searchengines-web/MeiliSearch
searchengines-web/MeiliSearch
1
0
Fork 0
mirror of https://github.com/meilisearch/MeiliSearch synced 2024-12-26 06:30:05 +01:00
Code Issues Releases Wiki Activity

Merge #4223

Browse Source 
4223: Update to heed 0.20 r=dureuill a=Kerollmops

This PR brings the v0.20-alpha.9 version of heed into Meilisearch 🎉 The main goal is to test it in a real environment to make the necessary changes if needed. We also want to merge it as soon as possible during the pre-release phase to ensure we catch bugs before the release.

Most of the calls to heed are the same as before, except:
 - The `PolyDatabase` has been replaced with a `Database<Unspecified, Unspecified>`. We replaced the `get<T, U>()` by a `remap<T, U>().get()` calls.
 - The `Database` `append(...)` method has been replaced with a `put_with_flags(PutFlags::APPEND, ...)`.
 - The `RwTxn<'e, 'p>` has been simplified into a `RwTxn<'e>`.
 - The `BytesEncode/Decode` traits return a `Result<_, BoxedError>` instead of an `Option<_>`.
 - We no longer need to wrap and unwrap the `BEU32` integer when storing/getting them from heed.

### TODO
 - [x] Create actual, simple error types instead of using strings in the codecs.

### Follow-up work
 - Move the codecs into another member crate (we depend on the uuid one in the meilitool crate).
 - Display the internal decoding error in the `SerializationError` internal error variant.

Co-authored-by: Clément Renault <clement@meilisearch.com>
This commit is contained in:
meili-bors[bot] 2023-11-28 13:39:44 +00:00 committed by GitHub
commit 6376c342c1
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
83 changed files with 962 additions and 904 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

103
Cargo.lock generated
View File

@ -520,6 +520,9 @@ name = "bitflags"
version = "2.3.3"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "630be753d4e58660abd17930c71b647fe46c27ea6b63cc59e1e3851406972e42"
dependencies = [
"serde",
]
[[package]]
name = "block-buffer"
@ -1255,6 +1258,15 @@ dependencies = [
"syn 2.0.28",
]
[[package]]
name = "doxygen-rs"
version = "0.2.2"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "bff670ea0c9bbb8414e7efa6e23ebde2b8f520a7eef78273a3918cf1903e7505"
dependencies = [
"phf",
]
[[package]]
name = "dump"
version = "1.5.0"
@ -1811,36 +1823,40 @@ checksum = "95505c38b4572b2d910cecb0281560f54b440a19336cbbcb27bf6ce6adc6f5a8"
[[package]]
name = "heed"
version = "0.12.7"
source = "git+https://github.com/meilisearch/heed?tag=v0.12.7#061a5276b1f336f5f3302bee291e336041d88632"
version = "0.20.0-alpha.9"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "9648a50991c86df7d00c56c268c27754fcf4c80be2ba57fc4a00dc928c6fe934"
dependencies = [
"bitflags 2.3.3",
"bytemuck",
"byteorder",
"heed-traits",
"heed-types",
"libc",
"lmdb-rkv-sys",
"lmdb-master-sys",
"once_cell",
"page_size 0.4.2",
"page_size 0.6.0",
"synchronoise",
"url",
"zerocopy",
]
[[package]]
name = "heed-traits"
version = "0.7.0"
source = "git+https://github.com/meilisearch/heed?tag=v0.12.7#061a5276b1f336f5f3302bee291e336041d88632"
version = "0.20.0-alpha.9"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "5ab0b7d9cde969ad36dde692e487dc89d97f7168bf6a7bd3b894ad4bf7278298"
[[package]]
name = "heed-types"
version = "0.7.2"
source = "git+https://github.com/meilisearch/heed?tag=v0.12.7#061a5276b1f336f5f3302bee291e336041d88632"
version = "0.20.0-alpha.9"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "f0cb3567a7363f28b597bf6e9897b9466397951dd0e52df2c8196dd8a71af44a"
dependencies = [
"bincode",
"byteorder",
"heed-traits",
"serde",
"serde_json",
"zerocopy",
]
[[package]]
@ -2968,11 +2984,13 @@ dependencies = [
]
[[package]]
name = "lmdb-rkv-sys"
version = "0.15.1"
source = "git+https://github.com/meilisearch/lmdb-rs#501aa34a1ab7f092e3ff54a6c22ff6c55931a2d8"
name = "lmdb-master-sys"
version = "0.1.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "629c123f5321b48fa4f8f4d3b868165b748d9ba79c7103fb58e3a94f736bcedd"
dependencies = [
"cc",
"doxygen-rs",
"libc",
"pkg-config",
]
@ -3472,9 +3490,9 @@ dependencies = [
[[package]]
name = "page_size"
version = "0.4.2"
version = "0.5.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "eebde548fbbf1ea81a99b128872779c437752fb99f217c45245e1a61dcd9edcd"
checksum = "1b7663cbd190cfd818d08efa8497f6cd383076688c49a391ef7c0d03cd12b561"
dependencies = [
"libc",
"winapi",
@ -3482,9 +3500,9 @@ dependencies = [
[[package]]
name = "page_size"
version = "0.5.0"
version = "0.6.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "1b7663cbd190cfd818d08efa8497f6cd383076688c49a391ef7c0d03cd12b561"
checksum = "30d5b2194ed13191c1999ae0704b7839fb18384fa22e49b57eeaa97d79ce40da"
dependencies = [
"libc",
"winapi",
@ -3630,6 +3648,7 @@ version = "0.11.2"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "ade2d8b8f33c7333b51bcf0428d37e217e9f32192ae4772156f65063b8ce03dc"
dependencies = [
"phf_macros",
"phf_shared",
]
@ -3653,6 +3672,19 @@ dependencies = [
"rand",
]
[[package]]
name = "phf_macros"
version = "0.11.2"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "3444646e286606587e49f3bcf1679b8cef1dc2c5ecc29ddacaffc305180d464b"
dependencies = [
"phf_generator",
"phf_shared",
"proc-macro2",
"quote",
"syn 2.0.28",
]
[[package]]
name = "phf_shared"
version = "0.11.2"
@ -4479,18 +4511,6 @@ dependencies = [
"crossbeam-queue",
]
[[package]]
name = "synstructure"
version = "0.12.6"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "f36bdaa60a83aca3921b5259d5400cbf5e90fc51931376a9bd4a0eb79aa7210f"
dependencies = [
"proc-macro2",
"quote",
"syn 1.0.109",
"unicode-xid",
]
[[package]]
name = "synstructure"
version = "0.13.0"
@ -5359,28 +5379,7 @@ dependencies = [
"proc-macro2",
"quote",
"syn 2.0.28",
"synstructure 0.13.0",
]
[[package]]
name = "zerocopy"
version = "0.3.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "6580539ad917b7c026220c4b3f2c08d52ce54d6ce0dc491e66002e35388fab46"
dependencies = [
"byteorder",
"zerocopy-derive",
]
[[package]]
name = "zerocopy-derive"
version = "0.2.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "d498dbd1fd7beb83c86709ae1c33ca50942889473473d287d56ce4770a18edfb"
dependencies = [
"proc-macro2",
"syn 1.0.109",
"synstructure 0.12.6",
"synstructure",
]
[[package]]
@ -5401,7 +5400,7 @@ dependencies = [
"proc-macro2",
"quote",
"syn 2.0.28",
"synstructure 0.13.0",
"synstructure",
]
[[package]]

2
benchmarks/benches/indexing.rs
View File

@ -36,7 +36,7 @@ fn setup_index() -> Index {
}
fn setup_settings<'t>(
wtxn: &mut RwTxn<'t, '_>,
wtxn: &mut RwTxn<'t>,
index: &'t Index,
primary_key: &str,
searchable_fields: &[&str],

12
dump/src/reader/mod.rs
View File

@ -13,12 +13,12 @@ use crate::{Result, Version};
mod compat;
pub(self) mod v1;
pub(self) mod v2;
pub(self) mod v3;
pub(self) mod v4;
pub(self) mod v5;
pub(self) mod v6;
mod v1;
mod v2;
mod v3;
mod v4;
mod v5;
mod v6;
pub type Document = serde_json::Map<String, serde_json::Value>;
pub type UpdateFile = dyn Iterator<Item = Result<Document>>;

3
dump/src/reader/v1/settings.rs
View File

@ -56,8 +56,7 @@ pub enum RankingRule {
Desc(String),
}
static ASC_DESC_REGEX: Lazy<Regex> =
Lazy::new(|| Regex::new(r#"(asc|desc)\(([\w_-]+)\)"#).unwrap());
static ASC_DESC_REGEX: Lazy<Regex> = Lazy::new(|| Regex::new(r"(asc|desc)\(([\w_-]+)\)").unwrap());
impl FromStr for RankingRule {
type Err = ();

8
filter-parser/src/lib.rs
View File

@ -564,10 +564,10 @@ pub mod tests {
#[test]
fn parse_escaped() {
insta::assert_display_snapshot!(p(r#"title = 'foo\\'"#), @r#"{title} = {foo\}"#);
insta::assert_display_snapshot!(p(r#"title = 'foo\\\\'"#), @r#"{title} = {foo\\}"#);
insta::assert_display_snapshot!(p(r#"title = 'foo\\\\\\'"#), @r#"{title} = {foo\\\}"#);
insta::assert_display_snapshot!(p(r#"title = 'foo\\\\\\\\'"#), @r#"{title} = {foo\\\\}"#);
insta::assert_display_snapshot!(p(r"title = 'foo\\'"), @r#"{title} = {foo\}"#);
insta::assert_display_snapshot!(p(r"title = 'foo\\\\'"), @r#"{title} = {foo\\}"#);
insta::assert_display_snapshot!(p(r"title = 'foo\\\\\\'"), @r#"{title} = {foo\\\}"#);
insta::assert_display_snapshot!(p(r"title = 'foo\\\\\\\\'"), @r#"{title} = {foo\\\\}"#);
// but it also works with other sequencies
insta::assert_display_snapshot!(p(r#"title = 'foo\x20\n\t\"\'"'"#), @"{title} = {foo \n\t\"\'\"}");
}

22
filter-parser/src/value.rs
View File

@ -270,8 +270,8 @@ pub mod test {
("aaaa", "", rtok("", "aaaa"), "aaaa"),
(r#"aa"aa"#, r#""aa"#, rtok("", "aa"), "aa"),
(r#"aa\"aa"#, r#""#, rtok("", r#"aa\"aa"#), r#"aa"aa"#),
(r#"aa\\\aa"#, r#""#, rtok("", r#"aa\\\aa"#), r#"aa\\\aa"#),
(r#"aa\\"\aa"#, r#""\aa"#, rtok("", r#"aa\\"#), r#"aa\\"#),
(r"aa\\\aa", r#""#, rtok("", r"aa\\\aa"), r"aa\\\aa"),
(r#"aa\\"\aa"#, r#""\aa"#, rtok("", r"aa\\"), r"aa\\"),
(r#"aa\\\"\aa"#, r#""#, rtok("", r#"aa\\\"\aa"#), r#"aa\\"\aa"#),
(r#"\"\""#, r#""#, rtok("", r#"\"\""#), r#""""#),
];
@ -301,12 +301,12 @@ pub mod test {
);
// simple quote
assert_eq!(
unescape(Span::new_extra(r#"Hello \'World\'"#, ""), '\''),
unescape(Span::new_extra(r"Hello \'World\'", ""), '\''),
r#"Hello 'World'"#.to_string()
);
assert_eq!(
unescape(Span::new_extra(r#"Hello \\\'World\\\'"#, ""), '\''),
r#"Hello \\'World\\'"#.to_string()
unescape(Span::new_extra(r"Hello \\\'World\\\'", ""), '\''),
r"Hello \\'World\\'".to_string()
);
}
@ -335,19 +335,19 @@ pub mod test {
("\"cha'nnel\"", "cha'nnel", false),
("I'm tamo", "I", false),
// escaped thing but not quote
(r#""\\""#, r#"\"#, true),
(r#""\\\\\\""#, r#"\\\"#, true),
(r#""aa\\aa""#, r#"aa\aa"#, true),
(r#""\\""#, r"\", true),
(r#""\\\\\\""#, r"\\\", true),
(r#""aa\\aa""#, r"aa\aa", true),
// with double quote
(r#""Hello \"world\"""#, r#"Hello "world""#, true),
(r#""Hello \\\"world\\\"""#, r#"Hello \"world\""#, true),
(r#""I'm \"super\" tamo""#, r#"I'm "super" tamo"#, true),
(r#""\"\"""#, r#""""#, true),
// with simple quote
(r#"'Hello \'world\''"#, r#"Hello 'world'"#, true),
(r#"'Hello \\\'world\\\''"#, r#"Hello \'world\'"#, true),
(r"'Hello \'world\''", r#"Hello 'world'"#, true),
(r"'Hello \\\'world\\\''", r"Hello \'world\'", true),
(r#"'I\'m "super" tamo'"#, r#"I'm "super" tamo"#, true),
(r#"'\'\''"#, r#"''"#, true),
(r"'\'\''", r#"''"#, true),
];
for (input, expected, escaped) in test_case {

2
fuzzers/src/bin/fuzz-indexing.rs
View File

@ -113,7 +113,7 @@ fn main() {
index.documents(&wtxn, res.documents_ids).unwrap();
progression.fetch_add(1, Ordering::Relaxed);
}
wtxn.abort().unwrap();
wtxn.abort();
});
if let err @ Err(_) = handle.join() {
stop.store(true, Ordering::Relaxed);

