mirror of
https://github.com/meilisearch/meilisearch.git
synced 2025-11-28 08:56:07 +00:00
Compare commits
3 Commits
release-v1
...
try-merge-
| Author | SHA1 | Date | |
|---|---|---|---|
|
14a980e54e | ||
|
|
cbc453c6d1 | ||
|
|
2fb065b9fb |
90
Cargo.lock
generated
90
Cargo.lock
generated
@@ -496,7 +496,7 @@ source = "git+https://github.com/meilisearch/bbqueue#cbb87cc707b5af415ef203bdaf2
|
||||
|
||||
[[package]]
|
||||
name = "benchmarks"
|
||||
version = "1.12.2"
|
||||
version = "1.12.0"
|
||||
dependencies = [
|
||||
"anyhow",
|
||||
"bumpalo",
|
||||
@@ -689,7 +689,7 @@ dependencies = [
|
||||
|
||||
[[package]]
|
||||
name = "build-info"
|
||||
version = "1.12.2"
|
||||
version = "1.12.0"
|
||||
dependencies = [
|
||||
"anyhow",
|
||||
"time",
|
||||
@@ -706,20 +706,6 @@ dependencies = [
|
||||
"serde",
|
||||
]
|
||||
|
||||
[[package]]
|
||||
name = "bumparaw-collections"
|
||||
version = "0.1.4"
|
||||
source = "registry+https://github.com/rust-lang/crates.io-index"
|
||||
checksum = "4ce682bdc86c2e25ef5cd95881d9d6a1902214eddf74cf9ffea88fe1464377e8"
|
||||
dependencies = [
|
||||
"allocator-api2",
|
||||
"bitpacking",
|
||||
"bumpalo",
|
||||
"hashbrown 0.15.1",
|
||||
"serde",
|
||||
"serde_json",
|
||||
]
|
||||
|
||||
[[package]]
|
||||
name = "byte-unit"
|
||||
version = "5.1.4"
|
||||
@@ -1664,7 +1650,7 @@ dependencies = [
|
||||
|
||||
[[package]]
|
||||
name = "dump"
|
||||
version = "1.12.2"
|
||||
version = "1.12.0"
|
||||
dependencies = [
|
||||
"anyhow",
|
||||
"big_s",
|
||||
@@ -1876,7 +1862,7 @@ checksum = "486f806e73c5707928240ddc295403b1b93c96a02038563881c4a2fd84b81ac4"
|
||||
|
||||
[[package]]
|
||||
name = "file-store"
|
||||
version = "1.12.2"
|
||||
version = "1.12.0"
|
||||
dependencies = [
|
||||
"tempfile",
|
||||
"thiserror",
|
||||
@@ -1898,7 +1884,7 @@ dependencies = [
|
||||
|
||||
[[package]]
|
||||
name = "filter-parser"
|
||||
version = "1.12.2"
|
||||
version = "1.12.0"
|
||||
dependencies = [
|
||||
"insta",
|
||||
"nom",
|
||||
@@ -1918,7 +1904,7 @@ dependencies = [
|
||||
|
||||
[[package]]
|
||||
name = "flatten-serde-json"
|
||||
version = "1.12.2"
|
||||
version = "1.12.0"
|
||||
dependencies = [
|
||||
"criterion",
|
||||
"serde_json",
|
||||
@@ -2057,7 +2043,7 @@ dependencies = [
|
||||
|
||||
[[package]]
|
||||
name = "fuzzers"
|
||||
version = "1.12.2"
|
||||
version = "1.12.0"
|
||||
dependencies = [
|
||||
"arbitrary",
|
||||
"bumpalo",
|
||||
@@ -2624,15 +2610,13 @@ checksum = "206ca75c9c03ba3d4ace2460e57b189f39f43de612c2f85836e65c929701bb2d"
|
||||
|
||||
[[package]]
|
||||
name = "index-scheduler"
|
||||
version = "1.12.2"
|
||||
version = "1.12.0"
|
||||
dependencies = [
|
||||
"anyhow",
|
||||
"arroy 0.5.0 (registry+https://github.com/rust-lang/crates.io-index)",
|
||||
"big_s",
|
||||
"bincode",
|
||||
"bumpalo",
|
||||
"bumparaw-collections",
|
||||
"convert_case 0.6.0",
|
||||
"crossbeam-channel",
|
||||
"csv",
|
||||
"derive_builder 0.20.0",
|
||||
@@ -2647,6 +2631,7 @@ dependencies = [
|
||||
"meilisearch-types",
|
||||
"memmap2",
|
||||
"page_size",
|
||||
"raw-collections",
|
||||
"rayon",
|
||||
"roaring",
|
||||
"serde",
|
||||
@@ -2662,12 +2647,12 @@ dependencies = [
|
||||
|
||||
[[package]]
|
||||
name = "indexmap"
|
||||
version = "2.7.0"
|
||||
version = "2.2.6"
|
||||
source = "registry+https://github.com/rust-lang/crates.io-index"
|
||||
checksum = "62f822373a4fe84d4bb149bf54e584a7f4abec90e072ed49cda0edea5b95471f"
|
||||
checksum = "168fb715dda47215e360912c096649d23d58bf392ac62f73919e831745e40f26"
|
||||
dependencies = [
|
||||
"equivalent",
|
||||
"hashbrown 0.15.1",
|
||||
"hashbrown 0.14.3",
|
||||
"serde",
|
||||
]
|
||||
|
||||
@@ -2822,7 +2807,7 @@ dependencies = [
|
||||
|
||||
[[package]]
|
||||
name = "json-depth-checker"
|
||||
version = "1.12.2"
|
||||
version = "1.12.0"
|
||||
dependencies = [
|
||||
"criterion",
|
||||
"serde_json",
|
||||
@@ -3441,7 +3426,7 @@ checksum = "490cc448043f947bae3cbee9c203358d62dbee0db12107a74be5c30ccfd09771"
|
||||
|
||||
[[package]]
|
||||
name = "meili-snap"
|
||||
version = "1.12.2"
|
||||
version = "1.12.0"
|
||||
dependencies = [
|
||||
"insta",
|
||||
"md5",
|
||||
@@ -3450,7 +3435,7 @@ dependencies = [
|
||||
|
||||
[[package]]
|
||||
name = "meilisearch"
|
||||
version = "1.12.2"
|
||||
version = "1.12.0"
|
||||
dependencies = [
|
||||
"actix-cors",
|
||||
"actix-http",
|
||||
@@ -3540,7 +3525,7 @@ dependencies = [
|
||||
|
||||
[[package]]
|
||||
name = "meilisearch-auth"
|
||||
version = "1.12.2"
|
||||
version = "1.12.0"
|
||||
dependencies = [
|
||||
"base64 0.22.1",
|
||||
"enum-iterator",
|
||||
@@ -3559,12 +3544,11 @@ dependencies = [
|
||||
|
||||
[[package]]
|
||||
name = "meilisearch-types"
|
||||
version = "1.12.2"
|
||||
version = "1.12.0"
|
||||
dependencies = [
|
||||
"actix-web",
|
||||
"anyhow",
|
||||
"bumpalo",
|
||||
"bumparaw-collections",
|
||||
"convert_case 0.6.0",
|
||||
"csv",
|
||||
"deserr",
|
||||
@@ -3577,8 +3561,8 @@ dependencies = [
|
||||
"meili-snap",
|
||||
"memmap2",
|
||||
"milli",
|
||||
"raw-collections",
|
||||
"roaring",
|
||||
"rustc-hash 2.1.0",
|
||||
"serde",
|
||||
"serde-cs",
|
||||
"serde_json",
|
||||
@@ -3592,19 +3576,16 @@ dependencies = [
|
||||
|
||||
[[package]]
|
||||
name = "meilitool"
|
||||
version = "1.12.2"
|
||||
version = "1.12.0"
|
||||
dependencies = [
|
||||
"anyhow",
|
||||
"arroy 0.5.0 (git+https://github.com/meilisearch/arroy/?tag=DO-NOT-DELETE-upgrade-v04-to-v05)",
|
||||
"clap",
|
||||
"dump",
|
||||
"file-store",
|
||||
"indexmap",
|
||||
"meilisearch-auth",
|
||||
"meilisearch-types",
|
||||
"serde",
|
||||
"serde_json",
|
||||
"tempfile",
|
||||
"time",
|
||||
"uuid",
|
||||
]
|
||||
@@ -3627,7 +3608,7 @@ dependencies = [
|
||||
|
||||
[[package]]
|
||||
name = "milli"
|
||||
version = "1.12.2"
|
||||
version = "1.12.0"
|
||||
dependencies = [
|
||||
"allocator-api2",
|
||||
"arroy 0.5.0 (registry+https://github.com/rust-lang/crates.io-index)",
|
||||
@@ -3637,7 +3618,6 @@ dependencies = [
|
||||
"bincode",
|
||||
"bstr",
|
||||
"bumpalo",
|
||||
"bumparaw-collections",
|
||||
"bytemuck",
|
||||
"byteorder",
|
||||
"candle-core",
|
||||
@@ -3676,12 +3656,13 @@ dependencies = [
|
||||
"once_cell",
|
||||
"ordered-float",
|
||||
"rand",
|
||||
"raw-collections",
|
||||
"rayon",
|
||||
"rayon-par-bridge",
|
||||
"rhai",
|
||||
"roaring",
|
||||
"rstar",
|
||||
"rustc-hash 2.1.0",
|
||||
"rustc-hash 2.0.0",
|
||||
"serde",
|
||||
"serde_json",
|
||||
"slice-group-by",
|
||||
@@ -4083,7 +4064,7 @@ checksum = "e3148f5046208a5d56bcfc03053e3ca6334e51da8dfb19b6cdc8b306fae3283e"
|
||||
|
||||
[[package]]
|
||||
name = "permissive-json-pointer"
|
||||
version = "1.12.2"
|
||||
version = "1.12.0"
|
||||
dependencies = [
|
||||
"big_s",
|
||||
"serde_json",
|
||||
@@ -4430,7 +4411,7 @@ dependencies = [
|
||||
"bytes",
|
||||
"rand",
|
||||
"ring",
|
||||
"rustc-hash 2.1.0",
|
||||
"rustc-hash 2.0.0",
|
||||
"rustls",
|
||||
"slab",
|
||||
"thiserror",
|
||||
@@ -4506,6 +4487,19 @@ dependencies = [
|
||||
"rand",
|
||||
]
|
||||
|
||||
[[package]]
|
||||
name = "raw-collections"
|
||||
version = "0.1.0"
|
||||
source = "git+https://github.com/meilisearch/raw-collections.git#15e5d7bdebc0c149b2a28b2454f307c717d07f8a"
|
||||
dependencies = [
|
||||
"allocator-api2",
|
||||
"bitpacking",
|
||||
"bumpalo",
|
||||
"hashbrown 0.15.1",
|
||||
"serde",
|
||||
"serde_json",
|
||||
]
|
||||
|
||||
[[package]]
|
||||
name = "raw-cpuid"
|
||||
version = "10.7.0"
|
||||
@@ -4803,9 +4797,9 @@ checksum = "08d43f7aa6b08d49f382cde6a7982047c3426db949b1424bc4b7ec9ae12c6ce2"
|
||||
|
||||
[[package]]
|
||||
name = "rustc-hash"
|
||||
version = "2.1.0"
|
||||
version = "2.0.0"
|
||||
source = "registry+https://github.com/rust-lang/crates.io-index"
|
||||
checksum = "c7fb8039b3032c191086b10f11f319a6e99e1e82889c5cc6046f515c9db1d497"
|
||||
checksum = "583034fd73374156e66797ed8e5b0d5690409c9226b22d87cb7f19821c05d152"
|
||||
|
||||
[[package]]
|
||||
name = "rustc_version"
|
||||
@@ -4974,9 +4968,9 @@ dependencies = [
|
||||
|
||||
[[package]]
|
||||
name = "serde_json"
|
||||
version = "1.0.133"
|
||||
version = "1.0.132"
|
||||
source = "registry+https://github.com/rust-lang/crates.io-index"
|
||||
checksum = "c7fceb2473b9166b2294ef05efcb65a3db80803f0b03ef86a5fc88a2b85ee377"
|
||||
checksum = "d726bfaff4b320266d395898905d0eba0345aae23b54aee3a737e260fd46db03"
|
||||
dependencies = [
|
||||
"indexmap",
|
||||
"itoa",
|
||||
@@ -6486,7 +6480,7 @@ dependencies = [
|
||||
|
||||
[[package]]
|
||||
name = "xtask"
|
||||
version = "1.12.2"
|
||||
version = "1.12.0"
|
||||
dependencies = [
|
||||
"anyhow",
|
||||
"build-info",
|
||||
|
||||
@@ -22,7 +22,7 @@ members = [
|
||||
]
|
||||
|
||||
[workspace.package]
|
||||
version = "1.12.2"
|
||||
version = "1.12.0"
|
||||
authors = [
|
||||
"Quentin de Quelen <quentin@dequelen.me>",
|
||||
"Clément Renault <clement@meilisearch.com>",
|
||||
|
||||
@@ -8,7 +8,6 @@ use bumpalo::Bump;
|
||||
use criterion::{criterion_group, criterion_main, Criterion};
|
||||
use milli::documents::PrimaryKey;
|
||||
use milli::heed::{EnvOpenOptions, RwTxn};
|
||||
use milli::progress::Progress;
|
||||
use milli::update::new::indexer;
|
||||
use milli::update::{IndexDocumentsMethod, IndexerConfig, Settings};
|
||||
use milli::vector::EmbeddingConfigs;
|
||||
@@ -152,7 +151,7 @@ fn indexing_songs_default(c: &mut Criterion) {
|
||||
None,
|
||||
&mut new_fields_ids_map,
|
||||
&|| false,
|
||||
Progress::default(),
|
||||
&|_progress| (),
|
||||
)
|
||||
.unwrap();
|
||||
|
||||
@@ -167,7 +166,7 @@ fn indexing_songs_default(c: &mut Criterion) {
|
||||
&document_changes,
|
||||
EmbeddingConfigs::default(),
|
||||
&|| false,
|
||||
&Progress::default(),
|
||||
&|_| (),
|
||||
)
|
||||
.unwrap();
|
||||
|
||||
@@ -219,7 +218,7 @@ fn reindexing_songs_default(c: &mut Criterion) {
|
||||
None,
|
||||
&mut new_fields_ids_map,
|
||||
&|| false,
|
||||
Progress::default(),
|
||||
&|_progress| (),
|
||||
)
|
||||
.unwrap();
|
||||
|
||||
@@ -234,7 +233,7 @@ fn reindexing_songs_default(c: &mut Criterion) {
|
||||
&document_changes,
|
||||
EmbeddingConfigs::default(),
|
||||
&|| false,
|
||||
&Progress::default(),
|
||||
&|_| (),
|
||||
)
|
||||
.unwrap();
|
||||
|
||||
@@ -264,7 +263,7 @@ fn reindexing_songs_default(c: &mut Criterion) {
|
||||
None,
|
||||
&mut new_fields_ids_map,
|
||||
&|| false,
|
||||
Progress::default(),
|
||||
&|_progress| (),
|
||||
)
|
||||
.unwrap();
|
||||
|
||||
@@ -279,7 +278,7 @@ fn reindexing_songs_default(c: &mut Criterion) {
|
||||
&document_changes,
|
||||
EmbeddingConfigs::default(),
|
||||
&|| false,
|
||||
&Progress::default(),
|
||||
&|_| (),
|
||||
)
|
||||
.unwrap();
|
||||
|
||||
@@ -333,7 +332,7 @@ fn deleting_songs_in_batches_default(c: &mut Criterion) {
|
||||
None,
|
||||
&mut new_fields_ids_map,
|
||||
&|| false,
|
||||
Progress::default(),
|
||||
&|_progress| (),
|
||||
)
|
||||
.unwrap();
|
||||
|
||||
@@ -348,7 +347,7 @@ fn deleting_songs_in_batches_default(c: &mut Criterion) {
|
||||
&document_changes,
|
||||
EmbeddingConfigs::default(),
|
||||
&|| false,
|
||||
&Progress::default(),
|
||||
&|_| (),
|
||||
)
|
||||
.unwrap();
|
||||
|
||||
@@ -410,7 +409,7 @@ fn indexing_songs_in_three_batches_default(c: &mut Criterion) {
|
||||
None,
|
||||
&mut new_fields_ids_map,
|
||||
&|| false,
|
||||
Progress::default(),
|
||||
&|_progress| (),
|
||||
)
|
||||
.unwrap();
|
||||
|
||||
@@ -425,7 +424,7 @@ fn indexing_songs_in_three_batches_default(c: &mut Criterion) {
|
||||
&document_changes,
|
||||
EmbeddingConfigs::default(),
|
||||
&|| false,
|
||||
&Progress::default(),
|
||||
&|_| (),
|
||||
)
|
||||
.unwrap();
|
||||
|
||||
@@ -455,7 +454,7 @@ fn indexing_songs_in_three_batches_default(c: &mut Criterion) {
|
||||
None,
|
||||
&mut new_fields_ids_map,
|
||||
&|| false,
|
||||
Progress::default(),
|
||||
&|_progress| (),
|
||||
)
|
||||
.unwrap();
|
||||
|
||||
@@ -470,7 +469,7 @@ fn indexing_songs_in_three_batches_default(c: &mut Criterion) {
|
||||
&document_changes,
|
||||
EmbeddingConfigs::default(),
|
||||
&|| false,
|
||||
&Progress::default(),
|
||||
&|_| (),
|
||||
)
|
||||
.unwrap();
|
||||
|
||||
@@ -496,7 +495,7 @@ fn indexing_songs_in_three_batches_default(c: &mut Criterion) {
|
||||
None,
|
||||
&mut new_fields_ids_map,
|
||||
&|| false,
|
||||
Progress::default(),
|
||||
&|_progress| (),
|
||||
)
|
||||
.unwrap();
|
||||
|
||||
@@ -511,7 +510,7 @@ fn indexing_songs_in_three_batches_default(c: &mut Criterion) {
|
||||
&document_changes,
|
||||
EmbeddingConfigs::default(),
|
||||
&|| false,
|
||||
&Progress::default(),
|
||||
&|_| (),
|
||||
)
|
||||
.unwrap();
|
||||
|
||||
@@ -564,7 +563,7 @@ fn indexing_songs_without_faceted_numbers(c: &mut Criterion) {
|
||||
None,
|
||||
&mut new_fields_ids_map,
|
||||
&|| false,
|
||||
Progress::default(),
|
||||
&|_progress| (),
|
||||
)
|
||||
.unwrap();
|
||||
|
||||
@@ -579,7 +578,7 @@ fn indexing_songs_without_faceted_numbers(c: &mut Criterion) {
|
||||
&document_changes,
|
||||
EmbeddingConfigs::default(),
|
||||
&|| false,
|
||||
&Progress::default(),
|
||||
&|_| (),
|
||||
)
|
||||
.unwrap();
|
||||
|
||||
@@ -631,7 +630,7 @@ fn indexing_songs_without_faceted_fields(c: &mut Criterion) {
|
||||
None,
|
||||
&mut new_fields_ids_map,
|
||||
&|| false,
|
||||
Progress::default(),
|
||||
&|_progress| (),
|
||||
)
|
||||
.unwrap();
|
||||
|
||||
@@ -646,7 +645,7 @@ fn indexing_songs_without_faceted_fields(c: &mut Criterion) {
|
||||
&document_changes,
|
||||
EmbeddingConfigs::default(),
|
||||
&|| false,
|
||||
&Progress::default(),
|
||||
&|_| (),
|
||||
)
|
||||
.unwrap();
|
||||
|
||||
@@ -698,7 +697,7 @@ fn indexing_wiki(c: &mut Criterion) {
|
||||
None,
|
||||
&mut new_fields_ids_map,
|
||||
&|| false,
|
||||
Progress::default(),
|
||||
&|_progress| (),
|
||||
)
|
||||
.unwrap();
|
||||
|
||||
@@ -713,7 +712,7 @@ fn indexing_wiki(c: &mut Criterion) {
|
||||
&document_changes,
|
||||
EmbeddingConfigs::default(),
|
||||
&|| false,
|
||||
&Progress::default(),
|
||||
&|_| (),
|
||||
)
|
||||
.unwrap();
|
||||
|
||||
@@ -764,7 +763,7 @@ fn reindexing_wiki(c: &mut Criterion) {
|
||||
None,
|
||||
&mut new_fields_ids_map,
|
||||
&|| false,
|
||||
Progress::default(),
|
||||
&|_progress| (),
|
||||
)
|
||||
.unwrap();
|
||||
|
||||
@@ -779,7 +778,7 @@ fn reindexing_wiki(c: &mut Criterion) {
|
||||
&document_changes,
|
||||
EmbeddingConfigs::default(),
|
||||
&|| false,
|
||||
&Progress::default(),
|
||||
&|_| (),
|
||||
)
|
||||
.unwrap();
|
||||
|
||||
@@ -809,7 +808,7 @@ fn reindexing_wiki(c: &mut Criterion) {
|
||||
None,
|
||||
&mut new_fields_ids_map,
|
||||
&|| false,
|
||||
Progress::default(),
|
||||
&|_progress| (),
|
||||
)
|
||||
.unwrap();
|
||||
|
||||
@@ -824,7 +823,7 @@ fn reindexing_wiki(c: &mut Criterion) {
|
||||
&document_changes,
|
||||
EmbeddingConfigs::default(),
|
||||
&|| false,
|
||||
&Progress::default(),
|
||||
&|_| (),
|
||||
)
|
||||
.unwrap();
|
||||
|
||||
@@ -877,7 +876,7 @@ fn deleting_wiki_in_batches_default(c: &mut Criterion) {
|
||||
None,
|
||||
&mut new_fields_ids_map,
|
||||
&|| false,
|
||||
Progress::default(),
|
||||
&|_progress| (),
|
||||
)
|
||||
.unwrap();
|
||||
|
||||
@@ -892,7 +891,7 @@ fn deleting_wiki_in_batches_default(c: &mut Criterion) {
|
||||
&document_changes,
|
||||
EmbeddingConfigs::default(),
|
||||
&|| false,
|
||||
&Progress::default(),
|
||||
&|_| (),
|
||||
)
|
||||
.unwrap();
|
||||
|
||||
@@ -954,7 +953,7 @@ fn indexing_wiki_in_three_batches(c: &mut Criterion) {
|
||||
None,
|
||||
&mut new_fields_ids_map,
|
||||
&|| false,
|
||||
Progress::default(),
|
||||
&|_progress| (),
|
||||
)
|
||||
.unwrap();
|
||||
|
||||
@@ -969,7 +968,7 @@ fn indexing_wiki_in_three_batches(c: &mut Criterion) {
|
||||
&document_changes,
|
||||
EmbeddingConfigs::default(),
|
||||
&|| false,
|
||||
&Progress::default(),
|
||||
&|_| (),
|
||||
)
|
||||
.unwrap();
|
||||
|
||||
@@ -1000,7 +999,7 @@ fn indexing_wiki_in_three_batches(c: &mut Criterion) {
|
||||
None,
|
||||
&mut new_fields_ids_map,
|
||||
&|| false,
|
||||
Progress::default(),
|
||||
&|_progress| (),
|
||||
)
|
||||
.unwrap();
|
||||
|
||||
@@ -1015,7 +1014,7 @@ fn indexing_wiki_in_three_batches(c: &mut Criterion) {
|
||||
&document_changes,
|
||||
EmbeddingConfigs::default(),
|
||||
&|| false,
|
||||
&Progress::default(),
|
||||
&|_| (),
|
||||
)
|
||||
.unwrap();
|
||||
|
||||
@@ -1042,7 +1041,7 @@ fn indexing_wiki_in_three_batches(c: &mut Criterion) {
|
||||
None,
|
||||
&mut new_fields_ids_map,
|
||||
&|| false,
|
||||
Progress::default(),
|
||||
&|_progress| (),
|
||||
)
|
||||
.unwrap();
|
||||
|
||||
@@ -1057,7 +1056,7 @@ fn indexing_wiki_in_three_batches(c: &mut Criterion) {
|
||||
&document_changes,
|
||||
EmbeddingConfigs::default(),
|
||||
&|| false,
|
||||
&Progress::default(),
|
||||
&|_| (),
|
||||
)
|
||||
.unwrap();
|
||||
|
||||
@@ -1109,7 +1108,7 @@ fn indexing_movies_default(c: &mut Criterion) {
|
||||
None,
|
||||
&mut new_fields_ids_map,
|
||||
&|| false,
|
||||
Progress::default(),
|
||||
&|_progress| (),
|
||||
)
|
||||
.unwrap();
|
||||
|
||||
@@ -1124,7 +1123,7 @@ fn indexing_movies_default(c: &mut Criterion) {
|
||||
&document_changes,
|
||||
EmbeddingConfigs::default(),
|
||||
&|| false,
|
||||
&Progress::default(),
|
||||
&|_| (),
|
||||
)
|
||||
.unwrap();
|
||||
|
||||
@@ -1175,7 +1174,7 @@ fn reindexing_movies_default(c: &mut Criterion) {
|
||||
None,
|
||||
&mut new_fields_ids_map,
|
||||
&|| false,
|
||||
Progress::default(),
|
||||
&|_progress| (),
|
||||
)
|
||||
.unwrap();
|
||||
|
||||
@@ -1190,7 +1189,7 @@ fn reindexing_movies_default(c: &mut Criterion) {
|
||||
&document_changes,
|
||||
EmbeddingConfigs::default(),
|
||||
&|| false,
|
||||
&Progress::default(),
|
||||
&|_| (),
|
||||
)
|
||||
.unwrap();
|
||||
|
||||
@@ -1220,7 +1219,7 @@ fn reindexing_movies_default(c: &mut Criterion) {
|
||||
None,
|
||||
&mut new_fields_ids_map,
|
||||
&|| false,
|
||||
Progress::default(),
|
||||
&|_progress| (),
|
||||
)
|
||||
.unwrap();
|
||||
|
||||
@@ -1235,7 +1234,7 @@ fn reindexing_movies_default(c: &mut Criterion) {
|
||||
&document_changes,
|
||||
EmbeddingConfigs::default(),
|
||||
&|| false,
|
||||
&Progress::default(),
|
||||
&|_| (),
|
||||
)
|
||||
.unwrap();
|
||||
|
||||
@@ -1288,7 +1287,7 @@ fn deleting_movies_in_batches_default(c: &mut Criterion) {
|
||||
None,
|
||||
&mut new_fields_ids_map,
|
||||
&|| false,
|
||||
Progress::default(),
|
||||
&|_progress| (),
|
||||
)
|
||||
.unwrap();
|
||||
|
||||
@@ -1303,7 +1302,7 @@ fn deleting_movies_in_batches_default(c: &mut Criterion) {
|
||||
&document_changes,
|
||||
EmbeddingConfigs::default(),
|
||||
&|| false,
|
||||
&Progress::default(),
|
||||
&|_| (),
|
||||
)
|
||||
.unwrap();
|
||||
|
||||
@@ -1351,7 +1350,7 @@ fn delete_documents_from_ids(index: Index, document_ids_to_delete: Vec<RoaringBi
|
||||
&document_changes,
|
||||
EmbeddingConfigs::default(),
|
||||
&|| false,
|
||||
&Progress::default(),
|
||||
&|_| (),
|
||||
)
|
||||
.unwrap();
|
||||
|
||||
@@ -1401,7 +1400,7 @@ fn indexing_movies_in_three_batches(c: &mut Criterion) {
|
||||
None,
|
||||
&mut new_fields_ids_map,
|
||||
&|| false,
|
||||
Progress::default(),
|
||||
&|_progress| (),
|
||||
)
|
||||
.unwrap();
|
||||
|
||||
@@ -1416,7 +1415,7 @@ fn indexing_movies_in_three_batches(c: &mut Criterion) {
|
||||
&document_changes,
|
||||
EmbeddingConfigs::default(),
|
||||
&|| false,
|
||||
&Progress::default(),
|
||||
&|_| (),
|
||||
)
|
||||
.unwrap();
|
||||
|
||||
@@ -1446,7 +1445,7 @@ fn indexing_movies_in_three_batches(c: &mut Criterion) {
|
||||
None,
|
||||
&mut new_fields_ids_map,
|
||||
&|| false,
|
||||
Progress::default(),
|
||||
&|_progress| (),
|
||||
)
|
||||
.unwrap();
|
||||
|
||||
@@ -1461,7 +1460,7 @@ fn indexing_movies_in_three_batches(c: &mut Criterion) {
|
||||
&document_changes,
|
||||
EmbeddingConfigs::default(),
|
||||
&|| false,
|
||||
&Progress::default(),
|
||||
&|_| (),
|
||||
)
|
||||
.unwrap();
|
||||
|
||||
@@ -1487,7 +1486,7 @@ fn indexing_movies_in_three_batches(c: &mut Criterion) {
|
||||
None,
|
||||
&mut new_fields_ids_map,
|
||||
&|| false,
|
||||
Progress::default(),
|
||||
&|_progress| (),
|
||||
)
|
||||
.unwrap();
|
||||
|
||||
@@ -1502,7 +1501,7 @@ fn indexing_movies_in_three_batches(c: &mut Criterion) {
|
||||
&document_changes,
|
||||
EmbeddingConfigs::default(),
|
||||
&|| false,
|
||||
&Progress::default(),
|
||||
&|_| (),
|
||||
)
|
||||
.unwrap();
|
||||
|
||||
@@ -1577,7 +1576,7 @@ fn indexing_nested_movies_default(c: &mut Criterion) {
|
||||
None,
|
||||
&mut new_fields_ids_map,
|
||||
&|| false,
|
||||
Progress::default(),
|
||||
&|_progress| (),
|
||||
)
|
||||
.unwrap();
|
||||
|
||||
@@ -1592,7 +1591,7 @@ fn indexing_nested_movies_default(c: &mut Criterion) {
|
||||
&document_changes,
|
||||
EmbeddingConfigs::default(),
|
||||
&|| false,
|
||||
&Progress::default(),
|
||||
&|_| (),
|
||||
)
|
||||
.unwrap();
|
||||
|
||||
@@ -1668,7 +1667,7 @@ fn deleting_nested_movies_in_batches_default(c: &mut Criterion) {
|
||||
None,
|
||||
&mut new_fields_ids_map,
|
||||
&|| false,
|
||||
Progress::default(),
|
||||
&|_progress| (),
|
||||
)
|
||||
.unwrap();
|
||||
|
||||
@@ -1683,7 +1682,7 @@ fn deleting_nested_movies_in_batches_default(c: &mut Criterion) {
|
||||
&document_changes,
|
||||
EmbeddingConfigs::default(),
|
||||
&|| false,
|
||||
&Progress::default(),
|
||||
&|_| (),
|
||||
)
|
||||
.unwrap();
|
||||
|
||||
@@ -1751,7 +1750,7 @@ fn indexing_nested_movies_without_faceted_fields(c: &mut Criterion) {
|
||||
None,
|
||||
&mut new_fields_ids_map,
|
||||
&|| false,
|
||||
Progress::default(),
|
||||
&|_progress| (),
|
||||
)
|
||||
.unwrap();
|
||||
|
||||
@@ -1766,7 +1765,7 @@ fn indexing_nested_movies_without_faceted_fields(c: &mut Criterion) {
|
||||
&document_changes,
|
||||
EmbeddingConfigs::default(),
|
||||
&|| false,
|
||||
&Progress::default(),
|
||||
&|_| (),
|
||||
)
|
||||
.unwrap();
|
||||
|
||||
@@ -1818,7 +1817,7 @@ fn indexing_geo(c: &mut Criterion) {
|
||||
None,
|
||||
&mut new_fields_ids_map,
|
||||
&|| false,
|
||||
Progress::default(),
|
||||
&|_progress| (),
|
||||
)
|
||||
.unwrap();
|
||||
|
||||
@@ -1833,7 +1832,7 @@ fn indexing_geo(c: &mut Criterion) {
|
||||
&document_changes,
|
||||
EmbeddingConfigs::default(),
|
||||
&|| false,
|
||||
&Progress::default(),
|
||||
&|_| (),
|
||||
)
|
||||
.unwrap();
|
||||
|
||||
@@ -1884,7 +1883,7 @@ fn reindexing_geo(c: &mut Criterion) {
|
||||
None,
|
||||
&mut new_fields_ids_map,
|
||||
&|| false,
|
||||
Progress::default(),
|
||||
&|_progress| (),
|
||||
)
|
||||
.unwrap();
|
||||
|
||||
@@ -1899,7 +1898,7 @@ fn reindexing_geo(c: &mut Criterion) {
|
||||
&document_changes,
|
||||
EmbeddingConfigs::default(),
|
||||
&|| false,
|
||||
&Progress::default(),
|
||||
&|_| (),
|
||||
)
|
||||
.unwrap();
|
||||
|
||||
@@ -1929,7 +1928,7 @@ fn reindexing_geo(c: &mut Criterion) {
|
||||
None,
|
||||
&mut new_fields_ids_map,
|
||||
&|| false,
|
||||
Progress::default(),
|
||||
&|_progress| (),
|
||||
)
|
||||
.unwrap();
|
||||
|
||||
@@ -1944,7 +1943,7 @@ fn reindexing_geo(c: &mut Criterion) {
|
||||
&document_changes,
|
||||
EmbeddingConfigs::default(),
|
||||
&|| false,
|
||||
&Progress::default(),
|
||||
&|_| (),
|
||||
)
|
||||
.unwrap();
|
||||
|
||||
@@ -1997,7 +1996,7 @@ fn deleting_geo_in_batches_default(c: &mut Criterion) {
|
||||
None,
|
||||
&mut new_fields_ids_map,
|
||||
&|| false,
|
||||
Progress::default(),
|
||||
&|_progress| (),
|
||||
)
|
||||
.unwrap();
|
||||
|
||||
@@ -2012,7 +2011,7 @@ fn deleting_geo_in_batches_default(c: &mut Criterion) {
|
||||
&document_changes,
|
||||
EmbeddingConfigs::default(),
|
||||
&|| false,
|
||||
&Progress::default(),
|
||||
&|_| (),
|
||||
)
|
||||
.unwrap();
|
||||
|
||||
|
||||
@@ -10,7 +10,6 @@ use bumpalo::Bump;
|
||||
use criterion::BenchmarkId;
|
||||
use memmap2::Mmap;
|
||||
use milli::heed::EnvOpenOptions;
|
||||
use milli::progress::Progress;
|
||||
use milli::update::new::indexer;
|
||||
use milli::update::{IndexDocumentsMethod, IndexerConfig, Settings};
|
||||
use milli::vector::EmbeddingConfigs;
|
||||
@@ -111,7 +110,7 @@ pub fn base_setup(conf: &Conf) -> Index {
|
||||
None,
|
||||
&mut new_fields_ids_map,
|
||||
&|| false,
|
||||
Progress::default(),
|
||||
&|_progress| (),
|
||||
)
|
||||
.unwrap();
|
||||
|
||||
@@ -126,7 +125,7 @@ pub fn base_setup(conf: &Conf) -> Index {
|
||||
&document_changes,
|
||||
EmbeddingConfigs::default(),
|
||||
&|| false,
|
||||
&Progress::default(),
|
||||
&|_| (),
|
||||
)
|
||||
.unwrap();
|
||||
|
||||
|
||||
@@ -136,14 +136,6 @@ pub struct File {
|
||||
}
|
||||
|
||||
impl File {
|
||||
pub fn from_parts(path: PathBuf, file: Option<NamedTempFile>) -> Self {
|
||||
Self { path, file }
|
||||
}
|
||||
|
||||
pub fn into_parts(self) -> (PathBuf, Option<NamedTempFile>) {
|
||||
(self.path, self.file)
|
||||
}
|
||||
|
||||
pub fn dry_file() -> Result<Self> {
|
||||
Ok(Self { path: PathBuf::new(), file: None })
|
||||
}
|
||||
|
||||
@@ -10,7 +10,6 @@ use either::Either;
|
||||
use fuzzers::Operation;
|
||||
use milli::documents::mmap_from_objects;
|
||||
use milli::heed::EnvOpenOptions;
|
||||
use milli::progress::Progress;
|
||||
use milli::update::new::indexer;
|
||||
use milli::update::{IndexDocumentsMethod, IndexerConfig};
|
||||
use milli::vector::EmbeddingConfigs;
|
||||
@@ -129,7 +128,7 @@ fn main() {
|
||||
None,
|
||||
&mut new_fields_ids_map,
|
||||
&|| false,
|
||||
Progress::default(),
|
||||
&|_progress| (),
|
||||
)
|
||||
.unwrap();
|
||||
|
||||
@@ -144,7 +143,7 @@ fn main() {
|
||||
&document_changes,
|
||||
embedders,
|
||||
&|| false,
|
||||
&Progress::default(),
|
||||
&|_| (),
|
||||
)
|
||||
.unwrap();
|
||||
|
||||
|
||||
@@ -13,9 +13,6 @@ license.workspace = true
|
||||
[dependencies]
|
||||
anyhow = "1.0.86"
|
||||
bincode = "1.3.3"
|
||||
bumpalo = "3.16.0"
|
||||
bumparaw-collections = "0.1.2"
|
||||
convert_case = "0.6.0"
|
||||
csv = "1.3.0"
|
||||
derive_builder = "0.20.0"
|
||||
dump = { path = "../dump" }
|
||||
@@ -24,8 +21,8 @@ file-store = { path = "../file-store" }
|
||||
flate2 = "1.0.30"
|
||||
meilisearch-auth = { path = "../meilisearch-auth" }
|
||||
meilisearch-types = { path = "../meilisearch-types" }
|
||||
memmap2 = "0.9.4"
|
||||
page_size = "0.6.0"
|
||||
raw-collections = { git = "https://github.com/meilisearch/raw-collections.git", version = "0.1.0" }
|
||||
rayon = "1.10.0"
|
||||
roaring = { version = "0.10.7", features = ["serde"] }
|
||||
serde = { version = "1.0.204", features = ["derive"] }
|
||||
@@ -33,6 +30,7 @@ serde_json = { version = "1.0.120", features = ["preserve_order"] }
|
||||
synchronoise = "1.0.1"
|
||||
tempfile = "3.10.1"
|
||||
thiserror = "1.0.61"
|
||||
memmap2 = "0.9.4"
|
||||
time = { version = "0.3.36", features = [
|
||||
"serde-well-known",
|
||||
"formatting",
|
||||
@@ -42,6 +40,7 @@ time = { version = "0.3.36", features = [
|
||||
tracing = "0.1.40"
|
||||
ureq = "2.10.0"
|
||||
uuid = { version = "1.10.0", features = ["serde", "v4"] }
|
||||
bumpalo = "3.16.0"
|
||||
|
||||
[dev-dependencies]
|
||||
arroy = "0.5.0"
|
||||
|
||||
@@ -22,7 +22,8 @@ use std::ffi::OsStr;
|
||||
use std::fmt;
|
||||
use std::fs::{self, File};
|
||||
use std::io::BufWriter;
|
||||
use std::sync::atomic::Ordering;
|
||||
use std::sync::atomic::{self, AtomicU64};
|
||||
use std::time::Duration;
|
||||
|
||||
use bumpalo::collections::CollectIn;
|
||||
use bumpalo::Bump;
|
||||
@@ -31,17 +32,16 @@ use meilisearch_types::batches::BatchId;
|
||||
use meilisearch_types::heed::{RoTxn, RwTxn};
|
||||
use meilisearch_types::milli::documents::{obkv_to_object, DocumentsBatchReader, PrimaryKey};
|
||||
use meilisearch_types::milli::heed::CompactionOption;
|
||||
use meilisearch_types::milli::progress::Progress;
|
||||
use meilisearch_types::milli::update::new::indexer::{self, UpdateByFunction};
|
||||
use meilisearch_types::milli::update::{
|
||||
DocumentAdditionResult, IndexDocumentsMethod, Settings as MilliSettings,
|
||||
};
|
||||
use meilisearch_types::milli::update::{IndexDocumentsMethod, Settings as MilliSettings};
|
||||
use meilisearch_types::milli::vector::parsed_vectors::{
|
||||
ExplicitVectors, VectorOrArrayOfVectors, RESERVED_VECTORS_FIELD_NAME,
|
||||
};
|
||||
use meilisearch_types::milli::{self, Filter, ThreadPoolNoAbortBuilder};
|
||||
use meilisearch_types::settings::{apply_settings_to_builder, Settings, Unchecked};
|
||||
use meilisearch_types::tasks::{Details, IndexSwap, Kind, KindWithContent, Status, Task};
|
||||
use meilisearch_types::tasks::{
|
||||
Details, IndexSwap, Kind, KindWithContent, Status, Task, TaskProgress,
|
||||
};
|
||||
use meilisearch_types::{compression, Index, VERSION_FILE_NAME};
|
||||
use roaring::RoaringBitmap;
|
||||
use time::macros::format_description;
|
||||
@@ -49,13 +49,6 @@ use time::OffsetDateTime;
|
||||
use uuid::Uuid;
|
||||
|
||||
use crate::autobatcher::{self, BatchKind};
|
||||
use crate::processing::{
|
||||
AtomicBatchStep, AtomicDocumentStep, AtomicTaskStep, AtomicUpdateFileStep, CreateIndexProgress,
|
||||
DeleteIndexProgress, DocumentDeletionProgress, DocumentEditionProgress,
|
||||
DocumentOperationProgress, DumpCreationProgress, InnerSwappingTwoIndexes, SettingsProgress,
|
||||
SnapshotCreationProgress, SwappingTheIndexes, TaskCancelationProgress, TaskDeletionProgress,
|
||||
UpdateIndexProgress, VariableNameStep,
|
||||
};
|
||||
use crate::utils::{self, swap_index_uid_in_task, ProcessingBatch};
|
||||
use crate::{Error, IndexScheduler, Result, TaskId};
|
||||
|
||||
@@ -566,12 +559,11 @@ impl IndexScheduler {
|
||||
/// The list of tasks that were processed. The metadata of each task in the returned
|
||||
/// list is updated accordingly, with the exception of the its date fields
|
||||
/// [`finished_at`](meilisearch_types::tasks::Task::finished_at) and [`started_at`](meilisearch_types::tasks::Task::started_at).
