ids_delta.register_from_key(key);
|
||||
let operation = if current.is_some() {
|
||||
FacetFieldIdOperation::InPlace
|
||||
} else {
|
||||
FacetFieldIdOperation::Insert
|
||||
};
|
||||
facet_field_ids_delta.register_from_key(key, operation);
|
||||
docids_sender.write(key, &bitmap)?;
|
||||
Ok(())
|
||||
}
|
||||
Operation::Delete => {
|
||||
facet_field_ids_delta.register_from_key(key);
|
||||
facet_field_ids_delta.register_from_key(key, FacetFieldIdOperation::Remove);
|
||||
docids_sender.delete(key)?;
|
||||
Ok(())
|
||||
}
|
||||
@ -126,7 +136,10 @@ pub fn merge_and_send_facet_docids<'extractor>(
|
||||
|
||||
Ok(facet_field_ids_delta)
|
||||
})
|
||||
.reduce(|| Ok(FacetFieldIdsDelta::default()), |lhs, rhs| Ok(lhs?.merge(rhs?)))
|
||||
.reduce(
|
||||
|| Ok(FacetFieldIdsDelta::new(max_string_count, max_number_count)),
|
||||
|lhs, rhs| Ok(lhs?.merge(rhs?)),
|
||||
)
|
||||
}
|
||||
|
||||
pub struct FacetDatabases<'a> {
|
||||
@ -155,60 +168,132 @@ impl<'a> FacetDatabases<'a> {
|
||||
}
|
||||
}
|
||||
|
||||
#[derive(Debug, Default)]
|
||||
#[derive(Debug)]
|
||||
pub enum FacetFieldIdDelta {
|
||||
Bulk,
|
||||
Incremental(Vec<FacetFieldIdChange>),
|
||||
}
|
||||
|
||||
impl FacetFieldIdDelta {
|
||||
fn push(&mut self, facet_value: &[u8], operation: FacetFieldIdOperation, max_count: usize) {
|
||||
*self = match std::mem::replace(self, FacetFieldIdDelta::Bulk) {
|
||||
FacetFieldIdDelta::Bulk => FacetFieldIdDelta::Bulk,
|
||||
FacetFieldIdDelta::Incremental(mut v) => {
|
||||
if v.len() >= max_count {
|
||||
FacetFieldIdDelta::Bulk
|
||||
} else {
|
||||
v.push(FacetFieldIdChange { facet_value: facet_value.into(), operation });
|
||||
FacetFieldIdDelta::Incremental(v)
|
||||
}
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
fn merge(&mut self, rhs: Option<Self>, max_count: usize) {
|
||||
let Some(rhs) = rhs else {
|
||||
return;
|
||||
};
|
||||
*self = match (std::mem::replace(self, FacetFieldIdDelta::Bulk), rhs) {
|
||||
(FacetFieldIdDelta::Bulk, _) | (_, FacetFieldIdDelta::Bulk) => FacetFieldIdDelta::Bulk,
|
||||
(
|
||||
FacetFieldIdDelta::Incremental(mut left),
|
||||
FacetFieldIdDelta::Incremental(mut right),
|
||||
) => {
|
||||
if left.len() + right.len() >= max_count {
|
||||
FacetFieldIdDelta::Bulk
|
||||
} else {
|
||||
left.append(&mut right);
|
||||
FacetFieldIdDelta::Incremental(left)
|
||||
}
|
||||
}
|
||||
};
|
||||
}
|
||||
}
|
||||
|
||||
#[derive(Debug)]
|
||||
pub struct FacetFieldIdsDelta {
|
||||
/// The field ids that have been modified
|
||||
modified_facet_string_ids: HashSet<FieldId>,
|
||||
modified_facet_number_ids: HashSet<FieldId>,
|
||||
modified_facet_string_ids: HashMap<FieldId, FacetFieldIdDelta, rustc_hash::FxBuildHasher>,
|
||||
modified_facet_number_ids: HashMap<FieldId, FacetFieldIdDelta, rustc_hash::FxBuildHasher>,
|
||||
max_string_count: usize,
|
||||
max_number_count: usize,
|
||||
}
|
||||
|
||||
impl FacetFieldIdsDelta {
|
||||
fn register_facet_string_id(&mut self, field_id: FieldId) {
|
||||
self.modified_facet_string_ids.insert(field_id);
|
||||
pub fn new(max_string_count: usize, max_number_count: usize) -> Self {
|
||||
Self {
|
||||
max_string_count,
|
||||
max_number_count,
|
||||
modified_facet_string_ids: Default::default(),
|
||||
modified_facet_number_ids: Default::default(),
|
||||
}
|
||||
}
|
||||
|
||||
fn register_facet_number_id(&mut self, field_id: FieldId) {
|
||||
self.modified_facet_number_ids.insert(field_id);
|
||||
fn register_facet_string_id(
|
||||
&mut self,
|
||||
field_id: FieldId,
|
||||
facet_value: &[u8],
|
||||
operation: FacetFieldIdOperation,
|
||||
) {
|
||||
self.modified_facet_string_ids