21
index-scheduler/src/batch.rs
View File

@ -32,7 +32,7 @@ use meilisearch_types::milli::heed::CompactionOption;
use meilisearch_types::milli::update::{
IndexDocumentsConfig, IndexDocumentsMethod, IndexerConfig, Settings as MilliSettings,
};
use meilisearch_types::milli::{self, Filter, BEU32};
use meilisearch_types::milli::{self, Filter};
use meilisearch_types::settings::{apply_settings_to_builder, Settings, Unchecked};
use meilisearch_types::tasks::{Details, IndexSwap, Kind, KindWithContent, Status, Task};
use meilisearch_types::{compression, Index, VERSION_FILE_NAME};
@ -715,7 +715,7 @@ impl IndexScheduler {
// 2. Snapshot the index-scheduler LMDB env
//
// When we call copy_to_path, LMDB opens a read transaction by itself,
// When we call copy_to_file, LMDB opens a read transaction by itself,
// we can't provide our own. It is an issue as we would like to know
// the update files to copy but new ones can be enqueued between the copy
// of the env and the new transaction we open to retrieve the enqueued tasks.
@ -728,7 +728,7 @@ impl IndexScheduler {
// 2.1 First copy the LMDB env of the index-scheduler
let dst = temp_snapshot_dir.path().join("tasks");
fs::create_dir_all(&dst)?;
self.env.copy_to_path(dst.join("data.mdb"), CompactionOption::Enabled)?;
self.env.copy_to_file(dst.join("data.mdb"), CompactionOption::Enabled)?;
// 2.2 Create a read transaction on the index-scheduler
let rtxn = self.env.read_txn()?;
@ -753,7 +753,7 @@ impl IndexScheduler {
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)?;
index.copy_to_path(dst.join("data.mdb"), CompactionOption::Enabled)?;
index.copy_to_file(dst.join("data.mdb"), CompactionOption::Enabled)?;
}
drop(rtxn);
@ -766,7 +766,7 @@ impl IndexScheduler {
.map_size(1024 * 1024 * 1024) // 1 GiB
.max_dbs(2)
.open(&self.auth_path)?;
auth.copy_to_path(dst.join("data.mdb"), CompactionOption::Enabled)?;
auth.copy_to_file(dst.join("data.mdb"), CompactionOption::Enabled)?;
// 5. Copy and tarball the flat snapshot
// 5.1 Find the original name of the database
@ -1106,7 +1106,7 @@ impl IndexScheduler {
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, &BEU32::new(task_id), &task)?;
self.all_tasks.put(wtxn, &task_id, &task)?;
}
// 4. remove the task from indexuid = before_name
@ -1132,7 +1132,7 @@ impl IndexScheduler {
/// The list of processed tasks.
fn apply_index_operation<'i>(
&self,
index_wtxn: &mut RwTxn<'i, '_>,
index_wtxn: &mut RwTxn<'i>,
index: &'i Index,
operation: IndexOperation,
) -> Result<Vec<Task>> {
@ -1479,10 +1479,9 @@ impl IndexScheduler {
}
for task in to_delete_tasks.iter() {
self.all_tasks.delete(wtxn, &BEU32::new(task))?;
self.all_tasks.delete(wtxn, &task)?;
}
for canceled_by in affected_canceled_by {
let canceled_by = BEU32::new(canceled_by);
if let Some(mut tasks) = self.canceled_by.get(wtxn, &canceled_by)? {
tasks -= &to_delete_tasks;
if tasks.is_empty() {
@ -1530,14 +1529,14 @@ impl IndexScheduler {
task.details = task.details.map(|d| d.to_failed());
self.update_task(wtxn, &task)?;
}
self.canceled_by.put(wtxn, &BEU32::new(cancel_task_id), &tasks_to_cancel)?;
self.canceled_by.put(wtxn, &cancel_task_id, &tasks_to_cancel)?;
Ok(content_files_to_delete)
}
}
fn delete_document_by_filter<'a>(
wtxn: &mut RwTxn<'a, '_>,
wtxn: &mut RwTxn<'a>,
filter: &serde_json::Value,
indexer_config: &IndexerConfig,
must_stop_processing: MustStopProcessing,

12
index-scheduler/src/index_mapper/index_map.rs
View File

@ -1,12 +1,8 @@
/// the map size to use when we don't succeed in reading it in indexes.
const DEFAULT_MAP_SIZE: usize = 10 * 1024 * 1024 * 1024; // 10 GiB
use std::collections::BTreeMap;
use std::path::Path;
use std::time::Duration;
use meilisearch_types::heed::flags::Flags;
use meilisearch_types::heed::{EnvClosingEvent, EnvOpenOptions};
use meilisearch_types::heed::{EnvClosingEvent, EnvFlags, EnvOpenOptions};
use meilisearch_types::milli::Index;
use time::OffsetDateTime;
use uuid::Uuid;
@ -236,7 +232,7 @@ impl IndexMap {
enable_mdb_writemap: bool,
map_size_growth: usize,
) {
let map_size = index.map_size().unwrap_or(DEFAULT_MAP_SIZE) + map_size_growth;
let map_size = index.map_size() + map_size_growth;
let closing_event = index.prepare_for_closing();
let generation = self.next_generation();
self.unavailable.insert(
@ -309,7 +305,7 @@ fn create_or_open_index(
options.map_size(clamp_to_page_size(map_size));
options.max_readers(1024);
if enable_mdb_writemap {
unsafe { options.flag(Flags::MdbWriteMap) };
unsafe { options.flags(EnvFlags::WRITE_MAP) };
}
if let Some((created, updated)) = date {
@ -388,7 +384,7 @@ mod tests {
fn assert_index_size(index: Index, expected: usize) {
let expected = clamp_to_page_size(expected);
let index_map_size = index.map_size().unwrap();
let index_map_size = index.map_size();
assert_eq!(index_map_size, expected);
}
}

14
index-scheduler/src/insta_snapshot.rs
View File

@ -1,7 +1,7 @@
use std::collections::BTreeSet;
use std::fmt::Write;
use meilisearch_types::heed::types::{OwnedType, SerdeBincode, SerdeJson, Str};
use meilisearch_types::heed::types::{SerdeBincode, SerdeJson, Str};
use meilisearch_types::heed::{Database, RoTxn};
use meilisearch_types::milli::{CboRoaringBitmapCodec, RoaringBitmapCodec, BEU32};
use meilisearch_types::tasks::{Details, Task};
@ -115,7 +115,7 @@ pub fn snapshot_bitmap(r: &RoaringBitmap) -> String {
snap
}
pub fn snapshot_all_tasks(rtxn: &RoTxn, db: Database<OwnedType<BEU32>, SerdeJson<Task>>) -> String {
pub fn snapshot_all_tasks(rtxn: &RoTxn, db: Database<BEU32, SerdeJson<Task>>) -> String {
let mut snap = String::new();
let iter = db.iter(rtxn).unwrap();
for next in iter {
@ -125,10 +125,7 @@ pub fn snapshot_all_tasks(rtxn: &RoTxn, db: Database<OwnedType<BEU32>, SerdeJson
snap
}
pub fn snapshot_date_db(
rtxn: &RoTxn,
db: Database<OwnedType<BEI128>, CboRoaringBitmapCodec>,
) -> String {
pub fn snapshot_date_db(rtxn: &RoTxn, db: Database<BEI128, CboRoaringBitmapCodec>) -> String {
let mut snap = String::new();
let iter = db.iter(rtxn).unwrap();
for next in iter {
@ -248,10 +245,7 @@ pub fn snapshot_index_tasks(rtxn: &RoTxn, db: Database<Str, RoaringBitmapCodec>)
}
snap
}
pub fn snapshot_canceled_by(
rtxn: &RoTxn,
db: Database<OwnedType<BEU32>, RoaringBitmapCodec>,
) -> String {
pub fn snapshot_canceled_by(rtxn: &RoTxn, db: Database<BEU32, RoaringBitmapCodec>) -> String {
let mut snap = String::new();
let iter = db.iter(rtxn).unwrap();
for next in iter {

38
index-scheduler/src/lib.rs
View File

@ -47,8 +47,9 @@ pub use features::RoFeatures;
use file_store::FileStore;
use meilisearch_types::error::ResponseError;
use meilisearch_types::features::{InstanceTogglableFeatures, RuntimeTogglableFeatures};
use meilisearch_types::heed::types::{OwnedType, SerdeBincode, SerdeJson, Str};
use meilisearch_types::heed::{self, Database, Env, RoTxn, RwTxn};
use meilisearch_types::heed::byteorder::BE;
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::update::IndexerConfig;
use meilisearch_types::milli::{self, CboRoaringBitmapCodec, Index, RoaringBitmapCodec, BEU32};
@ -64,8 +65,7 @@ use uuid::Uuid;
use crate::index_mapper::IndexMapper;
use crate::utils::{check_index_swap_validity, clamp_to_page_size};
pub(crate) type BEI128 =
meilisearch_types::heed::zerocopy::I128<meilisearch_types::heed::byteorder::BE>;
pub(crate) type BEI128 = I128<BE>;
/// Defines a subset of tasks to be retrieved from the [`IndexScheduler`].
///
@ -278,7 +278,7 @@ pub struct IndexScheduler {
pub(crate) file_store: FileStore,
// The main database, it contains all the tasks accessible by their Id.
pub(crate) all_tasks: Database<OwnedType<BEU32>, SerdeJson<Task>>,
pub(crate) all_tasks: Database<BEU32, SerdeJson<Task>>,
/// All the tasks ids grouped by their status.
// TODO we should not be able to serialize a `Status::Processing` in this database.
@ -289,16 +289,16 @@ pub struct IndexScheduler {
pub(crate) index_tasks: Database<Str, RoaringBitmapCodec>,
/// Store the tasks that were canceled by a task uid
pub(crate) canceled_by: Database<OwnedType<BEU32>, RoaringBitmapCodec>,
pub(crate) canceled_by: Database<BEU32, RoaringBitmapCodec>,
/// Store the task ids of tasks which were enqueued at a specific date
pub(crate) enqueued_at: Database<OwnedType<BEI128>, CboRoaringBitmapCodec>,
pub(crate) enqueued_at: Database<BEI128, CboRoaringBitmapCodec>,
/// Store the task ids of finished tasks which started being processed at a specific date
pub(crate) started_at: Database<OwnedType<BEI128>, CboRoaringBitmapCodec>,
pub(crate) started_at: Database<BEI128, CboRoaringBitmapCodec>,
/// Store the task ids of tasks which finished at a specific date
pub(crate) finished_at: Database<OwnedType<BEI128>, CboRoaringBitmapCodec>,
pub(crate) finished_at: Database<BEI128, CboRoaringBitmapCodec>,
/// In charge of creating, opening, storing and returning indexes.
pub(crate) index_mapper: IndexMapper,
@ -730,9 +730,7 @@ impl IndexScheduler {
if let Some(canceled_by) = &query.canceled_by {
let mut all_canceled_tasks = RoaringBitmap::new();
for cancel_task_uid in canceled_by {
if let Some(canceled_by_uid) =
self.canceled_by.get(rtxn, &BEU32::new(*cancel_task_uid))?
{
if let Some(canceled_by_uid) = self.canceled_by.get(rtxn, cancel_task_uid)? {
all_canceled_tasks |= canceled_by_uid;
}
}
@ -983,7 +981,7 @@ impl IndexScheduler {
// if the task doesn't delete anything and 50% of the task queue is full, we must refuse to enqueue the incomming task
if !matches!(&kind, KindWithContent::TaskDeletion { tasks, .. } if !tasks.is_empty())
&& (self.env.non_free_pages_size()? * 100) / self.env.map_size()? as u64 > 50
&& (self.env.non_free_pages_size()? * 100) / self.env.info().map_size as u64 > 50
{
return Err(Error::NoSpaceLeftInTaskQueue);
}
@ -1009,7 +1007,7 @@ impl IndexScheduler {
// Get rid of the mutability.
let task = task;
self.all_tasks.append(&mut wtxn, &BEU32::new(task.uid), &task)?;
self.all_tasks.put_with_flags(&mut wtxn, PutFlags::APPEND, &task.uid, &task)?;
for index in task.indexes() {
self.update_index(&mut wtxn, index, |bitmap| {
@ -1187,7 +1185,7 @@ impl IndexScheduler {
| Err(Error::AbortedTask) => {
#[cfg(test)]
self.breakpoint(Breakpoint::AbortedIndexation);
wtxn.abort().map_err(Error::HeedTransaction)?;
wtxn.abort();
// We make sure that we don't call `stop_processing` on the `processing_tasks`,
// this is because we want to let the next tick call `create_next_batch` and keep
@ -1208,7 +1206,7 @@ impl IndexScheduler {
let index_uid = index_uid.unwrap();
// fixme: handle error more gracefully? not sure when this could happen
self.index_mapper.resize_index(&wtxn, &index_uid)?;
wtxn.abort().map_err(Error::HeedTransaction)?;
wtxn.abort();
return Ok(TickOutcome::TickAgain(0));
}
@ -1354,7 +1352,7 @@ impl IndexScheduler {
pub struct Dump<'a> {
index_scheduler: &'a IndexScheduler,
wtxn: RwTxn<'a, 'a>,
wtxn: RwTxn<'a>,
indexes: HashMap<String, RoaringBitmap>,
statuses: HashMap<Status, RoaringBitmap>,
@ -1469,7 +1467,7 @@ impl<'a> Dump<'a> {
},
};
self.index_scheduler.all_tasks.put(&mut self.wtxn, &BEU32::new(task.uid), &task)?;
self.index_scheduler.all_tasks.put(&mut self.wtxn, &task.uid, &task)?;
for index in task.indexes() {
match self.indexes.get_mut(index) {
@ -1511,8 +1509,8 @@ impl<'a> Dump<'a> {
}
}
self.statuses.entry(task.status).or_insert(RoaringBitmap::new()).insert(task.uid);
self.kinds.entry(task.kind.as_kind()).or_insert(RoaringBitmap::new()).insert(task.uid);
self.statuses.entry(task.status).or_default().insert(task.uid);
self.kinds.entry(task.kind.as_kind()).or_default().insert(task.uid);
Ok(task)
}