|
||||
#[tracing::instrument(level = "trace", skip(self, batch, progress), target = "indexing::scheduler", fields(batch=batch.to_string()))]
|
||||
#[tracing::instrument(level = "trace", skip(self, batch), target = "indexing::scheduler", fields(batch=batch.to_string()))]
|
||||
pub(crate) fn process_batch(
|
||||
&self,
|
||||
batch: Batch,
|
||||
current_batch: &mut ProcessingBatch,
|
||||
progress: Progress,
|
||||
) -> Result<Vec<Task>> {
|
||||
#[cfg(test)]
|
||||
{
|
||||
@@ -591,13 +583,8 @@ impl IndexScheduler {
|
||||
};
|
||||
|
||||
let rtxn = self.env.read_txn()?;
|
||||
let mut canceled_tasks = self.cancel_matched_tasks(
|
||||
&rtxn,
|
||||
task.uid,
|
||||
current_batch,
|
||||
matched_tasks,
|
||||
&progress,
|
||||
)?;
|
||||
let mut canceled_tasks =
|
||||
self.cancel_matched_tasks(&rtxn, task.uid, current_batch, matched_tasks)?;
|
||||
|
||||
task.status = Status::Succeeded;
|
||||
match &mut task.details {
|
||||
@@ -628,8 +615,7 @@ impl IndexScheduler {
|
||||
}
|
||||
|
||||
let mut wtxn = self.env.write_txn()?;
|
||||
let mut deleted_tasks =
|
||||
self.delete_matched_tasks(&mut wtxn, &matched_tasks, &progress)?;
|
||||
let mut deleted_tasks = self.delete_matched_tasks(&mut wtxn, &matched_tasks)?;
|
||||
wtxn.commit()?;
|
||||
|
||||
for task in tasks.iter_mut() {
|
||||
@@ -655,8 +641,6 @@ impl IndexScheduler {
|
||||
Ok(tasks)
|
||||
}
|
||||
Batch::SnapshotCreation(mut tasks) => {
|
||||
progress.update_progress(SnapshotCreationProgress::StartTheSnapshotCreation);
|
||||
|
||||
fs::create_dir_all(&self.snapshots_path)?;
|
||||
let temp_snapshot_dir = tempfile::tempdir()?;
|
||||
|
||||
@@ -677,7 +661,6 @@ impl IndexScheduler {
|
||||
// two read operations as the task processing is synchronous.
|
||||
|
||||
// 2.1 First copy the LMDB env of the index-scheduler
|
||||
progress.update_progress(SnapshotCreationProgress::SnapshotTheIndexScheduler);
|
||||
let dst = temp_snapshot_dir.path().join("tasks");
|
||||
fs::create_dir_all(&dst)?;
|
||||
self.env.copy_to_file(dst.join("data.mdb"), CompactionOption::Enabled)?;
|
||||
@@ -690,29 +673,18 @@ impl IndexScheduler {
|
||||
fs::create_dir_all(&update_files_dir)?;
|
||||
|
||||
// 2.4 Only copy the update files of the enqueued tasks
|
||||
progress.update_progress(SnapshotCreationProgress::SnapshotTheUpdateFiles);
|
||||
let enqueued = self.get_status(&rtxn, Status::Enqueued)?;
|
||||
let (atomic, update_file_progress) =
|
||||
AtomicUpdateFileStep::new(enqueued.len() as u32);
|
||||
progress.update_progress(update_file_progress);
|
||||
for task_id in enqueued {
|
||||
for task_id in self.get_status(&rtxn, Status::Enqueued)? {
|
||||
let task = self.get_task(&rtxn, task_id)?.ok_or(Error::CorruptedTaskQueue)?;
|
||||
if let Some(content_uuid) = task.content_uuid() {
|
||||
let src = self.file_store.get_update_path(content_uuid);
|
||||
let dst = update_files_dir.join(content_uuid.to_string());
|
||||
fs::copy(src, dst)?;
|
||||
}
|
||||
atomic.fetch_add(1, Ordering::Relaxed);
|
||||
}
|
||||
|
||||
// 3. Snapshot every indexes
|
||||
progress.update_progress(SnapshotCreationProgress::SnapshotTheIndexes);
|
||||
let index_mapping = self.index_mapper.index_mapping;
|
||||
let nb_indexes = index_mapping.len(&rtxn)? as u32;
|
||||
|
||||
for (i, result) in index_mapping.iter(&rtxn)?.enumerate() {
|
||||
for result in self.index_mapper.index_mapping.iter(&rtxn)? {
|
||||
let (name, uuid) = result?;
|
||||
progress.update_progress(VariableNameStep::new(name, i as u32, nb_indexes));
|
||||
let index = self.index_mapper.index(&rtxn, name)?;
|
||||
let dst = temp_snapshot_dir.path().join("indexes").join(uuid.to_string());
|
||||
fs::create_dir_all(&dst)?;
|
||||
@@ -724,7 +696,6 @@ impl IndexScheduler {
|
||||
drop(rtxn);
|
||||
|
||||
// 4. Snapshot the auth LMDB env
|
||||
progress.update_progress(SnapshotCreationProgress::SnapshotTheApiKeys);
|
||||
let dst = temp_snapshot_dir.path().join("auth");
|
||||
fs::create_dir_all(&dst)?;
|
||||
// TODO We can't use the open_auth_store_env function here but we should
|
||||
@@ -737,7 +708,6 @@ impl IndexScheduler {
|
||||
auth.copy_to_file(dst.join("data.mdb"), CompactionOption::Enabled)?;
|
||||
|
||||
// 5. Copy and tarball the flat snapshot
|
||||
progress.update_progress(SnapshotCreationProgress::CreateTheTarball);
|
||||
// 5.1 Find the original name of the database
|
||||
// TODO find a better way to get this path
|
||||
let mut base_path = self.env.path().to_owned();
|
||||
@@ -770,7 +740,6 @@ impl IndexScheduler {
|
||||
Ok(tasks)
|
||||
}
|
||||
Batch::Dump(mut task) => {
|
||||
progress.update_progress(DumpCreationProgress::StartTheDumpCreation);
|
||||
let started_at = OffsetDateTime::now_utc();
|
||||
let (keys, instance_uid) =
|
||||
if let KindWithContent::DumpCreation { keys, instance_uid } = &task.kind {
|
||||
@@ -781,7 +750,6 @@ impl IndexScheduler {
|
||||
let dump = dump::DumpWriter::new(*instance_uid)?;
|
||||
|
||||
// 1. dump the keys
|
||||
progress.update_progress(DumpCreationProgress::DumpTheApiKeys);
|
||||
let mut dump_keys = dump.create_keys()?;
|
||||
for key in keys {
|
||||
dump_keys.push_key(key)?;
|
||||
@@ -791,13 +759,7 @@ impl IndexScheduler {
|
||||
let rtxn = self.env.read_txn()?;
|
||||
|
||||
// 2. dump the tasks
|
||||
progress.update_progress(DumpCreationProgress::DumpTheTasks);
|
||||
let mut dump_tasks = dump.create_tasks_queue()?;
|
||||
|
||||
let (atomic, update_task_progress) =
|
||||
AtomicTaskStep::new(self.all_tasks.len(&rtxn)? as u32);
|
||||
progress.update_progress(update_task_progress);
|
||||
|
||||
for ret in self.all_tasks.iter(&rtxn)? {
|
||||
if self.must_stop_processing.get() {
|
||||
return Err(Error::AbortedTask);
|
||||
@@ -847,22 +809,11 @@ impl IndexScheduler {
|
||||
dump_content_file.flush()?;
|
||||
}
|
||||
}
|
||||
atomic.fetch_add(1, Ordering::Relaxed);
|
||||
}
|
||||
dump_tasks.flush()?;
|
||||
|
||||
// 3. Dump the indexes
|
||||
progress.update_progress(DumpCreationProgress::DumpTheIndexes);
|
||||
let nb_indexes = self.index_mapper.index_mapping.len(&rtxn)? as u32;
|
||||
let mut count = 0;
|
||||
self.index_mapper.try_for_each_index(&rtxn, |uid, index| -> Result<()> {
|
||||
progress.update_progress(VariableNameStep::new(
|
||||
uid.to_string(),
|
||||
count,
|
||||
nb_indexes,
|
||||
));
|
||||
count += 1;
|
||||
|
||||
let rtxn = index.read_txn()?;
|
||||
let metadata = IndexMetadata {
|
||||
uid: uid.to_owned(),
|
||||
@@ -882,12 +833,6 @@ impl IndexScheduler {
|
||||
.embedding_configs(&rtxn)
|
||||
.map_err(|e| Error::from_milli(e, Some(uid.to_string())))?;
|
||||
|
||||
let nb_documents = index
|
||||
.number_of_documents(&rtxn)
|
||||
.map_err(|e| Error::from_milli(e, Some(uid.to_string())))?
|
||||
as u32;
|
||||
let (atomic, update_document_progress) = AtomicDocumentStep::new(nb_documents);
|
||||
progress.update_progress(update_document_progress);
|
||||
let documents = index
|
||||
.all_documents(&rtxn)
|
||||
.map_err(|e| Error::from_milli(e, Some(uid.to_string())))?;
|
||||
@@ -957,7 +902,6 @@ impl IndexScheduler {
|
||||
}
|
||||
|
||||
index_dumper.push_document(&document)?;
|
||||
atomic.fetch_add(1, Ordering::Relaxed);
|
||||
}
|
||||
|
||||
// 3.2. Dump the settings
|
||||
@@ -972,7 +916,6 @@ impl IndexScheduler {
|
||||
})?;
|
||||
|
||||
// 4. Dump experimental feature settings
|
||||
progress.update_progress(DumpCreationProgress::DumpTheExperimentalFeatures);
|
||||
let features = self.features().runtime_features();
|
||||
dump.create_experimental_features(features)?;
|
||||
|
||||
@@ -983,7 +926,6 @@ impl IndexScheduler {
|
||||
if self.must_stop_processing.get() {
|
||||
return Err(Error::AbortedTask);
|
||||
}
|
||||
progress.update_progress(DumpCreationProgress::CompressTheDump);
|
||||
let path = self.dumps_path.join(format!("{}.dump", dump_uid));
|
||||
let file = File::create(path)?;
|
||||
dump.persist_to(BufWriter::new(file))?;
|
||||
@@ -1009,7 +951,7 @@ impl IndexScheduler {
|
||||
.set_currently_updating_index(Some((index_uid.clone(), index.clone())));
|
||||
|
||||
let mut index_wtxn = index.write_txn()?;
|
||||
let tasks = self.apply_index_operation(&mut index_wtxn, &index, op, progress)?;
|
||||
let tasks = self.apply_index_operation(&mut index_wtxn, &index, op)?;
|
||||
|
||||
{
|
||||
let span = tracing::trace_span!(target: "indexing::scheduler", "commit");
|
||||
@@ -1043,8 +985,6 @@ impl IndexScheduler {
|
||||
Ok(tasks)
|
||||
}
|
||||
Batch::IndexCreation { index_uid, primary_key, task } => {
|
||||
progress.update_progress(CreateIndexProgress::CreatingTheIndex);
|
||||
|
||||
let wtxn = self.env.write_txn()?;
|
||||
if self.index_mapper.exists(&wtxn, &index_uid)? {
|
||||
return Err(Error::IndexAlreadyExists(index_uid));
|
||||
@@ -1054,11 +994,9 @@ impl IndexScheduler {
|
||||
self.process_batch(
|
||||
Batch::IndexUpdate { index_uid, primary_key, task },
|
||||
current_batch,
|
||||
progress,
|
||||
)
|
||||
}
|
||||
Batch::IndexUpdate { index_uid, primary_key, mut task } => {
|
||||
progress.update_progress(UpdateIndexProgress::UpdatingTheIndex);
|
||||
let rtxn = self.env.read_txn()?;
|
||||
let index = self.index_mapper.index(&rtxn, &index_uid)?;
|
||||
|
||||
@@ -1111,7 +1049,6 @@ impl IndexScheduler {
|
||||
Ok(vec![task])
|
||||
}
|
||||
Batch::IndexDeletion { index_uid, index_has_been_created, mut tasks } => {
|
||||
progress.update_progress(DeleteIndexProgress::DeletingTheIndex);
|
||||
let wtxn = self.env.write_txn()?;
|
||||
|
||||
// it's possible that the index doesn't exist
|
||||
@@ -1145,8 +1082,6 @@ impl IndexScheduler {
|
||||
Ok(tasks)
|
||||
}
|
||||
Batch::IndexSwap { mut task } => {
|
||||
progress.update_progress(SwappingTheIndexes::EnsuringCorrectnessOfTheSwap);
|
||||
|
||||
let mut wtxn = self.env.write_txn()?;
|
||||
let swaps = if let KindWithContent::IndexSwap { swaps } = &task.kind {
|
||||
swaps
|
||||
@@ -1173,20 +1108,8 @@ impl IndexScheduler {
|
||||
));
|
||||
}
|
||||
}
|
||||
progress.update_progress(SwappingTheIndexes::SwappingTheIndexes);
|
||||
for (step, swap) in swaps.iter().enumerate() {
|
||||
progress.update_progress(VariableNameStep::new(
|
||||
format!("swapping index {} and {}", swap.indexes.0, swap.indexes.1),
|
||||
step as u32,
|
||||
swaps.len() as u32,
|
||||
));
|
||||
self.apply_index_swap(
|
||||
&mut wtxn,
|
||||
&progress,
|
||||
task.uid,
|
||||
&swap.indexes.0,
|
||||
&swap.indexes.1,
|
||||
)?;
|
||||
for swap in swaps {
|
||||
self.apply_index_swap(&mut wtxn, task.uid, &swap.indexes.0, &swap.indexes.1)?;
|
||||
}
|
||||
wtxn.commit()?;
|
||||
task.status = Status::Succeeded;
|
||||
@@ -1196,15 +1119,7 @@ impl IndexScheduler {
|
||||
}
|
||||
|
||||
/// Swap the index `lhs` with the index `rhs`.
|
||||
fn apply_index_swap(
|
||||
&self,
|
||||
wtxn: &mut RwTxn,
|
||||
progress: &Progress,
|
||||
task_id: u32,
|
||||
lhs: &str,
|
||||
rhs: &str,
|
||||
) -> Result<()> {
|
||||
progress.update_progress(InnerSwappingTwoIndexes::RetrieveTheTasks);
|
||||
fn apply_index_swap(&self, wtxn: &mut RwTxn, task_id: u32, lhs: &str, rhs: &str) -> Result<()> {
|
||||
// 1. Verify that both lhs and rhs are existing indexes
|
||||
let index_lhs_exists = self.index_mapper.index_exists(wtxn, lhs)?;
|
||||
if !index_lhs_exists {
|
||||
@@ -1222,21 +1137,14 @@ impl IndexScheduler {
|
||||
index_rhs_task_ids.remove_range(task_id..);
|
||||
|
||||
// 3. before_name -> new_name in the task's KindWithContent
|
||||
progress.update_progress(InnerSwappingTwoIndexes::UpdateTheTasks);
|
||||
let tasks_to_update = &index_lhs_task_ids | &index_rhs_task_ids;
|
||||
let (atomic, task_progress) = AtomicTaskStep::new(tasks_to_update.len() as u32);
|
||||
progress.update_progress(task_progress);
|
||||
|
||||
for task_id in tasks_to_update {
|
||||
for task_id in &index_lhs_task_ids | &index_rhs_task_ids {
|
||||
let mut task = self.get_task(wtxn, task_id)?.ok_or(Error::CorruptedTaskQueue)?;
|
||||
swap_index_uid_in_task(&mut task, (lhs, rhs));
|
||||
self.all_tasks.put(wtxn, &task_id, &task)?;
|
||||
atomic.fetch_add(1, Ordering::Relaxed);
|
||||
}
|
||||
|
||||
// 4. remove the task from indexuid = before_name
|
||||
// 5. add the task to indexuid = after_name
|
||||
progress.update_progress(InnerSwappingTwoIndexes::UpdateTheIndexesMetadata);
|
||||
self.update_index(wtxn, lhs, |lhs_tasks| {
|
||||
*lhs_tasks -= &index_lhs_task_ids;
|
||||
*lhs_tasks |= &index_rhs_task_ids;
|
||||
@@ -1258,7 +1166,7 @@ impl IndexScheduler {
|
||||
/// The list of processed tasks.
|
||||
#[tracing::instrument(
|
||||
level = "trace",
|
||||
skip(self, index_wtxn, index, progress),
|
||||
skip(self, index_wtxn, index),
|
||||
target = "indexing::scheduler"
|
||||
)]
|
||||
fn apply_index_operation<'i>(
|
||||
@@ -1266,12 +1174,44 @@ impl IndexScheduler {
|
||||
index_wtxn: &mut RwTxn<'i>,
|
||||
index: &'i Index,
|
||||
operation: IndexOperation,
|
||||
progress: Progress,
|
||||
) -> Result<Vec<Task>> {
|
||||
let indexer_alloc = Bump::new();
|
||||
|
||||
let started_processing_at = std::time::Instant::now();
|
||||
let secs_since_started_processing_at = AtomicU64::new(0);
|
||||
const PRINT_SECS_DELTA: u64 = 5;
|
||||
|
||||
let processing_tasks = self.processing_tasks.clone();
|
||||
let must_stop_processing = self.must_stop_processing.clone();
|
||||
let send_progress = |progress| {
|
||||
let now = std::time::Instant::now();
|
||||
let elapsed = secs_since_started_processing_at.load(atomic::Ordering::Relaxed);
|
||||
let previous = started_processing_at + Duration::from_secs(elapsed);
|
||||
let elapsed = now - previous;
|
||||
|
||||
if elapsed.as_secs() < PRINT_SECS_DELTA {
|
||||
return;
|
||||
}
|
||||
|
||||
secs_since_started_processing_at
|
||||
.store((now - started_processing_at).as_secs(), atomic::Ordering::Relaxed);
|
||||
|
||||
let TaskProgress {
|
||||
current_step,
|
||||
finished_steps,
|
||||
total_steps,
|
||||
finished_substeps,
|
||||
total_substeps,
|
||||
} = processing_tasks.write().unwrap().update_progress(progress);
|
||||
|
||||
tracing::info!(
|
||||
current_step,
|
||||
finished_steps,
|
||||
total_steps,
|
||||
finished_substeps,
|
||||
total_substeps
|
||||
);
|
||||
};
|
||||
|
||||
match operation {
|
||||
IndexOperation::DocumentClear { index_uid, mut tasks } => {
|
||||
@@ -1303,7 +1243,6 @@ impl IndexScheduler {
|
||||
operations,
|
||||
mut tasks,
|
||||
} => {
|
||||
progress.update_progress(DocumentOperationProgress::RetrievingConfig);
|
||||
// TODO: at some point, for better efficiency we might want to reuse the bumpalo for successive batches.
|
||||
// this is made difficult by the fact we're doing private clones of the index scheduler and sending it
|
||||
// to a fresh thread.
|
||||
@@ -1312,7 +1251,9 @@ impl IndexScheduler {
|
||||
if let DocumentOperation::Add(content_uuid) = operation {
|
||||
let content_file = self.file_store.get_update(*content_uuid)?;
|
||||
let mmap = unsafe { memmap2::Mmap::map(&content_file)? };
|
||||
content_files.push(mmap);
|
||||
if !mmap.is_empty() {
|
||||
content_files.push(mmap);
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
@@ -1357,7 +1298,6 @@ impl IndexScheduler {
|
||||
}
|
||||
};
|
||||
|
||||
progress.update_progress(DocumentOperationProgress::ComputingDocumentChanges);
|
||||
let (document_changes, operation_stats, primary_key) = indexer
|
||||
.into_changes(
|
||||
&indexer_alloc,
|
||||
@@ -1366,13 +1306,13 @@ impl IndexScheduler {
|
||||
primary_key.as_deref(),
|
||||
&mut new_fields_ids_map,
|
||||
&|| must_stop_processing.get(),
|
||||
progress.clone(),
|
||||
&send_progress,
|
||||
)
|
||||
.map_err(|e| Error::from_milli(e, Some(index_uid.clone())))?;
|
||||
|
||||
let mut candidates_count = 0;
|
||||
let mut addition = 0;
|
||||
for (stats, task) in operation_stats.into_iter().zip(&mut tasks) {
|
||||
candidates_count += stats.document_count;
|
||||
addition += stats.document_count;
|
||||
match stats.error {
|
||||
Some(error) => {
|
||||
task.status = Status::Failed;
|
||||
@@ -1402,7 +1342,6 @@ impl IndexScheduler {
|
||||
}
|
||||
}
|
||||
|
||||
progress.update_progress(DocumentOperationProgress::Indexing);
|
||||
if tasks.iter().any(|res| res.error.is_none()) {
|
||||
indexer::index(
|
||||
index_wtxn,
|
||||
@@ -1415,25 +1354,16 @@ impl IndexScheduler {
|
||||
&document_changes,
|
||||
embedders,
|
||||
&|| must_stop_processing.get(),
|
||||
&progress,
|
||||
&send_progress,
|
||||
)
|
||||
.map_err(|e| Error::from_milli(e, Some(index_uid.clone())))?;
|
||||
|
||||
let addition = DocumentAdditionResult {
|
||||
indexed_documents: candidates_count,
|
||||
number_of_documents: index
|
||||
.number_of_documents(index_wtxn)
|
||||
.map_err(|err| Error::from_milli(err, Some(index_uid.clone())))?,
|
||||
};
|
||||
|
||||
tracing::info!(indexing_result = ?addition, processed_in = ?started_processing_at.elapsed(), "document indexing done");
|
||||
}
|
||||
|
||||
Ok(tasks)
|
||||
}
|
||||
IndexOperation::DocumentEdition { index_uid, mut task } => {
|
||||
progress.update_progress(DocumentEditionProgress::RetrievingConfig);
|
||||
|
||||
let (filter, code) = if let KindWithContent::DocumentEdition {
|
||||
filter_expr,
|
||||
context: _,
|
||||
@@ -1506,8 +1436,6 @@ impl IndexScheduler {
|
||||
}
|
||||
};
|
||||
|
||||
let candidates_count = candidates.len();
|
||||
progress.update_progress(DocumentEditionProgress::ComputingDocumentChanges);
|
||||
let indexer = UpdateByFunction::new(candidates, context.clone(), code.clone());
|
||||
let document_changes = pool
|
||||
.install(|| {
|
||||
@@ -1521,7 +1449,6 @@ impl IndexScheduler {
|
||||
.map_err(|err| Error::from_milli(err, Some(index_uid.clone())))?;
|
||||
let embedders = self.embedders(index_uid.clone(), embedders)?;
|
||||
|
||||
progress.update_progress(DocumentEditionProgress::Indexing);
|
||||
indexer::index(
|
||||
index_wtxn,
|
||||
index,
|
||||
@@ -1533,18 +1460,11 @@ impl IndexScheduler {
|
||||
&document_changes,
|
||||
embedders,
|
||||
&|| must_stop_processing.get(),
|
||||
&progress,
|
||||
&send_progress,
|
||||
)
|
||||
.map_err(|err| Error::from_milli(err, Some(index_uid.clone())))?;
|
||||
|
||||
let addition = DocumentAdditionResult {
|
||||
indexed_documents: candidates_count,
|
||||
number_of_documents: index
|
||||
.number_of_documents(index_wtxn)
|
||||
.map_err(|err| Error::from_milli(err, Some(index_uid.clone())))?,
|
||||
};
|
||||
|
||||
tracing::info!(indexing_result = ?addition, processed_in = ?started_processing_at.elapsed(), "document indexing done");
|
||||
// tracing::info!(indexing_result = ?addition, processed_in = ?started_processing_at.elapsed(), "document indexing done");
|
||||
}
|
||||
|
||||
match result_count {
|
||||
@@ -1574,8 +1494,6 @@ impl IndexScheduler {
|
||||
Ok(vec![task])
|
||||
}
|
||||
IndexOperation::DocumentDeletion { mut tasks, index_uid } => {
|
||||
progress.update_progress(DocumentDeletionProgress::RetrievingConfig);
|
||||
|
||||
let mut to_delete = RoaringBitmap::new();
|
||||
let external_documents_ids = index.external_documents_ids();
|
||||
|
||||
@@ -1666,9 +1584,7 @@ impl IndexScheduler {
|
||||
}
|
||||
};
|
||||
|
||||
progress.update_progress(DocumentDeletionProgress::DeleteDocuments);
|
||||
let mut indexer = indexer::DocumentDeletion::new();
|
||||
let candidates_count = to_delete.len();
|
||||
indexer.delete_documents_by_docids(to_delete);
|
||||
let document_changes = indexer.into_changes(&indexer_alloc, primary_key);
|
||||
let embedders = index
|
||||
@@ -1676,7 +1592,6 @@ impl IndexScheduler {
|
||||
.map_err(|err| Error::from_milli(err, Some(index_uid.clone())))?;
|
||||
let embedders = self.embedders(index_uid.clone(), embedders)?;
|
||||
|
||||
progress.update_progress(DocumentDeletionProgress::Indexing);
|
||||
indexer::index(
|
||||
index_wtxn,
|
||||
index,
|
||||
@@ -1688,24 +1603,16 @@ impl IndexScheduler {
|
||||
&document_changes,
|
||||
embedders,
|
||||
&|| must_stop_processing.get(),
|
||||
&progress,
|
||||
&send_progress,
|
||||
)
|
||||
.map_err(|err| Error::from_milli(err, Some(index_uid.clone())))?;
|
||||
|
||||
let addition = DocumentAdditionResult {
|
||||
indexed_documents: candidates_count,
|
||||
number_of_documents: index
|
||||
.number_of_documents(index_wtxn)
|
||||
.map_err(|err| Error::from_milli(err, Some(index_uid.clone())))?,
|
||||
};
|
||||
|
||||
tracing::info!(indexing_result = ?addition, processed_in = ?started_processing_at.elapsed(), "document indexing done");
|
||||
// tracing::info!(indexing_result = ?addition, processed_in = ?started_processing_at.elapsed(), "document indexing done");
|
||||
}
|
||||
|
||||
Ok(tasks)
|
||||
}
|
||||
IndexOperation::Settings { index_uid, settings, mut tasks } => {
|
||||
progress.update_progress(SettingsProgress::RetrievingAndMergingTheSettings);
|
||||
let indexer_config = self.index_mapper.indexer_config();
|
||||
let mut builder = milli::update::Settings::new(index_wtxn, index, indexer_config);
|
||||
|
||||
@@ -1719,7 +1626,6 @@ impl IndexScheduler {
|
||||
task.status = Status::Succeeded;
|
||||
}
|
||||
|
||||
progress.update_progress(SettingsProgress::ApplyTheSettings);
|
||||
builder
|
||||
.execute(
|
||||
|indexing_step| tracing::debug!(update = ?indexing_step),
|
||||
@@ -1742,14 +1648,12 @@ impl IndexScheduler {
|
||||
index_uid: index_uid.clone(),
|
||||
tasks: cleared_tasks,
|
||||
},
|
||||
progress.clone(),
|
||||
)?;
|
||||
|
||||
let settings_tasks = self.apply_index_operation(
|
||||
index_wtxn,
|
||||
index,
|
||||
IndexOperation::Settings { index_uid, settings, tasks: settings_tasks },
|
||||
progress,
|
||||
)?;
|
||||
|
||||
let mut tasks = settings_tasks;
|
||||
@@ -1766,18 +1670,15 @@ impl IndexScheduler {
|
||||
&self,
|
||||
wtxn: &mut RwTxn,
|
||||
matched_tasks: &RoaringBitmap,
|
||||
progress: &Progress,
|
||||
) -> Result<RoaringBitmap> {
|
||||
progress.update_progress(TaskDeletionProgress::DeletingTasksDateTime);
|
||||
|
||||
// 1. Remove from this list the tasks that we are not allowed to delete
|
||||
let enqueued_tasks = self.get_status(wtxn, Status::Enqueued)?;
|
||||
let processing_tasks = &self.processing_tasks.read().unwrap().processing.clone();
|
||||
|
||||
let all_task_ids = self.all_task_ids(wtxn)?;
|
||||
let mut to_delete_tasks = all_task_ids & matched_tasks;
|
||||
to_delete_tasks -= &**processing_tasks;
|
||||
to_delete_tasks -= &enqueued_tasks;
|
||||
to_delete_tasks -= processing_tasks;
|
||||
to_delete_tasks -= enqueued_tasks;
|
||||
|
||||
// 2. We now have a list of tasks to delete, delete them
|
||||
|
||||
@@ -1788,8 +1689,6 @@ impl IndexScheduler {
|
||||
// The tasks that have been removed *per batches*.