|
||||
.entry(field_id)
|
||||
.or_insert(FacetFieldIdDelta::Incremental(Default::default()))
|
||||
.push(facet_value, operation, self.max_string_count);
|
||||
}
|
||||
|
||||
fn register_from_key(&mut self, key: &[u8]) {
|
||||
let (facet_kind, field_id) = self.extract_key_data(key);
|
||||
fn register_facet_number_id(
|
||||
&mut self,
|
||||
field_id: FieldId,
|
||||
facet_value: &[u8],
|
||||
operation: FacetFieldIdOperation,
|
||||
) {
|
||||
self.modified_facet_number_ids
|
||||
.entry(field_id)
|
||||
.or_insert(FacetFieldIdDelta::Incremental(Default::default()))
|
||||
.push(facet_value, operation, self.max_number_count);
|
||||
}
|
||||
|
||||
fn register_from_key(&mut self, key: &[u8], operation: FacetFieldIdOperation) {
|
||||
let (facet_kind, field_id, facet_value) = self.extract_key_data(key);
|
||||
match facet_kind {
|
||||
FacetKind::Number => self.register_facet_number_id(field_id),
|
||||
FacetKind::String => self.register_facet_string_id(field_id),
|
||||
FacetKind::Number => self.register_facet_number_id(field_id, facet_value, operation),
|
||||
FacetKind::String => self.register_facet_string_id(field_id, facet_value, operation),
|
||||
_ => (),
|
||||
}
|
||||
}
|
||||
|
||||
fn extract_key_data(&self, key: &[u8]) -> (FacetKind, FieldId) {
|
||||
fn extract_key_data<'key>(&self, key: &'key [u8]) -> (FacetKind, FieldId, &'key [u8]) {
|
||||
let facet_kind = FacetKind::from(key[0]);
|
||||
let field_id = FieldId::from_be_bytes([key[1], key[2]]);
|
||||
(facet_kind, field_id)
|
||||
// level is also stored in the key at [3] (always 0)
|
||||
let facet_value = &key[4..];
|
||||
(facet_kind, field_id, facet_value)
|
||||
}
|
||||
|
||||
pub fn modified_facet_string_ids(&self) -> Option<Vec<FieldId>> {
|
||||
if self.modified_facet_string_ids.is_empty() {
|
||||
None
|
||||
} else {
|
||||
Some(self.modified_facet_string_ids.iter().copied().collect())
|
||||
}
|
||||
pub fn consume_facet_string_delta(
|
||||
&mut self,
|
||||
) -> impl Iterator<Item = (FieldId, FacetFieldIdDelta)> + '_ {
|
||||
self.modified_facet_string_ids.drain()
|
||||
}
|
||||
|
||||
pub fn modified_facet_number_ids(&self) -> Option<Vec<FieldId>> {
|
||||
if self.modified_facet_number_ids.is_empty() {
|
||||
None
|
||||
} else {
|
||||
Some(self.modified_facet_number_ids.iter().copied().collect())
|
||||
}
|
||||
pub fn consume_facet_number_delta(
|
||||
&mut self,
|
||||
) -> impl Iterator<Item = (FieldId, FacetFieldIdDelta)> + '_ {
|
||||
self.modified_facet_number_ids.drain()
|
||||
}
|
||||
|
||||
pub fn merge(mut self, rhs: Self) -> Self {
|
||||
let Self { modified_facet_number_ids, modified_facet_string_ids } = rhs;
|
||||
modified_facet_number_ids.into_iter().for_each(|fid| {
|
||||
self.modified_facet_number_ids.insert(fid);
|
||||
// rhs.max_xx_count is assumed to be equal to self.max_xx_count, and so gets unused
|
||||
let Self { modified_facet_number_ids, modified_facet_string_ids, .. } = rhs;
|
||||
modified_facet_number_ids.into_iter().for_each(|(fid, mut delta)| {
|
||||
let old_delta = self.modified_facet_number_ids.remove(&fid);
|
||||
delta.merge(old_delta, self.max_number_count);
|
||||
self.modified_facet_number_ids.insert(fid, delta);
|
||||
});
|
||||
modified_facet_string_ids.into_iter().for_each(|fid| {
|
||||
self.modified_facet_string_ids.insert(fid);
|
||||
modified_facet_string_ids.into_iter().for_each(|(fid, mut delta)| {
|
||||
let old_delta = self.modified_facet_string_ids.remove(&fid);
|
||||
delta.merge(old_delta, self.max_string_count);
|
||||
self.modified_facet_string_ids.insert(fid, delta);
|
||||
});
|
||||
self
|
||||
}
|
||||
|
Reference in New Issue
Block a user