37
index-scheduler/src/utils.rs
View File

@ -3,9 +3,9 @@
use std::collections::{BTreeSet, HashSet};
use std::ops::Bound;
use meilisearch_types::heed::types::{DecodeIgnore, OwnedType};
use meilisearch_types::heed::types::DecodeIgnore;
use meilisearch_types::heed::{Database, RoTxn, RwTxn};
use meilisearch_types::milli::{CboRoaringBitmapCodec, BEU32};
use meilisearch_types::milli::CboRoaringBitmapCodec;
use meilisearch_types::tasks::{Details, IndexSwap, Kind, KindWithContent, Status};
use roaring::{MultiOps, RoaringBitmap};
use time::OffsetDateTime;
@ -18,7 +18,7 @@ impl IndexScheduler {
}
pub(crate) fn last_task_id(&self, rtxn: &RoTxn) -> Result<Option<TaskId>> {
Ok(self.all_tasks.remap_data_type::<DecodeIgnore>().last(rtxn)?.map(|(k, _)| k.get() + 1))
Ok(self.all_tasks.remap_data_type::<DecodeIgnore>().last(rtxn)?.map(|(k, _)| k + 1))
}
pub(crate) fn next_task_id(&self, rtxn: &RoTxn) -> Result<TaskId> {
@ -26,7 +26,7 @@ impl IndexScheduler {
}
pub(crate) fn get_task(&self, rtxn: &RoTxn, task_id: TaskId) -> Result<Option<Task>> {
Ok(self.all_tasks.get(rtxn, &BEU32::new(task_id))?)
Ok(self.all_tasks.get(rtxn, &task_id)?)
}
/// Convert an iterator to a `Vec` of tasks. The tasks MUST exist or a
@ -88,7 +88,7 @@ impl IndexScheduler {
}
}
self.all_tasks.put(wtxn, &BEU32::new(task.uid), task)?;
self.all_tasks.put(wtxn, &task.uid, task)?;
Ok(())
}
@ -169,11 +169,11 @@ impl IndexScheduler {
pub(crate) fn insert_task_datetime(
wtxn: &mut RwTxn,
database: Database<OwnedType<BEI128>, CboRoaringBitmapCodec>,
database: Database<BEI128, CboRoaringBitmapCodec>,
time: OffsetDateTime,
task_id: TaskId,
) -> Result<()> {
let timestamp = BEI128::new(time.unix_timestamp_nanos());
let timestamp = time.unix_timestamp_nanos();
let mut task_ids = database.get(wtxn, &timestamp)?.unwrap_or_default();
task_ids.insert(task_id);
database.put(wtxn, &timestamp, &RoaringBitmap::from_iter(task_ids))?;
@ -182,11 +182,11 @@ pub(crate) fn insert_task_datetime(
pub(crate) fn remove_task_datetime(
wtxn: &mut RwTxn,
database: Database<OwnedType<BEI128>, CboRoaringBitmapCodec>,
database: Database<BEI128, CboRoaringBitmapCodec>,
time: OffsetDateTime,
task_id: TaskId,
) -> Result<()> {
let timestamp = BEI128::new(time.unix_timestamp_nanos());
let timestamp = time.unix_timestamp_nanos();
if let Some(mut existing) = database.get(wtxn, &timestamp)? {
existing.remove(task_id);
if existing.is_empty() {
@ -202,7 +202,7 @@ pub(crate) fn remove_task_datetime(
pub(crate) fn keep_tasks_within_datetimes(
rtxn: &RoTxn,
tasks: &mut RoaringBitmap,
database: Database<OwnedType<BEI128>, CboRoaringBitmapCodec>,
database: Database<BEI128, CboRoaringBitmapCodec>,
after: Option<OffsetDateTime>,
before: Option<OffsetDateTime>,
) -> Result<()> {
@ -213,8 +213,8 @@ pub(crate) fn keep_tasks_within_datetimes(
(Some(after), Some(before)) => (Bound::Excluded(*after), Bound::Excluded(*before)),
};
let mut collected_task_ids = RoaringBitmap::new();
let start = map_bound(start, |b| BEI128::new(b.unix_timestamp_nanos()));
let end = map_bound(end, |b| BEI128::new(b.unix_timestamp_nanos()));
let start = map_bound(start, |b| b.unix_timestamp_nanos());
let end = map_bound(end, |b| b.unix_timestamp_nanos());
let iter = database.range(rtxn, &(start, end))?;
for r in iter {
let (_timestamp, task_ids) = r?;
@ -337,8 +337,6 @@ impl IndexScheduler {
let rtxn = self.env.read_txn().unwrap();
for task in self.all_tasks.iter(&rtxn).unwrap() {
let (task_id, task) = task.unwrap();
let task_id = task_id.get();
let task_index_uid = task.index_uid().map(ToOwned::to_owned);
let Task {
@ -361,16 +359,13 @@ impl IndexScheduler {
.unwrap()
.contains(task.uid));
}
let db_enqueued_at = self
.enqueued_at
.get(&rtxn, &BEI128::new(enqueued_at.unix_timestamp_nanos()))
.unwrap()
.unwrap();
let db_enqueued_at =
self.enqueued_at.get(&rtxn, &enqueued_at.unix_timestamp_nanos()).unwrap().unwrap();
assert!(db_enqueued_at.contains(task_id));
if let Some(started_at) = started_at {
let db_started_at = self
.started_at
.get(&rtxn, &BEI128::new(started_at.unix_timestamp_nanos()))
.get(&rtxn, &started_at.unix_timestamp_nanos())
.unwrap()
.unwrap();
assert!(db_started_at.contains(task_id));
@ -378,7 +373,7 @@ impl IndexScheduler {
if let Some(finished_at) = finished_at {
let db_finished_at = self
.finished_at
.get(&rtxn, &BEI128::new(finished_at.unix_timestamp_nanos()))
.get(&rtxn, &finished_at.unix_timestamp_nanos())
.unwrap()
.unwrap();
assert!(db_finished_at.contains(task_id));

10
index-scheduler/src/uuid_codec.rs
View File

@ -1,7 +1,7 @@
use std::borrow::Cow;
use std::convert::TryInto;
use meilisearch_types::heed::{BytesDecode, BytesEncode};
use meilisearch_types::heed::{BoxedError, BytesDecode, BytesEncode};
use uuid::Uuid;
/// A heed codec for value of struct Uuid.
@ -10,15 +10,15 @@ pub struct UuidCodec;
impl<'a> BytesDecode<'a> for UuidCodec {
type DItem = Uuid;
fn bytes_decode(bytes: &'a [u8]) -> Option<Self::DItem> {
bytes.try_into().ok().map(Uuid::from_bytes)
fn bytes_decode(bytes: &'a [u8]) -> Result<Self::DItem, BoxedError> {
bytes.try_into().map(Uuid::from_bytes).map_err(Into::into)
}
}
impl BytesEncode<'_> for UuidCodec {
type EItem = Uuid;
fn bytes_encode(item: &Self::EItem) -> Option<Cow<[u8]>> {
Some(Cow::Borrowed(item.as_bytes()))
fn bytes_encode(item: &Self::EItem) -> Result<Cow<[u8]>, BoxedError> {
Ok(Cow::Borrowed(item.as_bytes()))
}
}

40
meilisearch-auth/src/store.rs
View File

@ -4,17 +4,20 @@ use std::collections::HashSet;
use std::convert::{TryFrom, TryInto};
use std::fs::create_dir_all;
use std::path::Path;
use std::result::Result as StdResult;
use std::str;
use std::str::FromStr;
use std::sync::Arc;
use hmac::{Hmac, Mac};
use meilisearch_types::heed::BoxedError;
use meilisearch_types::index_uid_pattern::IndexUidPattern;
use meilisearch_types::keys::KeyId;
use meilisearch_types::milli;
use meilisearch_types::milli::heed::types::{ByteSlice, DecodeIgnore, SerdeJson};
use meilisearch_types::milli::heed::types::{Bytes, DecodeIgnore, SerdeJson};
use meilisearch_types::milli::heed::{Database, Env, EnvOpenOptions, RwTxn};
use sha2::Sha256;
use thiserror::Error;
use time::OffsetDateTime;
use uuid::fmt::Hyphenated;
use uuid::Uuid;
@ -30,7 +33,7 @@ const KEY_ID_ACTION_INDEX_EXPIRATION_DB_NAME: &str = "keyid-action-index-expirat
#[derive(Clone)]
pub struct HeedAuthStore {
env: Arc<Env>,
keys: Database<ByteSlice, SerdeJson<Key>>,
keys: Database<Bytes, SerdeJson<Key>>,
action_keyid_index_expiration: Database<KeyIdActionCodec, SerdeJson<Option<OffsetDateTime>>>,
should_close_on_drop: bool,
}
@ -276,7 +279,7 @@ impl HeedAuthStore {
fn delete_key_from_inverted_db(&self, wtxn: &mut RwTxn, key: &KeyId) -> Result<()> {
let mut iter = self
.action_keyid_index_expiration
.remap_types::<ByteSlice, DecodeIgnore>()
.remap_types::<Bytes, DecodeIgnore>()
.prefix_iter_mut(wtxn, key.as_bytes())?;
while iter.next().transpose()?.is_some() {
// safety: we don't keep references from inside the LMDB database.
@ -294,23 +297,24 @@ pub struct KeyIdActionCodec;
impl<'a> milli::heed::BytesDecode<'a> for KeyIdActionCodec {
type DItem = (KeyId, Action, Option<&'a [u8]>);
fn bytes_decode(bytes: &'a [u8]) -> Option<Self::DItem> {
let (key_id_bytes, action_bytes) = try_split_array_at(bytes)?;
let (action_bytes, index) = match try_split_array_at(action_bytes)? {
(action, []) => (action, None),
(action, index) => (action, Some(index)),
};
fn bytes_decode(bytes: &'a [u8]) -> StdResult<Self::DItem, BoxedError> {
let (key_id_bytes, action_bytes) = try_split_array_at(bytes).ok_or(SliceTooShortError)?;
let (&action_byte, index) =
match try_split_array_at(action_bytes).ok_or(SliceTooShortError)? {
([action], []) => (action, None),
([action], index) => (action, Some(index)),
};
let key_id = Uuid::from_bytes(*key_id_bytes);
let action = Action::from_repr(u8::from_be_bytes(*action_bytes))?;
let action = Action::from_repr(action_byte).ok_or(InvalidActionError { action_byte })?;
Some((key_id, action, index))
Ok((key_id, action, index))
}
}
impl<'a> milli::heed::BytesEncode<'a> for KeyIdActionCodec {
type EItem = (&'a KeyId, &'a Action, Option<&'a [u8]>);
fn bytes_encode((key_id, action, index): &Self::EItem) -> Option<Cow<[u8]>> {
fn bytes_encode((key_id, action, index): &Self::EItem) -> StdResult<Cow<[u8]>, BoxedError> {
let mut bytes = Vec::new();
bytes.extend_from_slice(key_id.as_bytes());
@ -320,10 +324,20 @@ impl<'a> milli::heed::BytesEncode<'a> for KeyIdActionCodec {
bytes.extend_from_slice(index);
}
Some(Cow::Owned(bytes))
Ok(Cow::Owned(bytes))
}
}
#[derive(Error, Debug)]
#[error("the slice is too short")]
pub struct SliceTooShortError;
#[derive(Error, Debug)]
#[error("cannot construct a valid Action from {action_byte}")]
pub struct InvalidActionError {
pub action_byte: u8,
}
pub fn generate_key_as_hexa(uid: Uuid, master_key: &[u8]) -> String {
// format uid as hyphenated allowing user to generate their own keys.
let mut uid_buffer = [0; Hyphenated::LENGTH];

6
meilisearch-types/src/error.rs
View File

@ -386,11 +386,11 @@ impl ErrorCode for HeedError {
HeedError::Mdb(MdbError::Invalid) => Code::InvalidStoreFile,
HeedError::Io(e) => e.error_code(),
HeedError::Mdb(_)
| HeedError::Encoding
| HeedError::Decoding
| HeedError::Encoding(_)
| HeedError::Decoding(_)
| HeedError::InvalidDatabaseTyping
| HeedError::DatabaseClosing
| HeedError::BadOpenOptions => Code::Internal,
| HeedError::BadOpenOptions { .. } => Code::Internal,
}
}
}

10
meilisearch-types/src/settings.rs
View File

@ -532,7 +532,10 @@ pub fn settings(
let faceting = FacetingSettings {
max_values_per_facet: Setting::Set(
index.max_values_per_facet(rtxn)?.unwrap_or(DEFAULT_VALUES_PER_FACET),
index
.max_values_per_facet(rtxn)?
.map(|x| x as usize)
.unwrap_or(DEFAULT_VALUES_PER_FACET),
),
sort_facet_values_by: Setting::Set(
index
@ -545,7 +548,10 @@ pub fn settings(
let pagination = PaginationSettings {
max_total_hits: Setting::Set(
index.pagination_max_total_hits(rtxn)?.unwrap_or(DEFAULT_PAGINATION_MAX_TOTAL_HITS),
index
.pagination_max_total_hits(rtxn)?
.map(|x| x as usize)
.unwrap_or(DEFAULT_PAGINATION_MAX_TOTAL_HITS),
),
};