|
||||
let mut affected_batches: HashMap<BatchId, RoaringBitmap> = HashMap::new();
|
||||
|
||||
let (atomic_progress, task_progress) = AtomicTaskStep::new(to_delete_tasks.len() as u32);
|
||||
progress.update_progress(task_progress);
|
||||
for task_id in to_delete_tasks.iter() {
|
||||
let task = self.get_task(wtxn, task_id)?.ok_or(Error::CorruptedTaskQueue)?;
|
||||
|
||||
@@ -1813,35 +1712,22 @@ impl IndexScheduler {
|
||||
if let Some(batch_uid) = task.batch_uid {
|
||||
affected_batches.entry(batch_uid).or_default().insert(task_id);
|
||||
}
|
||||
atomic_progress.fetch_add(1, Ordering::Relaxed);
|
||||
}
|
||||
|
||||
progress.update_progress(TaskDeletionProgress::DeletingTasksMetadata);
|
||||
let (atomic_progress, task_progress) = AtomicTaskStep::new(
|
||||
(affected_indexes.len() + affected_statuses.len() + affected_kinds.len()) as u32,
|
||||
);
|
||||
progress.update_progress(task_progress);
|
||||
for index in affected_indexes.iter() {
|
||||
self.update_index(wtxn, index, |bitmap| *bitmap -= &to_delete_tasks)?;
|
||||
atomic_progress.fetch_add(1, Ordering::Relaxed);
|
||||
}
|
||||
|
||||
for status in affected_statuses.iter() {
|
||||
self.update_status(wtxn, *status, |bitmap| *bitmap -= &to_delete_tasks)?;
|
||||
atomic_progress.fetch_add(1, Ordering::Relaxed);
|
||||
}
|
||||
|
||||
for kind in affected_kinds.iter() {
|
||||
self.update_kind(wtxn, *kind, |bitmap| *bitmap -= &to_delete_tasks)?;
|
||||
atomic_progress.fetch_add(1, Ordering::Relaxed);
|
||||
}
|
||||
|
||||
progress.update_progress(TaskDeletionProgress::DeletingTasks);
|
||||
let (atomic_progress, task_progress) = AtomicTaskStep::new(to_delete_tasks.len() as u32);
|
||||
progress.update_progress(task_progress);
|
||||
for task in to_delete_tasks.iter() {
|
||||
self.all_tasks.delete(wtxn, &task)?;
|
||||
atomic_progress.fetch_add(1, Ordering::Relaxed);
|
||||
}
|
||||
for canceled_by in affected_canceled_by {
|
||||
if let Some(mut tasks) = self.canceled_by.get(wtxn, &canceled_by)? {
|
||||
@@ -1853,9 +1739,6 @@ impl IndexScheduler {
|
||||
}
|
||||
}
|
||||
}
|
||||
progress.update_progress(TaskDeletionProgress::DeletingBatches);
|
||||
let (atomic_progress, batch_progress) = AtomicBatchStep::new(affected_batches.len() as u32);
|
||||
progress.update_progress(batch_progress);
|
||||
for (batch_id, to_delete_tasks) in affected_batches {
|
||||
if let Some(mut tasks) = self.batch_to_tasks_mapping.get(wtxn, &batch_id)? {
|
||||
tasks -= &to_delete_tasks;
|
||||
@@ -1897,7 +1780,6 @@ impl IndexScheduler {
|
||||
}
|
||||
}
|
||||
}
|
||||
atomic_progress.fetch_add(1, Ordering::Relaxed);
|
||||
}
|
||||
|
||||
Ok(to_delete_tasks)
|
||||
@@ -1912,36 +1794,21 @@ impl IndexScheduler {
|
||||
cancel_task_id: TaskId,
|
||||
current_batch: &mut ProcessingBatch,
|
||||
matched_tasks: &RoaringBitmap,
|
||||
progress: &Progress,
|
||||
) -> Result<Vec<Task>> {
|
||||
progress.update_progress(TaskCancelationProgress::RetrievingTasks);
|
||||
|
||||
// 1. Remove from this list the tasks that we are not allowed to cancel
|
||||
// Notice that only the _enqueued_ ones are cancelable and we should
|
||||
// have already aborted the indexation of the _processing_ ones
|
||||
let cancelable_tasks = self.get_status(rtxn, Status::Enqueued)?;
|
||||
let tasks_to_cancel = cancelable_tasks & matched_tasks;
|
||||
|
||||
let (task_progress, progress_obj) = AtomicTaskStep::new(tasks_to_cancel.len() as u32);
|
||||
progress.update_progress(progress_obj);
|
||||
|
||||
// 2. We now have a list of tasks to cancel, cancel them
|
||||
let mut tasks = self.get_existing_tasks(
|
||||
rtxn,
|
||||
tasks_to_cancel.iter().inspect(|_| {
|
||||
task_progress.fetch_add(1, Ordering::Relaxed);
|
||||
}),
|
||||
)?;
|
||||
let mut tasks = self.get_existing_tasks(rtxn, tasks_to_cancel.iter())?;
|
||||
|
||||
progress.update_progress(TaskCancelationProgress::UpdatingTasks);
|
||||
let (task_progress, progress_obj) = AtomicTaskStep::new(tasks_to_cancel.len() as u32);
|
||||
progress.update_progress(progress_obj);
|
||||
for task in tasks.iter_mut() {
|
||||
task.status = Status::Canceled;
|
||||
task.canceled_by = Some(cancel_task_id);
|
||||
task.details = task.details.as_ref().map(|d| d.to_failed());
|
||||
current_batch.processing(Some(task));
|
||||
task_progress.fetch_add(1, Ordering::Relaxed);
|
||||
}
|
||||
|
||||
Ok(tasks)
|
||||
|
||||
@@ -1,13 +1,12 @@
|
||||
use std::fmt::Display;
|
||||
|
||||
use crate::TaskId;
|
||||
use meilisearch_types::batches::BatchId;
|
||||
use meilisearch_types::error::{Code, ErrorCode};
|
||||
use meilisearch_types::tasks::{Kind, Status};
|
||||
use meilisearch_types::{heed, milli};
|
||||
use thiserror::Error;
|
||||
|
||||
use crate::TaskId;
|
||||
|
||||
#[derive(Copy, Clone, Debug, PartialEq, Eq)]
|
||||
pub enum DateField {
|
||||
BeforeEnqueuedAt,
|
||||
@@ -104,7 +103,7 @@ pub enum Error {
|
||||
)]
|
||||
InvalidTaskCanceledBy { canceled_by: String },
|
||||
#[error(
|
||||
"{index_uid} is not a valid index uid. Index uid can be an integer or a string containing only alphanumeric characters, hyphens (-) and underscores (_), and can not be more than 400 bytes."
|
||||
"{index_uid} is not a valid index uid. Index uid can be an integer or a string containing only alphanumeric characters, hyphens (-) and underscores (_), and can not be more than 512 bytes."
|
||||
)]
|
||||
InvalidIndexUid { index_uid: String },
|
||||
#[error("Task `{0}` not found.")]
|
||||
|
||||
@@ -3,6 +3,10 @@ use std::sync::{Arc, RwLock};
|
||||
use std::time::Duration;
|
||||
use std::{fs, thread};
|
||||
|
||||
use self::index_map::IndexMap;
|
||||
use self::IndexStatus::{Available, BeingDeleted, Closing, Missing};
|
||||
use crate::uuid_codec::UuidCodec;
|
||||
use crate::{Error, Result};
|
||||
use meilisearch_types::heed::types::{SerdeJson, Str};
|
||||
use meilisearch_types::heed::{Database, Env, RoTxn, RwTxn};
|
||||
use meilisearch_types::milli;
|
||||
@@ -13,11 +17,6 @@ use time::OffsetDateTime;
|
||||
use tracing::error;
|
||||
use uuid::Uuid;
|
||||
|
||||
use self::index_map::IndexMap;
|
||||
use self::IndexStatus::{Available, BeingDeleted, Closing, Missing};
|
||||
use crate::uuid_codec::UuidCodec;
|
||||
use crate::{Error, Result};
|
||||
|
||||
mod index_map;
|
||||
|
||||
const INDEX_MAPPING: &str = "index-mapping";
|
||||
|
||||
@@ -353,7 +353,7 @@ pub fn snapshot_canceled_by(rtxn: &RoTxn, db: Database<BEU32, RoaringBitmapCodec
|
||||
|
||||
pub fn snapshot_batch(batch: &Batch) -> String {
|
||||
let mut snap = String::new();
|
||||
let Batch { uid, details, stats, started_at, finished_at, progress: _ } = batch;
|
||||
let Batch { uid, details, stats, started_at, finished_at } = batch;
|
||||
if let Some(finished_at) = finished_at {
|
||||
assert!(finished_at > started_at);
|
||||
}
|
||||
|
||||
@@ -26,7 +26,6 @@ mod index_mapper;
|
||||
#[cfg(test)]
|
||||
mod insta_snapshot;
|
||||
mod lru;
|
||||
mod processing;
|
||||
mod utils;
|
||||
pub mod uuid_codec;
|
||||
|
||||
@@ -57,12 +56,12 @@ use meilisearch_types::heed::types::{SerdeBincode, SerdeJson, Str, I128};
|
||||
use meilisearch_types::heed::{self, Database, Env, PutFlags, RoTxn, RwTxn};
|
||||
use meilisearch_types::milli::documents::DocumentsBatchBuilder;
|
||||
use meilisearch_types::milli::index::IndexEmbeddingConfig;
|
||||
use meilisearch_types::milli::update::new::indexer::document_changes::Progress;
|
||||
use meilisearch_types::milli::update::IndexerConfig;
|
||||
use meilisearch_types::milli::vector::{Embedder, EmbedderOptions, EmbeddingConfigs};
|
||||
use meilisearch_types::milli::{self, CboRoaringBitmapCodec, Index, RoaringBitmapCodec, BEU32};
|
||||
use meilisearch_types::task_view::TaskView;
|
||||
use meilisearch_types::tasks::{Kind, KindWithContent, Status, Task};
|
||||
use processing::ProcessingTasks;
|
||||
use meilisearch_types::tasks::{Kind, KindWithContent, Status, Task, TaskProgress};
|
||||
use rayon::current_num_threads;
|
||||
use rayon::prelude::{IntoParallelIterator, ParallelIterator};
|
||||
use roaring::RoaringBitmap;
|
||||
@@ -73,8 +72,7 @@ use utils::{filter_out_references_to_newer_tasks, keep_ids_within_datetimes, map
|
||||
use uuid::Uuid;
|
||||
|
||||
use crate::index_mapper::IndexMapper;
|
||||
use crate::processing::{AtomicTaskStep, BatchProgress};
|
||||
use crate::utils::{check_index_swap_validity, clamp_to_page_size};
|
||||
use crate::utils::{check_index_swap_validity, clamp_to_page_size, ProcessingBatch};
|
||||
|
||||
pub(crate) type BEI128 = I128<BE>;
|
||||
|
||||
@@ -165,6 +163,48 @@ impl Query {
|
||||
}
|
||||
}
|
||||
|
||||
#[derive(Debug, Clone)]
|
||||
pub struct ProcessingTasks {
|
||||
batch: Option<ProcessingBatch>,
|
||||
/// The list of tasks ids that are currently running.
|
||||
processing: RoaringBitmap,
|
||||
/// The progress on processing tasks
|
||||
progress: Option<TaskProgress>,
|
||||
}
|
||||
|
||||
impl ProcessingTasks {
|
||||
/// Creates an empty `ProcessingAt` struct.
|
||||
fn new() -> ProcessingTasks {
|
||||
ProcessingTasks { batch: None, processing: RoaringBitmap::new(), progress: None }
|
||||
}
|
||||
|
||||
/// Stores the currently processing tasks, and the date time at which it started.
|
||||
fn start_processing(&mut self, processing_batch: ProcessingBatch, processing: RoaringBitmap) {
|
||||
self.batch = Some(processing_batch);
|
||||
self.processing = processing;
|
||||
}
|
||||
|
||||
fn update_progress(&mut self, progress: Progress) -> TaskProgress {
|
||||
self.progress.get_or_insert_with(TaskProgress::default).update(progress)
|
||||
}
|
||||
|
||||
/// Set the processing tasks to an empty list
|
||||
fn stop_processing(&mut self) -> Self {
|
||||
self.progress = None;
|
||||
|
||||
Self {
|
||||
batch: std::mem::take(&mut self.batch),
|
||||
processing: std::mem::take(&mut self.processing),
|
||||
progress: None,
|
||||
}
|
||||
}
|
||||
|
||||
/// Returns `true` if there, at least, is one task that is currently processing that we must stop.
|
||||
fn must_cancel_processing_tasks(&self, canceled_tasks: &RoaringBitmap) -> bool {
|
||||
!self.processing.is_disjoint(canceled_tasks)
|
||||
}
|
||||
}
|
||||
|
||||
#[derive(Default, Clone, Debug)]
|
||||
struct MustStopProcessing(Arc<AtomicBool>);
|
||||
|
||||
@@ -773,7 +813,7 @@ impl IndexScheduler {
|
||||
let mut batch_tasks = RoaringBitmap::new();
|
||||
for batch_uid in batch_uids {
|
||||
if processing_batch.as_ref().map_or(false, |batch| batch.uid == *batch_uid) {
|
||||
batch_tasks |= &*processing_tasks;
|
||||
batch_tasks |= &processing_tasks;
|
||||
} else {
|
||||
batch_tasks |= self.tasks_in_batch(rtxn, *batch_uid)?;
|
||||
}
|
||||
@@ -787,13 +827,13 @@ impl IndexScheduler {
|
||||
match status {
|
||||
// special case for Processing tasks
|
||||
Status::Processing => {
|
||||
status_tasks |= &*processing_tasks;
|
||||
status_tasks |= &processing_tasks;
|
||||
}
|
||||
status => status_tasks |= &self.get_status(rtxn, *status)?,
|
||||
};
|
||||
}
|
||||
if !status.contains(&Status::Processing) {
|
||||
tasks -= &*processing_tasks;
|
||||
tasks -= &processing_tasks;
|
||||
}
|
||||
tasks &= status_tasks;
|
||||
}
|
||||
@@ -842,7 +882,7 @@ impl IndexScheduler {
|
||||
// Once we have filtered the two subsets, we put them back together and assign it back to `tasks`.
|
||||
tasks = {
|
||||
let (mut filtered_non_processing_tasks, mut filtered_processing_tasks) =
|
||||
(&tasks - &*processing_tasks, &tasks & &*processing_tasks);
|
||||
(&tasks - &processing_tasks, &tasks & &processing_tasks);
|
||||
|
||||
// special case for Processing tasks
|
||||
// A closure that clears the filtered_processing_tasks if their started_at date falls outside the given bounds
|
||||
@@ -1050,7 +1090,7 @@ impl IndexScheduler {
|
||||
// Once we have filtered the two subsets, we put them back together and assign it back to `batches`.
|
||||
batches = {
|
||||
let (mut filtered_non_processing_batches, mut filtered_processing_batches) =
|
||||
(&batches - &*processing.processing, &batches & &*processing.processing);
|
||||
(&batches - &processing.processing, &batches & &processing.processing);
|
||||
|
||||
// special case for Processing batches
|
||||
// A closure that clears the filtered_processing_batches if their started_at date falls outside the given bounds
|
||||
@@ -1566,8 +1606,7 @@ impl IndexScheduler {
|
||||
|
||||
// We reset the must_stop flag to be sure that we don't stop processing tasks
|
||||
self.must_stop_processing.reset();
|
||||
let progress = self
|
||||
.processing_tasks
|
||||
self.processing_tasks
|
||||
.write()
|
||||
.unwrap()
|
||||
// We can clone the processing batch here because we don't want its modification to affect the view of the processing batches
|
||||
@@ -1580,12 +1619,11 @@ impl IndexScheduler {
|
||||
let res = {
|
||||
let cloned_index_scheduler = self.private_clone();
|
||||
let processing_batch = &mut processing_batch;
|
||||
let progress = progress.clone();
|
||||
std::thread::scope(|s| {
|
||||
let handle = std::thread::Builder::new()
|
||||
.name(String::from("batch-operation"))
|
||||
.spawn_scoped(s, move || {
|
||||
cloned_index_scheduler.process_batch(batch, processing_batch, progress)
|
||||
cloned_index_scheduler.process_batch(batch, processing_batch)
|
||||
})
|
||||
.unwrap();
|
||||
handle.join().unwrap_or(Err(Error::ProcessBatchPanicked))
|
||||
@@ -1598,7 +1636,6 @@ impl IndexScheduler {
|
||||
#[cfg(test)]
|
||||
self.maybe_fail(tests::FailureLocation::AcquiringWtxn)?;
|
||||
|
||||
progress.update_progress(BatchProgress::WritingTasksToDisk);
|
||||
processing_batch.finished();
|
||||
let mut wtxn = self.env.write_txn().map_err(Error::HeedTransaction)?;
|
||||
let mut canceled = RoaringBitmap::new();
|
||||
@@ -1608,15 +1645,12 @@ impl IndexScheduler {
|
||||
#[cfg(test)]
|
||||
self.breakpoint(Breakpoint::ProcessBatchSucceeded);
|
||||
|
||||
let (task_progress, task_progress_obj) = AtomicTaskStep::new(tasks.len() as u32);
|
||||
progress.update_progress(task_progress_obj);
|
||||
let mut success = 0;
|
||||
let mut failure = 0;
|
||||
let mut canceled_by = None;
|
||||
|
||||
#[allow(unused_variables)]
|
||||
for (i, mut task) in tasks.into_iter().enumerate() {
|
||||
task_progress.fetch_add(1, Ordering::Relaxed);
|
||||
processing_batch.update(&mut task);
|
||||
if task.status == Status::Canceled {
|
||||
canceled.insert(task.uid);
|
||||
@@ -1684,12 +1718,8 @@ impl IndexScheduler {
|
||||
Err(err) => {
|
||||
#[cfg(test)]
|
||||
self.breakpoint(Breakpoint::ProcessBatchFailed);
|
||||
let (task_progress, task_progress_obj) = AtomicTaskStep::new(ids.len() as u32);
|
||||
progress.update_progress(task_progress_obj);
|
||||
|
||||
let error: ResponseError = err.into();
|
||||
for id in ids.iter() {
|
||||
task_progress.fetch_add(1, Ordering::Relaxed);
|
||||
let mut task = self
|
||||
.get_task(&wtxn, id)
|
||||
.map_err(|e| Error::TaskDatabaseUpdate(Box::new(e)))?
|
||||
|
||||
@@ -1,316 +0,0 @@
|
||||
use std::borrow::Cow;
|
||||
use std::sync::Arc;
|
||||
|
||||
use enum_iterator::Sequence;
|
||||
use meilisearch_types::milli::progress::{AtomicSubStep, NamedStep, Progress, ProgressView, Step};
|
||||
use meilisearch_types::milli::{make_atomic_progress, make_enum_progress};
|
||||
use roaring::RoaringBitmap;
|
||||
|
||||
use crate::utils::ProcessingBatch;
|
||||
|
||||
#[derive(Clone)]
|
||||
pub struct ProcessingTasks {
|
||||
pub batch: Option<Arc<ProcessingBatch>>,
|
||||
/// The list of tasks ids that are currently running.
|
||||
pub processing: Arc<RoaringBitmap>,
|
||||
/// The progress on processing tasks
|
||||
pub progress: Option<Progress>,
|
||||
}
|
||||
|
||||
impl ProcessingTasks {
|
||||
/// Creates an empty `ProcessingAt` struct.
|
||||
pub fn new() -> ProcessingTasks {
|
||||
ProcessingTasks { batch: None, processing: Arc::new(RoaringBitmap::new()), progress: None }
|
||||
}
|
||||
|
||||
pub fn get_progress_view(&self) -> Option<ProgressView> {
|
||||
Some(self.progress.as_ref()?.as_progress_view())
|
||||
}
|
||||
|
||||
/// Stores the currently processing tasks, and the date time at which it started.
|
||||
pub fn start_processing(
|
||||
&mut self,
|
||||
processing_batch: ProcessingBatch,
|
||||
processing: RoaringBitmap,
|
||||
) -> Progress {
|
||||
self.batch = Some(Arc::new(processing_batch));
|
||||
self.processing = Arc::new(processing);
|
||||
let progress = Progress::default();
|
||||
progress.update_progress(BatchProgress::ProcessingTasks);
|
||||
self.progress = Some(progress.clone());
|
||||
|
||||
progress
|
||||
}
|
||||
|
||||
/// Set the processing tasks to an empty list
|
||||
pub fn stop_processing(&mut self) -> Self {
|
||||
self.progress = None;
|
||||
|
||||
Self {
|
||||
batch: std::mem::take(&mut self.batch),
|
||||
processing: std::mem::take(&mut self.processing),
|
||||
progress: None,
|
||||
}
|
||||
}
|
||||
|
||||
/// Returns `true` if there, at least, is one task that is currently processing that we must stop.
|
||||
pub fn must_cancel_processing_tasks(&self, canceled_tasks: &RoaringBitmap) -> bool {
|
||||
!self.processing.is_disjoint(canceled_tasks)
|
||||
}
|
||||
}
|
||||
|
||||
make_enum_progress! {
|
||||
pub enum BatchProgress {
|
||||
ProcessingTasks,
|
||||
WritingTasksToDisk,
|
||||
}
|
||||
}
|
||||
|
||||
make_enum_progress! {
|
||||
pub enum TaskCancelationProgress {
|
||||
RetrievingTasks,
|
||||
UpdatingTasks,
|
||||
}
|
||||
}
|
||||
|
||||
make_enum_progress! {
|
||||
pub enum TaskDeletionProgress {
|
||||
DeletingTasksDateTime,
|
||||
DeletingTasksMetadata,
|
||||
DeletingTasks,
|
||||
DeletingBatches,
|
||||
}
|
||||
}
|
||||
|
||||
make_enum_progress! {
|
||||
pub enum SnapshotCreationProgress {
|
||||
StartTheSnapshotCreation,
|
||||
SnapshotTheIndexScheduler,
|
||||
SnapshotTheUpdateFiles,
|
||||
SnapshotTheIndexes,
|
||||
SnapshotTheApiKeys,
|
||||
CreateTheTarball,
|
||||
}
|
||||
}
|
||||
|
||||
make_enum_progress! {
|
||||
pub enum DumpCreationProgress {
|
||||
StartTheDumpCreation,
|
||||
DumpTheApiKeys,
|
||||
DumpTheTasks,
|
||||
DumpTheIndexes,
|
||||
DumpTheExperimentalFeatures,
|
||||
CompressTheDump,
|
||||
}
|
||||
}
|
||||
|
||||
make_enum_progress! {
|
||||
pub enum CreateIndexProgress {
|
||||
CreatingTheIndex,
|
||||
}
|
||||
}
|
||||
|
||||
make_enum_progress! {
|
||||
pub enum UpdateIndexProgress {
|
||||
UpdatingTheIndex,
|
||||
}
|
||||
}
|
||||
|
||||
make_enum_progress! {
|
||||
pub enum DeleteIndexProgress {
|
||||
DeletingTheIndex,
|
||||
}
|
||||
}
|
||||
|
||||
make_enum_progress! {
|
||||
pub enum SwappingTheIndexes {
|
||||
EnsuringCorrectnessOfTheSwap,
|
||||
SwappingTheIndexes,
|
||||
}
|
||||
}
|
||||
|
||||
make_enum_progress! {
|
||||
pub enum InnerSwappingTwoIndexes {
|
||||
RetrieveTheTasks,
|
||||
UpdateTheTasks,
|
||||
UpdateTheIndexesMetadata,
|
||||
}
|
||||
}
|
||||
|
||||
make_enum_progress! {
|
||||
pub enum DocumentOperationProgress {
|
||||
RetrievingConfig,
|
||||
ComputingDocumentChanges,
|
||||
Indexing,
|
||||
}
|
||||
}
|
||||
|
||||
make_enum_progress! {
|
||||
pub enum DocumentEditionProgress {
|
||||
RetrievingConfig,
|
||||
ComputingDocumentChanges,
|
||||
Indexing,
|
||||
}
|
||||
}
|
||||
|
||||
make_enum_progress! {
|
||||
pub enum DocumentDeletionProgress {
|
||||
RetrievingConfig,
|
||||
DeleteDocuments,
|
||||
Indexing,
|
||||
}
|
||||
}
|
||||
|
||||
make_enum_progress! {
|
||||
pub enum SettingsProgress {
|
||||
RetrievingAndMergingTheSettings,
|
||||
ApplyTheSettings,
|
||||
}
|
||||
}
|
||||
|
||||
make_atomic_progress!(Task alias AtomicTaskStep => "task" );
|
||||
make_atomic_progress!(Document alias AtomicDocumentStep => "document" );
|
||||
make_atomic_progress!(Batch alias AtomicBatchStep => "batch" );
|
||||
make_atomic_progress!(UpdateFile alias AtomicUpdateFileStep => "update file" );
|
||||
|
||||
pub struct VariableNameStep {
|
||||
name: String,
|
||||
current: u32,
|
||||
total: u32,
|
||||
}
|
||||
|
||||
impl VariableNameStep {
|
||||
pub fn new(name: impl Into<String>, current: u32, total: u32) -> Self {
|
||||
Self { name: name.into(), current, total }
|
||||
}
|
||||
}
|
||||
|
||||
impl Step for VariableNameStep {
|
||||
fn name(&self) -> Cow<'static, str> {
|
||||
self.name.clone().into()
|
||||
}
|
||||
|
||||
fn current(&self) -> u32 {
|
||||
self.current
|
||||
}
|
||||
|
||||
fn total(&self) -> u32 {
|
||||
self.total
|
||||
}
|
||||
}
|
||||
|
||||
#[cfg(test)]
|
||||
mod test {
|
||||
use std::sync::atomic::Ordering;
|
||||
|
||||
use meili_snap::{json_string, snapshot};
|
||||
|
||||
use super::*;
|
||||
|
||||
#[test]
|
||||
fn one_level() {
|
||||
let mut processing = ProcessingTasks::new();
|
||||
processing.start_processing(ProcessingBatch::new(0), RoaringBitmap::new());
|
||||
snapshot!(json_string!(processing.get_progress_view()), @r#"
|
||||
{
|
||||
"steps": [
|
||||
{
|
||||
"currentStep": "processing tasks",
|
||||
"finished": 0,
|
||||
"total": 2
|
||||
}
|
||||
],
|
||||
"percentage": 0.0
|
||||
}
|
||||
"#);
|
||||
processing.progress.as_ref().unwrap().update_progress(BatchProgress::WritingTasksToDisk);
|
||||
snapshot!(json_string!(processing.get_progress_view()), @r#"
|
||||
{
|
||||
"steps": [
|
||||
{
|
||||
"currentStep": "writing tasks to disk",
|
||||
"finished": 1,
|
||||
"total": 2
|
||||
}
|
||||
],
|
||||
"percentage": 50.0
|
||||
}
|
||||
"#);
|
||||
}
|
||||
|
||||
#[test]
|
||||
fn task_progress() {
|
||||
let mut processing = ProcessingTasks::new();
|
||||
processing.start_processing(ProcessingBatch::new(0), RoaringBitmap::new());
|
||||
let (atomic, tasks) = AtomicTaskStep::new(10);
|
||||
processing.progress.as_ref().unwrap().update_progress(tasks);
|
||||
snapshot!(json_string!(processing.get_progress_view()), @r#"
|
||||
{
|
||||
"steps": [
|
||||
{
|
||||
"currentStep": "processing tasks",
|
||||
"finished": 0,
|
||||
"total": 2
|
||||
},
|
||||
{
|
||||
"currentStep": "task",
|
||||
"finished": 0,
|
||||
"total": 10
|
||||
}
|
||||
],
|
||||
"percentage": 0.0
|
||||
}
|
||||
"#);
|
||||
atomic.fetch_add(6, Ordering::Relaxed);
|
||||
snapshot!(json_string!(processing.get_progress_view()), @r#"
|
||||
{
|
||||
"steps": [
|
||||
{
|
||||
"currentStep": "processing tasks",
|
||||
"finished": 0,
|
||||
"total": 2
|
||||
},
|
||||
{
|
||||
"currentStep": "task",
|
||||
"finished": 6,
|
||||
"total": 10
|
||||
}
|
||||
],
|
||||
"percentage": 30.000002
|
||||
}
|
||||
"#);
|
||||
processing.progress.as_ref().unwrap().update_progress(BatchProgress::WritingTasksToDisk);
|
||||
snapshot!(json_string!(processing.get_progress_view()), @r#"
|
||||
{
|
||||
"steps": [
|
||||
{
|
||||
"currentStep": "writing tasks to disk",
|
||||
"finished": 1,
|
||||
"total": 2
|
||||
}
|
||||
],
|
||||
"percentage": 50.0
|
||||
}
|
||||
"#);
|
||||
let (atomic, tasks) = AtomicTaskStep::new(5);
|
||||
processing.progress.as_ref().unwrap().update_progress(tasks);
|
||||
atomic.fetch_add(4, Ordering::Relaxed);
|
||||
snapshot!(json_string!(processing.get_progress_view()), @r#"
|
||||
{
|
||||
"steps": [
|
||||
{
|
||||
"currentStep": "writing tasks to disk",
|
||||
"finished": 1,
|
||||
"total": 2
|
||||
},
|
||||
{
|
||||
"currentStep": "task",
|
||||
"finished": 4,
|
||||
"total": 5
|
||||
}
|
||||
],
|
||||
"percentage": 90.0
|
||||
}
|
||||
"#);
|
||||
}
|
||||
}
|
||||
@@ -134,7 +134,6 @@ impl ProcessingBatch {
|
||||
pub fn to_batch(&self) -> Batch {
|
||||
Batch {
|
||||
uid: self.uid,
|
||||
progress: None,
|
||||
details: self.details.clone(),
|
||||
stats: self.stats.clone(),
|
||||
started_at: self.started_at,
|
||||
@@ -188,7 +187,6 @@ impl IndexScheduler {
|
||||
&batch.uid,
|
||||
&Batch {
|
||||
uid: batch.uid,
|
||||
progress: None,
|
||||
details: batch.details,
|
||||
stats: batch.stats,
|
||||
started_at: batch.started_at,
|
||||
@@ -275,9 +273,7 @@ impl IndexScheduler {
|
||||
.into_iter()
|
||||
.map(|batch_id| {
|
||||
if Some(batch_id) == processing.batch.as_ref().map(|batch| batch.uid) {
|
||||
let mut batch = processing.batch.as_ref().unwrap().to_batch();
|
||||
batch.progress = processing.get_progress_view();
|
||||
Ok(batch)
|
||||
Ok(processing.batch.as_ref().unwrap().to_batch())
|
||||
} else {
|
||||
self.get_batch(rtxn, batch_id)
|
||||
.and_then(|task| task.ok_or(Error::CorruptedTaskQueue))
|
||||
@@ -291,10 +287,7 @@ impl IndexScheduler {
|
||||
|
||||
debug_assert!(old_task != *task);
|
||||
debug_assert_eq!(old_task.uid, task.uid);
|
||||
debug_assert!(
|
||||
old_task.batch_uid.is_none() && task.batch_uid.is_some(),
|
||||
"\n==> old: {old_task:?}\n==> new: {task:?}"
|
||||
);
|
||||
debug_assert!(old_task.batch_uid.is_none() && task.batch_uid.is_some());
|
||||
|
||||
if old_task.status != task.status {
|
||||
self.update_status(wtxn, old_task.status, |bitmap| {
|
||||
|
||||
@@ -24,9 +24,8 @@ flate2 = "1.0.30"
|
||||
fst = "0.4.7"
|
||||
memmap2 = "0.9.4"
|
||||
milli = { path = "../milli" }
|
||||
bumparaw-collections = "0.1.2"
|
||||
raw-collections = { git = "https://github.com/meilisearch/raw-collections.git", version = "0.1.0" }
|
||||
roaring = { version = "0.10.7", features = ["serde"] }
|
||||
rustc-hash = "2.1.0"
|
||||
serde = { version = "1.0.204", features = ["derive"] }
|
||||
serde-cs = "0.2.4"
|
||||
serde_json = "1.0.120"
|
||||
|
||||
@@ -1,16 +1,16 @@
|
||||
use milli::progress::ProgressView;
|
||||
use serde::Serialize;
|
||||
use time::{Duration, OffsetDateTime};
|
||||
|
||||
use crate::batches::{Batch, BatchId, BatchStats};
|
||||
use crate::task_view::DetailsView;
|
||||
use crate::tasks::serialize_duration;
|
||||
use crate::{
|
||||
batches::{Batch, BatchId, BatchStats},
|
||||
task_view::DetailsView,
|
||||
tasks::serialize_duration,
|
||||
};
|
||||
|
||||
#[derive(Debug, Clone, Serialize)]
|
||||
#[serde(rename_all = "camelCase")]
|
||||
pub struct BatchView {
|
||||
pub uid: BatchId,
|
||||
pub progress: Option<ProgressView>,
|
||||
pub details: DetailsView,
|
||||
pub stats: BatchStats,
|
||||
#[serde(serialize_with = "serialize_duration", default)]
|
||||
@@ -25,7 +25,6 @@ impl BatchView {
|
||||
pub fn from_batch(batch: &Batch) -> Self {
|
||||
Self {
|
||||
uid: batch.uid,
|
||||
progress: batch.progress.clone(),
|
||||
details: batch.details.clone(),
|
||||
stats: batch.stats.clone(),
|
||||
duration: batch.finished_at.map(|finished_at| finished_at - batch.started_at),
|
||||
|
||||
@@ -1,11 +1,12 @@
|
||||
use std::collections::BTreeMap;
|
||||
|
||||
use milli::progress::ProgressView;
|
||||
use serde::{Deserialize, Serialize};
|
||||
use time::OffsetDateTime;
|
||||
|
||||
use crate::task_view::DetailsView;
|
||||
use crate::tasks::{Kind, Status};
|
||||
use crate::{
|
||||
task_view::DetailsView,
|
||||
tasks::{Kind, Status},
|
||||
};
|
||||
|
||||
pub type BatchId = u32;
|
||||
|
||||
@@ -14,8 +15,6 @@ pub type BatchId = u32;
|
||||
pub struct Batch {
|
||||
pub uid: BatchId,
|
||||
|
||||
#[serde(skip)]
|
||||
pub progress: Option<ProgressView>,
|
||||
pub details: DetailsView,
|
||||
pub stats: BatchStats,
|
||||
|
||||
|
||||
@@ -4,11 +4,10 @@ use std::io::{self, BufWriter};
|
||||
use std::marker::PhantomData;
|
||||
|
||||
use bumpalo::Bump;
|
||||
use bumparaw_collections::RawMap;
|
||||
use memmap2::Mmap;
|
||||
use milli::documents::Error;
|
||||
use milli::Object;
|
||||
use rustc_hash::FxBuildHasher;
|
||||
use raw_collections::RawMap;
|
||||
use serde::de::{SeqAccess, Visitor};
|
||||
use serde::{Deserialize, Deserializer};
|
||||
use serde_json::error::Category;
|
||||
@@ -221,7 +220,7 @@ pub fn read_json(input: &File, output: impl io::Write) -> Result<u64> {
|
||||
let mut deserializer = serde_json::Deserializer::from_slice(&input);
|
||||
let res = array_each(&mut deserializer, |obj: &RawValue| {
|
||||
doc_alloc.reset();
|
||||
let map = RawMap::from_raw_value_and_hasher(obj, FxBuildHasher, &doc_alloc)?;
|
||||
let map = RawMap::from_raw_value(obj, &doc_alloc)?;
|
||||
to_writer(&mut out, &map)
|
||||
});
|
||||
let count = match res {
|
||||
@@ -251,25 +250,26 @@ pub fn read_json(input: &File, output: impl io::Write) -> Result<u64> {
|
||||
}
|
||||
}
|
||||
|
||||
/// Reads NDJSON from file and checks it.