2
meilisearch/src/routes/indexes/documents.rs
View File

@ -3,7 +3,7 @@ use std::io::ErrorKind;
use actix_web::http::header::CONTENT_TYPE;
use actix_web::web::Data;
use actix_web::{web, HttpMessage, HttpRequest, HttpResponse};
use bstr::ByteSlice;
use bstr::ByteSlice as _;
use deserr::actix_web::{AwebJson, AwebQueryParameter};
use deserr::Deserr;
use futures::StreamExt;

1
meilisearch/src/routes/indexes/settings.rs
View File

@ -78,6 +78,7 @@ macro_rules! make_setting_route {
let body = body.into_inner();
#[allow(clippy::redundant_closure_call)]
$analytics(&body, &req);
let new_settings = Settings {

71
meilisearch/src/routes/multi_search.rs
View File

@ -46,49 +46,46 @@ pub async fn multi_search_with_post(
// Explicitly expect a `(ResponseError, usize)` for the error type rather than `ResponseError` only,
// so that `?` doesn't work if it doesn't use `with_index`, ensuring that it is not forgotten in case of code
// changes.
let search_results: Result<_, (ResponseError, usize)> = (|| {
async {
let mut search_results = Vec::with_capacity(queries.len());
for (query_index, (index_uid, mut query)) in
queries.into_iter().map(SearchQueryWithIndex::into_index_query).enumerate()
let search_results: Result<_, (ResponseError, usize)> = async {
let mut search_results = Vec::with_capacity(queries.len());
for (query_index, (index_uid, mut query)) in
queries.into_iter().map(SearchQueryWithIndex::into_index_query).enumerate()
{
debug!("multi-search #{query_index}: called with params: {:?}", query);
// Check index from API key
if !index_scheduler.filters().is_index_authorized(&index_uid) {
return Err(AuthenticationError::InvalidToken).with_index(query_index);
}
// Apply search rules from tenant token
if let Some(search_rules) = index_scheduler.filters().get_index_search_rules(&index_uid)
{
debug!("multi-search #{query_index}: called with params: {:?}", query);
add_search_rules(&mut query, search_rules);
}
// Check index from API key
if !index_scheduler.filters().is_index_authorized(&index_uid) {
return Err(AuthenticationError::InvalidToken).with_index(query_index);
}
// Apply search rules from tenant token
if let Some(search_rules) =
index_scheduler.filters().get_index_search_rules(&index_uid)
{
add_search_rules(&mut query, search_rules);
}
let index = index_scheduler
.index(&index_uid)
.map_err(|err| {
let mut err = ResponseError::from(err);
// Patch the HTTP status code to 400 as it defaults to 404 for `index_not_found`, but
// here the resource not found is not part of the URL.
err.code = StatusCode::BAD_REQUEST;
err
})
.with_index(query_index)?;
let index = index_scheduler
.index(&index_uid)
.map_err(|err| {
let mut err = ResponseError::from(err);
// Patch the HTTP status code to 400 as it defaults to 404 for `index_not_found`, but
// here the resource not found is not part of the URL.
err.code = StatusCode::BAD_REQUEST;
err
})
let search_result =
tokio::task::spawn_blocking(move || perform_search(&index, query, features))
.await
.with_index(query_index)?;
let search_result =
tokio::task::spawn_blocking(move || perform_search(&index, query, features))
.await
.with_index(query_index)?;
search_results.push(SearchResultWithIndex {
index_uid: index_uid.into_inner(),
result: search_result.with_index(query_index)?,
});
}
Ok(search_results)
search_results.push(SearchResultWithIndex {
index_uid: index_uid.into_inner(),
result: search_result.with_index(query_index)?,
});
}
})()
Ok(search_results)
}
.await;
if search_results.is_ok() {

2
meilisearch/src/search.rs
View File

@ -360,6 +360,7 @@ fn prepare_search<'t>(
let max_total_hits = index
.pagination_max_total_hits(rtxn)
.map_err(milli::Error::from)?
.map(|x| x as usize)
.unwrap_or(DEFAULT_PAGINATION_MAX_TOTAL_HITS);
search.exhaustive_number_hits(is_finite_pagination);
@ -586,6 +587,7 @@ pub fn perform_search(
let max_values_by_facet = index
.max_values_per_facet(&rtxn)
.map_err(milli::Error::from)?
.map(|x| x as usize)
.unwrap_or(DEFAULT_VALUES_PER_FACET);
facet_distribution.max_values_per_facet(max_values_by_facet);

6
meilisearch/tests/search/distinct.rs
View File

@ -4,7 +4,7 @@ use once_cell::sync::Lazy;
use crate::common::{Server, Value};
use crate::json;
pub(self) static DOCUMENTS: Lazy<Value> = Lazy::new(|| {
static DOCUMENTS: Lazy<Value> = Lazy::new(|| {
json!([
{
"id": 1,
@ -107,8 +107,8 @@ pub(self) static DOCUMENTS: Lazy<Value> = Lazy::new(|| {
])
});
pub(self) static DOCUMENT_PRIMARY_KEY: &str = "id";
pub(self) static DOCUMENT_DISTINCT_KEY: &str = "product_id";
static DOCUMENT_PRIMARY_KEY: &str = "id";
static DOCUMENT_DISTINCT_KEY: &str = "product_id";
/// testing: https://github.com/meilisearch/meilisearch/issues/4078
#[actix_rt::test]

2
meilisearch/tests/search/facet_search.rs
View File

@ -4,7 +4,7 @@ use once_cell::sync::Lazy;
use crate::common::{Server, Value};
use crate::json;
pub(self) static DOCUMENTS: Lazy<Value> = Lazy::new(|| {
static DOCUMENTS: Lazy<Value> = Lazy::new(|| {
json!([
{
"title": "Shazam!",

2
meilisearch/tests/search/geo.rs
View File

@ -4,7 +4,7 @@ use once_cell::sync::Lazy;
use crate::common::{Server, Value};
use crate::json;
pub(self) static DOCUMENTS: Lazy<Value> = Lazy::new(|| {
static DOCUMENTS: Lazy<Value> = Lazy::new(|| {
json!([
{
"id": 1,

4
meilisearch/tests/search/mod.rs
View File

@ -15,7 +15,7 @@ use once_cell::sync::Lazy;
use crate::common::{Server, Value};
use crate::json;
pub(self) static DOCUMENTS: Lazy<Value> = Lazy::new(|| {
static DOCUMENTS: Lazy<Value> = Lazy::new(|| {
json!([
{
"title": "Shazam!",
@ -40,7 +40,7 @@ pub(self) static DOCUMENTS: Lazy<Value> = Lazy::new(|| {
])
});
pub(self) static NESTED_DOCUMENTS: Lazy<Value> = Lazy::new(|| {
static NESTED_DOCUMENTS: Lazy<Value> = Lazy::new(|| {
json!([
{
"id": 852,

14
meilitool/src/main.rs
View File

@ -7,8 +7,8 @@ use clap::{Parser, Subcommand};
use dump::{DumpWriter, IndexMetadata};
use file_store::FileStore;
use meilisearch_auth::AuthController;
use meilisearch_types::heed::types::{OwnedType, SerdeJson, Str};
use meilisearch_types::heed::{Database, Env, EnvOpenOptions, PolyDatabase, RoTxn, RwTxn};
use meilisearch_types::heed::types::{SerdeJson, Str};
use meilisearch_types::heed::{Database, Env, EnvOpenOptions, RoTxn, RwTxn, Unspecified};
use meilisearch_types::milli::documents::{obkv_to_object, DocumentsBatchReader};
use meilisearch_types::milli::{obkv_to_json, BEU32};
use meilisearch_types::tasks::{Status, Task};
@ -148,15 +148,17 @@ fn try_opening_poly_database(
env: &Env,
rtxn: &RoTxn,
db_name: &str,
) -> anyhow::Result<PolyDatabase> {
env.open_poly_database(rtxn, Some(db_name))
) -> anyhow::Result<Database<Unspecified, Unspecified>> {
env.database_options()
.name(db_name)
.open(rtxn)
.with_context(|| format!("While opening the {db_name:?} poly database"))?
.with_context(|| format!("Missing the {db_name:?} poly database"))
}
fn try_clearing_poly_database(
wtxn: &mut RwTxn,
database: PolyDatabase,
database: Database<Unspecified, Unspecified>,
db_name: &str,
) -> anyhow::Result<()> {
database.clear(wtxn).with_context(|| format!("While clearing the {db_name:?} database"))
@ -212,7 +214,7 @@ fn export_a_dump(
eprintln!("Successfully dumped {count} keys!");
let rtxn = env.read_txn()?;
let all_tasks: Database<OwnedType<BEU32>, SerdeJson<Task>> =
let all_tasks: Database<BEU32, SerdeJson<Task>> =
try_opening_database(&env, &rtxn, "all-tasks")?;
let index_mapping: Database<Str, UuidCodec> =
try_opening_database(&env, &rtxn, "index-mapping")?;

10
meilitool/src/uuid_codec.rs
View File

@ -1,7 +1,7 @@
use std::borrow::Cow;
use std::convert::TryInto;
use meilisearch_types::heed::{BytesDecode, BytesEncode};
use meilisearch_types::heed::{BoxedError, BytesDecode, BytesEncode};
use uuid::Uuid;
/// A heed codec for value of struct Uuid.
@ -10,15 +10,15 @@ pub struct UuidCodec;
impl<'a> BytesDecode<'a> for UuidCodec {
type DItem = Uuid;
fn bytes_decode(bytes: &'a [u8]) -> Option<Self::DItem> {
bytes.try_into().ok().map(Uuid::from_bytes)
fn bytes_decode(bytes: &'a [u8]) -> Result<Self::DItem, BoxedError> {
bytes.try_into().map(Uuid::from_bytes).map_err(Into::into)
}
}
impl BytesEncode<'_> for UuidCodec {
type EItem = Uuid;
fn bytes_encode(item: &Self::EItem) -> Option<Cow<[u8]>> {
Some(Cow::Borrowed(item.as_bytes()))
fn bytes_encode(item: &Self::EItem) -> Result<Cow<[u8]>, BoxedError> {
Ok(Cow::Borrowed(item.as_bytes()))
}
}

4
milli/Cargo.toml
View File

@ -29,8 +29,8 @@ geoutils = "0.5.1"
grenad = { version = "0.4.5", default-features = false, features = [
"rayon", "tempfile"
] }
heed = { git = "https://github.com/meilisearch/heed", tag = "v0.12.7", default-features = false, features = [
"lmdb", "read-txn-no-tls"
heed = { version = "0.20.0-alpha.9", default-features = false, features = [
"serde-json", "serde-bincode", "read-txn-no-tls"
] }
indexmap = { version = "2.0.0", features = ["serde"] }
instant-distance = { version = "0.6.1", features = ["with-serde"] }

9
milli/src/error.rs
View File

@ -152,7 +152,7 @@ only composed of alphanumeric characters (a-z A-Z 0-9), hyphens (-) and undersco
valid_fields: BTreeSet<String>,
hidden_fields: bool,
},
#[error("{}", HeedError::BadOpenOptions)]
#[error("an environment is already opened with different options")]
InvalidLmdbOpenOptions,
#[error("You must specify where `sort` is listed in the rankingRules setting to use the sort parameter at search time.")]
SortRankingRuleMissing,
@ -326,11 +326,12 @@ impl From<HeedError> for Error {
HeedError::Mdb(MdbError::MapFull) => UserError(MaxDatabaseSizeReached),
HeedError::Mdb(MdbError::Invalid) => UserError(InvalidStoreFile),
HeedError::Mdb(error) => InternalError(Store(error)),
HeedError::Encoding => InternalError(Serialization(Encoding { db_name: None })),
HeedError::Decoding => InternalError(Serialization(Decoding { db_name: None })),
// TODO use the encoding
HeedError::Encoding(_) => InternalError(Serialization(Encoding { db_name: None })),
HeedError::Decoding(_) => InternalError(Serialization(Decoding { db_name: None })),
HeedError::InvalidDatabaseTyping => InternalError(InvalidDatabaseTyping),
HeedError::DatabaseClosing => InternalError(DatabaseClosing),
HeedError::BadOpenOptions => UserError(InvalidLmdbOpenOptions),
HeedError::BadOpenOptions { .. } => UserError(InvalidLmdbOpenOptions),
}
}
}

14
milli/src/external_documents_ids.rs
View File

@ -1,6 +1,6 @@
use std::collections::HashMap;
use heed::types::{OwnedType, Str};
use heed::types::Str;
use heed::{Database, RoIter, RoTxn, RwTxn};
use crate::{DocumentId, BEU32};
@ -16,10 +16,10 @@ pub struct DocumentOperation {
pub kind: DocumentOperationKind,
}
pub struct ExternalDocumentsIds(Database<Str, OwnedType<BEU32>>);
pub struct ExternalDocumentsIds(Database<Str, BEU32>);
impl ExternalDocumentsIds {
pub fn new(db: Database<Str, OwnedType<BEU32>>) -> ExternalDocumentsIds {
pub fn new(db: Database<Str, BEU32>) -> ExternalDocumentsIds {
ExternalDocumentsIds(db)
}
@ -29,7 +29,7 @@ impl ExternalDocumentsIds {
}
pub fn get<A: AsRef<str>>(&self, rtxn: &RoTxn, external_id: A) -> heed::Result<Option<u32>> {
Ok(self.0.get(rtxn, external_id.as_ref())?.map(|x| x.get()))
self.0.get(rtxn, external_id.as_ref())
}
/// An helper function to debug this type, returns an `HashMap` of both,
@ -38,7 +38,7 @@ impl ExternalDocumentsIds {
let mut map = HashMap::default();
for result in self.0.iter(rtxn)? {
let (external, internal) = result?;
map.insert(external.to_owned(), internal.get());
map.insert(external.to_owned(), internal);
}
Ok(map)
}
@ -55,7 +55,7 @@ impl ExternalDocumentsIds {
for DocumentOperation { external_id, internal_id, kind } in operations {
match kind {
DocumentOperationKind::Create => {
self.0.put(wtxn, &external_id, &BEU32::new(internal_id))?;
self.0.put(wtxn, &external_id, &internal_id)?;
}
DocumentOperationKind::Delete => {
if !self.0.delete(wtxn, &external_id)? {
@ -69,7 +69,7 @@ impl ExternalDocumentsIds {
}
/// Returns an iterator over all the external ids.
pub fn iter<'t>(&self, rtxn: &'t RoTxn) -> heed::Result<RoIter<'t, Str, OwnedType<BEU32>>> {
pub fn iter<'t>(&self, rtxn: &'t RoTxn) -> heed::Result<RoIter<'t, Str, BEU32>> {
self.0.iter(rtxn)
}
}

14
milli/src/heed_codec/beu16_str_codec.rs
View File

@ -2,26 +2,28 @@ use std::borrow::Cow;
use std::convert::TryInto;
use std::str;
use heed::BoxedError;
pub struct BEU16StrCodec;
impl<'a> heed::BytesDecode<'a> for BEU16StrCodec {
type DItem = (u16, &'a str);
fn bytes_decode(bytes: &'a [u8]) -> Option<Self::DItem> {
fn bytes_decode(bytes: &'a [u8]) -> Result<Self::DItem, BoxedError> {
let (n_bytes, str_bytes) = bytes.split_at(2);
let n = n_bytes.try_into().map(u16::from_be_bytes).ok()?;
let s = str::from_utf8(str_bytes).ok()?;
Some((n, s))
let n = n_bytes.try_into().map(u16::from_be_bytes)?;
let s = str::from_utf8(str_bytes)?;
Ok((n, s))
}
}
impl<'a> heed::BytesEncode<'a> for BEU16StrCodec {
type EItem = (u16, &'a str);
fn bytes_encode((n, s): &Self::EItem) -> Option<Cow<[u8]>> {
fn bytes_encode((n, s): &Self::EItem) -> Result<Cow<[u8]>, BoxedError> {
let mut bytes = Vec::with_capacity(s.len() + 2);
bytes.extend_from_slice(&n.to_be_bytes());
bytes.extend_from_slice(s.as_bytes());
Some(Cow::Owned(bytes))
Ok(Cow::Owned(bytes))
}
}

14
milli/src/heed_codec/beu32_str_codec.rs
View File

@ -2,26 +2,28 @@ use std::borrow::Cow;
use std::convert::TryInto;
use std::str;
use heed::BoxedError;
pub struct BEU32StrCodec;
impl<'a> heed::BytesDecode<'a> for BEU32StrCodec {
type DItem = (u32, &'a str);
fn bytes_decode(bytes: &'a [u8]) -> Option<Self::DItem> {
fn bytes_decode(bytes: &'a [u8]) -> Result<Self::DItem, BoxedError> {
let (n_bytes, str_bytes) = bytes.split_at(4);
let n = n_bytes.try_into().map(u32::from_be_bytes).ok()?;
let s = str::from_utf8(str_bytes).ok()?;
Some((n, s))
let n = n_bytes.try_into().map(u32::from_be_bytes)?;
let s = str::from_utf8(str_bytes)?;
Ok((n, s))
}
}
impl<'a> heed::BytesEncode<'a> for BEU32StrCodec {
type EItem = (u32, &'a str);
fn bytes_encode((n, s): &Self::EItem) -> Option<Cow<[u8]>> {
fn bytes_encode((n, s): &Self::EItem) -> Result<Cow<[u8]>, BoxedError> {
let mut bytes = Vec::with_capacity(s.len() + 4);
bytes.extend_from_slice(&n.to_be_bytes());
bytes.extend_from_slice(s.as_bytes());
Some(Cow::Owned(bytes))
Ok(Cow::Owned(bytes))
}
}

18
milli/src/heed_codec/byte_slice_ref.rs
View File

@ -1,23 +1,23 @@
use std::borrow::Cow;
use heed::{BytesDecode, BytesEncode};
use heed::{BoxedError, BytesDecode, BytesEncode};
/// A codec for values of type `&[u8]`. Unlike `ByteSlice`, its `EItem` and `DItem` associated
/// A codec for values of type `&[u8]`. Unlike `Bytes`, its `EItem` and `DItem` associated
/// types are equivalent (= `&'a [u8]`) and these values can reside within another structure.
pub struct ByteSliceRefCodec;
pub struct BytesRefCodec;
impl<'a> BytesEncode<'a> for ByteSliceRefCodec {
impl<'a> BytesEncode<'a> for BytesRefCodec {
type EItem = &'a [u8];
fn bytes_encode(item: &'a Self::EItem) -> Option<Cow<'a, [u8]>> {
Some(Cow::Borrowed(item))
fn bytes_encode(item: &'a Self::EItem) -> Result<Cow<'a, [u8]>, BoxedError> {
Ok(Cow::Borrowed(item))
}
}
impl<'a> BytesDecode<'a> for ByteSliceRefCodec {
impl<'a> BytesDecode<'a> for BytesRefCodec {
type DItem = &'a [u8];
fn bytes_decode(bytes: &'a [u8]) -> Option<Self::DItem> {
Some(bytes)
fn bytes_decode(bytes: &'a [u8]) -> Result<Self::DItem, BoxedError> {
Ok(bytes)
}
}

17
milli/src/heed_codec/facet/field_doc_id_facet_codec.rs
View File

@ -1,8 +1,9 @@
use std::borrow::Cow;
use std::marker::PhantomData;
use heed::{BytesDecode, BytesEncode};
use heed::{BoxedError, BytesDecode, BytesEncode};
use crate::heed_codec::SliceTooShortError;
use crate::{try_split_array_at, DocumentId, FieldId};
pub struct FieldDocIdFacetCodec<C>(PhantomData<C>);
@ -13,16 +14,16 @@ where
{
type DItem = (FieldId, DocumentId, C::DItem);
fn bytes_decode(bytes: &'a [u8]) -> Option<Self::DItem> {
let (field_id_bytes, bytes) = try_split_array_at(bytes)?;
fn bytes_decode(bytes: &'a [u8]) -> Result<Self::DItem, BoxedError> {
let (field_id_bytes, bytes) = try_split_array_at(bytes).ok_or(SliceTooShortError)?;
let field_id = u16::from_be_bytes(field_id_bytes);
let (document_id_bytes, bytes) = try_split_array_at(bytes)?;
let (document_id_bytes, bytes) = try_split_array_at(bytes).ok_or(SliceTooShortError)?;
let document_id = u32::from_be_bytes(document_id_bytes);
let value = C::bytes_decode(bytes)?;
Some((field_id, document_id, value))
Ok((field_id, document_id, value))
}
}
@ -32,13 +33,15 @@ where
{
type EItem = (FieldId, DocumentId, C::EItem);
fn bytes_encode((field_id, document_id, value): &'a Self::EItem) -> Option<Cow<[u8]>> {
fn bytes_encode(
(field_id, document_id, value): &'a Self::EItem,
) -> Result<Cow<[u8]>, BoxedError> {
let mut bytes = Vec::with_capacity(32);
bytes.extend_from_slice(&field_id.to_be_bytes()); // 2 bytes
bytes.extend_from_slice(&document_id.to_be_bytes()); // 4 bytes
let value_bytes = C::bytes_encode(value)?;
// variable length, if f64 -> 16 bytes, if string -> large, potentially
bytes.extend_from_slice(&value_bytes);
Some(Cow::Owned(bytes))
Ok(Cow::Owned(bytes))
}
}

26
milli/src/heed_codec/facet/mod.rs
View File

@ -5,8 +5,8 @@ use std::borrow::Cow;
use std::convert::TryFrom;
use std::marker::PhantomData;
use heed::types::{DecodeIgnore, OwnedType};
use heed::{BytesDecode, BytesEncode};
use heed::types::DecodeIgnore;
use heed::{BoxedError, BytesDecode, BytesEncode};
use roaring::RoaringBitmap;
pub use self::field_doc_id_facet_codec::FieldDocIdFacetCodec;
@ -18,7 +18,7 @@ pub type FieldDocIdFacetF64Codec = FieldDocIdFacetCodec<OrderedF64Codec>;
pub type FieldDocIdFacetStringCodec = FieldDocIdFacetCodec<StrRefCodec>;
pub type FieldDocIdFacetIgnoreCodec = FieldDocIdFacetCodec<DecodeIgnore>;
pub type FieldIdCodec = OwnedType<BEU16>;
pub type FieldIdCodec = BEU16;
/// Tries to split a slice in half at the given middle point,
/// `None` if the slice is too short.
@ -58,7 +58,7 @@ where
{
type EItem = FacetGroupKey<T::EItem>;
fn bytes_encode(value: &'a Self::EItem) -> Option<Cow<'a, [u8]>> {
fn bytes_encode(value: &'a Self::EItem) -> Result<Cow<'a, [u8]>, BoxedError> {
let mut v = vec![];
v.extend_from_slice(&value.field_id.to_be_bytes());
v.extend_from_slice(&[value.level]);
@ -66,7 +66,7 @@ where
let bound = T::bytes_encode(&value.left_bound)?;
v.extend_from_slice(&bound);
Some(Cow::Owned(v))
Ok(Cow::Owned(v))
}
}
impl<'a, T> heed::BytesDecode<'a> for FacetGroupKeyCodec<T>
@ -75,11 +75,11 @@ where
{
type DItem = FacetGroupKey<T::DItem>;
fn bytes_decode(bytes: &'a [u8]) -> Option<Self::DItem> {
let fid = u16::from_be_bytes(<[u8; 2]>::try_from(&bytes[0..=1]).ok()?);
fn bytes_decode(bytes: &'a [u8]) -> Result<Self::DItem, BoxedError> {
let fid = u16::from_be_bytes(<[u8; 2]>::try_from(&bytes[0..=1])?);
let level = bytes[2];
let bound = T::bytes_decode(&bytes[3..])?;
Some(FacetGroupKey { field_id: fid, level, left_bound: bound })
Ok(FacetGroupKey { field_id: fid, level, left_bound: bound })
}
}
@ -87,17 +87,17 @@ pub struct FacetGroupValueCodec;
impl<'a> heed::BytesEncode<'a> for FacetGroupValueCodec {
type EItem = FacetGroupValue;
fn bytes_encode(value: &'a Self::EItem) -> Option<Cow<'a, [u8]>> {
fn bytes_encode(value: &'a Self::EItem) -> Result<Cow<'a, [u8]>, BoxedError> {
let mut v = vec![value.size];
CboRoaringBitmapCodec::serialize_into(&value.bitmap, &mut v);
Some(Cow::Owned(v))
Ok(Cow::Owned(v))
}
}
impl<'a> heed::BytesDecode<'a> for FacetGroupValueCodec {
type DItem = FacetGroupValue;
fn bytes_decode(bytes: &'a [u8]) -> Option<Self::DItem> {
fn bytes_decode(bytes: &'a [u8]) -> Result<Self::DItem, BoxedError> {
let size = bytes[0];
let bitmap = CboRoaringBitmapCodec::deserialize_from(&bytes[1..]).ok()?;
Some(FacetGroupValue { size, bitmap })
let bitmap = CboRoaringBitmapCodec::deserialize_from(&bytes[1..])?;
Ok(FacetGroupValue { size, bitmap })
}
}

24
milli/src/heed_codec/facet/ordered_f64_codec.rs
View File

@ -1,37 +1,45 @@
use std::borrow::Cow;
use std::convert::TryInto;
use heed::BytesDecode;
use heed::{BoxedError, BytesDecode};
use thiserror::Error;
use crate::facet::value_encoding::f64_into_bytes;
use crate::heed_codec::SliceTooShortError;
pub struct OrderedF64Codec;
impl<'a> BytesDecode<'a> for OrderedF64Codec {
type DItem = f64;
fn bytes_decode(bytes: &'a [u8]) -> Option<Self::DItem> {
fn bytes_decode(bytes: &'a [u8]) -> Result<Self::DItem, BoxedError> {
if bytes.len() < 16 {
return None;
Err(SliceTooShortError.into())
} else {
bytes[8..].try_into().map(f64::from_be_bytes).map_err(Into::into)
}
let f = bytes[8..].try_into().ok().map(f64::from_be_bytes)?;
Some(f)
}
}
impl heed::BytesEncode<'_> for OrderedF64Codec {
type EItem = f64;
fn bytes_encode(f: &Self::EItem) -> Option<Cow<[u8]>> {
fn bytes_encode(f: &Self::EItem) -> Result<Cow<[u8]>, BoxedError> {
let mut buffer = [0u8; 16];
// write the globally ordered float
let bytes = f64_into_bytes(*f)?;
let bytes = f64_into_bytes(*f).ok_or(InvalidGloballyOrderedFloatError { float: *f })?;
buffer[..8].copy_from_slice(&bytes[..]);
// Then the f64 value just to be able to read it back
let bytes = f.to_be_bytes();
buffer[8..16].copy_from_slice(&bytes[..]);
Some(Cow::Owned(buffer.to_vec()))
Ok(Cow::Owned(buffer.to_vec()))
}
}
#[derive(Error, Debug)]
#[error("the float {float} cannot be converted to a globally ordered representation")]
pub struct InvalidGloballyOrderedFloatError {
float: f64,
}