|
||||
pub fn read_ndjson(input: &File) -> Result<u64> {
|
||||
/// Reads NDJSON from file and write it in NDJSON in a file checking it along the way.
|
||||
pub fn read_ndjson(input: &File, output: impl io::Write) -> Result<u64> {
|
||||
// We memory map to be able to deserialize into a RawMap that
|
||||
// does not allocate when possible and only materialize the first/top level.
|
||||
let input = unsafe { Mmap::map(input).map_err(DocumentFormatError::Io)? };
|
||||
let mut output = BufWriter::new(output);
|
||||
|
||||
let mut bump = Bump::with_capacity(1024 * 1024);
|
||||
|
||||
let mut count = 0;
|
||||
for result in serde_json::Deserializer::from_slice(&input).into_iter() {
|
||||
bump.reset();
|
||||
match result {
|
||||
Ok(raw) => {
|
||||
count += 1;
|
||||
result
|
||||
.and_then(|raw: &RawValue| {
|
||||
// try to deserialize as a map
|
||||
RawMap::from_raw_value_and_hasher(raw, FxBuildHasher, &bump)
|
||||
.map_err(|e| DocumentFormatError::from((PayloadType::Ndjson, e)))?;
|
||||
count += 1;
|
||||
}
|
||||
Err(e) => return Err(DocumentFormatError::from((PayloadType::Ndjson, e))),
|
||||
}
|
||||
let map = RawMap::from_raw_value(raw, &bump)?;
|
||||
to_writer(&mut output, &map)
|
||||
})
|
||||
.map_err(|e| DocumentFormatError::from((PayloadType::Ndjson, e)))?;
|
||||
}
|
||||
|
||||
Ok(count)
|
||||
|
||||
@@ -550,7 +550,7 @@ impl fmt::Display for deserr_codes::InvalidSimilarId {
|
||||
"the value of `id` is invalid. \
|
||||
A document identifier can be of type integer or string, \
|
||||
only composed of alphanumeric characters (a-z A-Z 0-9), hyphens (-) and underscores (_), \
|
||||
and can not be more than 511 bytes."
|
||||
and can not be more than 512 bytes."
|
||||
)
|
||||
}
|
||||
}
|
||||
|
||||
@@ -4,6 +4,7 @@ use std::fmt::{Display, Write};
|
||||
use std::str::FromStr;
|
||||
|
||||
use enum_iterator::Sequence;
|
||||
use milli::update::new::indexer::document_changes::Progress;
|
||||
use milli::update::IndexDocumentsMethod;
|
||||
use milli::Object;
|
||||
use roaring::RoaringBitmap;
|
||||
@@ -40,6 +41,62 @@ pub struct Task {
|
||||
pub kind: KindWithContent,
|
||||
}
|
||||
|
||||
#[derive(Clone, Copy, Debug, PartialEq, Eq, Serialize, Deserialize)]
|
||||
#[serde(rename_all = "camelCase")]
|
||||
pub struct TaskProgress {
|
||||
pub current_step: &'static str,
|
||||
pub finished_steps: u16,
|
||||
pub total_steps: u16,
|
||||
pub finished_substeps: Option<u32>,
|
||||
pub total_substeps: Option<u32>,
|
||||
}
|
||||
|
||||
impl Default for TaskProgress {
|
||||
fn default() -> Self {
|
||||
Self::new()
|
||||
}
|
||||
}
|
||||
|
||||
impl TaskProgress {
|
||||
pub fn new() -> Self {
|
||||
Self {
|
||||
current_step: "start",
|
||||
finished_steps: 0,
|
||||
total_steps: 1,
|
||||
finished_substeps: None,
|
||||
total_substeps: None,
|
||||
}
|
||||
}
|
||||
|
||||
pub fn update(&mut self, progress: Progress) -> TaskProgress {
|
||||
if self.finished_steps > progress.finished_steps {
|
||||
return *self;
|
||||
}
|
||||
|
||||
if self.current_step != progress.step_name {
|
||||
self.current_step = progress.step_name
|
||||
}
|
||||
|
||||
self.total_steps = progress.total_steps;
|
||||
|
||||
if self.finished_steps < progress.finished_steps {
|
||||
self.finished_substeps = None;
|
||||
self.total_substeps = None;
|
||||
}
|
||||
self.finished_steps = progress.finished_steps;
|
||||
if let Some((finished_substeps, total_substeps)) = progress.finished_total_substep {
|
||||
if let Some(task_finished_substeps) = self.finished_substeps {
|
||||
if task_finished_substeps > finished_substeps {
|
||||
return *self;
|
||||
}
|
||||
}
|
||||
self.finished_substeps = Some(finished_substeps);
|
||||
self.total_substeps = Some(total_substeps);
|
||||
}
|
||||
*self
|
||||
}
|
||||
}
|
||||
|
||||
impl Task {
|
||||
pub fn index_uid(&self) -> Option<&str> {
|
||||
use KindWithContent::*;
|
||||
|
||||
@@ -129,11 +129,6 @@ async fn try_main() -> anyhow::Result<()> {
|
||||
|
||||
print_launch_resume(&opt, analytics.clone(), config_read_from);
|
||||
|
||||
tokio::spawn(async move {
|
||||
tokio::signal::ctrl_c().await.unwrap();
|
||||
std::process::exit(130);
|
||||
});
|
||||
|
||||
run_http(index_scheduler, auth_controller, opt, log_handle, Arc::new(analytics)).await?;
|
||||
|
||||
Ok(())
|
||||
|
||||
@@ -1,18 +1,18 @@
|
||||
use actix_web::web::{self, Data};
|
||||
use actix_web::HttpResponse;
|
||||
use actix_web::{
|
||||
web::{self, Data},
|
||||
HttpResponse,
|
||||
};
|
||||
use deserr::actix_web::AwebQueryParameter;
|
||||
use index_scheduler::{IndexScheduler, Query};
|
||||
use meilisearch_types::batch_view::BatchView;
|
||||
use meilisearch_types::batches::BatchId;
|
||||
use meilisearch_types::deserr::DeserrQueryParamError;
|
||||
use meilisearch_types::error::ResponseError;
|
||||
use meilisearch_types::keys::actions;
|
||||
use meilisearch_types::{
|
||||
batch_view::BatchView, batches::BatchId, deserr::DeserrQueryParamError, error::ResponseError,
|
||||
keys::actions,
|
||||
};
|
||||
use serde::Serialize;
|
||||
|
||||
use super::tasks::TasksFilterQuery;
|
||||
use super::ActionPolicy;
|
||||
use crate::extractors::authentication::GuardedData;
|
||||
use crate::extractors::sequential_extractor::SeqHandler;
|
||||
use crate::extractors::{authentication::GuardedData, sequential_extractor::SeqHandler};
|
||||
|
||||
use super::{tasks::TasksFilterQuery, ActionPolicy};
|
||||
|
||||
pub fn configure(cfg: &mut web::ServiceConfig) {
|
||||
cfg.service(web::resource("").route(web::get().to(SeqHandler(get_batches))))
|
||||
|
||||
@@ -1,5 +1,5 @@
|
||||
use std::collections::HashSet;
|
||||
use std::io::{ErrorKind, Seek as _};
|
||||
use std::io::ErrorKind;
|
||||
use std::marker::PhantomData;
|
||||
|
||||
use actix_web::http::header::CONTENT_TYPE;
|
||||
@@ -572,7 +572,7 @@ async fn document_addition(
|
||||
index_uid: IndexUid,
|
||||
primary_key: Option<String>,
|
||||
csv_delimiter: Option<u8>,
|
||||
body: Payload,
|
||||
mut body: Payload,
|
||||
method: IndexDocumentsMethod,
|
||||
task_id: Option<TaskId>,
|
||||
dry_run: bool,
|
||||
@@ -609,60 +609,54 @@ async fn document_addition(
|
||||
};
|
||||
|
||||
let (uuid, mut update_file) = index_scheduler.create_update_file(dry_run)?;
|
||||
let documents_count = match format {
|
||||
PayloadType::Ndjson => {
|
||||
let (path, file) = update_file.into_parts();
|
||||
let file = match file {
|
||||
Some(file) => {
|
||||
let (file, path) = file.into_parts();
|
||||
let mut file = copy_body_to_file(file, body, format).await?;
|
||||
file.rewind().map_err(|e| {
|
||||
index_scheduler::Error::FileStore(file_store::Error::IoError(e))
|
||||
})?;
|
||||
Some(tempfile::NamedTempFile::from_parts(file, path))
|
||||
}
|
||||
None => None,
|
||||
};
|
||||
|
||||
let documents_count = tokio::task::spawn_blocking(move || {
|
||||
let documents_count = file.as_ref().map_or(Ok(0), |ntf| {
|
||||
read_ndjson(ntf.as_file()).map_err(MeilisearchHttpError::DocumentFormat)
|
||||
})?;
|
||||
|
||||
let update_file = file_store::File::from_parts(path, file);
|
||||
update_file.persist()?;
|
||||
|
||||
Ok(documents_count)
|
||||
})
|
||||
.await?;
|
||||
|
||||
Ok(documents_count)
|
||||
}
|
||||
PayloadType::Json | PayloadType::Csv { delimiter: _ } => {
|
||||
let temp_file = match tempfile() {
|
||||
Ok(file) => file,
|
||||
Err(e) => return Err(MeilisearchHttpError::Payload(ReceivePayload(Box::new(e)))),
|
||||
};
|
||||
|
||||
let read_file = copy_body_to_file(temp_file, body, format).await?;
|
||||
tokio::task::spawn_blocking(move || {
|
||||
let documents_count = match format {
|
||||
PayloadType::Json => read_json(&read_file, &mut update_file)?,
|
||||
PayloadType::Csv { delimiter } => {
|
||||
read_csv(&read_file, &mut update_file, delimiter)?
|
||||
}
|
||||
PayloadType::Ndjson => {
|
||||
unreachable!("We already wrote the user content into the update file")
|
||||
}
|
||||
};
|
||||
// we NEED to persist the file here because we moved the `udpate_file` in another task.
|
||||
update_file.persist()?;
|
||||
Ok(documents_count)
|
||||
})
|
||||
.await
|
||||
}
|
||||
let temp_file = match tempfile() {
|
||||
Ok(file) => file,
|
||||
Err(e) => return Err(MeilisearchHttpError::Payload(ReceivePayload(Box::new(e)))),
|
||||
};
|
||||
|
||||
let async_file = File::from_std(temp_file);
|
||||
let mut buffer = BufWriter::new(async_file);
|
||||
|
||||
let mut buffer_write_size: usize = 0;
|
||||
while let Some(result) = body.next().await {
|
||||
let byte = result?;
|
||||
|
||||
if byte.is_empty() && buffer_write_size == 0 {
|
||||
return Err(MeilisearchHttpError::MissingPayload(format));
|
||||
}
|
||||
|
||||
match buffer.write_all(&byte).await {
|
||||
Ok(()) => buffer_write_size += 1,
|
||||
Err(e) => return Err(MeilisearchHttpError::Payload(ReceivePayload(Box::new(e)))),
|
||||
}
|
||||
}
|
||||
|
||||
if let Err(e) = buffer.flush().await {
|
||||
return Err(MeilisearchHttpError::Payload(ReceivePayload(Box::new(e))));
|
||||
}
|
||||
|
||||
if buffer_write_size == 0 {
|
||||
return Err(MeilisearchHttpError::MissingPayload(format));
|
||||
}
|
||||
|
||||
if let Err(e) = buffer.seek(std::io::SeekFrom::Start(0)).await {
|
||||
return Err(MeilisearchHttpError::Payload(ReceivePayload(Box::new(e))));
|
||||
}
|
||||
|
||||
let read_file = buffer.into_inner().into_std().await;
|
||||
let documents_count = tokio::task::spawn_blocking(move || {
|
||||
let documents_count = match format {
|
||||
PayloadType::Json => read_json(&read_file, &mut update_file)?,
|
||||
PayloadType::Csv { delimiter } => read_csv(&read_file, &mut update_file, delimiter)?,
|
||||
PayloadType::Ndjson => read_ndjson(&read_file, &mut update_file)?,
|
||||
};
|
||||
// we NEED to persist the file here because we moved the `udpate_file` in another task.
|
||||
update_file.persist()?;
|
||||
Ok(documents_count)
|
||||
})
|
||||
.await;
|
||||
|
||||
let documents_count = match documents_count {
|
||||
Ok(Ok(documents_count)) => documents_count,
|
||||
// in this case the file has not possibly be persisted.
|
||||
@@ -709,39 +703,6 @@ async fn document_addition(
|
||||
Ok(task.into())
|
||||
}
|
||||
|
||||
async fn copy_body_to_file(
|
||||
output: std::fs::File,
|
||||
mut body: Payload,
|
||||
format: PayloadType,
|
||||
) -> Result<std::fs::File, MeilisearchHttpError> {
|
||||
let async_file = File::from_std(output);
|
||||
let mut buffer = BufWriter::new(async_file);
|
||||
let mut buffer_write_size: usize = 0;
|
||||
while let Some(result) = body.next().await {
|
||||
let byte = result?;
|
||||
|
||||
if byte.is_empty() && buffer_write_size == 0 {
|
||||
return Err(MeilisearchHttpError::MissingPayload(format));
|
||||
}
|
||||
|
||||
match buffer.write_all(&byte).await {
|
||||
Ok(()) => buffer_write_size += 1,
|
||||
Err(e) => return Err(MeilisearchHttpError::Payload(ReceivePayload(Box::new(e)))),
|
||||
}
|
||||
}
|
||||
if let Err(e) = buffer.flush().await {
|
||||
return Err(MeilisearchHttpError::Payload(ReceivePayload(Box::new(e))));
|
||||
}
|
||||
if buffer_write_size == 0 {
|
||||
return Err(MeilisearchHttpError::MissingPayload(format));
|
||||
}
|
||||
if let Err(e) = buffer.seek(std::io::SeekFrom::Start(0)).await {
|
||||
return Err(MeilisearchHttpError::Payload(ReceivePayload(Box::new(e))));
|
||||
}
|
||||
let read_file = buffer.into_inner().into_std().await;
|
||||
Ok(read_file)
|
||||
}
|
||||
|
||||
pub async fn delete_documents_batch(
|
||||
index_scheduler: GuardedData<ActionPolicy<{ actions::DOCUMENTS_DELETE }>, Data<IndexScheduler>>,
|
||||
index_uid: web::Path<String>,
|
||||
|
||||
@@ -284,7 +284,6 @@ async fn test_summarized_document_addition_or_update() {
|
||||
@r#"
|
||||
{
|
||||
"uid": 0,
|
||||
"progress": null,
|
||||
"details": {
|
||||
"receivedDocuments": 1,
|
||||
"indexedDocuments": 1
|
||||
@@ -315,7 +314,6 @@ async fn test_summarized_document_addition_or_update() {
|
||||
@r#"
|
||||
{
|
||||
"uid": 1,
|
||||
"progress": null,
|
||||
"details": {
|
||||
"receivedDocuments": 1,
|
||||
"indexedDocuments": 1
|
||||
@@ -351,7 +349,6 @@ async fn test_summarized_delete_documents_by_batch() {
|
||||
@r#"
|
||||
{
|
||||
"uid": 0,
|
||||
"progress": null,
|
||||
"details": {
|
||||
"providedIds": 3,
|
||||
"deletedDocuments": 0
|
||||
@@ -383,7 +380,6 @@ async fn test_summarized_delete_documents_by_batch() {
|
||||
@r#"
|
||||
{
|
||||
"uid": 2,
|
||||
"progress": null,
|
||||
"details": {
|
||||
"providedIds": 1,
|
||||
"deletedDocuments": 0
|
||||
@@ -420,7 +416,6 @@ async fn test_summarized_delete_documents_by_filter() {
|
||||
@r#"
|
||||
{
|
||||
"uid": 0,
|
||||
"progress": null,
|
||||
"details": {
|
||||
"providedIds": 0,
|
||||
"deletedDocuments": 0,
|
||||
@@ -453,7 +448,6 @@ async fn test_summarized_delete_documents_by_filter() {
|
||||
@r#"
|
||||
{
|
||||
"uid": 2,
|
||||
"progress": null,
|
||||
"details": {
|
||||
"providedIds": 0,
|
||||
"deletedDocuments": 0,
|
||||
@@ -486,7 +480,6 @@ async fn test_summarized_delete_documents_by_filter() {
|
||||
@r#"
|
||||
{
|
||||
"uid": 4,
|
||||
"progress": null,
|
||||
"details": {
|
||||
"providedIds": 0,
|
||||
"deletedDocuments": 0,
|
||||
@@ -523,7 +516,6 @@ async fn test_summarized_delete_document_by_id() {
|
||||
@r#"
|
||||
{
|
||||
"uid": 0,
|
||||
"progress": null,
|
||||
"details": {
|
||||
"providedIds": 1,
|
||||
"deletedDocuments": 0
|
||||
@@ -555,7 +547,6 @@ async fn test_summarized_delete_document_by_id() {
|
||||
@r#"
|
||||
{
|
||||
"uid": 2,
|
||||
"progress": null,
|
||||
"details": {
|
||||
"providedIds": 1,
|
||||
"deletedDocuments": 0
|
||||
@@ -603,7 +594,6 @@ async fn test_summarized_settings_update() {
|
||||
@r#"
|
||||
{
|
||||
"uid": 0,
|
||||
"progress": null,
|
||||
"details": {
|
||||
"displayedAttributes": [
|
||||
"doggos",
|
||||
@@ -648,7 +638,6 @@ async fn test_summarized_index_creation() {
|
||||
@r#"
|
||||
{
|
||||
"uid": 0,
|
||||
"progress": null,
|
||||
"details": {},
|
||||
"stats": {
|
||||
"totalNbTasks": 1,
|
||||
@@ -676,7 +665,6 @@ async fn test_summarized_index_creation() {
|
||||
@r#"
|
||||
{
|
||||
"uid": 1,
|
||||
"progress": null,
|
||||
"details": {
|
||||
"primaryKey": "doggos"
|
||||
},
|
||||
@@ -821,7 +809,6 @@ async fn test_summarized_index_update() {
|
||||
@r#"
|
||||
{
|
||||
"uid": 0,
|
||||
"progress": null,
|
||||
"details": {},
|
||||
"stats": {
|
||||
"totalNbTasks": 1,
|
||||
@@ -849,7 +836,6 @@ async fn test_summarized_index_update() {
|
||||
@r#"
|
||||
{
|
||||
"uid": 1,
|
||||
"progress": null,
|
||||
"details": {
|
||||
"primaryKey": "bones"
|
||||
},
|
||||
@@ -882,7 +868,6 @@ async fn test_summarized_index_update() {
|
||||
@r#"
|
||||
{
|
||||
"uid": 3,
|
||||
"progress": null,
|
||||
"details": {},
|
||||
"stats": {
|
||||
"totalNbTasks": 1,
|
||||
@@ -910,7 +895,6 @@ async fn test_summarized_index_update() {
|
||||
@r#"
|
||||
{
|
||||
"uid": 4,
|
||||
"progress": null,
|
||||
"details": {
|
||||
"primaryKey": "bones"
|
||||
},
|
||||
@@ -948,7 +932,6 @@ async fn test_summarized_index_swap() {
|
||||
@r#"
|
||||
{
|
||||
"uid": 0,
|
||||
"progress": null,
|
||||
"details": {
|
||||
"swaps": [
|
||||
{
|
||||
@@ -989,7 +972,6 @@ async fn test_summarized_index_swap() {
|
||||
@r#"
|
||||
{
|
||||
"uid": 3,
|
||||
"progress": null,
|
||||
"details": {
|
||||
"swaps": [
|
||||
{
|
||||
@@ -1032,7 +1014,6 @@ async fn test_summarized_batch_cancelation() {
|
||||
@r#"
|
||||
{
|
||||
"uid": 1,
|
||||
"progress": null,
|
||||
"details": {
|
||||
"matchedTasks": 1,
|
||||
"canceledTasks": 0,
|
||||
@@ -1070,7 +1051,6 @@ async fn test_summarized_batch_deletion() {
|
||||
@r#"
|
||||
{
|
||||
"uid": 1,
|
||||
"progress": null,
|
||||
"details": {
|
||||
"matchedTasks": 1,
|
||||
"deletedTasks": 1,
|
||||
@@ -1104,7 +1084,6 @@ async fn test_summarized_dump_creation() {
|
||||
@r#"
|
||||
{
|
||||
"uid": 0,
|
||||
"progress": null,
|
||||
"details": {
|
||||
"dumpUid": "[dumpUid]"
|
||||
},
|
||||
|
||||
@@ -1220,89 +1220,9 @@ async fn replace_document() {
|
||||
#[actix_rt::test]
|
||||
async fn add_no_documents() {
|
||||
let server = Server::new().await;
|
||||
let index = server.index("kefir");
|
||||
let (task, code) = index.add_documents(json!([]), None).await;
|
||||
let index = server.index("test");
|
||||
let (_response, code) = index.add_documents(json!([]), None).await;
|
||||
snapshot!(code, @"202 Accepted");
|
||||
let task = server.wait_task(task.uid()).await;
|
||||
let task = task.succeeded();
|
||||
snapshot!(task, @r#"
|
||||
{
|
||||
"uid": "[uid]",
|
||||
"batchUid": "[batch_uid]",
|
||||
"indexUid": "kefir",
|
||||
"status": "succeeded",
|
||||
"type": "documentAdditionOrUpdate",
|
||||
"canceledBy": null,
|
||||
"details": {
|
||||
"receivedDocuments": 0,
|
||||
"indexedDocuments": 0
|
||||
},
|
||||
"error": null,
|
||||
"duration": "[duration]",
|
||||
"enqueuedAt": "[date]",
|
||||
"startedAt": "[date]",
|
||||
"finishedAt": "[date]"
|
||||
}
|
||||
"#);
|
||||
|
||||
let (task, _code) = index.add_documents(json!([]), Some("kefkef")).await;
|
||||
let task = server.wait_task(task.uid()).await;
|
||||
let task = task.succeeded();
|
||||
snapshot!(task, @r#"
|
||||
{
|
||||
"uid": "[uid]",
|
||||
"batchUid": "[batch_uid]",
|
||||
"indexUid": "kefir",
|
||||
"status": "succeeded",
|
||||
"type": "documentAdditionOrUpdate",
|
||||
"canceledBy": null,
|
||||
"details": {
|
||||
"receivedDocuments": 0,
|
||||
"indexedDocuments": 0
|
||||
},
|
||||
"error": null,
|
||||
"duration": "[duration]",
|
||||
"enqueuedAt": "[date]",
|
||||
"startedAt": "[date]",
|
||||
"finishedAt": "[date]"
|
||||
}
|
||||
"#);
|
||||
|
||||
let (task, _code) = index.add_documents(json!([{ "kefkef": 1 }]), None).await;
|
||||
let task = server.wait_task(task.uid()).await;
|
||||
let task = task.succeeded();
|
||||
snapshot!(task, @r#"
|
||||
{
|
||||
"uid": "[uid]",
|
||||
"batchUid": "[batch_uid]",
|
||||
"indexUid": "kefir",
|
||||
"status": "succeeded",
|
||||
"type": "documentAdditionOrUpdate",
|
||||
"canceledBy": null,
|
||||
"details": {
|
||||
"receivedDocuments": 1,
|
||||
"indexedDocuments": 1
|
||||
},
|
||||
"error": null,
|
||||
"duration": "[duration]",
|
||||
"enqueuedAt": "[date]",
|
||||
"startedAt": "[date]",
|
||||
"finishedAt": "[date]"
|
||||
}
|
||||
"#);
|
||||
let (documents, _status) = index.get_all_documents(GetAllDocumentsOptions::default()).await;
|
||||
snapshot!(documents, @r#"
|
||||
{
|
||||
"results": [
|
||||
{
|
||||
"kefkef": 1
|
||||
}
|
||||
],
|
||||
"offset": 0,
|
||||
"limit": 20,
|
||||
"total": 1
|
||||
}
|
||||
"#);
|
||||
}
|
||||
|
||||
#[actix_rt::test]
|
||||
@@ -1344,18 +1264,15 @@ async fn error_add_documents_bad_document_id() {
|
||||
let server = Server::new().await;
|
||||
let index = server.index("test");
|
||||
index.create(Some("docid")).await;
|
||||
|
||||
// unsupported characters
|
||||
|
||||
let documents = json!([
|
||||
{
|
||||
"docid": "foo & bar",
|
||||
"content": "foobar"
|
||||
}
|
||||
]);
|
||||
let (value, _code) = index.add_documents(documents, None).await;
|
||||
index.wait_task(value.uid()).await;
|
||||
let (response, code) = index.get_task(value.uid()).await;
|
||||
index.add_documents(documents, None).await;
|
||||
index.wait_task(1).await;
|
||||
let (response, code) = index.get_task(1).await;
|
||||
snapshot!(code, @"200 OK");
|
||||
snapshot!(json_string!(response, { ".duration" => "[duration]", ".enqueuedAt" => "[date]", ".startedAt" => "[date]", ".finishedAt" => "[date]" }),
|
||||
@r###"
|
||||
@@ -1371,81 +1288,7 @@ async fn error_add_documents_bad_document_id() {
|
||||
"indexedDocuments": 0
|
||||
},
|
||||
"error": {
|
||||
"message": "Document identifier `\"foo & bar\"` is invalid. A document identifier can be of type integer or string, only composed of alphanumeric characters (a-z A-Z 0-9), hyphens (-) and underscores (_), and can not be more than 511 bytes.",
|
||||
"code": "invalid_document_id",
|
||||
"type": "invalid_request",
|
||||
"link": "https://docs.meilisearch.com/errors#invalid_document_id"
|
||||
},
|
||||
"duration": "[duration]",
|
||||
"enqueuedAt": "[date]",
|
||||
"startedAt": "[date]",
|
||||
"finishedAt": "[date]"
|
||||
}
|
||||
"###);
|
||||
|
||||
// More than 512 bytes
|
||||
let documents = json!([
|
||||
{
|
||||
"docid": "a".repeat(600),
|
||||
"content": "foobar"
|
||||
}
|
||||
]);
|
||||
let (value, _code) = index.add_documents(documents, None).await;
|
||||
index.wait_task(value.uid()).await;
|
||||
let (response, code) = index.get_task(value.uid()).await;
|
||||
snapshot!(code, @"200 OK");
|
||||
snapshot!(json_string!(response, { ".duration" => "[duration]", ".enqueuedAt" => "[date]", ".startedAt" => "[date]", ".finishedAt" => "[date]" }),
|
||||
@r###"
|
||||
{
|
||||
"uid": 2,
|
||||
"batchUid": 2,
|
||||
"indexUid": "test",
|
||||
"status": "failed",
|
||||
"type": "documentAdditionOrUpdate",
|
||||
"canceledBy": null,
|
||||
"details": {
|
||||
"receivedDocuments": 1,
|
||||
"indexedDocuments": 0
|
||||
},
|
||||
"error": {
|
||||
"message": "Document identifier `\"aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa\"` is invalid. A document identifier can be of type integer or string, only composed of alphanumeric characters (a-z A-Z 0-9), hyphens (-) and underscores (_), and can not be more than 511 bytes.",
|
||||
"code": "invalid_document_id",
|
||||
"type": "invalid_request",
|
||||
"link": "https://docs.meilisearch.com/errors#invalid_document_id"
|
||||
},
|
||||
"duration": "[duration]",
|
||||
"enqueuedAt": "[date]",
|
||||
"startedAt": "[date]",
|
||||
"finishedAt": "[date]"
|
||||
}
|
||||
"###);
|
||||
|
||||
// Exactly 512 bytes
|
||||
let documents = json!([
|
||||
{
|
||||
"docid": "a".repeat(512),
|
||||
"content": "foobar"
|
||||
}
|
||||
]);
|
||||
let (value, _code) = index.add_documents(documents, None).await;
|
||||
index.wait_task(value.uid()).await;
|
||||
let (response, code) = index.get_task(value.uid()).await;
|
||||
snapshot!(code, @"200 OK");
|
||||
snapshot!(json_string!(response, { ".duration" => "[duration]", ".enqueuedAt" => "[date]", ".startedAt" => "[date]", ".finishedAt" => "[date]" }),
|
||||
@r###"
|
||||
{
|
||||
"uid": 3,
|
||||
"batchUid": 3,
|
||||
"indexUid": "test",
|
||||
"status": "failed",
|
||||
"type": "documentAdditionOrUpdate",
|
||||
"canceledBy": null,
|
||||
"details": {
|
||||
"receivedDocuments": 1,
|
||||
"indexedDocuments": 0
|
||||
},
|
||||
"error": {
|
||||
"message": "Document identifier `\"aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa\"` is invalid. A document identifier can be of type integer or string, only composed of alphanumeric characters (a-z A-Z 0-9), hyphens (-) and underscores (_), and can not be more than 511 bytes.",
|
||||
"message": "Document identifier `\"foo & bar\"` is invalid. A document identifier can be of type integer or string, only composed of alphanumeric characters (a-z A-Z 0-9), hyphens (-) and underscores (_), and can not be more than 512 bytes.",
|
||||
"code": "invalid_document_id",
|
||||
"type": "invalid_request",
|
||||
"link": "https://docs.meilisearch.com/errors#invalid_document_id"
|
||||
|
||||
@@ -172,7 +172,7 @@ async fn error_update_documents_bad_document_id() {
|
||||
assert_eq!(
|
||||
response["error"]["message"],
|
||||
json!(
|
||||
r#"Document identifier `"foo & bar"` is invalid. A document identifier can be of type integer or string, only composed of alphanumeric characters (a-z A-Z 0-9), hyphens (-) and underscores (_), and can not be more than 511 bytes."#
|
||||
r#"Document identifier `"foo & bar"` is invalid. A document identifier can be of type integer or string, only composed of alphanumeric characters (a-z A-Z 0-9), hyphens (-) and underscores (_), and can not be more than 512 bytes."#
|
||||
)
|
||||
);
|
||||
assert_eq!(response["error"]["code"], json!("invalid_document_id"));
|
||||
|
||||
@@ -57,116 +57,6 @@ async fn simple_facet_search() {
|
||||
assert_eq!(response["facetHits"].as_array().unwrap().len(), 1);
|
||||
}
|
||||
|
||||
#[actix_rt::test]
|
||||
async fn simple_facet_search_on_movies() {
|
||||
let server = Server::new().await;
|
||||
let index = server.index("test");
|
||||
|
||||
let documents = json!([
|
||||
{
|
||||
"id": 1,
|
||||
"title": "Carol",
|
||||
"genres": [
|
||||
"Romance",
|
||||
"Drama"
|
||||
],
|
||||
"color": [
|
||||
"red"
|
||||
],
|
||||
"platforms": [
|
||||
"MacOS",
|
||||
"Linux",
|
||||
"Windows"
|
||||
]
|
||||
},
|
||||
{
|
||||
"id": 2,
|
||||
"title": "Wonder Woman",
|
||||
"genres": [
|
||||
"Action",
|
||||
"Adventure"
|
||||
],
|
||||
"color": [
|
||||
"green"
|
||||
],
|
||||
"platforms": [
|
||||
"MacOS"
|
||||
]
|
||||
},
|
||||
{
|
||||
"id": 3,
|
||||
"title": "Life of Pi",
|
||||
"genres": [
|
||||
"Adventure",
|
||||
"Drama"
|
||||
],
|
||||
"color": [
|
||||
"blue"
|
||||
],
|
||||
"platforms": [
|
||||
"Windows"
|
||||
]
|
||||
},
|
||||
{
|
||||
"id": 4,
|
||||
"title": "Mad Max: Fury Road",
|
||||
"genres": [
|
||||
"Adventure",
|
||||
"Science Fiction"
|
||||
],
|
||||
"color": [
|
||||
"red"
|
||||
],
|
||||
"platforms": [
|
||||
"MacOS",
|
||||
"Linux"
|
||||
]
|
||||
},
|
||||
{
|
||||
"id": 5,
|
||||
"title": "Moana",
|
||||
"genres": [
|
||||
"Fantasy",
|
||||
"Action"
|
||||
],
|
||||
"color": [
|
||||
"red"
|
||||
],
|
||||
"platforms": [
|
||||
"Windows"
|
||||
]
|
||||
},
|
||||
{
|
||||
"id": 6,
|
||||
"title": "Philadelphia",
|
||||
"genres": [
|
||||
"Drama"
|
||||
],
|
||||
"color": [
|
||||
"blue"
|
||||
],
|
||||
"platforms": [
|
||||
"MacOS",
|
||||
"Linux",
|
||||
"Windows"
|
||||
]
|
||||
}
|
||||
]);
|
||||
let (response, code) =
|
||||
index.update_settings_filterable_attributes(json!(["genres", "color"])).await;
|
||||
assert_eq!(202, code, "{:?}", response);
|
||||
index.wait_task(response.uid()).await;
|
||||
|
||||
let (response, _code) = index.add_documents(documents, None).await;
|
||||
index.wait_task(response.uid()).await;
|
||||
|
||||
let (response, code) =
|
||||
index.facet_search(json!({"facetQuery": "", "facetName": "genres", "q": "" })).await;
|
||||
|
||||
assert_eq!(code, 200, "{}", response);
|
||||
snapshot!(response["facetHits"], @r###"[{"value":"Action","count":2},{"value":"Adventure","count":3},{"value":"Drama","count":3},{"value":"Fantasy","count":1},{"value":"Romance","count":1},{"value":"Science Fiction","count":1}]"###);
|
||||
}
|
||||
|
||||
#[actix_rt::test]
|
||||
async fn advanced_facet_search() {
|
||||
let server = Server::new().await;
|
||||
|
||||
@@ -79,7 +79,7 @@ async fn similar_bad_id() {
|
||||
snapshot!(code, @"400 Bad Request");
|
||||
snapshot!(json_string!(response), @r###"
|
||||
{
|
||||
"message": "Invalid value at `.id`: the value of `id` is invalid. A document identifier can be of type integer or string, only composed of alphanumeric characters (a-z A-Z 0-9), hyphens (-) and underscores (_), and can not be more than 511 bytes.",
|
||||
"message": "Invalid value at `.id`: the value of `id` is invalid. A document identifier can be of type integer or string, only composed of alphanumeric characters (a-z A-Z 0-9), hyphens (-) and underscores (_), and can not be more than 512 bytes.",
|
||||
"code": "invalid_similar_id",
|
||||
"type": "invalid_request",
|
||||
"link": "https://docs.meilisearch.com/errors#invalid_similar_id"
|
||||
@@ -172,7 +172,7 @@ async fn similar_invalid_id() {
|
||||
snapshot!(code, @"400 Bad Request");
|
||||
snapshot!(json_string!(response), @r###"
|
||||
{
|
||||
"message": "Invalid value at `.id`: the value of `id` is invalid. A document identifier can be of type integer or string, only composed of alphanumeric characters (a-z A-Z 0-9), hyphens (-) and underscores (_), and can not be more than 511 bytes.",
|
||||
"message": "Invalid value at `.id`: the value of `id` is invalid. A document identifier can be of type integer or string, only composed of alphanumeric characters (a-z A-Z 0-9), hyphens (-) and underscores (_), and can not be more than 512 bytes.",
|
||||
"code": "invalid_similar_id",
|
||||
"type": "invalid_request",
|
||||
"link": "https://docs.meilisearch.com/errors#invalid_similar_id"
|
||||
|
||||
@@ -10,15 +10,12 @@ license.workspace = true
|
||||
|
||||
[dependencies]
|
||||
anyhow = "1.0.86"
|
||||
arroy_v04_to_v05 = { package = "arroy", git = "https://github.com/meilisearch/arroy/", tag = "DO-NOT-DELETE-upgrade-v04-to-v05" }
|
||||
clap = { version = "4.5.9", features = ["derive"] }
|
||||
dump = { path = "../dump" }
|
||||
file-store = { path = "../file-store" }
|
||||
indexmap = {version = "2.7.0", features = ["serde"]}
|
||||
meilisearch-auth = { path = "../meilisearch-auth" }
|
||||
meilisearch-types = { path = "../meilisearch-types" }
|
||||
serde = { version = "1.0.209", features = ["derive"] }
|
||||
serde_json = {version = "1.0.133", features = ["preserve_order"]}
|
||||
tempfile = "3.14.0"
|
||||
time = { version = "0.3.36", features = ["formatting", "parsing", "alloc"] }
|
||||
uuid = { version = "1.10.0", features = ["v4"], default-features = false }
|
||||
arroy_v04_to_v05 = { package = "arroy", git = "https://github.com/meilisearch/arroy/", tag = "DO-NOT-DELETE-upgrade-v04-to-v05" }
|
||||
|
||||
@@ -73,7 +73,7 @@ enum Command {
|
||||
///
|
||||
/// Supported upgrade paths:
|
||||
///
|
||||
/// - v1.9.x -> v1.10.x -> v1.11.x -> v1.12.x
|
||||
/// - v1.9.x -> v1.10.x -> v1.11.x
|
||||
OfflineUpgrade {
|
||||
#[arg(long)]
|
||||
target_version: String,
|
||||
|
||||
@@ -1,14 +1,13 @@
|
||||
mod v1_10;
|
||||
mod v1_11;
|
||||
mod v1_12;
|
||||
mod v1_9;
|
||||
|
||||
use std::path::{Path, PathBuf};
|
||||
|
||||
use anyhow::{bail, Context};
|
||||
use meilisearch_types::versioning::create_version_file;
|
||||
|
||||
use v1_10::v1_9_to_v1_10;
|
||||
use v1_12::v1_11_to_v1_12;
|
||||
|
||||
use crate::upgrade::v1_11::v1_10_to_v1_11;
|
||||
|
||||
@@ -23,7 +22,6 @@ impl OfflineUpgrade {
|
||||
let upgrade_list = [
|
||||
(v1_9_to_v1_10 as fn(&Path) -> Result<(), anyhow::Error>, "1", "10", "0"),
|
||||
(v1_10_to_v1_11, "1", "11", "0"),
|
||||
(v1_11_to_v1_12, "1", "12", "0"),
|
||||
];
|
||||
|
||||
let (current_major, current_minor, current_patch) = &self.current_version;
|
||||
@@ -35,7 +33,6 @@ impl OfflineUpgrade {
|
||||
) {
|
||||
("1", "9", _) => 0,
|
||||
("1", "10", _) => 1,
|
||||
("1", "11", _) => 2,
|
||||
_ => {
|
||||
bail!("Unsupported current version {current_major}.{current_minor}.{current_patch}. Can only upgrade from v1.9 and v1.10")
|
||||
}
|
||||
@@ -46,7 +43,6 @@ impl OfflineUpgrade {
|
||||
let ends_at = match (target_major.as_str(), target_minor.as_str(), target_patch.as_str()) {
|
||||
("1", "10", _) => 0,
|
||||
("1", "11", _) => 1,
|
||||
("1", "12", _) => 2,
|
||||
(major, _, _) if major.starts_with('v') => {
|
||||
bail!("Target version must not starts with a `v`. Instead of writing `v1.9.0` write `1.9.0` for example.")