15
milli/src/heed_codec/field_id_word_count_codec.rs
View File

@ -1,5 +1,8 @@
use std::borrow::Cow;
use heed::BoxedError;
use super::SliceTooShortError;
use crate::{try_split_array_at, FieldId};
pub struct FieldIdWordCountCodec;
@ -7,21 +10,21 @@ pub struct FieldIdWordCountCodec;
impl<'a> heed::BytesDecode<'a> for FieldIdWordCountCodec {
type DItem = (FieldId, u8);
fn bytes_decode(bytes: &'a [u8]) -> Option<Self::DItem> {
let (field_id_bytes, bytes) = try_split_array_at(bytes)?;
fn bytes_decode(bytes: &'a [u8]) -> Result<Self::DItem, BoxedError> {
let (field_id_bytes, bytes) = try_split_array_at(bytes).ok_or(SliceTooShortError)?;
let field_id = u16::from_be_bytes(field_id_bytes);
let ([word_count], _nothing) = try_split_array_at(bytes)?;
Some((field_id, word_count))
let ([word_count], _nothing) = try_split_array_at(bytes).ok_or(SliceTooShortError)?;
Ok((field_id, word_count))
}
}
impl<'a> heed::BytesEncode<'a> for FieldIdWordCountCodec {
type EItem = (FieldId, u8);
fn bytes_encode((field_id, word_count): &Self::EItem) -> Option<Cow<[u8]>> {
fn bytes_encode((field_id, word_count): &Self::EItem) -> Result<Cow<[u8]>, BoxedError> {
let mut bytes = Vec::with_capacity(2 + 1);
bytes.extend_from_slice(&field_id.to_be_bytes());
bytes.push(*word_count);
Some(Cow::Owned(bytes))
Ok(Cow::Owned(bytes))
}
}

10
milli/src/heed_codec/fst_set_codec.rs
View File

@ -1,7 +1,7 @@
use std::borrow::Cow;
use fst::Set;
use heed::{BytesDecode, BytesEncode};
use heed::{BoxedError, BytesDecode, BytesEncode};
/// A codec for values of type `Set<&[u8]>`.
pub struct FstSetCodec;
@ -9,15 +9,15 @@ pub struct FstSetCodec;
impl<'a> BytesEncode<'a> for FstSetCodec {
type EItem = Set<Vec<u8>>;
fn bytes_encode(item: &'a Self::EItem) -> Option<Cow<'a, [u8]>> {
Some(Cow::Borrowed(item.as_fst().as_bytes()))
fn bytes_encode(item: &'a Self::EItem) -> Result<Cow<'a, [u8]>, BoxedError> {
Ok(Cow::Borrowed(item.as_fst().as_bytes()))
}
}
impl<'a> BytesDecode<'a> for FstSetCodec {
type DItem = Set<&'a [u8]>;
fn bytes_decode(bytes: &'a [u8]) -> Option<Self::DItem> {
Set::new(bytes).ok()
fn bytes_decode(bytes: &'a [u8]) -> Result<Self::DItem, BoxedError> {
Set::new(bytes).map_err(Into::into)
}
}

10
milli/src/heed_codec/mod.rs
View File

@ -12,8 +12,10 @@ mod str_beu32_codec;
mod str_ref;
mod str_str_u8_codec;
pub use byte_slice_ref::ByteSliceRefCodec;
pub use byte_slice_ref::BytesRefCodec;
use heed::BoxedError;
pub use str_ref::StrRefCodec;
use thiserror::Error;
pub use self::beu16_str_codec::BEU16StrCodec;
pub use self::beu32_str_codec::BEU32StrCodec;
@ -31,5 +33,9 @@ pub use self::str_str_u8_codec::{U8StrStrCodec, UncheckedU8StrStrCodec};
pub trait BytesDecodeOwned {
type DItem;
fn bytes_decode_owned(bytes: &[u8]) -> Option<Self::DItem>;
fn bytes_decode_owned(bytes: &[u8]) -> Result<Self::DItem, BoxedError>;
}
#[derive(Error, Debug)]
#[error("the slice is too short")]
pub struct SliceTooShortError;

9
milli/src/heed_codec/obkv_codec.rs
View File

@ -1,5 +1,6 @@
use std::borrow::Cow;
use heed::BoxedError;
use obkv::{KvReaderU16, KvWriterU16};
pub struct ObkvCodec;
@ -7,15 +8,15 @@ pub struct ObkvCodec;
impl<'a> heed::BytesDecode<'a> for ObkvCodec {
type DItem = KvReaderU16<'a>;
fn bytes_decode(bytes: &'a [u8]) -> Option<Self::DItem> {
Some(KvReaderU16::new(bytes))
fn bytes_decode(bytes: &'a [u8]) -> Result<Self::DItem, BoxedError> {
Ok(KvReaderU16::new(bytes))
}
}
impl heed::BytesEncode<'_> for ObkvCodec {
type EItem = KvWriterU16<Vec<u8>>;
fn bytes_encode(item: &Self::EItem) -> Option<Cow<[u8]>> {
item.clone().into_inner().map(Cow::Owned).ok()
fn bytes_encode(item: &Self::EItem) -> Result<Cow<[u8]>, BoxedError> {
item.clone().into_inner().map(Cow::Owned).map_err(Into::into)
}
}

14
milli/src/heed_codec/roaring_bitmap/bo_roaring_bitmap_codec.rs
View File

@ -2,7 +2,7 @@ use std::borrow::Cow;
use std::convert::TryInto;
use std::mem::size_of;
use heed::BytesDecode;
use heed::{BoxedError, BytesDecode};
use roaring::RoaringBitmap;
use crate::heed_codec::BytesDecodeOwned;
@ -19,22 +19,22 @@ impl BoRoaringBitmapCodec {
impl BytesDecode<'_> for BoRoaringBitmapCodec {
type DItem = RoaringBitmap;
fn bytes_decode(bytes: &[u8]) -> Option<Self::DItem> {
fn bytes_decode(bytes: &[u8]) -> Result<Self::DItem, BoxedError> {
let mut bitmap = RoaringBitmap::new();
for chunk in bytes.chunks(size_of::<u32>()) {
let bytes = chunk.try_into().ok()?;
let bytes = chunk.try_into()?;
bitmap.push(u32::from_ne_bytes(bytes));
}
Some(bitmap)
Ok(bitmap)
}
}
impl BytesDecodeOwned for BoRoaringBitmapCodec {
type DItem = RoaringBitmap;
fn bytes_decode_owned(bytes: &[u8]) -> Option<Self::DItem> {
fn bytes_decode_owned(bytes: &[u8]) -> Result<Self::DItem, BoxedError> {
Self::bytes_decode(bytes)
}
}
@ -42,9 +42,9 @@ impl BytesDecodeOwned for BoRoaringBitmapCodec {
impl heed::BytesEncode<'_> for BoRoaringBitmapCodec {
type EItem = RoaringBitmap;
fn bytes_encode(item: &Self::EItem) -> Option<Cow<[u8]>> {
fn bytes_encode(item: &Self::EItem) -> Result<Cow<[u8]>, BoxedError> {
let mut out = Vec::new();
BoRoaringBitmapCodec::serialize_into(item, &mut out);
Some(Cow::Owned(out))
Ok(Cow::Owned(out))
}
}

13
milli/src/heed_codec/roaring_bitmap/cbo_roaring_bitmap_codec.rs
View File

@ -3,6 +3,7 @@ use std::io;
use std::mem::size_of;
use byteorder::{NativeEndian, ReadBytesExt, WriteBytesExt};
use heed::BoxedError;
use roaring::RoaringBitmap;
use crate::heed_codec::BytesDecodeOwned;
@ -132,26 +133,26 @@ impl CboRoaringBitmapCodec {
impl heed::BytesDecode<'_> for CboRoaringBitmapCodec {
type DItem = RoaringBitmap;
fn bytes_decode(bytes: &[u8]) -> Option<Self::DItem> {
Self::deserialize_from(bytes).ok()
fn bytes_decode(bytes: &[u8]) -> Result<Self::DItem, BoxedError> {
Self::deserialize_from(bytes).map_err(Into::into)
}
}
impl BytesDecodeOwned for CboRoaringBitmapCodec {
type DItem = RoaringBitmap;
fn bytes_decode_owned(bytes: &[u8]) -> Option<Self::DItem> {
Self::deserialize_from(bytes).ok()
fn bytes_decode_owned(bytes: &[u8]) -> Result<Self::DItem, BoxedError> {
Self::deserialize_from(bytes).map_err(Into::into)
}
}
impl heed::BytesEncode<'_> for CboRoaringBitmapCodec {
type EItem = RoaringBitmap;
fn bytes_encode(item: &Self::EItem) -> Option<Cow<[u8]>> {
fn bytes_encode(item: &Self::EItem) -> Result<Cow<[u8]>, BoxedError> {
let mut vec = Vec::with_capacity(Self::serialized_size(item));
Self::serialize_into(item, &mut vec);
Some(Cow::Owned(vec))
Ok(Cow::Owned(vec))
}
}

15
milli/src/heed_codec/roaring_bitmap/roaring_bitmap_codec.rs
View File

@ -1,5 +1,6 @@
use std::borrow::Cow;
use heed::BoxedError;
use roaring::RoaringBitmap;
use crate::heed_codec::BytesDecodeOwned;
@ -9,25 +10,25 @@ pub struct RoaringBitmapCodec;
impl heed::BytesDecode<'_> for RoaringBitmapCodec {
type DItem = RoaringBitmap;
fn bytes_decode(bytes: &[u8]) -> Option<Self::DItem> {
RoaringBitmap::deserialize_unchecked_from(bytes).ok()
fn bytes_decode(bytes: &[u8]) -> Result<Self::DItem, BoxedError> {
RoaringBitmap::deserialize_unchecked_from(bytes).map_err(Into::into)
}
}
impl BytesDecodeOwned for RoaringBitmapCodec {
type DItem = RoaringBitmap;
fn bytes_decode_owned(bytes: &[u8]) -> Option<Self::DItem> {
RoaringBitmap::deserialize_from(bytes).ok()
fn bytes_decode_owned(bytes: &[u8]) -> Result<Self::DItem, BoxedError> {
RoaringBitmap::deserialize_from(bytes).map_err(Into::into)
}
}
impl heed::BytesEncode<'_> for RoaringBitmapCodec {
type EItem = RoaringBitmap;
fn bytes_encode(item: &Self::EItem) -> Option<Cow<[u8]>> {
fn bytes_encode(item: &Self::EItem) -> Result<Cow<[u8]>, BoxedError> {
let mut bytes = Vec::with_capacity(item.serialized_size());
item.serialize_into(&mut bytes).ok()?;
Some(Cow::Owned(bytes))
item.serialize_into(&mut bytes)?;
Ok(Cow::Owned(bytes))
}
}

8
milli/src/heed_codec/roaring_bitmap_length/bo_roaring_bitmap_len_codec.rs
View File

@ -1,6 +1,6 @@
use std::mem;
use heed::BytesDecode;
use heed::{BoxedError, BytesDecode};
use crate::heed_codec::BytesDecodeOwned;
@ -9,15 +9,15 @@ pub struct BoRoaringBitmapLenCodec;
impl BytesDecode<'_> for BoRoaringBitmapLenCodec {
type DItem = u64;
fn bytes_decode(bytes: &[u8]) -> Option<Self::DItem> {
Some((bytes.len() / mem::size_of::<u32>()) as u64)
fn bytes_decode(bytes: &[u8]) -> Result<Self::DItem, BoxedError> {
Ok((bytes.len() / mem::size_of::<u32>()) as u64)
}
}
impl BytesDecodeOwned for BoRoaringBitmapLenCodec {
type DItem = u64;
fn bytes_decode_owned(bytes: &[u8]) -> Option<Self::DItem> {
fn bytes_decode_owned(bytes: &[u8]) -> Result<Self::DItem, BoxedError> {
Self::bytes_decode(bytes)
}
}

6
milli/src/heed_codec/roaring_bitmap_length/cbo_roaring_bitmap_len_codec.rs
View File

@ -1,6 +1,6 @@
use std::mem;
use heed::BytesDecode;
use heed::{BoxedError, BytesDecode};
use super::{BoRoaringBitmapLenCodec, RoaringBitmapLenCodec};
use crate::heed_codec::roaring_bitmap::cbo_roaring_bitmap_codec::THRESHOLD;
@ -11,7 +11,7 @@ pub struct CboRoaringBitmapLenCodec;
impl BytesDecode<'_> for CboRoaringBitmapLenCodec {
type DItem = u64;
fn bytes_decode(bytes: &[u8]) -> Option<Self::DItem> {
fn bytes_decode(bytes: &[u8]) -> Result<Self::DItem, BoxedError> {
if bytes.len() <= THRESHOLD * mem::size_of::<u32>() {
// If there is threshold or less than threshold integers that can fit into this array
// of bytes it means that we used the ByteOrder codec serializer.
@ -27,7 +27,7 @@ impl BytesDecode<'_> for CboRoaringBitmapLenCodec {
impl BytesDecodeOwned for CboRoaringBitmapLenCodec {
type DItem = u64;
fn bytes_decode_owned(bytes: &[u8]) -> Option<Self::DItem> {
fn bytes_decode_owned(bytes: &[u8]) -> Result<Self::DItem, BoxedError> {
Self::bytes_decode(bytes)
}
}