|
||||
}
|
||||
|
||||
@@ -1,13 +1,18 @@
|
||||
use anyhow::bail;
|
||||
use std::path::Path;
|
||||
|
||||
use anyhow::{bail, Context};
|
||||
use meilisearch_types::heed::types::{SerdeJson, Str};
|
||||
use meilisearch_types::heed::{Database, Env, EnvOpenOptions, RoTxn, RwTxn, Unspecified};
|
||||
use meilisearch_types::milli::index::{db_name, main_key};
|
||||
use anyhow::Context;
|
||||
use meilisearch_types::{
|
||||
heed::{
|
||||
types::{SerdeJson, Str},
|
||||
Database, Env, EnvOpenOptions, RoTxn, RwTxn, Unspecified,
|
||||
},
|
||||
milli::index::{db_name, main_key},
|
||||
};
|
||||
|
||||
use crate::{try_opening_database, try_opening_poly_database, uuid_codec::UuidCodec};
|
||||
|
||||
use super::v1_9;
|
||||
use crate::uuid_codec::UuidCodec;
|
||||
use crate::{try_opening_database, try_opening_poly_database};
|
||||
|
||||
pub type FieldDistribution = std::collections::BTreeMap<String, u64>;
|
||||
|
||||
|
||||
@@ -7,12 +7,12 @@
|
||||
use std::path::Path;
|
||||
|
||||
use anyhow::Context;
|
||||
use meilisearch_types::heed::types::Str;
|
||||
use meilisearch_types::heed::{Database, EnvOpenOptions};
|
||||
use meilisearch_types::milli::index::db_name;
|
||||
use meilisearch_types::{
|
||||
heed::{types::Str, Database, EnvOpenOptions},
|
||||
milli::index::db_name,
|
||||
};
|
||||
|
||||
use crate::uuid_codec::UuidCodec;
|
||||
use crate::{try_opening_database, try_opening_poly_database};
|
||||
use crate::{try_opening_database, try_opening_poly_database, uuid_codec::UuidCodec};
|
||||
|
||||
pub fn v1_10_to_v1_11(db_path: &Path) -> anyhow::Result<()> {
|
||||
println!("Upgrading from v1.10.0 to v1.11.0");
|
||||
|
||||
@@ -1,79 +0,0 @@
|
||||
//! The breaking changes that happened between the v1.11 and the v1.12 are:
|
||||
//! - The new indexer changed the update files format from OBKV to ndjson. https://github.com/meilisearch/meilisearch/pull/4900
|
||||
|
||||
use std::io::BufWriter;
|
||||
use std::path::Path;
|
||||
|
||||
use anyhow::Context;
|
||||
use file_store::FileStore;
|
||||
use indexmap::IndexMap;
|
||||
use meilisearch_types::milli::documents::DocumentsBatchReader;
|
||||
use serde_json::value::RawValue;
|
||||
use tempfile::NamedTempFile;
|
||||
|
||||
pub fn v1_11_to_v1_12(db_path: &Path) -> anyhow::Result<()> {
|
||||
println!("Upgrading from v1.11.0 to v1.12.0");
|
||||
|
||||
convert_update_files(db_path)?;
|
||||
|
||||
Ok(())
|
||||
}
|
||||
|
||||
/// Convert the update files from OBKV to ndjson format.
|
||||
///
|
||||
/// 1) List all the update files using the file store.
|
||||
/// 2) For each update file, read the update file into a DocumentsBatchReader.
|
||||
/// 3) For each document in the update file, convert the document to a JSON object.
|
||||
/// 4) Write the JSON object to a tmp file in the update files directory.
|
||||
/// 5) Persist the tmp file replacing the old update file.
|
||||
fn convert_update_files(db_path: &Path) -> anyhow::Result<()> {
|
||||
let update_files_dir_path = db_path.join("update_files");
|
||||
let file_store = FileStore::new(&update_files_dir_path).with_context(|| {
|
||||
format!("while creating file store for update files dir {update_files_dir_path:?}")
|
||||
})?;
|
||||
|
||||
for uuid in file_store.all_uuids().context("while retrieving uuids from file store")? {
|
||||
let uuid = uuid.context("while retrieving uuid from file store")?;
|
||||
let update_file_path = file_store.get_update_path(uuid);
|
||||
let update_file = file_store
|
||||
.get_update(uuid)
|
||||
.with_context(|| format!("while getting update file for uuid {uuid:?}"))?;
|
||||
|
||||
let mut file =
|
||||
NamedTempFile::new_in(&update_files_dir_path).map(BufWriter::new).with_context(
|
||||
|| format!("while creating bufwriter for update file {update_file_path:?}"),
|
||||
)?;
|
||||
|
||||
let reader = DocumentsBatchReader::from_reader(update_file).with_context(|| {
|
||||
format!("while creating documents batch reader for update file {update_file_path:?}")
|
||||
})?;
|
||||
let (mut cursor, index) = reader.into_cursor_and_fields_index();
|
||||
|
||||
while let Some(document) = cursor.next_document().with_context(|| {
|
||||
format!(
|
||||
"while reading documents from batch reader for update file {update_file_path:?}"
|
||||
)
|
||||
})? {
|
||||
let mut json_document = IndexMap::new();
|
||||
for (fid, value) in document {
|
||||
let field_name = index
|
||||
.name(fid)
|
||||
.with_context(|| format!("while getting field name for fid {fid} for update file {update_file_path:?}"))?;
|
||||
let value: &RawValue = serde_json::from_slice(value)?;
|
||||
json_document.insert(field_name, value);
|
||||
}
|
||||
|
||||
serde_json::to_writer(&mut file, &json_document)?;
|
||||
}
|
||||
|
||||
let file = file.into_inner().map_err(|e| e.into_error()).context(format!(
|
||||
"while flushing update file bufwriter for update file {update_file_path:?}"
|
||||
))?;
|
||||
let _ = file
|
||||
// atomically replace the obkv file with the rewritten NDJSON file
|
||||
.persist(&update_file_path)
|
||||
.with_context(|| format!("while persisting update file {update_file_path:?}"))?;
|
||||
}
|
||||
|
||||
Ok(())
|
||||
}
|
||||
@@ -91,8 +91,8 @@ ureq = { version = "2.10.0", features = ["json"] }
|
||||
url = "2.5.2"
|
||||
rayon-par-bridge = "0.1.0"
|
||||
hashbrown = "0.15.0"
|
||||
raw-collections = { git = "https://github.com/meilisearch/raw-collections.git", version = "0.1.0" }
|
||||
bumpalo = "3.16.0"
|
||||
bumparaw-collections = "0.1.2"
|
||||
thread_local = "1.1.8"
|
||||
allocator-api2 = "0.2.18"
|
||||
rustc-hash = "2.0.0"
|
||||
|
||||
@@ -280,7 +280,7 @@ fn starts_with(selector: &str, key: &str) -> bool {
|
||||
|
||||
pub fn validate_document_id_str(document_id: &str) -> Option<&str> {
|
||||
if document_id.is_empty()
|
||||
|| document_id.len() >= 512
|
||||
|| document_id.len() > 512
|
||||
|| !document_id.chars().all(|c| c.is_ascii_alphanumeric() || c == '-' || c == '_')
|
||||
{
|
||||
None
|
||||
|
||||
@@ -114,7 +114,7 @@ pub enum UserError {
|
||||
"Document identifier `{}` is invalid. \
|
||||
A document identifier can be of type integer or string, \
|
||||
only composed of alphanumeric characters (a-z A-Z 0-9), hyphens (-) and underscores (_), \
|
||||
and can not be more than 511 bytes.", .document_id.to_string()
|
||||
and can not be more than 512 bytes.", .document_id.to_string()
|
||||
)]
|
||||
InvalidDocumentId { document_id: Value },
|
||||
#[error("Invalid facet distribution, {}", format_invalid_filter_distribution(.invalid_facets_name, .valid_facets_name))]
|
||||
|
||||
@@ -1734,7 +1734,6 @@ pub(crate) mod tests {
|
||||
|
||||
use crate::error::{Error, InternalError};
|
||||
use crate::index::{DEFAULT_MIN_WORD_LEN_ONE_TYPO, DEFAULT_MIN_WORD_LEN_TWO_TYPOS};
|
||||
use crate::progress::Progress;
|
||||
use crate::update::new::indexer;
|
||||
use crate::update::settings::InnerIndexSettings;
|
||||
use crate::update::{
|
||||
@@ -1811,7 +1810,7 @@ pub(crate) mod tests {
|
||||
None,
|
||||
&mut new_fields_ids_map,
|
||||
&|| false,
|
||||
Progress::default(),
|
||||
&|_progress| (),
|
||||
)?;
|
||||
|
||||
if let Some(error) = operation_stats.into_iter().find_map(|stat| stat.error) {
|
||||
@@ -1830,7 +1829,7 @@ pub(crate) mod tests {
|
||||
&document_changes,
|
||||
embedders,
|
||||
&|| false,
|
||||
&Progress::default(),
|
||||
&|_| (),
|
||||
)
|
||||
})
|
||||
.unwrap()?;
|
||||
@@ -1902,7 +1901,7 @@ pub(crate) mod tests {
|
||||
None,
|
||||
&mut new_fields_ids_map,
|
||||
&|| false,
|
||||
Progress::default(),
|
||||
&|_progress| (),
|
||||
)?;
|
||||
|
||||
if let Some(error) = operation_stats.into_iter().find_map(|stat| stat.error) {
|
||||
@@ -1921,7 +1920,7 @@ pub(crate) mod tests {
|
||||
&document_changes,
|
||||
embedders,
|
||||
&|| false,
|
||||
&Progress::default(),
|
||||
&|_| (),
|
||||
)
|
||||
})
|
||||
.unwrap()?;
|
||||
@@ -1983,7 +1982,7 @@ pub(crate) mod tests {
|
||||
None,
|
||||
&mut new_fields_ids_map,
|
||||
&|| false,
|
||||
Progress::default(),
|
||||
&|_progress| (),
|
||||
)
|
||||
.unwrap();
|
||||
|
||||
@@ -2002,7 +2001,7 @@ pub(crate) mod tests {
|
||||
&document_changes,
|
||||
embedders,
|
||||
&|| should_abort.load(Relaxed),
|
||||
&Progress::default(),
|
||||
&|_| (),
|
||||
)
|
||||
})
|
||||
.unwrap()
|
||||
|
||||
@@ -31,7 +31,6 @@ pub mod vector;
|
||||
#[macro_use]
|
||||
pub mod snapshot_tests;
|
||||
mod fieldids_weights_map;
|
||||
pub mod progress;
|
||||
|
||||
use std::collections::{BTreeMap, HashMap};
|
||||
use std::convert::{TryFrom, TryInto};
|
||||
|
||||
@@ -1,152 +0,0 @@
|
||||
use std::any::TypeId;
|
||||
use std::borrow::Cow;
|
||||
use std::sync::atomic::{AtomicU32, Ordering};
|
||||
use std::sync::{Arc, RwLock};
|
||||
|
||||
use serde::Serialize;
|
||||
|
||||
pub trait Step: 'static + Send + Sync {
|
||||
fn name(&self) -> Cow<'static, str>;
|
||||
fn current(&self) -> u32;
|
||||
fn total(&self) -> u32;
|
||||
}
|
||||
|
||||
#[derive(Clone, Default)]
|
||||
pub struct Progress {
|
||||
steps: Arc<RwLock<Vec<(TypeId, Box<dyn Step>)>>>,
|
||||
}
|
||||
|
||||
impl Progress {
|
||||
pub fn update_progress<P: Step>(&self, sub_progress: P) {
|
||||
let mut steps = self.steps.write().unwrap();
|
||||
let step_type = TypeId::of::<P>();
|
||||
if let Some(idx) = steps.iter().position(|(id, _)| *id == step_type) {
|
||||
steps.truncate(idx);
|
||||
}
|
||||
steps.push((step_type, Box::new(sub_progress)));
|
||||
}
|
||||
|
||||
// TODO: This code should be in meilisearch_types but cannot because milli can't depend on meilisearch_types
|
||||
pub fn as_progress_view(&self) -> ProgressView {
|
||||
let steps = self.steps.read().unwrap();
|
||||
|
||||
let mut percentage = 0.0;
|
||||
let mut prev_factors = 1.0;
|
||||
|
||||
let mut step_view = Vec::with_capacity(steps.len());
|
||||
for (_, step) in steps.iter() {
|
||||
prev_factors *= step.total() as f32;
|
||||
percentage += step.current() as f32 / prev_factors;
|
||||
|
||||
step_view.push(ProgressStepView {
|
||||
current_step: step.name(),
|
||||
finished: step.current(),
|
||||
total: step.total(),
|
||||
});
|
||||
}
|
||||
|
||||
ProgressView { steps: step_view, percentage: percentage * 100.0 }
|
||||
}
|
||||
}
|
||||
|
||||
/// This trait lets you use the AtomicSubStep defined right below.
|
||||
/// The name must be a const that never changed but that can't be enforced by the type system because it make the trait non object-safe.
|
||||
/// By forcing the Default trait + the &'static str we make it harder to miss-use the trait.
|
||||
pub trait NamedStep: 'static + Send + Sync + Default {
|
||||
fn name(&self) -> &'static str;
|
||||
}
|
||||
|
||||
/// Structure to quickly define steps that need very quick, lockless updating of their current step.
|
||||
/// You can use this struct if:
|
||||
/// - The name of the step doesn't change
|
||||
/// - The total number of steps doesn't change
|
||||
pub struct AtomicSubStep<Name: NamedStep> {
|
||||
unit_name: Name,
|
||||
current: Arc<AtomicU32>,
|
||||
total: u32,
|
||||
}
|
||||
|
||||
impl<Name: NamedStep> AtomicSubStep<Name> {
|
||||
pub fn new(total: u32) -> (Arc<AtomicU32>, Self) {
|
||||
let current = Arc::new(AtomicU32::new(0));
|
||||
(current.clone(), Self { current, total, unit_name: Name::default() })
|
||||
}
|
||||
}
|
||||
|
||||
impl<Name: NamedStep> Step for AtomicSubStep<Name> {
|
||||
fn name(&self) -> Cow<'static, str> {
|
||||
self.unit_name.name().into()
|
||||
}
|
||||
|
||||
fn current(&self) -> u32 {
|
||||
self.current.load(Ordering::Relaxed)
|
||||
}
|
||||
|
||||
fn total(&self) -> u32 {
|
||||
self.total
|
||||
}
|
||||
}
|
||||
|
||||
#[macro_export]
|
||||
macro_rules! make_enum_progress {
|
||||
($visibility:vis enum $name:ident { $($variant:ident,)+ }) => {
|
||||
#[repr(u8)]
|
||||
#[derive(Debug, Clone, Copy, PartialEq, Eq, Sequence)]
|
||||
#[allow(clippy::enum_variant_names)]
|
||||
$visibility enum $name {
|
||||
$($variant),+
|
||||
}
|
||||
|
||||
impl Step for $name {
|
||||
fn name(&self) -> Cow<'static, str> {
|
||||
use convert_case::Casing;
|
||||
|
||||
match self {
|
||||
$(
|
||||
$name::$variant => stringify!($variant).from_case(convert_case::Case::Camel).to_case(convert_case::Case::Lower).into()
|
||||
),+
|
||||
}
|
||||
}
|
||||
|
||||
fn current(&self) -> u32 {
|
||||
*self as u32
|
||||
}
|
||||
|
||||
fn total(&self) -> u32 {
|
||||
Self::CARDINALITY as u32
|
||||
}
|
||||
}
|
||||
};
|
||||
}
|
||||
|
||||
#[macro_export]
|
||||
macro_rules! make_atomic_progress {
|
||||
($struct_name:ident alias $atomic_struct_name:ident => $step_name:literal) => {
|
||||
#[derive(Default, Debug, Clone, Copy)]
|
||||
pub struct $struct_name {}
|
||||
impl NamedStep for $struct_name {
|
||||
fn name(&self) -> &'static str {
|
||||
$step_name
|
||||
}
|
||||
}
|
||||
pub type $atomic_struct_name = AtomicSubStep<$struct_name>;
|
||||
};
|
||||
}
|
||||
|
||||
make_atomic_progress!(Document alias AtomicDocumentStep => "document" );
|
||||
make_atomic_progress!(Payload alias AtomicPayloadStep => "payload" );
|
||||
|
||||
#[derive(Debug, Serialize, Clone)]
|
||||
#[serde(rename_all = "camelCase")]
|
||||
pub struct ProgressView {
|
||||
pub steps: Vec<ProgressStepView>,
|
||||
pub percentage: f32,
|
||||
}
|
||||
|
||||
#[derive(Debug, Serialize, Clone)]
|
||||
#[serde(rename_all = "camelCase")]
|
||||
pub struct ProgressStepView {
|
||||
pub current_step: Cow<'static, str>,
|
||||
pub finished: u32,
|
||||
pub total: u32,
|
||||
}
|
||||
@@ -3,13 +3,12 @@ use std::collections::BTreeMap;
|
||||
use std::fmt::{self, Debug};
|
||||
|
||||
use bumpalo::Bump;
|
||||
use bumparaw_collections::{RawMap, RawVec, Value};
|
||||
use liquid::model::{
|
||||
ArrayView, DisplayCow, KString, KStringCow, ObjectRender, ObjectSource, ScalarCow, State,
|
||||
Value as LiquidValue,
|
||||
};
|
||||
use liquid::{ObjectView, ValueView};
|
||||
use rustc_hash::FxBuildHasher;
|
||||
use raw_collections::{RawMap, RawVec};
|
||||
use serde_json::value::RawValue;
|
||||
|
||||
use crate::update::del_add::{DelAdd, KvReaderDelAdd};
|
||||
@@ -196,7 +195,7 @@ impl<'doc, D: DocumentTrait<'doc> + Debug> ObjectView for ParseableDocument<'doc
|
||||
}
|
||||
|
||||
impl<'doc, D: DocumentTrait<'doc> + Debug> ValueView for ParseableDocument<'doc, D> {
|
||||
fn as_debug(&self) -> &dyn Debug {
|
||||
fn as_debug(&self) -> &dyn fmt::Debug {
|
||||
self
|
||||
}
|
||||
fn render(&self) -> liquid::model::DisplayCow<'_> {
|
||||
@@ -244,13 +243,14 @@ impl<'doc, D: DocumentTrait<'doc> + Debug> ValueView for ParseableDocument<'doc,
|
||||
}
|
||||
}
|
||||
|
||||
#[derive(Debug)]
|
||||
struct ParseableValue<'doc> {
|
||||
value: Value<'doc, FxBuildHasher>,
|
||||
value: raw_collections::Value<'doc>,
|
||||
}
|
||||
|
||||
impl<'doc> ParseableValue<'doc> {
|
||||
pub fn new(value: &'doc RawValue, doc_alloc: &'doc Bump) -> Self {
|
||||
let value = Value::from_raw_value_and_hasher(value, FxBuildHasher, doc_alloc).unwrap();
|
||||
let value = raw_collections::Value::from_raw_value(value, doc_alloc).unwrap();
|
||||
Self { value }
|
||||
}
|
||||
|
||||
@@ -260,19 +260,19 @@ impl<'doc> ParseableValue<'doc> {
|
||||
}
|
||||
|
||||
// transparent newtype for implementing ValueView
|
||||
#[derive(Debug)]
|
||||
#[repr(transparent)]
|
||||
struct ParseableMap<'doc>(RawMap<'doc, FxBuildHasher>);
|
||||
#[derive(Debug)]
|
||||
struct ParseableMap<'doc>(RawMap<'doc>);
|
||||
|
||||
// transparent newtype for implementing ValueView
|
||||
#[derive(Debug)]
|
||||
#[repr(transparent)]
|
||||
#[derive(Debug)]
|
||||
struct ParseableArray<'doc>(RawVec<'doc>);
|
||||
|
||||
impl<'doc> ParseableMap<'doc> {
|
||||
pub fn as_parseable<'a>(map: &'a RawMap<'doc, FxBuildHasher>) -> &'a ParseableMap<'doc> {
|
||||
pub fn as_parseable<'a>(map: &'a RawMap<'doc>) -> &'a ParseableMap<'doc> {
|
||||
// SAFETY: repr(transparent)
|
||||
unsafe { &*(map as *const RawMap<FxBuildHasher> as *const Self) }
|
||||
unsafe { &*(map as *const RawMap as *const Self) }
|
||||
}
|
||||
}
|
||||
|
||||
@@ -447,9 +447,8 @@ impl<'doc> ValueView for ParseableValue<'doc> {
|
||||
}
|
||||
|
||||
fn render(&self) -> DisplayCow<'_> {
|
||||
use bumparaw_collections::value::Number;
|
||||
use bumparaw_collections::Value;
|
||||
|
||||
use raw_collections::value::Number;
|
||||
use raw_collections::Value;
|
||||
match &self.value {
|
||||
Value::Null => LiquidValue::Nil.render(),
|
||||
Value::Bool(v) => v.render(),
|
||||
@@ -465,9 +464,8 @@ impl<'doc> ValueView for ParseableValue<'doc> {
|
||||
}
|
||||
|
||||
fn source(&self) -> DisplayCow<'_> {
|
||||
use bumparaw_collections::value::Number;
|
||||
use bumparaw_collections::Value;
|
||||
|
||||
use raw_collections::value::Number;
|
||||
use raw_collections::Value;
|
||||
match &self.value {
|
||||
Value::Null => LiquidValue::Nil.source(),
|
||||
Value::Bool(v) => ValueView::source(v),
|
||||
@@ -483,9 +481,8 @@ impl<'doc> ValueView for ParseableValue<'doc> {
|
||||
}
|
||||
|
||||
fn type_name(&self) -> &'static str {
|
||||
use bumparaw_collections::value::Number;
|
||||
use bumparaw_collections::Value;
|
||||
|
||||
use raw_collections::value::Number;
|
||||
use raw_collections::Value;
|
||||
match &self.value {
|
||||
Value::Null => LiquidValue::Nil.type_name(),
|
||||
Value::Bool(v) => v.type_name(),
|
||||
@@ -501,8 +498,7 @@ impl<'doc> ValueView for ParseableValue<'doc> {
|
||||
}
|
||||
|
||||
fn query_state(&self, state: State) -> bool {
|
||||
use bumparaw_collections::Value;
|
||||
|
||||
use raw_collections::Value;
|
||||
match &self.value {
|
||||
Value::Null => ValueView::query_state(&LiquidValue::Nil, state),
|
||||
Value::Bool(v) => ValueView::query_state(v, state),
|
||||
@@ -519,8 +515,7 @@ impl<'doc> ValueView for ParseableValue<'doc> {
|
||||
}
|
||||
|
||||
fn to_kstr(&self) -> KStringCow<'_> {
|
||||
use bumparaw_collections::Value;
|
||||
|
||||
use raw_collections::Value;
|
||||
match &self.value {
|
||||
Value::Null => ValueView::to_kstr(&LiquidValue::Nil),
|
||||
Value::Bool(v) => ValueView::to_kstr(v),
|
||||
@@ -532,14 +527,12 @@ impl<'doc> ValueView for ParseableValue<'doc> {
|
||||
}
|
||||
|
||||
fn to_value(&self) -> LiquidValue {
|
||||
use bumparaw_collections::value::Number;
|
||||
use bumparaw_collections::Value;
|
||||
|
||||
use raw_collections::Value;
|
||||
match &self.value {
|
||||
Value::Null => LiquidValue::Nil,
|
||||
Value::Bool(v) => LiquidValue::Scalar(liquid::model::ScalarCow::new(*v)),
|
||||
Value::Number(number) => match number {
|
||||
Number::PosInt(number) => {
|
||||
raw_collections::value::Number::PosInt(number) => {
|
||||
let number: i64 = match (*number).try_into() {
|
||||
Ok(number) => number,
|
||||
Err(_) => {
|
||||
@@ -548,8 +541,12 @@ impl<'doc> ValueView for ParseableValue<'doc> {
|
||||
};
|
||||
LiquidValue::Scalar(ScalarCow::new(number))
|
||||
}
|
||||
Number::NegInt(number) => LiquidValue::Scalar(ScalarCow::new(*number)),
|
||||
Number::Finite(number) => LiquidValue::Scalar(ScalarCow::new(*number)),
|
||||
raw_collections::value::Number::NegInt(number) => {
|
||||
LiquidValue::Scalar(ScalarCow::new(*number))
|
||||
}
|
||||
raw_collections::value::Number::Finite(number) => {
|
||||
LiquidValue::Scalar(ScalarCow::new(*number))
|
||||
}
|
||||
},
|
||||
Value::String(s) => LiquidValue::Scalar(liquid::model::ScalarCow::new(s.to_string())),
|
||||
Value::Array(raw_vec) => ParseableArray::as_parseable(raw_vec).to_value(),
|
||||
@@ -558,9 +555,8 @@ impl<'doc> ValueView for ParseableValue<'doc> {
|
||||
}
|
||||
|
||||
fn as_scalar(&self) -> Option<liquid::model::ScalarCow<'_>> {
|
||||
use bumparaw_collections::value::Number;
|
||||
use bumparaw_collections::Value;
|
||||
|
||||
use raw_collections::value::Number;
|
||||
use raw_collections::Value;
|
||||
match &self.value {
|
||||
Value::Bool(v) => Some(liquid::model::ScalarCow::new(*v)),
|
||||
Value::Number(number) => match number {
|
||||
@@ -580,41 +576,34 @@ impl<'doc> ValueView for ParseableValue<'doc> {
|
||||
}
|
||||
|
||||
fn is_scalar(&self) -> bool {
|
||||
use bumparaw_collections::Value;
|
||||
|
||||
use raw_collections::Value;
|
||||
matches!(&self.value, Value::Bool(_) | Value::Number(_) | Value::String(_))
|
||||
}
|
||||
|
||||
fn as_array(&self) -> Option<&dyn liquid::model::ArrayView> {
|
||||
if let Value::Array(array) = &self.value {
|
||||
if let raw_collections::Value::Array(array) = &self.value {
|
||||
return Some(ParseableArray::as_parseable(array) as _);
|
||||
}
|
||||
None
|
||||
}
|
||||
|
||||
fn is_array(&self) -> bool {
|
||||
matches!(&self.value, bumparaw_collections::Value::Array(_))
|
||||
matches!(&self.value, raw_collections::Value::Array(_))
|
||||
}
|
||||
|
||||
fn as_object(&self) -> Option<&dyn ObjectView> {
|
||||
if let Value::Object(object) = &self.value {
|
||||
if let raw_collections::Value::Object(object) = &self.value {
|
||||
return Some(ParseableMap::as_parseable(object) as _);
|
||||
}
|
||||
None
|
||||
}
|
||||
|
||||
fn is_object(&self) -> bool {
|
||||
matches!(&self.value, bumparaw_collections::Value::Object(_))
|
||||
matches!(&self.value, raw_collections::Value::Object(_))
|
||||
}
|
||||
|
||||
fn is_nil(&self) -> bool {
|
||||
matches!(&self.value, bumparaw_collections::Value::Null)
|
||||
}
|
||||
}
|
||||
|
||||
impl Debug for ParseableValue<'_> {
|
||||
fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
|
||||
f.debug_struct("ParseableValue").field("value", &self.value).finish()
|
||||
matches!(&self.value, raw_collections::Value::Null)
|
||||
}
|
||||
}
|
||||
|
||||
|
||||
@@ -38,16 +38,6 @@ pub struct RenderPromptError {
|
||||
pub fault: FaultSource,
|
||||
}
|
||||
impl RenderPromptError {
|
||||
pub(crate) fn missing_context_with_external_docid(
|
||||
external_docid: String,
|
||||
inner: liquid::Error,
|
||||
) -> RenderPromptError {
|
||||
Self {
|
||||
kind: RenderPromptErrorKind::MissingContextWithExternalDocid(external_docid, inner),
|
||||
fault: FaultSource::User,
|
||||
}
|
||||
}
|
||||
|
||||
pub(crate) fn missing_context(inner: liquid::Error) -> RenderPromptError {
|
||||
Self { kind: RenderPromptErrorKind::MissingContext(inner), fault: FaultSource::User }
|
||||
}
|
||||
@@ -57,8 +47,6 @@ impl RenderPromptError {
|
||||
pub enum RenderPromptErrorKind {
|
||||
#[error("missing field in document: {0}")]
|
||||
MissingContext(liquid::Error),
|
||||
#[error("missing field in document `{0}`: {1}")]
|
||||
MissingContextWithExternalDocid(String, liquid::Error),
|
||||
}
|
||||
|
||||
impl From<RenderPromptError> for crate::Error {
|
||||
|
||||
@@ -119,7 +119,6 @@ impl Prompt {
|
||||
'doc: 'a, // lifetime of the allocator, will live for an entire chunk of documents
|
||||
>(
|
||||
&self,
|
||||
external_docid: &str,
|
||||
document: impl crate::update::new::document::Document<'a> + Debug,
|
||||
field_id_map: &RefCell<GlobalFieldsIdsMap>,
|
||||
doc_alloc: &'doc Bump,
|
||||
@@ -131,12 +130,9 @@ impl Prompt {
|
||||
self.max_bytes.unwrap_or_else(default_max_bytes).get(),
|
||||
doc_alloc,
|
||||
);
|
||||
self.template.render_to(&mut rendered, &context).map_err(|liquid_error| {
|
||||
RenderPromptError::missing_context_with_external_docid(
|
||||
external_docid.to_owned(),
|
||||
liquid_error,
|
||||
)
|
||||
})?;
|
||||
self.template
|
||||
.render_to(&mut rendered, &context)
|
||||
.map_err(RenderPromptError::missing_context)?;
|
||||
Ok(std::str::from_utf8(rendered.into_bump_slice())
|
||||
.expect("render can only write UTF-8 because all inputs and processing preserve utf-8"))
|
||||
}
|
||||
|
||||
@@ -5,7 +5,6 @@ use bumpalo::Bump;
|
||||
use heed::EnvOpenOptions;
|
||||
use maplit::{btreemap, hashset};
|
||||
|
||||
use crate::progress::Progress;
|
||||
use crate::update::new::indexer;
|
||||
use crate::update::{IndexDocumentsMethod, IndexerConfig, Settings};
|
||||
use crate::vector::EmbeddingConfigs;
|
||||
@@ -73,7 +72,7 @@ pub fn setup_search_index_with_criteria(criteria: &[Criterion]) -> Index {
|
||||
None,
|
||||
&mut new_fields_ids_map,
|
||||
&|| false,
|
||||
Progress::default(),
|
||||
&|_progress| (),
|
||||
)
|
||||
.unwrap();
|
||||
|
||||
@@ -92,7 +91,7 @@ pub fn setup_search_index_with_criteria(criteria: &[Criterion]) -> Index {
|
||||
&document_changes,
|
||||
embedders,
|
||||
&|| false,
|
||||
&Progress::default(),
|
||||
&|_| (),
|
||||
)
|
||||
.unwrap();
|
||||
|
||||
|
||||
@@ -79,29 +79,22 @@ 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, OrderedF64Codec,
|
||||
};
|
||||
use crate::heed_codec::facet::{FacetGroupKey, FacetGroupKeyCodec, FacetGroupValueCodec};
|
||||
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.
|
||||
///
|
||||
@@ -653,194 +646,3 @@ 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, found_level={}, expected_level={}",
|
||||
parent_facet_value.level, current_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:?}"),
|
||||
},
|
||||
}
|
||||
}
|
||||
|
||||
@@ -1,498 +0,0 @@
|
||||
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::update::valid_facet_value;
|
||||
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)
|
||||
// sort in **reverse** lexicographic order
|
||||
.reverse()
|
||||
},
|
||||
);
|
||||
|
||||
self.inner.find_changed_parents(wtxn, self.delta_data)?;
|
||||
|
||||
self.inner.add_or_delete_level(wtxn)
|
||||
}
|
||||
}
|
||||
|
||||
impl FacetsUpdateIncrementalInner {
|
||||
/// WARNING: `changed_children` must be sorted in **reverse** lexicographic order.
|
||||
fn find_changed_parents(
|
||||
&self,
|
||||
wtxn: &mut RwTxn,
|
||||
mut changed_children: Vec<FacetFieldIdChange>,
|
||||
) -> Result<()> {
|
||||
let mut changed_parents = vec![];
|
||||
for child_level in 0u8..u8::MAX {
|
||||
// child_level < u8::MAX by construction
|
||||
let parent_level = child_level + 1;
|
||||
let parent_level_left_bound: FacetGroupKey<&[u8]> =
|
||||
FacetGroupKey { field_id: self.field_id, level: parent_level, left_bound: &[] };
|
||||
|
||||
let mut last_parent: Option<Box<[u8]>> = None;
|
||||
let mut child_it = changed_children
|
||||
// drain all changed children
|
||||
.drain(..)
|
||||
// keep only children whose value is valid in the LMDB sense
|
||||
.filter(|child| valid_facet_value(&child.facet_value));
|
||||
// `while let` rather than `for` because we advance `child_it` inside of the loop
|
||||
'current_level: while let Some(child) = child_it.next() {
|
||||
if let Some(last_parent) = &last_parent {
|
||||
if &child.facet_value >= last_parent {
|
||||
self.compute_parent_group(wtxn, child_level, child.facet_value)?;
|
||||
continue 'current_level;
|
||||
}
|
||||
}
|
||||
|
||||
// need to find a new parent
|
||||
let parent_key_prefix = FacetGroupKey {
|
||||
field_id: self.field_id,
|
||||
level: parent_level,
|
||||
left_bound: &*child.facet_value,
|
||||
};
|
||||
|
||||
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().into_boxed_slice());
|
||||
|
||||
// add to modified list for parent level
|
||||
changed_parents.push(FacetFieldIdChange {
|
||||
facet_value: parent_key.left_bound.to_owned().into_boxed_slice(),
|
||||
});
|
||||
self.compute_parent_group(wtxn, child_level, child.facet_value)?;
|
||||
}
|
||||
Some(Err(err)) => return Err(err.into()),
|
||||
None => {
|
||||
// no parent for that key
|
||||
let mut parent_it = self
|
||||
.db
|
||||
.remap_data_type::<DecodeIgnore>()
|
||||
.prefix_iter_mut(wtxn, &parent_level_left_bound)?;
|
||||
match parent_it.next() {
|
||||
// 1. left of the current left bound, or
|
||||
Some(Ok((first_key, _first_value))) => {
|
||||
// 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
|
||||
{
|
||||
// max level reached, exit
|
||||
drop(parent_it);
|
||||
self.compute_parent_group(
|
||||
wtxn,
|
||||
child_level,
|
||||
child.facet_value,
|
||||
)?;
|
||||
for child in child_it.by_ref() {
|
||||
self.compute_parent_group(
|
||||
wtxn,
|
||||
child_level,
|
||||
child.facet_value,
|
||||
)?;
|
||||
}
|
||||
return Ok(());
|
||||
}
|
||||
// remove old left bound
|
||||
unsafe { parent_it.del_current()? };
|
||||
drop(parent_it);
|
||||
changed_parents.push(FacetFieldIdChange {
|
||||
facet_value: child.facet_value.clone(),
|
||||
});
|
||||
self.compute_parent_group(wtxn, child_level, child.facet_value)?;
|
||||
// pop all elements in order to visit the new left bound
|
||||
let new_left_bound =
|
||||
&mut changed_parents.last_mut().unwrap().facet_value;
|
||||
for child in child_it.by_ref() {
|
||||
new_left_bound.clone_from(&child.facet_value);
|
||||
|
||||
self.compute_parent_group(
|
||||
wtxn,
|
||||
child_level,
|
||||
child.facet_value,
|
||||
)?;
|
||||
}
|
||||
}
|
||||
Some(Err(err)) => return Err(err.into()),
|
||||
// 2. max level reached, exit
|
||||
None => {
|
||||
drop(parent_it);
|
||||
self.compute_parent_group(wtxn, child_level, child.facet_value)?;
|
||||
for child in child_it.by_ref() {
|
||||
self.compute_parent_group(
|
||||
wtxn,
|
||||
child_level,
|
||||
child.facet_value,
|
||||
)?;
|
||||
}
|
||||
return Ok(());
|
||||
}
|
||||
}
|
||||
}
|
||||
}
|
||||
}
|
||||
if changed_parents.is_empty() {
|
||||
return Ok(());
|
||||
}
|
||||
drop(child_it);
|
||||
std::mem::swap(&mut changed_children, &mut changed_parents);
|
||||
// changed_parents is now empty because changed_children was emptied by the drain
|
||||
}
|
||||
Ok(())
|
||||
}
|
||||
|
||||
fn compute_parent_group(
|
||||
&self,
|
||||
wtxn: &mut RwTxn<'_>,
|
||||
parent_level: u8,
|
||||
parent_left_bound: Box<[u8]>,
|
||||
) -> Result<()> {
|
||||
let mut range_left_bound: Vec<u8> = parent_left_bound.into();
|
||||
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,
|
||||
};
|
||||
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,
|
||||
};
|
||||
let mut child_left_bound = Bound::Included(child_left_key);
|
||||
|
||||
loop {
|
||||
// do a first pass on the range to find the number of children
|
||||
let child_count = self
|
||||
.db
|
||||
.remap_data_type::<DecodeIgnore>()
|
||||
.range(wtxn, &(child_left_bound, child_right_bound))?
|
||||
.take(self.max_group_size as usize * 2)
|
||||
.count();
|
||||
let mut child_it = self.db.range(wtxn, &(child_left_bound, child_right_bound))?;
|
||||
|
||||
// pick the right group_size depending on the number of children
|
||||
let group_size = if child_count >= self.max_group_size as usize * 2 {
|
||||
// more than twice the max_group_size => there will be space for at least 2 groups of max_group_size
|
||||
self.max_group_size as usize
|
||||
} else if child_count >= self.group_size as usize {
|
||||
// size in [group_size, max_group_size * 2[
|
||||
// divided by 2 it is between [group_size / 2, max_group_size[
|
||||
// this ensures that the tree is balanced
|
||||
child_count / 2
|
||||
} else {
|
||||
// take everything
|
||||
child_count
|
||||
};
|
||||
|
||||
let res: Result<_> = child_it
|
||||
.by_ref()
|
||||
.take(group_size)
|
||||
// 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,
|
||||
};
|
||||
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]>,
|
||||
}
|
||||
@@ -10,14 +10,10 @@ use fst::{IntoStreamer, Streamer};
|
||||
pub use grenad_helpers::*;
|
||||
pub use merge_functions::*;
|
||||
|
||||
use crate::MAX_LMDB_KEY_LENGTH;
|
||||
use crate::MAX_WORD_LENGTH;
|
||||
|
||||
pub fn valid_lmdb_key(key: impl AsRef<[u8]>) -> bool {
|
||||
key.as_ref().len() <= MAX_LMDB_KEY_LENGTH - 3 && !key.as_ref().is_empty()
|
||||
}
|
||||
|
||||
pub fn valid_facet_value(facet_value: impl AsRef<[u8]>) -> bool {
|
||||
facet_value.as_ref().len() <= MAX_LMDB_KEY_LENGTH - 3 && !facet_value.as_ref().is_empty()
|
||||
key.as_ref().len() <= MAX_WORD_LENGTH * 2 && !key.as_ref().is_empty()
|
||||
}
|
||||
|
||||
/// Divides one slice into two at an index, returns `None` if mid is out of bounds.
|
||||
|
||||
@@ -766,7 +766,6 @@ mod tests {
|
||||
use crate::documents::mmap_from_objects;
|
||||
use crate::index::tests::TempIndex;
|
||||
use crate::index::IndexEmbeddingConfig;
|
||||
use crate::progress::Progress;
|
||||
use crate::search::TermsMatchingStrategy;
|
||||
use crate::update::new::indexer;
|
||||
use crate::update::Setting;
|
||||
@@ -1965,7 +1964,7 @@ mod tests {
|
||||
None,
|
||||
&mut new_fields_ids_map,
|
||||
&|| false,
|
||||
Progress::default(),
|
||||
&|_progress| (),
|
||||
)
|
||||
.unwrap();
|
||||
|
||||
@@ -2149,7 +2148,7 @@ mod tests {
|
||||
None,
|
||||
&mut new_fields_ids_map,
|
||||
&|| false,
|
||||
Progress::default(),
|
||||
&|_progress| (),
|
||||
)
|
||||
.unwrap();
|
||||
|
||||
@@ -2164,7 +2163,7 @@ mod tests {
|
||||
&document_changes,
|
||||
embedders,
|
||||
&|| false,
|
||||
&Progress::default(),
|
||||
&|_| (),
|
||||
)
|
||||
.unwrap();
|
||||
wtxn.commit().unwrap();
|
||||
@@ -2211,7 +2210,7 @@ mod tests {
|
||||
None,
|
||||
&mut new_fields_ids_map,
|
||||
&|| false,
|
||||
Progress::default(),
|
||||
&|_progress| (),
|
||||
)
|
||||
.unwrap();
|
||||
|
||||
@@ -2226,7 +2225,7 @@ mod tests {
|
||||
&document_changes,
|
||||
embedders,
|
||||
&|| false,
|
||||
&Progress::default(),
|
||||
&|_| (),
|
||||
)
|
||||
.unwrap();
|
||||
wtxn.commit().unwrap();
|
||||
@@ -2264,7 +2263,7 @@ mod tests {
|
||||
None,
|
||||
&mut new_fields_ids_map,
|
||||
&|| false,
|
||||
Progress::default(),
|
||||
&|_progress| (),
|
||||
)
|
||||
.unwrap();
|
||||
|
||||
@@ -2279,7 +2278,7 @@ mod tests {
|
||||
&document_changes,
|
||||
embedders,
|
||||
&|| false,
|
||||
&Progress::default(),
|
||||
&|_| (),
|
||||
)
|
||||
.unwrap();
|
||||
wtxn.commit().unwrap();
|
||||
@@ -2316,7 +2315,7 @@ mod tests {
|
||||
None,
|
||||
&mut new_fields_ids_map,
|
||||
&|| false,
|
||||
Progress::default(),
|
||||
&|_progress| (),
|
||||
)
|
||||
.unwrap();
|
||||
|
||||
@@ -2331,7 +2330,7 @@ mod tests {
|
||||
&document_changes,
|
||||
embedders,
|
||||
&|| false,
|
||||
&Progress::default(),
|
||||
&|_| (),
|
||||
)
|
||||
.unwrap();
|
||||
wtxn.commit().unwrap();
|
||||
@@ -2370,7 +2369,7 @@ mod tests {
|
||||
None,
|
||||
&mut new_fields_ids_map,
|
||||
&|| false,
|
||||
Progress::default(),
|
||||
&|_progress| (),
|
||||
)
|
||||
.unwrap();
|
||||
|
||||
@@ -2385,7 +2384,7 @@ mod tests {
|
||||
&document_changes,
|
||||
embedders,
|
||||
&|| false,
|
||||
&Progress::default(),
|
||||
&|_| (),
|
||||
)
|
||||
.unwrap();
|
||||
wtxn.commit().unwrap();
|
||||
@@ -2429,7 +2428,7 @@ mod tests {
|
||||
None,
|
||||
&mut new_fields_ids_map,
|
||||
&|| false,
|
||||
Progress::default(),
|
||||
&|_progress| (),
|
||||
)
|
||||
.unwrap();
|
||||
|
||||
@@ -2444,7 +2443,7 @@ mod tests {
|
||||
&document_changes,
|
||||
embedders,
|
||||
&|| false,
|
||||
&Progress::default(),
|
||||
&|_| (),
|
||||
)
|
||||
.unwrap();
|
||||
wtxn.commit().unwrap();
|
||||
@@ -2481,7 +2480,7 @@ mod tests {
|
||||
None,
|
||||
&mut new_fields_ids_map,
|
||||
&|| false,
|
||||
Progress::default(),
|
||||
&|_progress| (),
|
||||
)
|
||||
.unwrap();
|
||||
|
||||
@@ -2496,7 +2495,7 @@ mod tests {
|
||||
&document_changes,
|
||||
embedders,
|
||||
&|| false,
|
||||
&Progress::default(),
|
||||
&|_| (),
|
||||
)
|
||||
.unwrap();
|
||||
wtxn.commit().unwrap();
|
||||
@@ -2533,7 +2532,7 @@ mod tests {
|
||||
None,
|
||||
&mut new_fields_ids_map,
|
||||
&|| false,
|
||||
Progress::default(),
|
||||
&|_progress| (),
|
||||
)
|
||||
.unwrap();
|
||||
|
||||
@@ -2548,7 +2547,7 @@ mod tests {
|
||||
&document_changes,
|
||||
embedders,
|
||||
&|| false,
|
||||
&Progress::default(),
|
||||
&|_| (),
|
||||
)
|
||||
.unwrap();
|
||||
wtxn.commit().unwrap();
|
||||
@@ -2727,7 +2726,7 @@ mod tests {
|
||||
None,
|
||||
&mut new_fields_ids_map,
|
||||
&|| false,
|
||||
Progress::default(),
|
||||
&|_progress| (),
|
||||
)
|
||||
.unwrap();
|
||||
|
||||
@@ -2742,7 +2741,7 @@ mod tests {
|
||||
&document_changes,
|
||||
embedders,
|
||||
&|| false,
|
||||
&Progress::default(),
|
||||
&|_| (),
|
||||
)
|
||||
.unwrap();
|
||||
wtxn.commit().unwrap();
|
||||
@@ -2786,7 +2785,7 @@ mod tests {
|
||||
None,
|
||||
&mut new_fields_ids_map,
|
||||
&|| false,
|
||||
Progress::default(),
|
||||
&|_progress| (),
|
||||
)
|
||||
.unwrap();
|
||||
|
||||
@@ -2801,7 +2800,7 @@ mod tests {
|
||||
&document_changes,
|
||||
embedders,
|
||||
&|| false,
|
||||
&Progress::default(),
|
||||
&|_| (),
|
||||
)
|
||||
.unwrap();
|
||||
wtxn.commit().unwrap();
|
||||
@@ -2842,7 +2841,7 @@ mod tests {
|
||||
None,
|
||||
&mut new_fields_ids_map,
|
||||
&|| false,
|
||||
Progress::default(),
|
||||
&|_progress| (),
|
||||
)
|
||||
.unwrap();
|
||||
|
||||
@@ -2857,7 +2856,7 @@ mod tests {
|
||||
&document_changes,
|
||||
embedders,
|
||||
&|| false,
|
||||
&Progress::default(),
|
||||
&|_| (),
|
||||
)
|
||||
.unwrap();
|
||||
wtxn.commit().unwrap();
|
||||
|
||||
@@ -1,5 +1,5 @@
|
||||
---
|
||||
source: crates/milli/src/update/index_documents/mod.rs
|
||||
source: milli/src/update/index_documents/mod.rs
|
||||
---
|
||||
3 0 48.9021 1 [19, ]
|
||||
3 0 49.9314 1 [17, ]
|
||||
@@ -15,11 +15,6 @@ source: crates/milli/src/update/index_documents/mod.rs
|
||||
3 0 50.7453 1 [7, ]
|
||||
3 0 50.8466 1 [10, ]
|
||||
3 0 51.0537 1 [9, ]
|
||||
3 1 48.9021 2 [17, 19, ]
|
||||
3 1 50.1793 3 [13, 14, 15, ]
|
||||
3 1 50.4502 4 [0, 3, 8, 12, ]
|
||||
3 1 50.6312 2 [1, 2, ]
|
||||
3 1 50.7453 3 [7, 9, 10, ]
|
||||
4 0 2.271 1 [17, ]
|
||||
4 0 2.3708 1 [19, ]
|
||||
4 0 2.7637 1 [14, ]
|
||||
@@ -33,3 +28,4 @@ source: crates/milli/src/update/index_documents/mod.rs
|
||||
4 0 3.6957 1 [9, ]
|
||||
4 0 3.9623 1 [12, ]
|
||||
4 0 4.337 1 [10, ]
|
||||
|
||||
|
||||
@@ -21,11 +21,15 @@ use super::ref_cell_ext::RefCellExt;
|
||||
use super::thread_local::{FullySend, ThreadLocal};
|
||||
use super::StdResult;
|
||||
use crate::heed_codec::facet::{FieldDocIdFacetF64Codec, FieldDocIdFacetStringCodec};
|
||||
use crate::heed_codec::StrBEU16Codec;
|
||||
use crate::index::db_name;
|
||||
use crate::index::main_key::{GEO_FACETED_DOCUMENTS_IDS_KEY, GEO_RTREE_KEY};
|
||||
use crate::update::new::KvReaderFieldId;
|
||||
use crate::vector::Embedding;
|
||||
use crate::{CboRoaringBitmapCodec, DocumentId, Error, Index, InternalError};
|
||||
use crate::{
|
||||
CboRoaringBitmapCodec, DocumentId, Error, FieldIdWordCountCodec, Index, InternalError,
|
||||
U8StrStrCodec,
|
||||
};
|
||||
|
||||
/// Creates a tuple of senders/receiver to be used by
|
||||
/// the extractors and the writer loop.
|
||||
@@ -403,6 +407,32 @@ impl Database {
|
||||
Database::FieldIdDocidFacetF64s => db_name::FIELD_ID_DOCID_FACET_F64S,
|
||||
}
|
||||
}
|
||||
|
||||
pub fn stringify_key(&self, key: &[u8]) -> String {
|
||||
use heed::types::*;
|
||||
|
||||
match self {
|
||||
Database::WordDocids => format!("{:?}", Str::bytes_decode(key).unwrap()),
|
||||
Database::WordFidDocids => format!("{:?}", StrBEU16Codec::bytes_decode(key).unwrap()),
|
||||
Database::WordPositionDocids => {
|
||||
format!("{:?}", StrBEU16Codec::bytes_decode(key).unwrap())
|
||||
}
|
||||
Database::WordPairProximityDocids => {
|
||||
format!("{:?}", U8StrStrCodec::bytes_decode(key).unwrap())
|
||||
}
|
||||
Database::ExactWordDocids => format!("{:?}", Str::bytes_decode(key).unwrap()),
|
||||
Database::FidWordCountDocids => {
|
||||
format!("{:?}", FieldIdWordCountCodec::bytes_decode(key).unwrap())
|
||||
}
|
||||
Database::FieldIdDocidFacetStrings => {
|
||||
format!("{:?}", FieldDocIdFacetStringCodec::bytes_decode(key).unwrap())
|
||||
}
|
||||
Database::FieldIdDocidFacetF64s => {
|
||||
format!("{:?}", FieldDocIdFacetF64Codec::bytes_decode(key).unwrap())
|
||||
}
|
||||
d => unimplemented!("stringify_key for {:?}", d),
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
impl From<FacetKind> for Database {
|
||||
|
||||
@@ -1,8 +1,7 @@
|
||||
use std::collections::{BTreeMap, BTreeSet};
|
||||
|
||||
use bumparaw_collections::RawMap;
|
||||
use heed::RoTxn;
|
||||
use rustc_hash::FxBuildHasher;
|
||||
use raw_collections::RawMap;
|
||||
use serde_json::value::RawValue;
|
||||
|
||||
use super::vector_document::VectorDocument;
|
||||
@@ -386,12 +385,12 @@ pub type Entry<'doc> = (&'doc str, &'doc RawValue);
|
||||
|
||||
#[derive(Debug)]
|
||||
pub struct Versions<'doc> {
|
||||
data: RawMap<'doc, FxBuildHasher>,
|
||||
data: RawMap<'doc>,
|
||||
}
|
||||
|
||||
impl<'doc> Versions<'doc> {
|
||||
pub fn multiple(
|
||||
mut versions: impl Iterator<Item = Result<RawMap<'doc, FxBuildHasher>>>,
|
||||
mut versions: impl Iterator<Item = Result<RawMap<'doc>>>,
|
||||
) -> Result<Option<Self>> {
|
||||
let Some(data) = versions.next() else { return Ok(None) };
|
||||
let mut data = data?;
|
||||
@@ -404,7 +403,7 @@ impl<'doc> Versions<'doc> {
|
||||
Ok(Some(Self::single(data)))
|
||||
}
|
||||
|
||||
pub fn single(version: RawMap<'doc, FxBuildHasher>) -> Self {
|
||||
pub fn single(version: RawMap<'doc>) -> Self {
|
||||
Self { data: version }
|
||||
}
|
||||
|
||||
|
||||
@@ -69,12 +69,12 @@ use std::io::BufReader;
|
||||
use std::{io, iter, mem};
|
||||
|
||||
use bumpalo::Bump;
|
||||
use bumparaw_collections::bbbul::{BitPacker, BitPacker4x};
|
||||
use bumparaw_collections::map::FrozenMap;
|
||||
use bumparaw_collections::{Bbbul, FrozenBbbul};
|
||||
use grenad::ReaderCursor;
|
||||
use hashbrown::hash_map::RawEntryMut;
|
||||
use hashbrown::HashMap;
|
||||
use raw_collections::bbbul::{BitPacker, BitPacker4x};
|
||||
use raw_collections::map::FrozenMap;
|
||||
use raw_collections::{Bbbul, FrozenBbbul};
|
||||
use roaring::RoaringBitmap;
|
||||
use rustc_hash::FxBuildHasher;
|
||||
|
||||
@@ -177,12 +177,12 @@ impl<'extractor> BalancedCaches<'extractor> {
|
||||
Ok(())
|
||||
}
|
||||
|
||||
pub fn freeze(&mut self, source_id: usize) -> Result<Vec<FrozenCache<'_, 'extractor>>> {
|
||||
pub fn freeze(&mut self) -> Result<Vec<FrozenCache<'_, 'extractor>>> {
|
||||
match &mut self.caches {
|
||||
InnerCaches::Normal(NormalCaches { caches }) => caches
|
||||
.iter_mut()
|
||||
.enumerate()
|
||||
.map(|(bucket_id, map)| {
|
||||
.map(|(bucket, map)| {
|
||||
// safety: we are transmuting the Bbbul into a FrozenBbbul
|
||||
// that are the same size.
|
||||
let map = unsafe {
|
||||
@@ -201,19 +201,14 @@ impl<'extractor> BalancedCaches<'extractor> {
|
||||
>,
|
||||
>(map)
|
||||
};
|
||||
Ok(FrozenCache {
|
||||
source_id,
|
||||
bucket_id,
|
||||
cache: FrozenMap::new(map),
|
||||
spilled: Vec::new(),
|
||||
})
|
||||
Ok(FrozenCache { bucket, cache: FrozenMap::new(map), spilled: Vec::new() })
|
||||
})
|
||||
.collect(),
|
||||
InnerCaches::Spilling(SpillingCaches { caches, spilled_entries, .. }) => caches
|
||||
.iter_mut()
|
||||
.zip(mem::take(spilled_entries))
|
||||
.enumerate()
|
||||
.map(|(bucket_id, (map, sorter))| {
|
||||
.map(|(bucket, (map, sorter))| {
|
||||
let spilled = sorter
|
||||
.into_reader_cursors()?
|
||||
.into_iter()
|
||||
@@ -239,7 +234,7 @@ impl<'extractor> BalancedCaches<'extractor> {
|
||||
>,
|
||||
>(map)
|
||||
};
|
||||
Ok(FrozenCache { source_id, bucket_id, cache: FrozenMap::new(map), spilled })
|
||||
Ok(FrozenCache { bucket, cache: FrozenMap::new(map), spilled })
|
||||
})
|
||||
.collect(),
|
||||
}
|
||||
@@ -445,8 +440,7 @@ fn spill_entry_to_sorter(
|
||||
}
|
||||
|
||||
pub struct FrozenCache<'a, 'extractor> {
|
||||
bucket_id: usize,
|
||||
source_id: usize,
|
||||
bucket: usize,
|
||||
cache: FrozenMap<
|
||||
'a,
|
||||
'extractor,
|
||||
@@ -463,9 +457,9 @@ pub fn transpose_and_freeze_caches<'a, 'extractor>(
|
||||
let width = caches.first().map(BalancedCaches::buckets).unwrap_or(0);
|
||||
let mut bucket_caches: Vec<_> = iter::repeat_with(Vec::new).take(width).collect();
|
||||
|
||||
for (thread_index, thread_cache) in caches.iter_mut().enumerate() {
|
||||
for frozen in thread_cache.freeze(thread_index)? {
|
||||
bucket_caches[frozen.bucket_id].push(frozen);
|
||||
for thread_cache in caches {
|
||||
for frozen in thread_cache.freeze()? {
|
||||
bucket_caches[frozen.bucket].push(frozen);
|
||||
}
|
||||
}
|
||||
|
||||
@@ -483,16 +477,21 @@ where
|
||||
F: for<'a> FnMut(&'a [u8], DelAddRoaringBitmap) -> Result<()>,
|
||||
{
|
||||
let mut maps = Vec::new();
|
||||
let mut heap = BinaryHeap::new();
|
||||
let mut readers = Vec::new();
|
||||
let mut current_bucket = None;
|
||||
for FrozenCache { source_id, bucket_id, cache, spilled } in frozen {
|
||||
assert_eq!(*current_bucket.get_or_insert(bucket_id), bucket_id);
|
||||
maps.push((source_id, cache));
|
||||
for reader in spilled {
|
||||
let mut cursor = reader.into_cursor()?;
|
||||
if cursor.move_on_next()?.is_some() {
|
||||
heap.push(Entry { cursor, source_id });
|
||||
}
|
||||
for FrozenCache { bucket, cache, ref mut spilled } in frozen {
|
||||
assert_eq!(*current_bucket.get_or_insert(bucket), bucket);
|
||||
maps.push(cache);
|
||||
readers.append(spilled);
|
||||
}
|
||||
|
||||
// First manage the spilled entries by looking into the HashMaps,
|
||||
// merge them and mark them as dummy.
|
||||
let mut heap = BinaryHeap::new();
|
||||
for (source_index, source) in readers.into_iter().enumerate() {
|
||||
let mut cursor = source.into_cursor()?;
|
||||
if cursor.move_on_next()?.is_some() {
|
||||
heap.push(Entry { cursor, source_index });
|
||||
}
|
||||
}
|
||||
|
||||
@@ -509,29 +508,25 @@ where
|
||||
|
||||
let mut output = DelAddRoaringBitmap::from_bytes(first_value)?;
|
||||
while let Some(mut entry) = heap.peek_mut() {
|
||||
if let Some((key, value)) = entry.cursor.current() {
|
||||
if first_key != key {
|
||||
if let Some((key, _value)) = entry.cursor.current() {
|
||||
if first_key == key {
|
||||
let new = DelAddRoaringBitmap::from_bytes(first_value)?;
|
||||
output = output.merge(new);
|
||||
// When we are done we the current value of this entry move make
|
||||
// it move forward and let the heap reorganize itself (on drop)
|
||||
if entry.cursor.move_on_next()?.is_none() {
|
||||
PeekMut::pop(entry);
|
||||
}
|
||||
} else {
|
||||
break;
|
||||
}
|
||||
|
||||
let new = DelAddRoaringBitmap::from_bytes(value)?;
|
||||
output = output.merge(new);
|
||||
// When we are done we the current value of this entry move make
|
||||
// it move forward and let the heap reorganize itself (on drop)
|
||||
if entry.cursor.move_on_next()?.is_none() {
|
||||
PeekMut::pop(entry);
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
// Once we merged all of the spilled bitmaps we must also
|
||||
// fetch the entries from the non-spilled entries (the HashMaps).
|
||||
for (source_id, map) in maps.iter_mut() {
|
||||
debug_assert!(
|
||||
!(map.get(first_key).is_some() && first_entry.source_id == *source_id),
|
||||
"A thread should not have spiled a key that has been inserted in the cache"
|
||||
);
|
||||
if first_entry.source_id != *source_id {
|
||||
for (map_index, map) in maps.iter_mut().enumerate() {
|
||||
if first_entry.source_index != map_index {
|
||||
if let Some(new) = map.get_mut(first_key) {
|
||||
output.union_and_clear_bbbul(new);
|
||||
}
|
||||
@@ -543,12 +538,12 @@ where
|
||||
|
||||
// Don't forget to put the first entry back into the heap.
|
||||
if first_entry.cursor.move_on_next()?.is_some() {
|
||||
heap.push(first_entry);
|
||||
heap.push(first_entry)
|
||||
}
|
||||
}
|
||||
|
||||
// Then manage the content on the HashMap entries that weren't taken (mem::take).