9
milli/src/heed_codec/roaring_bitmap_length/roaring_bitmap_len_codec.rs
View File

@ -2,6 +2,7 @@ use std::io::{self, BufRead, Read};
use std::mem;
use byteorder::{LittleEndian, ReadBytesExt};
use heed::BoxedError;
use crate::heed_codec::BytesDecodeOwned;
@ -56,16 +57,16 @@ impl RoaringBitmapLenCodec {
impl heed::BytesDecode<'_> for RoaringBitmapLenCodec {
type DItem = u64;
fn bytes_decode(bytes: &[u8]) -> Option<Self::DItem> {
RoaringBitmapLenCodec::deserialize_from_slice(bytes).ok()
fn bytes_decode(bytes: &[u8]) -> Result<Self::DItem, BoxedError> {
RoaringBitmapLenCodec::deserialize_from_slice(bytes).map_err(Into::into)
}
}
impl BytesDecodeOwned for RoaringBitmapLenCodec {
type DItem = u64;
fn bytes_decode_owned(bytes: &[u8]) -> Option<Self::DItem> {
RoaringBitmapLenCodec::deserialize_from_slice(bytes).ok()
fn bytes_decode_owned(bytes: &[u8]) -> Result<Self::DItem, BoxedError> {
RoaringBitmapLenCodec::deserialize_from_slice(bytes).map_err(Into::into)
}
}

19
milli/src/heed_codec/script_language_codec.rs
View File

@ -1,30 +1,31 @@
use std::borrow::Cow;
use std::ffi::CStr;
use std::str;
use charabia::{Language, Script};
use heed::BoxedError;
pub struct ScriptLanguageCodec;
impl<'a> heed::BytesDecode<'a> for ScriptLanguageCodec {
type DItem = (Script, Language);
fn bytes_decode(bytes: &'a [u8]) -> Option<Self::DItem> {
let sep = bytes.iter().position(|b| *b == 0)?;
let (s_bytes, l_bytes) = bytes.split_at(sep);
let script = str::from_utf8(s_bytes).ok()?;
fn bytes_decode(bytes: &'a [u8]) -> Result<Self::DItem, BoxedError> {
let cstr = CStr::from_bytes_until_nul(bytes)?;
let script = cstr.to_str()?;
let script_name = Script::from_name(script);
let lan = str::from_utf8(l_bytes).ok()?;
// skip '\0' byte between the two strings.
let lan_name = Language::from_name(&lan[1..]);
let lan = str::from_utf8(&bytes[script.len() + 1..])?;
let lan_name = Language::from_name(lan);
Some((script_name, lan_name))
Ok((script_name, lan_name))
}
}
impl<'a> heed::BytesEncode<'a> for ScriptLanguageCodec {
type EItem = (Script, Language);
fn bytes_encode((script, lan): &Self::EItem) -> Option<Cow<[u8]>> {
fn bytes_encode((script, lan): &Self::EItem) -> Result<Cow<[u8]>, BoxedError> {
let script_name = script.name().as_bytes();
let lan_name = lan.name().as_bytes();
@ -33,6 +34,6 @@ impl<'a> heed::BytesEncode<'a> for ScriptLanguageCodec {
bytes.push(0);
bytes.extend_from_slice(lan_name);
Some(Cow::Owned(bytes))
Ok(Cow::Owned(bytes))
}
}

35
milli/src/heed_codec/str_beu32_codec.rs
View File

@ -3,37 +3,41 @@ use std::convert::TryInto;
use std::mem::size_of;
use std::str;
use heed::BoxedError;
use super::SliceTooShortError;
pub struct StrBEU32Codec;
impl<'a> heed::BytesDecode<'a> for StrBEU32Codec {
type DItem = (&'a str, u32);
fn bytes_decode(bytes: &'a [u8]) -> Option<Self::DItem> {
fn bytes_decode(bytes: &'a [u8]) -> Result<Self::DItem, BoxedError> {
let footer_len = size_of::<u32>();
if bytes.len() < footer_len {
return None;
return Err(SliceTooShortError.into());
}
let (word, bytes) = bytes.split_at(bytes.len() - footer_len);
let word = str::from_utf8(word).ok()?;
let pos = bytes.try_into().map(u32::from_be_bytes).ok()?;
let word = str::from_utf8(word)?;
let pos = bytes.try_into().map(u32::from_be_bytes)?;
Some((word, pos))
Ok((word, pos))
}
}
impl<'a> heed::BytesEncode<'a> for StrBEU32Codec {
type EItem = (&'a str, u32);
fn bytes_encode((word, pos): &Self::EItem) -> Option<Cow<[u8]>> {
fn bytes_encode((word, pos): &Self::EItem) -> Result<Cow<[u8]>, BoxedError> {
let pos = pos.to_be_bytes();
let mut bytes = Vec::with_capacity(word.len() + pos.len());
bytes.extend_from_slice(word.as_bytes());
bytes.extend_from_slice(&pos[..]);
Some(Cow::Owned(bytes))
Ok(Cow::Owned(bytes))
}
}
@ -42,26 +46,27 @@ pub struct StrBEU16Codec;
impl<'a> heed::BytesDecode<'a> for StrBEU16Codec {
type DItem = (&'a str, u16);
fn bytes_decode(bytes: &'a [u8]) -> Option<Self::DItem> {
fn bytes_decode(bytes: &'a [u8]) -> Result<Self::DItem, BoxedError> {
let footer_len = size_of::<u16>();
if bytes.len() < footer_len + 1 {
return None;
return Err(SliceTooShortError.into());
}
let (word_plus_nul_byte, bytes) = bytes.split_at(bytes.len() - footer_len);
let (_, word) = word_plus_nul_byte.split_last()?;
let word = str::from_utf8(word).ok()?;
let pos = bytes.try_into().map(u16::from_be_bytes).ok()?;
// unwrap: we just checked the footer + 1 above.
let (_, word) = word_plus_nul_byte.split_last().unwrap();
let word = str::from_utf8(word)?;
let pos = bytes.try_into().map(u16::from_be_bytes)?;
Some((word, pos))
Ok((word, pos))
}
}
impl<'a> heed::BytesEncode<'a> for StrBEU16Codec {
type EItem = (&'a str, u16);
fn bytes_encode((word, pos): &Self::EItem) -> Option<Cow<[u8]>> {
fn bytes_encode((word, pos): &Self::EItem) -> Result<Cow<[u8]>, BoxedError> {
let pos = pos.to_be_bytes();
let mut bytes = Vec::with_capacity(word.len() + 1 + pos.len());
@ -69,6 +74,6 @@ impl<'a> heed::BytesEncode<'a> for StrBEU16Codec {
bytes.push(0);
bytes.extend_from_slice(&pos[..]);
Some(Cow::Owned(bytes))
Ok(Cow::Owned(bytes))
}
}

11
milli/src/heed_codec/str_ref.rs
View File

@ -1,6 +1,6 @@
use std::borrow::Cow;
use heed::{BytesDecode, BytesEncode};
use heed::{BoxedError, BytesDecode, BytesEncode};
/// A codec for values of type `&str`. Unlike `Str`, its `EItem` and `DItem` associated
/// types are equivalent (= `&'a str`) and these values can reside within another structure.
@ -8,15 +8,14 @@ pub struct StrRefCodec;
impl<'a> BytesEncode<'a> for StrRefCodec {
type EItem = &'a str;
fn bytes_encode(item: &'a &'a str) -> Option<Cow<'a, [u8]>> {
Some(Cow::Borrowed(item.as_bytes()))
fn bytes_encode(item: &'a &'a str) -> Result<Cow<'a, [u8]>, BoxedError> {
Ok(Cow::Borrowed(item.as_bytes()))
}
}
impl<'a> BytesDecode<'a> for StrRefCodec {
type DItem = &'a str;
fn bytes_decode(bytes: &'a [u8]) -> Option<Self::DItem> {
let s = std::str::from_utf8(bytes).ok()?;
Some(s)
fn bytes_decode(bytes: &'a [u8]) -> Result<Self::DItem, BoxedError> {
std::str::from_utf8(bytes).map_err(Into::into)
}
}

41
milli/src/heed_codec/str_str_u8_codec.rs
View File

@ -1,32 +1,36 @@
use std::borrow::Cow;
use std::ffi::CStr;
use std::str;
use heed::BoxedError;
use super::SliceTooShortError;
pub struct U8StrStrCodec;
impl<'a> heed::BytesDecode<'a> for U8StrStrCodec {
type DItem = (u8, &'a str, &'a str);
fn bytes_decode(bytes: &'a [u8]) -> Option<Self::DItem> {
let (n, bytes) = bytes.split_first()?;
let s1_end = bytes.iter().position(|b| *b == 0)?;
let (s1_bytes, rest) = bytes.split_at(s1_end);
let s2_bytes = &rest[1..];
let s1 = str::from_utf8(s1_bytes).ok()?;
let s2 = str::from_utf8(s2_bytes).ok()?;
Some((*n, s1, s2))
fn bytes_decode(bytes: &'a [u8]) -> Result<Self::DItem, BoxedError> {
let (n, bytes) = bytes.split_first().ok_or(SliceTooShortError)?;
let cstr = CStr::from_bytes_until_nul(bytes)?;
let s1 = cstr.to_str()?;
// skip '\0' byte between the two strings.
let s2 = str::from_utf8(&bytes[s1.len() + 1..])?;
Ok((*n, s1, s2))
}
}
impl<'a> heed::BytesEncode<'a> for U8StrStrCodec {
type EItem = (u8, &'a str, &'a str);
fn bytes_encode((n, s1, s2): &Self::EItem) -> Option<Cow<[u8]>> {
fn bytes_encode((n, s1, s2): &Self::EItem) -> Result<Cow<[u8]>, BoxedError> {
let mut bytes = Vec::with_capacity(s1.len() + s2.len() + 1);
bytes.push(*n);
bytes.extend_from_slice(s1.as_bytes());
bytes.push(0);
bytes.extend_from_slice(s2.as_bytes());
Some(Cow::Owned(bytes))
Ok(Cow::Owned(bytes))
}
}
pub struct UncheckedU8StrStrCodec;
@ -34,24 +38,25 @@ pub struct UncheckedU8StrStrCodec;
impl<'a> heed::BytesDecode<'a> for UncheckedU8StrStrCodec {
type DItem = (u8, &'a [u8], &'a [u8]);
fn bytes_decode(bytes: &'a [u8]) -> Option<Self::DItem> {
let (n, bytes) = bytes.split_first()?;
let s1_end = bytes.iter().position(|b| *b == 0)?;
let (s1_bytes, rest) = bytes.split_at(s1_end);
let s2_bytes = &rest[1..];
Some((*n, s1_bytes, s2_bytes))
fn bytes_decode(bytes: &'a [u8]) -> Result<Self::DItem, BoxedError> {
let (n, bytes) = bytes.split_first().ok_or(SliceTooShortError)?;
let cstr = CStr::from_bytes_until_nul(bytes)?;
let s1_bytes = cstr.to_bytes();
// skip '\0' byte between the two strings.
let s2_bytes = &bytes[s1_bytes.len() + 1..];
Ok((*n, s1_bytes, s2_bytes))
}
}
impl<'a> heed::BytesEncode<'a> for UncheckedU8StrStrCodec {
type EItem = (u8, &'a [u8], &'a [u8]);
fn bytes_encode((n, s1, s2): &Self::EItem) -> Option<Cow<[u8]>> {
fn bytes_encode((n, s1, s2): &Self::EItem) -> Result<Cow<[u8]>, BoxedError> {
let mut bytes = Vec::with_capacity(s1.len() + s2.len() + 1);
bytes.push(*n);
bytes.extend_from_slice(s1);
bytes.push(0);
bytes.extend_from_slice(s2);
Some(Cow::Owned(bytes))
Ok(Cow::Owned(bytes))
}
}