|
||||
while let Some((_, mut map)) = maps.pop() {
|
||||
while let Some(mut map) = maps.pop() {
|
||||
// Make sure we don't try to work with entries already managed by the spilled
|
||||
let mut ordered_entries: Vec<_> =
|
||||
map.iter_mut().filter(|(_, bbbul)| !bbbul.is_empty()).collect();
|
||||
@@ -558,7 +553,7 @@ where
|
||||
let mut output = DelAddRoaringBitmap::empty();
|
||||
output.union_and_clear_bbbul(bbbul);
|
||||
|
||||
for (_, rhs) in maps.iter_mut() {
|
||||
for rhs in maps.iter_mut() {
|
||||
if let Some(new) = rhs.get_mut(key) {
|
||||
output.union_and_clear_bbbul(new);
|
||||
}
|
||||
@@ -574,14 +569,14 @@ where
|
||||
|
||||
struct Entry<R> {
|
||||
cursor: ReaderCursor<R>,
|
||||
source_id: usize,
|
||||
source_index: usize,
|
||||
}
|
||||
|
||||
impl<R> Ord for Entry<R> {
|
||||
fn cmp(&self, other: &Entry<R>) -> Ordering {
|
||||
let skey = self.cursor.current().map(|(k, _)| k);
|
||||
let okey = other.cursor.current().map(|(k, _)| k);
|
||||
skey.cmp(&okey).then(self.source_id.cmp(&other.source_id)).reverse()
|
||||
skey.cmp(&okey).then(self.source_index.cmp(&other.source_index)).reverse()
|
||||
}
|
||||
}
|
||||
|
||||
|
||||
@@ -16,10 +16,10 @@ use crate::update::del_add::DelAdd;
|
||||
use crate::update::new::channel::FieldIdDocidFacetSender;
|
||||
use crate::update::new::extract::perm_json_p;
|
||||
use crate::update::new::indexer::document_changes::{
|
||||
extract, DocumentChangeContext, DocumentChanges, Extractor, IndexingContext,
|
||||
extract, DocumentChangeContext, DocumentChanges, Extractor, IndexingContext, Progress,
|
||||
};
|
||||
use crate::update::new::ref_cell_ext::RefCellExt as _;
|
||||
use crate::update::new::steps::IndexingStep;
|
||||
use crate::update::new::steps::Step;
|
||||
use crate::update::new::thread_local::{FullySend, ThreadLocal};
|
||||
use crate::update::new::DocumentChange;
|
||||
use crate::update::GrenadParameters;
|
||||
@@ -373,16 +373,26 @@ fn truncate_str(s: &str) -> &str {
|
||||
|
||||
impl FacetedDocidsExtractor {
|
||||
#[tracing::instrument(level = "trace", skip_all, target = "indexing::extract::faceted")]
|
||||
pub fn run_extraction<'pl, 'fid, 'indexer, 'index, 'extractor, DC: DocumentChanges<'pl>, MSP>(
|
||||
pub fn run_extraction<
|
||||
'pl,
|
||||
'fid,
|
||||
'indexer,
|
||||
'index,
|
||||
'extractor,
|
||||
DC: DocumentChanges<'pl>,
|
||||
MSP,
|
||||
SP,
|
||||
>(
|
||||
grenad_parameters: GrenadParameters,
|
||||
document_changes: &DC,
|
||||
indexing_context: IndexingContext<'fid, 'indexer, 'index, MSP>,
|
||||
indexing_context: IndexingContext<'fid, 'indexer, 'index, MSP, SP>,
|
||||
extractor_allocs: &'extractor mut ThreadLocal<FullySend<Bump>>,
|
||||
sender: &FieldIdDocidFacetSender,
|
||||
step: IndexingStep,
|
||||
step: Step,
|
||||
) -> Result<Vec<BalancedCaches<'extractor>>>
|
||||
where
|
||||
MSP: Fn() -> bool + Sync,
|
||||
SP: Fn(Progress) + Sync,
|
||||
{
|
||||
let index = indexing_context.index;
|
||||
let rtxn = index.read_txn()?;
|
||||
|
||||
@@ -15,22 +15,23 @@ pub use geo::*;
|
||||
pub use searchable::*;
|
||||
pub use vectors::EmbeddingExtractor;
|
||||
|
||||
use super::indexer::document_changes::{DocumentChanges, IndexingContext};
|
||||
use super::steps::IndexingStep;
|
||||
use super::indexer::document_changes::{DocumentChanges, IndexingContext, Progress};
|
||||
use super::steps::Step;
|
||||
use super::thread_local::{FullySend, ThreadLocal};
|
||||
use crate::update::GrenadParameters;
|
||||
use crate::Result;
|
||||
|
||||
pub trait DocidsExtractor {
|
||||
fn run_extraction<'pl, 'fid, 'indexer, 'index, 'extractor, DC: DocumentChanges<'pl>, MSP>(
|
||||
fn run_extraction<'pl, 'fid, 'indexer, 'index, 'extractor, DC: DocumentChanges<'pl>, MSP, SP>(
|
||||
grenad_parameters: GrenadParameters,
|
||||
document_changes: &DC,
|
||||
indexing_context: IndexingContext<'fid, 'indexer, 'index, MSP>,
|
||||
indexing_context: IndexingContext<'fid, 'indexer, 'index, MSP, SP>,
|
||||
extractor_allocs: &'extractor mut ThreadLocal<FullySend<Bump>>,
|
||||
step: IndexingStep,
|
||||
step: Step,
|
||||
) -> Result<Vec<BalancedCaches<'extractor>>>
|
||||
where
|
||||
MSP: Fn() -> bool + Sync;
|
||||
MSP: Fn() -> bool + Sync,
|
||||
SP: Fn(Progress) + Sync;
|
||||
}
|
||||
|
||||
/// TODO move in permissive json pointer
|
||||
|
||||
@@ -11,10 +11,10 @@ use super::tokenize_document::{tokenizer_builder, DocumentTokenizer};
|
||||
use crate::update::new::extract::cache::BalancedCaches;
|
||||
use crate::update::new::extract::perm_json_p::contained_in;
|
||||
use crate::update::new::indexer::document_changes::{
|
||||
extract, DocumentChangeContext, DocumentChanges, Extractor, IndexingContext,
|
||||
extract, DocumentChangeContext, DocumentChanges, Extractor, IndexingContext, Progress,
|
||||
};
|
||||
use crate::update::new::ref_cell_ext::RefCellExt as _;
|
||||
use crate::update::new::steps::IndexingStep;
|
||||
use crate::update::new::steps::Step;
|
||||
use crate::update::new::thread_local::{FullySend, MostlySend, ThreadLocal};
|
||||
use crate::update::new::DocumentChange;
|
||||
use crate::update::GrenadParameters;
|
||||
@@ -239,15 +239,25 @@ impl<'a, 'extractor> Extractor<'extractor> for WordDocidsExtractorData<'a> {
|
||||
pub struct WordDocidsExtractors;
|
||||
|
||||
impl WordDocidsExtractors {
|
||||
pub fn run_extraction<'pl, 'fid, 'indexer, 'index, 'extractor, DC: DocumentChanges<'pl>, MSP>(
|
||||
pub fn run_extraction<
|
||||
'pl,
|
||||
'fid,
|
||||
'indexer,
|
||||
'index,
|
||||
'extractor,
|
||||
DC: DocumentChanges<'pl>,
|
||||
MSP,
|
||||
SP,
|
||||
>(
|
||||
grenad_parameters: GrenadParameters,
|
||||
document_changes: &DC,
|
||||
indexing_context: IndexingContext<'fid, 'indexer, 'index, MSP>,
|
||||
indexing_context: IndexingContext<'fid, 'indexer, 'index, MSP, SP>,
|
||||
extractor_allocs: &'extractor mut ThreadLocal<FullySend<Bump>>,
|
||||
step: IndexingStep,
|
||||
step: Step,
|
||||
) -> Result<WordDocidsCaches<'extractor>>
|
||||
where
|
||||
MSP: Fn() -> bool + Sync,
|
||||
SP: Fn(Progress) + Sync,
|
||||
{
|
||||
let index = indexing_context.index;
|
||||
let rtxn = index.read_txn()?;
|
||||
|
||||
@@ -14,9 +14,9 @@ use tokenize_document::{tokenizer_builder, DocumentTokenizer};
|
||||
use super::cache::BalancedCaches;
|
||||
use super::DocidsExtractor;
|
||||
use crate::update::new::indexer::document_changes::{
|
||||
extract, DocumentChangeContext, DocumentChanges, Extractor, IndexingContext,
|
||||
extract, DocumentChangeContext, DocumentChanges, Extractor, IndexingContext, Progress,
|
||||
};
|
||||
use crate::update::new::steps::IndexingStep;
|
||||
use crate::update::new::steps::Step;
|
||||
use crate::update::new::thread_local::{FullySend, ThreadLocal};
|
||||
use crate::update::new::DocumentChange;
|
||||
use crate::update::GrenadParameters;
|
||||
@@ -56,15 +56,16 @@ impl<'a, 'extractor, EX: SearchableExtractor + Sync> Extractor<'extractor>
|
||||
}
|
||||
|
||||
pub trait SearchableExtractor: Sized + Sync {
|
||||
fn run_extraction<'pl, 'fid, 'indexer, 'index, 'extractor, DC: DocumentChanges<'pl>, MSP>(
|
||||
fn run_extraction<'pl, 'fid, 'indexer, 'index, 'extractor, DC: DocumentChanges<'pl>, MSP, SP>(
|
||||
grenad_parameters: GrenadParameters,
|
||||
document_changes: &DC,
|
||||
indexing_context: IndexingContext<'fid, 'indexer, 'index, MSP>,
|
||||
indexing_context: IndexingContext<'fid, 'indexer, 'index, MSP, SP>,
|
||||
extractor_allocs: &'extractor mut ThreadLocal<FullySend<Bump>>,
|
||||
step: IndexingStep,
|
||||
step: Step,
|
||||
) -> Result<Vec<BalancedCaches<'extractor>>>
|
||||
where
|
||||
MSP: Fn() -> bool + Sync,
|
||||
SP: Fn(Progress) + Sync,
|
||||
{
|
||||
let rtxn = indexing_context.index.read_txn()?;
|
||||
let stop_words = indexing_context.index.stop_words(&rtxn)?;
|
||||
@@ -133,15 +134,16 @@ pub trait SearchableExtractor: Sized + Sync {
|
||||
}
|
||||
|
||||
impl<T: SearchableExtractor> DocidsExtractor for T {
|
||||
fn run_extraction<'pl, 'fid, 'indexer, 'index, 'extractor, DC: DocumentChanges<'pl>, MSP>(
|
||||
fn run_extraction<'pl, 'fid, 'indexer, 'index, 'extractor, DC: DocumentChanges<'pl>, MSP, SP>(
|
||||
grenad_parameters: GrenadParameters,
|
||||
document_changes: &DC,
|
||||
indexing_context: IndexingContext<'fid, 'indexer, 'index, MSP>,
|
||||
indexing_context: IndexingContext<'fid, 'indexer, 'index, MSP, SP>,
|
||||
extractor_allocs: &'extractor mut ThreadLocal<FullySend<Bump>>,
|
||||
step: IndexingStep,
|
||||
step: Step,
|
||||
) -> Result<Vec<BalancedCaches<'extractor>>>
|
||||
where
|
||||
MSP: Fn() -> bool + Sync,
|
||||
SP: Fn(Progress) + Sync,
|
||||
{
|
||||
Self::run_extraction(
|
||||
grenad_parameters,
|
||||
|
||||
@@ -176,10 +176,9 @@ pub fn tokenizer_builder<'a>(
|
||||
#[cfg(test)]
|
||||
mod test {
|
||||
use bumpalo::Bump;
|
||||
use bumparaw_collections::RawMap;
|
||||
use charabia::TokenizerBuilder;
|
||||
use meili_snap::snapshot;
|
||||
use rustc_hash::FxBuildHasher;
|
||||
use raw_collections::RawMap;
|
||||
use serde_json::json;
|
||||
use serde_json::value::RawValue;
|
||||
|
||||
@@ -235,7 +234,7 @@ mod test {
|
||||
|
||||
let bump = Bump::new();
|
||||
let document: &RawValue = serde_json::from_str(&document).unwrap();
|
||||
let document = RawMap::from_raw_value_and_hasher(document, FxBuildHasher, &bump).unwrap();
|
||||
let document = RawMap::from_raw_value(document, &bump).unwrap();
|
||||
|
||||
let document = Versions::single(document);
|
||||
let document = DocumentFromVersions::new(&document);
|
||||
|
||||
@@ -130,7 +130,6 @@ impl<'a, 'b, 'extractor> Extractor<'extractor> for EmbeddingExtractor<'a, 'b> {
|
||||
);
|
||||
} else if new_vectors.regenerate {
|
||||
let new_rendered = prompt.render_document(
|
||||
update.external_document_id(),
|
||||
update.current(
|
||||
&context.rtxn,
|
||||
context.index,
|
||||
@@ -140,7 +139,6 @@ impl<'a, 'b, 'extractor> Extractor<'extractor> for EmbeddingExtractor<'a, 'b> {
|
||||
&context.doc_alloc,
|
||||
)?;
|
||||
let old_rendered = prompt.render_document(
|
||||
update.external_document_id(),
|
||||
update.merged(
|
||||
&context.rtxn,
|
||||
context.index,
|
||||
@@ -160,7 +158,6 @@ impl<'a, 'b, 'extractor> Extractor<'extractor> for EmbeddingExtractor<'a, 'b> {
|
||||
}
|
||||
} else if old_vectors.regenerate {
|
||||
let old_rendered = prompt.render_document(
|
||||
update.external_document_id(),
|
||||
update.current(
|
||||
&context.rtxn,
|
||||
context.index,
|
||||
@@ -170,7 +167,6 @@ impl<'a, 'b, 'extractor> Extractor<'extractor> for EmbeddingExtractor<'a, 'b> {
|
||||
&context.doc_alloc,
|
||||
)?;
|
||||
let new_rendered = prompt.render_document(
|
||||
update.external_document_id(),
|
||||
update.merged(
|
||||
&context.rtxn,
|
||||
context.index,
|
||||
@@ -220,7 +216,6 @@ impl<'a, 'b, 'extractor> Extractor<'extractor> for EmbeddingExtractor<'a, 'b> {
|
||||
);
|
||||
} else if new_vectors.regenerate {
|
||||
let rendered = prompt.render_document(
|
||||
insertion.external_document_id(),
|
||||
insertion.inserted(),
|
||||
context.new_fields_ids_map,
|
||||
&context.doc_alloc,
|
||||
@@ -234,7 +229,6 @@ impl<'a, 'b, 'extractor> Extractor<'extractor> for EmbeddingExtractor<'a, 'b> {
|
||||
}
|
||||
} else {
|
||||
let rendered = prompt.render_document(
|
||||
insertion.external_document_id(),
|
||||
insertion.inserted(),
|
||||
context.new_fields_ids_map,
|
||||
&context.doc_alloc,
|
||||
|
||||
@@ -103,8 +103,6 @@ impl<'indexer> FacetSearchBuilder<'indexer> {
|
||||
|
||||
#[tracing::instrument(level = "trace", skip_all, target = "indexing::facet_fst")]
|
||||
pub fn merge_and_write(self, index: &Index, wtxn: &mut RwTxn, rtxn: &RoTxn) -> Result<()> {
|
||||
tracing::trace!("merge facet strings for facet search: {:?}", self.registered_facets);
|
||||
|
||||
let reader = self.normalized_facet_string_docids_sorter.into_reader_cursors()?;
|
||||
let mut builder = grenad::MergerBuilder::new(MergeDeladdBtreesetString);
|
||||
builder.extend(reader);
|
||||
@@ -120,15 +118,12 @@ impl<'indexer> FacetSearchBuilder<'indexer> {
|
||||
BEU16StrCodec::bytes_decode(key).map_err(heed::Error::Encoding)?;
|
||||
|
||||
if current_field_id != Some(field_id) {
|
||||
if let (Some(current_field_id), Some(fst_merger_builder)) =
|
||||
(current_field_id, fst_merger_builder)
|
||||
{
|
||||
if let Some(fst_merger_builder) = fst_merger_builder {
|
||||
let mmap = fst_merger_builder.build(&mut callback)?;
|
||||
index.facet_id_string_fst.remap_data_type::<Bytes>().put(
|
||||
wtxn,
|
||||
¤t_field_id,
|
||||
&mmap,
|
||||
)?;
|
||||
index
|
||||
.facet_id_string_fst
|
||||
.remap_data_type::<Bytes>()
|
||||
.put(wtxn, &field_id, &mmap)?;
|
||||
}
|
||||
|
||||
fst = index.facet_id_string_fst.get(rtxn, &field_id)?;
|
||||
|
||||
@@ -1,8 +1,6 @@
|
||||
use std::ops::ControlFlow;
|
||||
|
||||
use bumpalo::Bump;
|
||||
use bumparaw_collections::RawVec;
|
||||
use rustc_hash::FxBuildHasher;
|
||||
use serde::de::{DeserializeSeed, Deserializer as _, Visitor};
|
||||
use serde_json::value::RawValue;
|
||||
|
||||
@@ -362,7 +360,7 @@ impl<'a> DeserrRawValue<'a> {
|
||||
}
|
||||
|
||||
pub struct DeserrRawVec<'a> {
|
||||
vec: RawVec<'a>,
|
||||
vec: raw_collections::RawVec<'a>,
|
||||
alloc: &'a Bump,
|
||||
}
|
||||
|
||||
@@ -381,7 +379,7 @@ impl<'a> deserr::Sequence for DeserrRawVec<'a> {
|
||||
}
|
||||
|
||||
pub struct DeserrRawVecIter<'a> {
|
||||
it: bumparaw_collections::vec::iter::IntoIter<'a>,
|
||||
it: raw_collections::vec::iter::IntoIter<'a>,
|
||||
alloc: &'a Bump,
|
||||
}
|
||||
|
||||
@@ -395,7 +393,7 @@ impl<'a> Iterator for DeserrRawVecIter<'a> {
|
||||
}
|
||||
|
||||
pub struct DeserrRawMap<'a> {
|
||||
map: bumparaw_collections::RawMap<'a, FxBuildHasher>,
|
||||
map: raw_collections::RawMap<'a>,
|
||||
alloc: &'a Bump,
|
||||
}
|
||||
|
||||
@@ -418,7 +416,7 @@ impl<'a> deserr::Map for DeserrRawMap<'a> {
|
||||
}
|
||||
|
||||
pub struct DeserrRawMapIter<'a> {
|
||||
it: bumparaw_collections::map::iter::IntoIter<'a>,
|
||||
it: raw_collections::map::iter::IntoIter<'a>,
|
||||
alloc: &'a Bump,
|
||||
}
|
||||
|
||||
@@ -617,7 +615,7 @@ impl<'de> Visitor<'de> for DeserrRawValueVisitor<'de> {
|
||||
where
|
||||
A: serde::de::SeqAccess<'de>,
|
||||
{
|
||||
let mut raw_vec = RawVec::new_in(self.alloc);
|
||||
let mut raw_vec = raw_collections::RawVec::new_in(self.alloc);
|
||||
while let Some(next) = seq.next_element()? {
|
||||
raw_vec.push(next);
|
||||
}
|
||||
|
||||
@@ -1,5 +1,4 @@
|
||||
use std::cell::{Cell, RefCell};
|
||||
use std::sync::atomic::Ordering;
|
||||
use std::sync::{Arc, RwLock};
|
||||
|
||||
use bumpalo::Bump;
|
||||
@@ -8,9 +7,8 @@ use rayon::iter::IndexedParallelIterator;
|
||||
|
||||
use super::super::document_change::DocumentChange;
|
||||
use crate::fields_ids_map::metadata::FieldIdMapWithMetadata;
|
||||
use crate::progress::{AtomicDocumentStep, Progress};
|
||||
use crate::update::new::parallel_iterator_ext::ParallelIteratorExt as _;
|
||||
use crate::update::new::steps::IndexingStep;
|
||||
use crate::update::new::steps::Step;
|
||||
use crate::update::new::thread_local::{FullySend, MostlySend, ThreadLocal};
|
||||
use crate::{FieldsIdsMap, GlobalFieldsIdsMap, Index, InternalError, Result};
|
||||
|
||||
@@ -135,8 +133,10 @@ pub struct IndexingContext<
|
||||
'indexer, // covariant lifetime of objects that are borrowed during the entire indexing operation
|
||||
'index, // covariant lifetime of the index
|
||||
MSP,
|
||||
SP,
|
||||
> where
|
||||
MSP: Fn() -> bool + Sync,
|
||||
SP: Fn(Progress) + Sync,
|
||||
{
|
||||
pub index: &'index Index,
|
||||
pub db_fields_ids_map: &'indexer FieldsIdsMap,
|
||||
@@ -144,7 +144,7 @@ pub struct IndexingContext<
|
||||
pub doc_allocs: &'indexer ThreadLocal<FullySend<Cell<Bump>>>,
|
||||
pub fields_ids_map_store: &'indexer ThreadLocal<FullySend<RefCell<GlobalFieldsIdsMap<'fid>>>>,
|
||||
pub must_stop_processing: &'indexer MSP,
|
||||
pub progress: &'indexer Progress,
|
||||
pub send_progress: &'indexer SP,
|
||||
}
|
||||
|
||||
impl<
|
||||
@@ -152,15 +152,18 @@ impl<
|
||||
'indexer, // covariant lifetime of objects that are borrowed during the entire indexing operation
|
||||
'index, // covariant lifetime of the index
|
||||
MSP,
|
||||
SP,
|
||||
> Copy
|
||||
for IndexingContext<
|
||||
'fid, // invariant lifetime of fields ids map
|
||||
'indexer, // covariant lifetime of objects that are borrowed during the entire indexing operation
|
||||
'index, // covariant lifetime of the index
|
||||
MSP,
|
||||
SP,
|
||||
>
|
||||
where
|
||||
MSP: Fn() -> bool + Sync,
|
||||
SP: Fn(Progress) + Sync,
|
||||
{
|
||||
}
|
||||
|
||||
@@ -169,15 +172,18 @@ impl<
|
||||
'indexer, // covariant lifetime of objects that are borrowed during the entire indexing operation
|
||||
'index, // covariant lifetime of the index
|
||||
MSP,
|
||||
SP,
|
||||
> Clone
|
||||
for IndexingContext<
|
||||
'fid, // invariant lifetime of fields ids map
|
||||
'indexer, // covariant lifetime of objects that are borrowed during the entire indexing operation
|
||||
'index, // covariant lifetime of the index
|
||||
MSP,
|
||||
SP,
|
||||
>
|
||||
where
|
||||
MSP: Fn() -> bool + Sync,
|
||||
SP: Fn(Progress) + Sync,
|
||||
{
|
||||
fn clone(&self) -> Self {
|
||||
*self
|
||||
@@ -196,6 +202,7 @@ pub fn extract<
|
||||
EX,
|
||||
DC: DocumentChanges<'pl>,
|
||||
MSP,
|
||||
SP,
|
||||
>(
|
||||
document_changes: &DC,
|
||||
extractor: &EX,
|
||||
@@ -206,18 +213,18 @@ pub fn extract<
|
||||
doc_allocs,
|
||||
fields_ids_map_store,
|
||||
must_stop_processing,
|
||||
progress,
|
||||
}: IndexingContext<'fid, 'indexer, 'index, MSP>,
|
||||
send_progress,
|
||||
}: IndexingContext<'fid, 'indexer, 'index, MSP, SP>,
|
||||
extractor_allocs: &'extractor mut ThreadLocal<FullySend<Bump>>,
|
||||
datastore: &'data ThreadLocal<EX::Data>,
|
||||
step: IndexingStep,
|
||||
step: Step,
|
||||
) -> Result<()>
|
||||
where
|
||||
EX: Extractor<'extractor>,
|
||||
MSP: Fn() -> bool + Sync,
|
||||
SP: Fn(Progress) + Sync,
|
||||
{
|
||||
tracing::trace!("We are resetting the extractor allocators");
|
||||
progress.update_progress(step);
|
||||
// Clean up and reuse the extractor allocs
|
||||
for extractor_alloc in extractor_allocs.iter_mut() {
|
||||
tracing::trace!("\tWith {} bytes reset", extractor_alloc.0.allocated_bytes());
|
||||
@@ -225,11 +232,9 @@ where
|
||||
}
|
||||
|
||||
let total_documents = document_changes.len() as u32;
|
||||
let (step, progress_step) = AtomicDocumentStep::new(total_documents);
|
||||
progress.update_progress(progress_step);
|
||||
|
||||
let pi = document_changes.iter(CHUNK_SIZE);
|
||||
pi.try_arc_for_each_try_init(
|
||||
pi.enumerate().try_arc_for_each_try_init(
|
||||
|| {
|
||||
DocumentChangeContext::new(
|
||||
index,
|
||||
@@ -242,10 +247,13 @@ where
|
||||
move |index_alloc| extractor.init_data(index_alloc),
|
||||
)
|
||||
},
|
||||
|context, items| {
|
||||
|context, (finished_documents, items)| {
|
||||
if (must_stop_processing)() {
|
||||
return Err(Arc::new(InternalError::AbortedIndexation.into()));
|
||||
}
|
||||
let finished_documents = (finished_documents * CHUNK_SIZE) as u32;
|
||||
|
||||
(send_progress)(Progress::from_step_substep(step, finished_documents, total_documents));
|
||||
|
||||
// Clean up and reuse the document-specific allocator
|
||||
context.doc_alloc.reset();
|
||||
@@ -256,7 +264,6 @@ where
|
||||
});
|
||||
|
||||
let res = extractor.process(changes, context).map_err(Arc::new);
|
||||
step.fetch_add(items.as_ref().len() as u32, Ordering::Relaxed);
|
||||
|
||||
// send back the doc_alloc in the pool
|
||||
context.doc_allocs.get_or_default().0.set(std::mem::take(&mut context.doc_alloc));
|
||||
@@ -264,7 +271,32 @@ where
|
||||
res
|
||||
},
|
||||
)?;
|
||||
step.store(total_documents, Ordering::Relaxed);
|
||||
|
||||
(send_progress)(Progress::from_step_substep(step, total_documents, total_documents));
|
||||
|
||||
Ok(())
|
||||
}
|
||||
|
||||
pub struct Progress {
|
||||
pub finished_steps: u16,
|
||||
pub total_steps: u16,
|
||||
pub step_name: &'static str,
|
||||
pub finished_total_substep: Option<(u32, u32)>,
|
||||
}
|
||||
|
||||
impl Progress {
|
||||
pub fn from_step(step: Step) -> Self {
|
||||
Self {
|
||||
finished_steps: step.finished_steps(),
|
||||
total_steps: Step::total_steps(),
|
||||
step_name: step.name(),
|
||||
finished_total_substep: None,
|
||||
}
|
||||
}
|
||||
pub fn from_step_substep(step: Step, finished_substep: u32, total_substep: u32) -> Self {
|
||||
Self {
|
||||
finished_total_substep: Some((finished_substep, total_substep)),
|
||||
..Progress::from_step(step)
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
@@ -92,12 +92,11 @@ mod test {
|
||||
|
||||
use crate::fields_ids_map::metadata::{FieldIdMapWithMetadata, MetadataBuilder};
|
||||
use crate::index::tests::TempIndex;
|
||||
use crate::progress::Progress;
|
||||
use crate::update::new::indexer::document_changes::{
|
||||
extract, DocumentChangeContext, Extractor, IndexingContext,
|
||||
};
|
||||
use crate::update::new::indexer::DocumentDeletion;
|
||||
use crate::update::new::steps::IndexingStep;
|
||||
use crate::update::new::steps::Step;
|
||||
use crate::update::new::thread_local::{MostlySend, ThreadLocal};
|
||||
use crate::update::new::DocumentChange;
|
||||
use crate::DocumentId;
|
||||
@@ -165,7 +164,7 @@ mod test {
|
||||
doc_allocs: &doc_allocs,
|
||||
fields_ids_map_store: &fields_ids_map_store,
|
||||
must_stop_processing: &(|| false),
|
||||
progress: &Progress::default(),
|
||||
send_progress: &(|_progress| {}),
|
||||
};
|
||||
|
||||
for _ in 0..3 {
|
||||
@@ -177,7 +176,7 @@ mod test {
|
||||
context,
|
||||
&mut extractor_allocs,
|
||||
&datastore,
|
||||
IndexingStep::ExtractingDocuments,
|
||||
Step::ExtractingDocuments,
|
||||
)
|
||||
.unwrap();
|
||||
|
||||
|
||||
@@ -1,23 +1,19 @@
|
||||
use std::sync::atomic::Ordering;
|
||||
|
||||
use bumpalo::collections::CollectIn;
|
||||
use bumpalo::Bump;
|
||||
use bumparaw_collections::RawMap;
|
||||
use hashbrown::hash_map::Entry;
|
||||
use heed::RoTxn;
|
||||
use memmap2::Mmap;
|
||||
use raw_collections::RawMap;
|
||||
use rayon::slice::ParallelSlice;
|
||||
use rustc_hash::FxBuildHasher;
|
||||
use serde_json::value::RawValue;
|
||||
use serde_json::Deserializer;
|
||||
|
||||
use super::super::document_change::DocumentChange;
|
||||
use super::document_changes::{DocumentChangeContext, DocumentChanges};
|
||||
use super::document_changes::{DocumentChangeContext, DocumentChanges, Progress};
|
||||
use super::retrieve_or_guess_primary_key;
|
||||
use crate::documents::PrimaryKey;
|
||||
use crate::progress::{AtomicPayloadStep, Progress};
|
||||
use crate::update::new::document::Versions;
|
||||
use crate::update::new::steps::IndexingStep;
|
||||
use crate::update::new::steps::Step;
|
||||
use crate::update::new::thread_local::MostlySend;
|
||||
use crate::update::new::{Deletion, Insertion, Update};
|
||||
use crate::update::{AvailableIds, IndexDocumentsMethod};
|
||||
@@ -48,7 +44,7 @@ impl<'pl> DocumentOperation<'pl> {
|
||||
|
||||
#[allow(clippy::too_many_arguments)]
|
||||
#[tracing::instrument(level = "trace", skip_all, target = "indexing::document_operation")]
|
||||
pub fn into_changes<MSP>(
|
||||
pub fn into_changes<MSP, SP>(
|
||||
self,
|
||||
indexer: &'pl Bump,
|
||||
index: &Index,
|
||||
@@ -56,12 +52,12 @@ impl<'pl> DocumentOperation<'pl> {
|
||||
primary_key_from_op: Option<&'pl str>,
|
||||
new_fields_ids_map: &mut FieldsIdsMap,
|
||||
must_stop_processing: &MSP,
|
||||
progress: Progress,
|
||||
send_progress: &SP,
|
||||
) -> Result<(DocumentOperationChanges<'pl>, Vec<PayloadStats>, Option<PrimaryKey<'pl>>)>
|
||||
where
|
||||
MSP: Fn() -> bool,
|
||||
SP: Fn(Progress),
|
||||
{
|
||||
progress.update_progress(IndexingStep::PreparingPayloads);
|
||||
let Self { operations, method } = self;
|
||||
|
||||
let documents_ids = index.documents_ids(rtxn)?;
|
||||
@@ -71,14 +67,16 @@ impl<'pl> DocumentOperation<'pl> {
|
||||
let mut primary_key = None;
|
||||
|
||||
let payload_count = operations.len();
|
||||
let (step, progress_step) = AtomicPayloadStep::new(payload_count as u32);
|
||||
progress.update_progress(progress_step);
|
||||
|
||||
for (payload_index, operation) in operations.into_iter().enumerate() {
|
||||
if must_stop_processing() {
|
||||
return Err(InternalError::AbortedIndexation.into());
|
||||
}
|
||||
step.store(payload_index as u32, Ordering::Relaxed);
|
||||
send_progress(Progress::from_step_substep(
|
||||
Step::PreparingPayloads,
|
||||
payload_index as u32,
|
||||
payload_count as u32,
|
||||
));
|
||||
|
||||
let mut bytes = 0;
|
||||
let result = match operation {
|
||||
@@ -119,7 +117,12 @@ impl<'pl> DocumentOperation<'pl> {
|
||||
};
|
||||
operations_stats.push(PayloadStats { document_count, bytes, error });
|
||||
}
|
||||
step.store(payload_count as u32, Ordering::Relaxed);
|
||||
|
||||
send_progress(Progress::from_step_substep(
|
||||
Step::PreparingPayloads,
|
||||
payload_count as u32,
|
||||
payload_count as u32,
|
||||
));
|
||||
|
||||
// TODO We must drain the HashMap into a Vec because rayon::hash_map::IntoIter: !Clone
|
||||
let mut docids_version_offsets: bumpalo::collections::vec::Vec<_> =
|
||||
@@ -163,9 +166,8 @@ fn extract_addition_payload_changes<'r, 'pl: 'r>(
|
||||
|
||||
// Only guess the primary key if it is the first document
|
||||
let retrieved_primary_key = if previous_offset == 0 {
|
||||
let doc = RawMap::from_raw_value_and_hasher(doc, FxBuildHasher, indexer)
|
||||
.map(Some)
|
||||
.map_err(UserError::SerdeJson)?;
|
||||
let doc =
|
||||
RawMap::from_raw_value(doc, indexer).map(Some).map_err(UserError::SerdeJson)?;
|
||||
|
||||
let result = retrieve_or_guess_primary_key(
|
||||
rtxn,
|
||||
@@ -252,24 +254,6 @@ fn extract_addition_payload_changes<'r, 'pl: 'r>(
|
||||
previous_offset = iter.byte_offset();
|
||||
}
|
||||
|
||||
if payload.is_empty() {
|
||||
let result = retrieve_or_guess_primary_key(
|
||||
rtxn,
|
||||
index,
|
||||
new_fields_ids_map,
|
||||
primary_key_from_op,
|
||||
None,
|
||||
);
|
||||
match result {
|
||||
Ok(Ok((pk, _))) => {
|
||||
primary_key.get_or_insert(pk);
|
||||
}
|
||||
Ok(Err(UserError::NoPrimaryKeyCandidateFound)) => (),
|
||||
Ok(Err(user_error)) => return Err(Error::UserError(user_error)),
|
||||
Err(error) => return Err(error),
|
||||
};
|
||||
}
|
||||
|
||||
Ok(new_docids_version_offsets)
|
||||
}
|
||||
|
||||
@@ -561,9 +545,8 @@ impl MergeChanges for MergeDocumentForReplacement {
|
||||
match operations.last() {
|
||||
Some(InnerDocOp::Addition(DocumentOffset { content })) => {
|
||||
let document = serde_json::from_slice(content).unwrap();
|
||||
let document =
|
||||
RawMap::from_raw_value_and_hasher(document, FxBuildHasher, doc_alloc)
|
||||
.map_err(UserError::SerdeJson)?;
|
||||
let document = raw_collections::RawMap::from_raw_value(document, doc_alloc)
|
||||
.map_err(UserError::SerdeJson)?;
|
||||
|
||||
if is_new {
|
||||
Ok(Some(DocumentChange::Insertion(Insertion::create(
|
||||
@@ -649,9 +632,8 @@ impl MergeChanges for MergeDocumentForUpdates {
|
||||
}
|
||||
};
|
||||
let document = serde_json::from_slice(content).unwrap();
|
||||
let document =
|
||||
RawMap::from_raw_value_and_hasher(document, FxBuildHasher, doc_alloc)
|
||||
.map_err(UserError::SerdeJson)?;
|
||||
let document = raw_collections::RawMap::from_raw_value(document, doc_alloc)
|
||||
.map_err(UserError::SerdeJson)?;
|
||||
|
||||
Some(Versions::single(document))
|
||||
}
|
||||
@@ -665,9 +647,8 @@ impl MergeChanges for MergeDocumentForUpdates {
|
||||
};
|
||||
|
||||
let document = serde_json::from_slice(content).unwrap();
|
||||
let document =
|
||||
RawMap::from_raw_value_and_hasher(document, FxBuildHasher, doc_alloc)
|
||||
.map_err(UserError::SerdeJson)?;
|
||||
let document = raw_collections::RawMap::from_raw_value(document, doc_alloc)
|
||||
.map_err(UserError::SerdeJson)?;
|
||||
Ok(document)
|
||||
});
|
||||
Versions::multiple(versions)?
|
||||
|
||||
@@ -4,8 +4,7 @@ use std::sync::{OnceLock, RwLock};
|
||||
use std::thread::{self, Builder};
|
||||
|
||||
use big_s::S;
|
||||
use bumparaw_collections::RawMap;
|
||||
use document_changes::{extract, DocumentChanges, IndexingContext};
|
||||
use document_changes::{extract, DocumentChanges, IndexingContext, Progress};
|
||||
pub use document_deletion::DocumentDeletion;
|
||||
pub use document_operation::{DocumentOperation, PayloadStats};
|
||||
use hashbrown::HashMap;
|
||||
@@ -14,7 +13,7 @@ use heed::{RoTxn, RwTxn};
|
||||
use itertools::{merge_join_by, EitherOrBoth};
|
||||
pub use partial_dump::PartialDump;
|
||||
use rand::SeedableRng as _;
|
||||
use rustc_hash::FxBuildHasher;
|
||||
use raw_collections::RawMap;
|
||||
use time::OffsetDateTime;
|
||||
pub use update_by_function::UpdateByFunction;
|
||||
|
||||
@@ -22,7 +21,7 @@ use super::channel::*;
|
||||
use super::extract::*;
|
||||
use super::facet_search_builder::FacetSearchBuilder;
|
||||
use super::merger::FacetFieldIdsDelta;
|
||||
use super::steps::IndexingStep;
|
||||
use super::steps::Step;
|
||||
use super::thread_local::ThreadLocal;
|
||||
use super::word_fst_builder::{PrefixData, PrefixDelta, WordFstBuilder};
|
||||
use super::words_prefix_docids::{
|
||||
@@ -33,11 +32,8 @@ use crate::documents::{PrimaryKey, DEFAULT_PRIMARY_KEY};
|
||||
use crate::facet::FacetType;
|
||||
use crate::fields_ids_map::metadata::{FieldIdMapWithMetadata, MetadataBuilder};
|
||||
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;
|
||||
@@ -63,7 +59,7 @@ mod update_by_function;
|
||||
///
|
||||
/// TODO return stats
|
||||
#[allow(clippy::too_many_arguments)] // clippy: 😝
|
||||
pub fn index<'pl, 'indexer, 'index, DC, MSP>(
|
||||
pub fn index<'pl, 'indexer, 'index, DC, MSP, SP>(
|
||||
wtxn: &mut RwTxn,
|
||||
index: &'index Index,
|
||||
pool: &ThreadPoolNoAbort,
|
||||
@@ -74,11 +70,12 @@ pub fn index<'pl, 'indexer, 'index, DC, MSP>(
|
||||
document_changes: &DC,
|
||||
embedders: EmbeddingConfigs,
|
||||
must_stop_processing: &'indexer MSP,
|
||||
progress: &'indexer Progress,
|
||||
send_progress: &'indexer SP,
|
||||
) -> Result<()>
|
||||
where
|
||||
DC: DocumentChanges<'pl>,
|
||||
MSP: Fn() -> bool + Sync,
|
||||
SP: Fn(Progress) + Sync,
|
||||
{
|
||||
let mut bbbuffers = Vec::new();
|
||||
let finished_extraction = AtomicBool::new(false);
|
||||
@@ -127,7 +124,7 @@ where
|
||||
doc_allocs: &doc_allocs,
|
||||
fields_ids_map_store: &fields_ids_map_store,
|
||||
must_stop_processing,
|
||||
progress,
|
||||
send_progress,
|
||||
};
|
||||
|
||||
let mut index_embeddings = index.embedding_configs(wtxn)?;
|
||||
@@ -161,7 +158,7 @@ where
|
||||
indexing_context,
|
||||
&mut extractor_allocs,
|
||||
&datastore,
|
||||
IndexingStep::ExtractingDocuments,
|
||||
Step::ExtractingDocuments,
|
||||
)?;
|
||||
}
|
||||
{
|
||||
@@ -193,7 +190,7 @@ where
|
||||
indexing_context,
|
||||
&mut extractor_allocs,
|
||||
&extractor_sender.field_id_docid_facet_sender(),
|
||||
IndexingStep::ExtractingFacets
|
||||
Step::ExtractingFacets
|
||||
)?
|
||||
};
|
||||
|
||||
@@ -205,7 +202,6 @@ where
|
||||
caches,
|
||||
FacetDatabases::new(index),
|
||||
index,
|
||||
&rtxn,
|
||||
extractor_sender.facet_docids(),
|
||||
)?;
|
||||
}
|
||||
@@ -227,7 +223,7 @@ where
|
||||
document_changes,
|
||||
indexing_context,
|
||||
&mut extractor_allocs,
|
||||
IndexingStep::ExtractingWords
|
||||
Step::ExtractingWords
|
||||
)?
|
||||
};
|
||||
|
||||
@@ -305,7 +301,7 @@ where
|
||||
document_changes,
|
||||
indexing_context,
|
||||
&mut extractor_allocs,
|
||||
IndexingStep::ExtractingWordProximity,
|
||||
Step::ExtractingWordProximity,
|
||||
)?
|
||||
};
|
||||
|
||||
@@ -341,7 +337,7 @@ where
|
||||
indexing_context,
|
||||
&mut extractor_allocs,
|
||||
&datastore,
|
||||
IndexingStep::ExtractingEmbeddings,
|
||||
Step::ExtractingEmbeddings,
|
||||
)?;
|
||||
}
|
||||
{
|
||||
@@ -374,7 +370,7 @@ where
|
||||
indexing_context,
|
||||
&mut extractor_allocs,
|
||||
&datastore,
|
||||
IndexingStep::WritingGeoPoints
|
||||
Step::WritingGeoPoints
|
||||
)?;
|
||||
}
|
||||
|
||||
@@ -386,7 +382,9 @@ where
|
||||
&indexing_context.must_stop_processing,
|
||||
)?;
|
||||
}
|
||||
indexing_context.progress.update_progress(IndexingStep::WritingToDatabase);
|
||||
|
||||
(indexing_context.send_progress)(Progress::from_step(Step::WritingToDatabase));
|
||||
|
||||
finished_extraction.store(true, std::sync::atomic::Ordering::Relaxed);
|
||||
|
||||
Result::Ok((facet_field_ids_delta, index_embeddings))
|
||||
@@ -486,7 +484,7 @@ where
|
||||
)?;
|
||||
}
|
||||
|
||||
indexing_context.progress.update_progress(IndexingStep::WaitingForExtractors);
|
||||
(indexing_context.send_progress)(Progress::from_step(Step::WaitingForExtractors));
|
||||
|
||||
let (facet_field_ids_delta, index_embeddings) = extractor_handle.join().unwrap()?;
|
||||
|
||||
@@ -499,7 +497,10 @@ where
|
||||
break 'vectors;
|
||||
}
|
||||
|
||||
indexing_context.progress.update_progress(IndexingStep::WritingEmbeddingsToDatabase);
|
||||
(indexing_context.send_progress)(Progress::from_step(
|
||||
Step::WritingEmbeddingsToDatabase,
|
||||
));
|
||||
|
||||
let mut rng = rand::rngs::StdRng::seed_from_u64(42);
|
||||
for (_index, (_embedder_name, _embedder, writer, dimensions)) in &mut arroy_writers {
|
||||
let dimensions = *dimensions;
|
||||
@@ -515,19 +516,21 @@ where
|
||||
index.put_embedding_configs(wtxn, index_embeddings)?;
|
||||
}
|
||||
|
||||
indexing_context.progress.update_progress(IndexingStep::PostProcessingFacets);
|
||||
(indexing_context.send_progress)(Progress::from_step(Step::PostProcessingFacets));
|
||||
|
||||
if index.facet_search(wtxn)? {
|
||||
compute_facet_search_database(index, wtxn, global_fields_ids_map)?;
|
||||
}
|
||||
|
||||
compute_facet_level_database(index, wtxn, facet_field_ids_delta)?;
|
||||
|
||||
indexing_context.progress.update_progress(IndexingStep::PostProcessingWords);
|
||||
(indexing_context.send_progress)(Progress::from_step(Step::PostProcessingWords));
|
||||
|
||||
if let Some(prefix_delta) = compute_word_fst(index, wtxn)? {
|
||||
compute_prefix_database(index, wtxn, prefix_delta, grenad_parameters)?;
|
||||
}
|
||||
|
||||
indexing_context.progress.update_progress(IndexingStep::Finalizing);
|
||||
(indexing_context.send_progress)(Progress::from_step(Step::Finalizing));
|
||||
|
||||
Ok(()) as Result<_>
|
||||
})?;
|
||||
@@ -738,66 +741,27 @@ fn compute_facet_search_database(
|
||||
fn compute_facet_level_database(
|
||||
index: &Index,
|
||||
wtxn: &mut RwTxn,
|
||||
mut facet_field_ids_delta: FacetFieldIdsDelta,
|
||||
facet_field_ids_delta: FacetFieldIdsDelta,
|
||||
) -> Result<()> {
|
||||
for (fid, delta) in facet_field_ids_delta.consume_facet_string_delta() {
|
||||
if let Some(modified_facet_string_ids) = facet_field_ids_delta.modified_facet_string_ids() {
|
||||
let span = tracing::trace_span!(target: "indexing::facet_field_ids", "string");
|
||||
let _entered = span.enter();
|
||||
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)?
|
||||
}
|
||||
}
|
||||
FacetsUpdateBulk::new_not_updating_level_0(
|
||||
index,
|
||||
modified_facet_string_ids,
|
||||
FacetType::String,
|
||||
)
|
||||
.execute(wtxn)?;
|
||||
}
|
||||
|
||||
for (fid, delta) in facet_field_ids_delta.consume_facet_number_delta() {
|
||||
if let Some(modified_facet_number_ids) = facet_field_ids_delta.modified_facet_number_ids() {
|
||||
let span = tracing::trace_span!(target: "indexing::facet_field_ids", "number");
|
||||
let _entered = span.enter();
|
||||
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");
|
||||
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(
|
||||
FacetsUpdateBulk::new_not_updating_level_0(
|
||||
index,
|
||||
wtxn,
|
||||
fid,
|
||||
modified_facet_number_ids,
|
||||
FacetType::Number,
|
||||
FACET_GROUP_SIZE as usize,
|
||||
FACET_MIN_LEVEL_SIZE as usize,
|
||||
FACET_MAX_GROUP_SIZE as usize,
|
||||
)
|
||||
.is_ok());
|
||||
.execute(wtxn)?;
|
||||
}
|
||||
|
||||
Ok(())
|
||||
@@ -812,7 +776,7 @@ pub fn retrieve_or_guess_primary_key<'a>(
|
||||
index: &Index,
|
||||
new_fields_ids_map: &mut FieldsIdsMap,
|
||||
primary_key_from_op: Option<&'a str>,
|
||||
first_document: Option<RawMap<'a, FxBuildHasher>>,
|
||||
first_document: Option<RawMap<'a>>,
|
||||
) -> Result<StdResult<(PrimaryKey<'a>, bool), UserError>> {
|
||||
// make sure that we have a declared primary key, either fetching it from the index or attempting to guess it.
|
||||
|
||||
|
||||
@@ -1,8 +1,6 @@
|
||||
use std::ops::DerefMut;
|
||||
|
||||
use bumparaw_collections::RawMap;
|
||||
use rayon::iter::IndexedParallelIterator;
|
||||
use rustc_hash::FxBuildHasher;
|
||||
use serde_json::value::RawValue;
|
||||
|
||||
use super::document_changes::{DocumentChangeContext, DocumentChanges};
|
||||
@@ -77,7 +75,7 @@ where
|
||||
self.primary_key.extract_fields_and_docid(document, fields_ids_map, doc_alloc)?;
|
||||
let external_document_id = external_document_id.to_de();
|
||||
|
||||
let document = RawMap::from_raw_value_and_hasher(document, FxBuildHasher, doc_alloc)
|
||||
let document = raw_collections::RawMap::from_raw_value(document, doc_alloc)
|
||||
.map_err(InternalError::SerdeJson)?;
|
||||
|
||||
let insertion = Insertion::create(docid, external_document_id, Versions::single(document));
|
||||
|
||||
@@ -1,9 +1,8 @@
|
||||
use bumparaw_collections::RawMap;
|
||||
use raw_collections::RawMap;
|
||||
use rayon::iter::IndexedParallelIterator;
|
||||
use rayon::slice::ParallelSlice as _;
|
||||
use rhai::{Dynamic, Engine, OptimizationLevel, Scope, AST};
|
||||
use roaring::RoaringBitmap;
|
||||
use rustc_hash::FxBuildHasher;
|
||||
|
||||
use super::document_changes::DocumentChangeContext;
|
||||
use super::DocumentChanges;
|
||||
@@ -161,12 +160,8 @@ impl<'index> DocumentChanges<'index> for UpdateByFunctionChanges<'index> {
|
||||
if document_id != new_document_id {
|
||||
Err(Error::UserError(UserError::DocumentEditionCannotModifyPrimaryKey))
|
||||
} else {
|
||||
let raw_new_doc = RawMap::from_raw_value_and_hasher(
|
||||
raw_new_doc,
|
||||
FxBuildHasher,
|
||||
doc_alloc,
|
||||
)
|
||||
.map_err(InternalError::SerdeJson)?;
|
||||
let raw_new_doc = RawMap::from_raw_value(raw_new_doc, doc_alloc)
|
||||
.map_err(InternalError::SerdeJson)?;
|
||||
|
||||
Ok(Some(DocumentChange::Update(Update::create(
|
||||
docid,
|
||||
|
||||
@@ -1,6 +1,6 @@
|
||||
use std::cell::RefCell;
|
||||
|
||||
use hashbrown::HashMap;
|
||||
use hashbrown::HashSet;
|
||||
use heed::types::Bytes;
|
||||
use heed::{Database, RoTxn};
|
||||
use memmap2::Mmap;
|
||||
@@ -12,7 +12,6 @@ use super::extract::{
|
||||
merge_caches_sorted, transpose_and_freeze_caches, BalancedCaches, DelAddRoaringBitmap,
|
||||
FacetKind, GeoExtractorData,
|
||||
};
|
||||
use crate::update::facet::new_incremental::FacetFieldIdChange;
|
||||
use crate::{CboRoaringBitmapCodec, FieldId, GeoPoint, Index, InternalError, Result};
|
||||
|
||||
#[tracing::instrument(level = "trace", skip_all, target = "indexing::merge")]
|
||||
@@ -81,16 +80,35 @@ where
|
||||
}
|
||||
merge_caches_sorted(frozen, |key, DelAddRoaringBitmap { del, add }| {
|
||||
let current = database.get(&rtxn, key)?;
|
||||
match merge_cbo_bitmaps(current, del, add)? {
|
||||
Operation::Write(bitmap) => {
|
||||
|
||||
if let (Some(del), Some(current)) = (&del, ¤t) {
|
||||
let current = CboRoaringBitmapCodec::deserialize_from(current).unwrap();
|
||||
let diff = del - ¤t;
|
||||
let external_ids = index.external_id_of(&rtxn, &diff).unwrap().into_iter().map(|id| id.unwrap()).collect::<Vec<_>>();
|
||||
if !del.is_subset(¤t) {
|
||||
eprintln!(
|
||||
"======================== {:?}: {} -> c: {:?} d: {:?} a: {:?} extra: {:?} extra_external_ids: {:?}",
|
||||
D::DATABASE,
|
||||
D::DATABASE.stringify_key(key),
|
||||
¤t,
|
||||
del,
|
||||
add,
|
||||
diff,
|
||||
external_ids
|
||||
);
|
||||
}
|
||||
}
|
||||
match merge_cbo_bitmaps(current, del, add) {
|
||||
Ok(Operation::Write(bitmap)) => {
|
||||
docids_sender.write(key, &bitmap)?;
|
||||
Ok(())
|
||||
}
|
||||
Operation::Delete => {
|
||||
Ok(Operation::Delete) => {
|
||||
docids_sender.delete(key)?;
|
||||
Ok(())
|
||||
}
|
||||
Operation::Ignore => Ok(()),
|
||||
Ok(Operation::Ignore) => Ok(()),
|
||||
Err(e) => Err(e),
|
||||
}
|
||||
})
|
||||
})
|
||||
@@ -101,18 +119,12 @@ 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;
|
||||
let max_string_count = max_string_count.clamp(1000, 100_000);
|
||||
let max_number_count = max_number_count.clamp(1000, 100_000);
|
||||
transpose_and_freeze_caches(&mut caches)?
|
||||
.into_par_iter()
|
||||
.map(|frozen| {
|
||||
let mut facet_field_ids_delta =
|
||||
FacetFieldIdsDelta::new(max_string_count, max_number_count);
|
||||
let mut facet_field_ids_delta = FacetFieldIdsDelta::default();
|
||||
let rtxn = index.read_txn()?;
|
||||
merge_caches_sorted(frozen, |key, DelAddRoaringBitmap { del, add }| {
|
||||
let current = database.get_cbo_roaring_bytes_value(&rtxn, key)?;
|
||||
@@ -133,10 +145,7 @@ pub fn merge_and_send_facet_docids<'extractor>(
|
||||
|
||||
Ok(facet_field_ids_delta)
|
||||
})
|
||||
.reduce(
|
||||
|| Ok(FacetFieldIdsDelta::new(max_string_count, max_number_count)),
|
||||
|lhs, rhs| Ok(lhs?.merge(rhs?)),
|
||||
)
|
||||
.reduce(|| Ok(FacetFieldIdsDelta::default()), |lhs, rhs| Ok(lhs?.merge(rhs?)))
|
||||
}
|
||||
|
||||
pub struct FacetDatabases<'a> {
|
||||
@@ -165,131 +174,60 @@ impl<'a> FacetDatabases<'a> {
|
||||
}
|
||||
}
|
||||
|
||||
#[derive(Debug)]
|
||||
pub enum FacetFieldIdDelta {
|
||||
Bulk,
|
||||
Incremental(Vec<FacetFieldIdChange>),
|
||||
}
|
||||
|
||||
impl FacetFieldIdDelta {
|
||||
fn push(&mut self, facet_value: &[u8], 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() });
|
||||
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)]
|
||||
#[derive(Debug, Default)]
|
||||
pub struct FacetFieldIdsDelta {
|
||||
/// The field ids that have been modified
|
||||
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,
|
||||
modified_facet_string_ids: HashSet<FieldId>,
|
||||
modified_facet_number_ids: HashSet<FieldId>,
|
||||
}
|
||||
|
||||
impl FacetFieldIdsDelta {
|
||||
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_string_id(&mut self, field_id: FieldId) {
|
||||
self.modified_facet_string_ids.insert(field_id);
|
||||
}
|
||||
|
||||
fn register_facet_string_id(&mut self, field_id: FieldId, facet_value: &[u8]) {
|
||||
self.modified_facet_string_ids
|
||||
.entry(field_id)
|
||||
.or_insert(FacetFieldIdDelta::Incremental(Default::default()))
|
||||
.push(facet_value, self.max_string_count);
|
||||
}
|
||||
|
||||
fn register_facet_number_id(&mut self, field_id: FieldId, facet_value: &[u8]) {
|
||||
self.modified_facet_number_ids
|
||||
.entry(field_id)
|
||||
.or_insert(FacetFieldIdDelta::Incremental(Default::default()))
|
||||
.push(facet_value, self.max_number_count);
|
||||
fn register_facet_number_id(&mut self, field_id: FieldId) {
|
||||
self.modified_facet_number_ids.insert(field_id);
|
||||
}
|
||||
|
||||
fn register_from_key(&mut self, key: &[u8]) {
|
||||
let (facet_kind, field_id, facet_value) = self.extract_key_data(key);
|
||||
match (facet_kind, facet_value) {
|
||||
(FacetKind::Number, Some(facet_value)) => {
|
||||
self.register_facet_number_id(field_id, facet_value)
|
||||
}
|
||||
(FacetKind::String, Some(facet_value)) => {
|
||||
self.register_facet_string_id(field_id, facet_value)
|
||||
}
|
||||
let (facet_kind, field_id) = 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),
|
||||
_ => (),
|
||||
}
|
||||
}
|
||||
|
||||
fn extract_key_data<'key>(&self, key: &'key [u8]) -> (FacetKind, FieldId, Option<&'key [u8]>) {
|
||||
fn extract_key_data(&self, key: &[u8]) -> (FacetKind, FieldId) {
|
||||
let facet_kind = FacetKind::from(key[0]);
|
||||
let field_id = FieldId::from_be_bytes([key[1], key[2]]);
|
||||
let facet_value = if key.len() >= 4 {
|
||||
// level is also stored in the key at [3] (always 0)
|
||||
Some(&key[4..])
|
||||
} else {
|
||||
(facet_kind, field_id)
|
||||
}
|
||||
|
||||
pub fn modified_facet_string_ids(&self) -> Option<Vec<FieldId>> {
|
||||
if self.modified_facet_string_ids.is_empty() {
|
||||
None
|
||||
};
|
||||
|
||||
(facet_kind, field_id, facet_value)
|
||||
} 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 consume_facet_number_delta(
|
||||
&mut self,
|
||||
) -> impl Iterator<Item = (FieldId, FacetFieldIdDelta)> + '_ {
|
||||
self.modified_facet_number_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 merge(mut self, rhs: Self) -> Self {
|
||||
// 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);
|
||||
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);
|
||||
});
|
||||
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);
|
||||
modified_facet_string_ids.into_iter().for_each(|fid| {
|
||||
self.modified_facet_string_ids.insert(fid);
|
||||
});
|
||||
self
|
||||
}
|
||||
@@ -315,21 +253,46 @@ fn merge_cbo_bitmaps(
|
||||
(None, Some(_del), Some(add)) => Ok(Operation::Write(add)),
|
||||
(Some(_current), None, None) => Ok(Operation::Ignore), // but it's strange
|
||||
(Some(current), None, Some(add)) => Ok(Operation::Write(current | add)),
|
||||
(Some(current), Some(del), add) => {
|
||||
(Some(current), Some(mut del), add) => {
|
||||
debug_assert!(
|
||||
del.is_subset(¤t),
|
||||
"del is not a subset of current, which must be impossible."
|
||||
);
|
||||
let output = match add {
|
||||
Some(add) => (¤t - (&del - &add)) | (add - del),
|
||||
None => ¤t - del,
|
||||
Some(add) => {
|
||||
del -= &add;
|
||||
|
||||
if del.is_empty() {
|
||||
if add.is_subset(¤t) {
|
||||
// no changes, no allocation
|
||||
None
|
||||
} else {
|
||||
// addition
|
||||
Some(current | add)
|
||||
}
|
||||
} else {
|
||||
if add.is_subset(¤t) {
|
||||
// deletion only, no union
|
||||
Some(current - del)
|
||||
} else {
|
||||
// deletion and addition
|
||||
Some((current - del) | add)
|
||||
}
|
||||
}
|
||||
}
|
||||
// deletion only, no union
|
||||
None => Some(current - del),
|
||||
};
|
||||
if output.is_empty() {
|
||||
Ok(Operation::Delete)
|
||||
} else if current == output {
|
||||
Ok(Operation::Ignore)
|
||||
} else {
|
||||
Ok(Operation::Write(output))
|
||||
|
||||
match output {
|
||||
Some(output) => {
|
||||
if output.is_empty() {
|
||||
Ok(Operation::Delete)
|
||||
} else {
|
||||
Ok(Operation::Write(output))
|
||||
}
|
||||
}
|
||||
None => Ok(Operation::Ignore),
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
@@ -1,12 +1,8 @@
|
||||
use std::borrow::Cow;
|
||||
|
||||
use enum_iterator::Sequence;
|
||||
|
||||
use crate::progress::Step;
|
||||
|
||||
#[derive(Debug, Clone, Copy, PartialEq, Eq, Sequence)]
|
||||
#[repr(u8)]
|
||||
pub enum IndexingStep {
|
||||
#[repr(u16)]
|
||||
pub enum Step {
|
||||
PreparingPayloads,
|
||||
ExtractingDocuments,
|
||||
ExtractingFacets,
|
||||
@@ -22,31 +18,30 @@ pub enum IndexingStep {
|
||||
Finalizing,
|
||||
}
|
||||
|
||||
impl Step for IndexingStep {
|
||||
fn name(&self) -> Cow<'static, str> {
|
||||
impl Step {
|
||||
pub fn name(&self) -> &'static str {
|
||||
match self {
|
||||
IndexingStep::PreparingPayloads => "preparing update file",
|
||||
IndexingStep::ExtractingDocuments => "extracting documents",
|
||||
IndexingStep::ExtractingFacets => "extracting facets",
|
||||
IndexingStep::ExtractingWords => "extracting words",
|
||||
IndexingStep::ExtractingWordProximity => "extracting word proximity",
|
||||
IndexingStep::ExtractingEmbeddings => "extracting embeddings",
|
||||
IndexingStep::WritingGeoPoints => "writing geo points",
|
||||
IndexingStep::WritingToDatabase => "writing to database",
|
||||
IndexingStep::WaitingForExtractors => "waiting for extractors",
|
||||
IndexingStep::WritingEmbeddingsToDatabase => "writing embeddings to database",
|
||||
IndexingStep::PostProcessingFacets => "post-processing facets",
|
||||
IndexingStep::PostProcessingWords => "post-processing words",
|
||||
IndexingStep::Finalizing => "finalizing",
|
||||
Step::PreparingPayloads => "preparing update file",
|
||||
Step::ExtractingDocuments => "extracting documents",
|
||||
Step::ExtractingFacets => "extracting facets",
|
||||
Step::ExtractingWords => "extracting words",
|
||||
Step::ExtractingWordProximity => "extracting word proximity",
|
||||
Step::ExtractingEmbeddings => "extracting embeddings",
|
||||
Step::WritingGeoPoints => "writing geo points",
|
||||
Step::WritingToDatabase => "writing to database",
|
||||
Step::WaitingForExtractors => "waiting for extractors",
|
||||
Step::WritingEmbeddingsToDatabase => "writing embeddings to database",
|
||||
Step::PostProcessingFacets => "post-processing facets",
|
||||
Step::PostProcessingWords => "post-processing words",
|
||||
Step::Finalizing => "finalizing",
|
||||
}
|
||||
.into()
|
||||
}
|
||||
|
||||
fn current(&self) -> u32 {
|
||||
*self as u32
|
||||
pub fn finished_steps(self) -> u16 {
|
||||
self as u16
|
||||
}
|
||||
|
||||
fn total(&self) -> u32 {
|
||||
Self::CARDINALITY as u32
|
||||
pub const fn total_steps() -> u16 {
|
||||
Self::CARDINALITY as u16
|
||||
}
|
||||
}
|
||||
|
||||
@@ -1,10 +1,9 @@
|
||||
use std::collections::BTreeSet;
|
||||
|
||||
use bumpalo::Bump;
|
||||
use bumparaw_collections::RawMap;
|
||||
use deserr::{Deserr, IntoValue};
|
||||
use heed::RoTxn;
|
||||
use rustc_hash::FxBuildHasher;
|
||||
use raw_collections::RawMap;
|
||||
use serde::Serialize;
|
||||
use serde_json::value::RawValue;
|
||||
|
||||
@@ -85,7 +84,7 @@ pub struct VectorDocumentFromDb<'t> {
|
||||
docid: DocumentId,
|
||||
embedding_config: Vec<IndexEmbeddingConfig>,
|
||||
index: &'t Index,
|
||||
vectors_field: Option<RawMap<'t, FxBuildHasher>>,
|
||||
vectors_field: Option<RawMap<'t>>,
|
||||
rtxn: &'t RoTxn<'t>,
|
||||
doc_alloc: &'t Bump,
|
||||
}
|
||||
@@ -103,10 +102,9 @@ impl<'t> VectorDocumentFromDb<'t> {
|
||||
};
|
||||
let vectors = document.vectors_field()?;
|
||||
let vectors_field = match vectors {
|
||||
Some(vectors) => Some(
|
||||
RawMap::from_raw_value_and_hasher(vectors, FxBuildHasher, doc_alloc)
|
||||
.map_err(InternalError::SerdeJson)?,
|
||||
),
|
||||
Some(vectors) => {
|
||||
Some(RawMap::from_raw_value(vectors, doc_alloc).map_err(InternalError::SerdeJson)?)
|
||||
}
|
||||
None => None,
|
||||
};
|
||||
|
||||
@@ -222,7 +220,7 @@ fn entry_from_raw_value(
|
||||
|
||||
pub struct VectorDocumentFromVersions<'doc> {
|
||||
external_document_id: &'doc str,
|
||||
vectors: RawMap<'doc, FxBuildHasher>,
|
||||
vectors: RawMap<'doc>,
|
||||
embedders: &'doc EmbeddingConfigs,
|
||||
}
|
||||
|
||||
@@ -235,8 +233,8 @@ impl<'doc> VectorDocumentFromVersions<'doc> {
|
||||
) -> Result<Option<Self>> {
|
||||
let document = DocumentFromVersions::new(versions);
|
||||
if let Some(vectors_field) = document.vectors_field()? {
|
||||
let vectors = RawMap::from_raw_value_and_hasher(vectors_field, FxBuildHasher, bump)
|
||||
.map_err(UserError::SerdeJson)?;
|
||||
let vectors =
|
||||
RawMap::from_raw_value(vectors_field, bump).map_err(UserError::SerdeJson)?;
|
||||
Ok(Some(Self { external_document_id, vectors, embedders }))
|
||||
} else {
|
||||
Ok(None)
|
||||
|
||||
@@ -3,7 +3,6 @@ use bumpalo::Bump;
|
||||
use heed::EnvOpenOptions;
|
||||
use maplit::hashset;
|
||||
use milli::documents::mmap_from_objects;
|
||||
use milli::progress::Progress;
|
||||
use milli::update::new::indexer;
|
||||
use milli::update::{IndexDocumentsMethod, IndexerConfig, Settings};
|
||||
use milli::vector::EmbeddingConfigs;
|
||||
@@ -58,7 +57,7 @@ fn test_facet_distribution_with_no_facet_values() {
|
||||
None,
|
||||
&mut new_fields_ids_map,
|
||||
&|| false,
|
||||
Progress::default(),
|
||||
&|_progress| (),
|
||||
)
|
||||
.unwrap();
|
||||
|
||||
@@ -73,7 +72,7 @@ fn test_facet_distribution_with_no_facet_values() {
|
||||
&document_changes,
|
||||
embedders,
|
||||
&|| false,
|
||||
&Progress::default(),
|
||||
&|_| (),
|
||||
)
|
||||
.unwrap();
|
||||
|
||||
|
||||
@@ -7,7 +7,6 @@ use bumpalo::Bump;
|
||||
use either::{Either, Left, Right};
|
||||
use heed::EnvOpenOptions;
|
||||
use maplit::{btreemap, hashset};
|
||||
use milli::progress::Progress;
|
||||
use milli::update::new::indexer;
|
||||
use milli::update::{IndexDocumentsMethod, IndexerConfig, Settings};
|
||||
use milli::vector::EmbeddingConfigs;
|
||||
@@ -91,7 +90,7 @@ pub fn setup_search_index_with_criteria(criteria: &[Criterion]) -> Index {
|
||||
None,
|
||||
&mut new_fields_ids_map,
|
||||
&|| false,
|
||||
Progress::default(),
|
||||
&|_progress| (),
|
||||
)
|
||||
.unwrap();
|
||||
|
||||
@@ -110,7 +109,7 @@ pub fn setup_search_index_with_criteria(criteria: &[Criterion]) -> Index {
|
||||
&document_changes,
|
||||
embedders,
|
||||
&|| false,
|
||||
&Progress::default(),
|
||||
&|_| (),
|
||||
)
|
||||
.unwrap();
|
||||
|
||||
|
||||
@@ -5,7 +5,6 @@ use bumpalo::Bump;
|
||||
use heed::EnvOpenOptions;
|
||||
use itertools::Itertools;
|
||||
use maplit::hashset;
|
||||
use milli::progress::Progress;
|
||||
use milli::update::new::indexer;
|
||||
use milli::update::{IndexDocumentsMethod, IndexerConfig, Settings};
|
||||
use milli::vector::EmbeddingConfigs;
|
||||
@@ -327,7 +326,7 @@ fn criteria_ascdesc() {
|
||||
None,
|
||||
&mut new_fields_ids_map,
|
||||
&|| false,
|
||||
Progress::default(),
|
||||
&|_progress| (),
|
||||
)
|
||||
.unwrap();
|
||||
|
||||
@@ -342,7 +341,7 @@ fn criteria_ascdesc() {
|
||||
&document_changes,
|
||||
embedders,
|
||||
&|| false,
|
||||
&Progress::default(),
|
||||
&|_| (),
|
||||
)
|
||||
.unwrap();
|
||||
|
||||
|
||||
@@ -3,7 +3,6 @@ use std::collections::BTreeSet;
|
||||
use bumpalo::Bump;
|
||||
use heed::EnvOpenOptions;
|
||||
use milli::documents::mmap_from_objects;
|
||||
use milli::progress::Progress;
|
||||
use milli::update::new::indexer;
|
||||
use milli::update::{IndexDocumentsMethod, IndexerConfig, Settings};
|
||||
use milli::vector::EmbeddingConfigs;
|
||||
@@ -136,7 +135,7 @@ fn test_typo_disabled_on_word() {
|
||||
None,
|
||||
&mut new_fields_ids_map,
|
||||
&|| false,
|
||||
Progress::default(),
|
||||
&|_progress| (),
|
||||
)
|
||||
.unwrap();
|
||||
|
||||
@@ -151,7 +150,7 @@ fn test_typo_disabled_on_word() {
|
||||
&document_changes,
|
||||
embedders,
|
||||
&|| false,
|
||||
&Progress::default(),
|
||||
&|_| (),
|
||||
)
|
||||
.unwrap();
|
||||
|
||||
|
||||
@@ -1,56 +1,23 @@
|
||||
use std::collections::BTreeMap;
|
||||
use std::time::Duration;
|
||||
|
||||
use anyhow::{bail, Context as _};
|
||||
use tokio::process::Command;
|
||||
use tokio::time;
|
||||
|
||||
use super::assets::Asset;
|
||||
use super::client::Client;
|
||||
use super::workload::Workload;
|
||||
|
||||
pub async fn kill(mut meilisearch: tokio::process::Child) {
|
||||
let Some(id) = meilisearch.id() else { return };
|
||||
|
||||
match Command::new("kill").args(["--signal=TERM", &id.to_string()]).spawn() {
|
||||
Ok(mut cmd) => {
|
||||
let Err(error) = cmd.wait().await else { return };
|
||||
tracing::warn!(
|
||||
error = &error as &dyn std::error::Error,
|
||||
"while awaiting the Meilisearch server kill"
|
||||
);
|
||||
}
|
||||
Err(error) => {
|
||||
tracing::warn!(
|
||||
error = &error as &dyn std::error::Error,
|
||||
"while terminating Meilisearch server with a kill -s TERM"
|
||||
);
|
||||
if let Err(error) = meilisearch.kill().await {
|
||||
tracing::warn!(
|
||||
error = &error as &dyn std::error::Error,
|
||||
"while terminating Meilisearch server"
|
||||
)
|
||||
}
|
||||
return;
|
||||
}
|
||||
};
|
||||
|
||||
match time::timeout(Duration::from_secs(5), meilisearch.wait()).await {
|
||||
Ok(_) => (),
|
||||
Err(_) => {
|
||||
if let Err(error) = meilisearch.kill().await {
|
||||
tracing::warn!(
|
||||
error = &error as &dyn std::error::Error,
|
||||
"while terminating Meilisearch server"
|
||||
)
|
||||
}
|
||||
}
|
||||
if let Err(error) = meilisearch.kill().await {
|
||||
tracing::warn!(
|
||||
error = &error as &dyn std::error::Error,
|
||||
"while terminating Meilisearch server"
|
||||
)
|
||||
}
|
||||
}
|
||||
|
||||
#[tracing::instrument]
|
||||
pub async fn build() -> anyhow::Result<()> {
|
||||
let mut command = Command::new("cargo");
|
||||
let mut command = tokio::process::Command::new("cargo");
|
||||
command.arg("build").arg("--release").arg("-p").arg("meilisearch");
|
||||
|
||||
command.kill_on_drop(true);
|
||||
@@ -70,8 +37,17 @@ pub async fn start(
|
||||
master_key: Option<&str>,
|
||||
workload: &Workload,
|
||||
asset_folder: &str,
|
||||
mut command: Command,
|
||||
) -> anyhow::Result<tokio::process::Child> {
|
||||
let mut command = tokio::process::Command::new("cargo");
|
||||
command
|
||||
.arg("run")
|
||||
.arg("--release")
|
||||
.arg("-p")
|
||||
.arg("meilisearch")
|
||||
.arg("--bin")
|
||||
.arg("meilisearch")
|
||||
.arg("--");
|
||||
|
||||
command.arg("--db-path").arg("./_xtask_benchmark.ms");
|
||||
if let Some(master_key) = master_key {
|
||||
command.arg("--master-key").arg(master_key);
|
||||
@@ -110,7 +86,7 @@ async fn wait_for_health(
|
||||
|
||||
return Ok(());
|
||||
}
|
||||
time::sleep(Duration::from_millis(500)).await;
|
||||
tokio::time::sleep(std::time::Duration::from_millis(500)).await;
|
||||
// check whether the Meilisearch instance exited early (cut the wait)
|
||||
if let Some(exit_code) =
|
||||
meilisearch.try_wait().context("cannot check Meilisearch server process status")?
|
||||
|
||||
@@ -86,12 +86,6 @@ pub struct BenchDeriveArgs {
|
||||
/// The maximum time in seconds we allow for fetching the task queue before timing out.
|
||||
#[arg(long, default_value_t = 60)]
|
||||
tasks_queue_timeout_secs: u64,
|
||||
|
||||
/// The path to the binary to run.
|
||||
///
|
||||
/// If unspecified, runs `cargo run` after building Meilisearch with `cargo build`.
|
||||
#[arg(long)]
|
||||
binary_path: Option<PathBuf>,
|
||||
}
|
||||
|
||||
pub fn run(args: BenchDeriveArgs) -> anyhow::Result<()> {
|
||||
@@ -176,7 +170,6 @@ pub fn run(args: BenchDeriveArgs) -> anyhow::Result<()> {
|
||||
args.master_key.as_deref(),
|
||||
workload,
|
||||
&args,
|
||||
args.binary_path.as_deref(),
|
||||
)
|
||||
.await?;
|
||||
|
||||
|
||||
@@ -1,7 +1,6 @@
|
||||
use std::collections::BTreeMap;
|
||||
use std::fs::File;
|
||||
use std::io::{Seek as _, Write as _};
|
||||
use std::path::Path;
|
||||
|
||||
use anyhow::{bail, Context as _};
|
||||
use futures_util::TryStreamExt as _;
|
||||
@@ -86,13 +85,13 @@ pub async fn execute(
|
||||
master_key: Option<&str>,
|
||||
workload: Workload,
|
||||
args: &BenchDeriveArgs,
|
||||
binary_path: Option<&Path>,
|
||||
) -> anyhow::Result<()> {
|
||||
assets::fetch_assets(assets_client, &workload.assets, &args.asset_folder).await?;
|
||||
|
||||
let workload_uuid = dashboard_client.create_workload(invocation_uuid, &workload).await?;
|
||||
|
||||
let mut tasks = Vec::new();
|
||||
|
||||
for i in 0..workload.run_count {
|
||||
tasks.push(
|
||||
execute_run(
|
||||
@@ -103,7 +102,6 @@ pub async fn execute(
|
||||
master_key,
|
||||
&workload,
|
||||
args,
|
||||
binary_path,
|
||||
i,
|
||||
)
|
||||
.await?,
|
||||
@@ -111,6 +109,7 @@ pub async fn execute(
|
||||
}
|
||||
|
||||
let mut reports = Vec::with_capacity(workload.run_count as usize);
|
||||
|
||||
for task in tasks {
|
||||
reports.push(
|
||||
task.await
|
||||
@@ -134,31 +133,13 @@ async fn execute_run(
|
||||
master_key: Option<&str>,
|
||||
workload: &Workload,
|
||||
args: &BenchDeriveArgs,
|
||||
binary_path: Option<&Path>,
|
||||
run_number: u16,
|
||||
) -> anyhow::Result<tokio::task::JoinHandle<anyhow::Result<std::fs::File>>> {
|
||||
meili_process::delete_db();
|
||||
|
||||
let run_command = match binary_path {
|
||||
Some(binary_path) => tokio::process::Command::new(binary_path),
|
||||
None => {
|
||||
meili_process::build().await?;
|
||||
let mut command = tokio::process::Command::new("cargo");
|
||||
command
|
||||
.arg("run")
|
||||
.arg("--release")
|
||||
.arg("-p")
|
||||
.arg("meilisearch")
|
||||
.arg("--bin")
|
||||
.arg("meilisearch")
|
||||
.arg("--");
|
||||
command
|
||||
}
|
||||
};
|
||||
|
||||
meili_process::build().await?;
|
||||
let meilisearch =
|
||||
meili_process::start(meili_client, master_key, workload, &args.asset_folder, run_command)
|
||||
.await?;
|
||||
meili_process::start(meili_client, master_key, workload, &args.asset_folder).await?;
|
||||
|
||||
let processor = run_commands(
|
||||
dashboard_client,
|
||||
|
||||
Reference in New Issue
Block a user