354
milli/src/index.rs
View File

@ -4,9 +4,8 @@ use std::fs::File;
use std::path::Path;
use charabia::{Language, Script};
use heed::flags::Flags;
use heed::types::*;
use heed::{CompactionOption, Database, PolyDatabase, RoTxn, RwTxn};
use heed::{CompactionOption, Database, RoTxn, RwTxn, Unspecified};
use roaring::RoaringBitmap;
use rstar::RTree;
use time::OffsetDateTime;
@ -27,7 +26,7 @@ use crate::{
default_criteria, CboRoaringBitmapCodec, Criterion, DocumentId, ExternalDocumentsIds,
FacetDistribution, FieldDistribution, FieldId, FieldIdWordCountCodec, GeoPoint, ObkvCodec,
OrderBy, Result, RoaringBitmapCodec, RoaringBitmapLenCodec, Search, U8StrStrCodec, BEU16,
BEU32,
BEU32, BEU64,
};
/// The HNSW data-structure that we serialize, fill and search in.
@ -109,10 +108,10 @@ pub struct Index {
pub(crate) env: heed::Env,
/// Contains many different types (e.g. the fields ids map).
pub(crate) main: PolyDatabase,
pub(crate) main: Database<Unspecified, Unspecified>,
/// Maps the external documents ids with the internal document id.
pub external_documents_ids: Database<Str, OwnedType<BEU32>>,
pub external_documents_ids: Database<Str, BEU32>,
/// A word and all the documents ids containing the word.
pub word_docids: Database<Str, CboRoaringBitmapCodec>,
@ -158,7 +157,7 @@ pub struct Index {
/// Maps the facet field id of the normalized-for-search string facets with their original versions.
pub facet_id_normalized_string_strings: Database<BEU16StrCodec, SerdeJson<BTreeSet<String>>>,
/// Maps the facet field id of the string facets with an FST containing all the facets values.
pub facet_id_string_fst: Database<OwnedType<BEU16>, FstSetCodec>,
pub facet_id_string_fst: Database<BEU16, FstSetCodec>,
/// Maps the document id, the facet field id and the numbers.
pub field_id_docid_facet_f64s: Database<FieldDocIdFacetF64Codec, Unit>,
@ -166,10 +165,10 @@ pub struct Index {
pub field_id_docid_facet_strings: Database<FieldDocIdFacetStringCodec, Str>,
/// Maps a vector id to the document id that have it.
pub vector_id_docid: Database<OwnedType<BEU32>, OwnedType<BEU32>>,
pub vector_id_docid: Database<BEU32, BEU32>,
/// Maps the document id to the document as an obkv store.
pub(crate) documents: Database<OwnedType<BEU32>, ObkvCodec>,
pub(crate) documents: Database<BEU32, ObkvCodec>,
}
impl Index {
@ -182,11 +181,10 @@ impl Index {
use db_name::*;
options.max_dbs(24);
unsafe { options.flag(Flags::MdbAlwaysFreePages) };
let env = options.open(path)?;
let mut wtxn = env.write_txn()?;
let main = env.create_poly_database(&mut wtxn, Some(MAIN))?;
let main = env.database_options().name(MAIN).create(&mut wtxn)?;
let word_docids = env.create_database(&mut wtxn, Some(WORD_DOCIDS))?;
let external_documents_ids =
env.create_database(&mut wtxn, Some(EXTERNAL_DOCUMENTS_IDS))?;
@ -264,24 +262,16 @@ impl Index {
fn set_creation_dates(
env: &heed::Env,
main: PolyDatabase,
main: Database<Unspecified, Unspecified>,
created_at: OffsetDateTime,
updated_at: OffsetDateTime,
) -> heed::Result<()> {
let mut txn = env.write_txn()?;
// The db was just created, we update its metadata with the relevant information.
if main.get::<_, Str, SerdeJson<OffsetDateTime>>(&txn, main_key::CREATED_AT_KEY)?.is_none()
{
main.put::<_, Str, SerdeJson<OffsetDateTime>>(
&mut txn,
main_key::UPDATED_AT_KEY,
&updated_at,
)?;
main.put::<_, Str, SerdeJson<OffsetDateTime>>(
&mut txn,
main_key::CREATED_AT_KEY,
&created_at,
)?;
let main = main.remap_types::<Str, SerdeJson<OffsetDateTime>>();
if main.get(&txn, main_key::CREATED_AT_KEY)?.is_none() {
main.put(&mut txn, main_key::UPDATED_AT_KEY, &updated_at)?;
main.put(&mut txn, main_key::CREATED_AT_KEY, &created_at)?;
txn.commit()?;
}
Ok(())
@ -318,12 +308,12 @@ impl Index {
///
/// This value is the maximum between the map size passed during the opening of the index
/// and the on-disk size of the index at the time of opening.
pub fn map_size(&self) -> Result<usize> {
Ok(self.env.map_size()?)
pub fn map_size(&self) -> usize {
self.env.info().map_size
}
pub fn copy_to_path<P: AsRef<Path>>(&self, path: P, option: CompactionOption) -> Result<File> {
self.env.copy_to_path(path, option).map_err(Into::into)
pub fn copy_to_file<P: AsRef<Path>>(&self, path: P, option: CompactionOption) -> Result<File> {
self.env.copy_to_file(path, option).map_err(Into::into)
}
/// Returns an `EnvClosingEvent` that can be used to wait for the closing event,
@ -343,21 +333,28 @@ impl Index {
wtxn: &mut RwTxn,
docids: &RoaringBitmap,
) -> heed::Result<()> {
self.main.put::<_, Str, RoaringBitmapCodec>(wtxn, main_key::DOCUMENTS_IDS_KEY, docids)
self.main.remap_types::<Str, RoaringBitmapCodec>().put(
wtxn,
main_key::DOCUMENTS_IDS_KEY,
docids,
)
}
/// Returns the internal documents ids.
pub fn documents_ids(&self, rtxn: &RoTxn) -> heed::Result<RoaringBitmap> {
Ok(self
.main
.get::<_, Str, RoaringBitmapCodec>(rtxn, main_key::DOCUMENTS_IDS_KEY)?
.remap_types::<Str, RoaringBitmapCodec>()
.get(rtxn, main_key::DOCUMENTS_IDS_KEY)?
.unwrap_or_default())
}
/// Returns the number of documents indexed in the database.
pub fn number_of_documents(&self, rtxn: &RoTxn) -> Result<u64> {
let count =
self.main.get::<_, Str, RoaringBitmapLenCodec>(rtxn, main_key::DOCUMENTS_IDS_KEY)?;
let count = self
.main
.remap_types::<Str, RoaringBitmapLenCodec>()
.get(rtxn, main_key::DOCUMENTS_IDS_KEY)?;
Ok(count.unwrap_or_default())
}
@ -366,17 +363,17 @@ impl Index {
/// Writes the documents primary key, this is the field name that is used to store the id.
pub(crate) fn put_primary_key(&self, wtxn: &mut RwTxn, primary_key: &str) -> heed::Result<()> {
self.set_updated_at(wtxn, &OffsetDateTime::now_utc())?;
self.main.put::<_, Str, Str>(wtxn, main_key::PRIMARY_KEY_KEY, primary_key)
self.main.remap_types::<Str, Str>().put(wtxn, main_key::PRIMARY_KEY_KEY, primary_key)
}
/// Deletes the primary key of the documents, this can be done to reset indexes settings.
pub(crate) fn delete_primary_key(&self, wtxn: &mut RwTxn) -> heed::Result<bool> {
self.main.delete::<_, Str>(wtxn, main_key::PRIMARY_KEY_KEY)
self.main.remap_key_type::<Str>().delete(wtxn, main_key::PRIMARY_KEY_KEY)
}
/// Returns the documents primary key, `None` if it hasn't been defined.
pub fn primary_key<'t>(&self, rtxn: &'t RoTxn) -> heed::Result<Option<&'t str>> {
self.main.get::<_, Str, Str>(rtxn, main_key::PRIMARY_KEY_KEY)
self.main.remap_types::<Str, Str>().get(rtxn, main_key::PRIMARY_KEY_KEY)
}
/* external documents ids */
@ -396,7 +393,11 @@ impl Index {
wtxn: &mut RwTxn,
map: &FieldsIdsMap,
) -> heed::Result<()> {
self.main.put::<_, Str, SerdeJson<FieldsIdsMap>>(wtxn, main_key::FIELDS_IDS_MAP_KEY, map)
self.main.remap_types::<Str, SerdeJson<FieldsIdsMap>>().put(
wtxn,
main_key::FIELDS_IDS_MAP_KEY,
map,
)
}
/// Returns the fields ids map which associate the documents keys with an internal field id
@ -404,7 +405,8 @@ impl Index {
pub fn fields_ids_map(&self, rtxn: &RoTxn) -> heed::Result<FieldsIdsMap> {
Ok(self
.main
.get::<_, Str, SerdeJson<FieldsIdsMap>>(rtxn, main_key::FIELDS_IDS_MAP_KEY)?
.remap_types::<Str, SerdeJson<FieldsIdsMap>>()
.get(rtxn, main_key::FIELDS_IDS_MAP_KEY)?
.unwrap_or_default())
}
@ -416,19 +418,24 @@ impl Index {
wtxn: &mut RwTxn,
rtree: &RTree<GeoPoint>,
) -> heed::Result<()> {
self.main.put::<_, Str, SerdeBincode<RTree<GeoPoint>>>(wtxn, main_key::GEO_RTREE_KEY, rtree)
self.main.remap_types::<Str, SerdeBincode<RTree<GeoPoint>>>().put(
wtxn,
main_key::GEO_RTREE_KEY,
rtree,
)
}
/// Delete the `rtree` which associates coordinates to documents ids.
pub(crate) fn delete_geo_rtree(&self, wtxn: &mut RwTxn) -> heed::Result<bool> {
self.main.delete::<_, Str>(wtxn, main_key::GEO_RTREE_KEY)
self.main.remap_key_type::<Str>().delete(wtxn, main_key::GEO_RTREE_KEY)
}
/// Returns the `rtree` which associates coordinates to documents ids.
pub fn geo_rtree(&self, rtxn: &RoTxn) -> Result<Option<RTree<GeoPoint>>> {
match self
.main
.get::<_, Str, SerdeBincode<RTree<GeoPoint>>>(rtxn, main_key::GEO_RTREE_KEY)?
.remap_types::<Str, SerdeBincode<RTree<GeoPoint>>>()
.get(rtxn, main_key::GEO_RTREE_KEY)?
{
Some(rtree) => Ok(Some(rtree)),
None => Ok(None),
@ -443,7 +450,7 @@ impl Index {
wtxn: &mut RwTxn,
docids: &RoaringBitmap,
) -> heed::Result<()> {
self.main.put::<_, Str, RoaringBitmapCodec>(
self.main.remap_types::<Str, RoaringBitmapCodec>().put(
wtxn,
main_key::GEO_FACETED_DOCUMENTS_IDS_KEY,
docids,
@ -452,14 +459,15 @@ impl Index {
/// Delete the documents ids that are faceted with a _geo field.
pub(crate) fn delete_geo_faceted_documents_ids(&self, wtxn: &mut RwTxn) -> heed::Result<bool> {
self.main.delete::<_, Str>(wtxn, main_key::GEO_FACETED_DOCUMENTS_IDS_KEY)
self.main.remap_key_type::<Str>().delete(wtxn, main_key::GEO_FACETED_DOCUMENTS_IDS_KEY)
}
/// Retrieve all the documents ids that are faceted with a _geo field.
pub fn geo_faceted_documents_ids(&self, rtxn: &RoTxn) -> heed::Result<RoaringBitmap> {
match self
.main
.get::<_, Str, RoaringBitmapCodec>(rtxn, main_key::GEO_FACETED_DOCUMENTS_IDS_KEY)?
.remap_types::<Str, RoaringBitmapCodec>()
.get(rtxn, main_key::GEO_FACETED_DOCUMENTS_IDS_KEY)?
{
Some(docids) => Ok(docids),
None => Ok(RoaringBitmap::new()),
@ -474,22 +482,22 @@ impl Index {
self.delete_vector_hnsw(wtxn)?;
let chunk_size = 1024 * 1024 * (1024 + 512); // 1.5 GiB
let bytes = bincode::serialize(hnsw).map_err(|_| heed::Error::Encoding)?;
let bytes = bincode::serialize(hnsw).map_err(Into::into).map_err(heed::Error::Encoding)?;
for (i, chunk) in bytes.chunks(chunk_size).enumerate() {
let i = i as u32;
let mut key = main_key::VECTOR_HNSW_KEY_PREFIX.as_bytes().to_vec();
key.extend_from_slice(&i.to_be_bytes());
self.main.put::<_, ByteSlice, ByteSlice>(wtxn, &key, chunk)?;
self.main.remap_types::<Bytes, Bytes>().put(wtxn, &key, chunk)?;
}
Ok(())
}
/// Delete the `hnsw`.
pub(crate) fn delete_vector_hnsw(&self, wtxn: &mut RwTxn) -> heed::Result<bool> {
let mut iter = self.main.prefix_iter_mut::<_, ByteSlice, DecodeIgnore>(
wtxn,
main_key::VECTOR_HNSW_KEY_PREFIX.as_bytes(),
)?;
let mut iter = self
.main
.remap_types::<Bytes, DecodeIgnore>()
.prefix_iter_mut(wtxn, main_key::VECTOR_HNSW_KEY_PREFIX.as_bytes())?;
let mut deleted = false;
while iter.next().transpose()?.is_some() {
// We do not keep a reference to the key or the value.
@ -501,8 +509,10 @@ impl Index {
/// Returns the `hnsw`.
pub fn vector_hnsw(&self, rtxn: &RoTxn) -> Result<Option<Hnsw>> {
let mut slices = Vec::new();
for result in
self.main.prefix_iter::<_, Str, ByteSlice>(rtxn, main_key::VECTOR_HNSW_KEY_PREFIX)?
for result in self
.main
.remap_types::<Str, Bytes>()
.prefix_iter(rtxn, main_key::VECTOR_HNSW_KEY_PREFIX)?
{
let (_, slice) = result?;
slices.push(slice);
@ -512,7 +522,11 @@ impl Index {
Ok(None)
} else {
let readable_slices: ReadableSlices<_> = slices.into_iter().collect();
Ok(Some(bincode::deserialize_from(readable_slices).map_err(|_| heed::Error::Decoding)?))
Ok(Some(
bincode::deserialize_from(readable_slices)
.map_err(Into::into)
.map_err(heed::Error::Decoding)?,
))
}
}
@ -525,7 +539,7 @@ impl Index {
wtxn: &mut RwTxn,
distribution: &FieldDistribution,
) -> heed::Result<()> {
self.main.put::<_, Str, SerdeJson<FieldDistribution>>(
self.main.remap_types::<Str, SerdeJson<FieldDistribution>>().put(
wtxn,
main_key::FIELD_DISTRIBUTION_KEY,
distribution,
@ -537,7 +551,8 @@ impl Index {
pub fn field_distribution(&self, rtxn: &RoTxn) -> heed::Result<FieldDistribution> {
Ok(self
.main
.get::<_, Str, SerdeJson<FieldDistribution>>(rtxn, main_key::FIELD_DISTRIBUTION_KEY)?
.remap_types::<Str, SerdeJson<FieldDistribution>>()
.get(rtxn, main_key::FIELD_DISTRIBUTION_KEY)?
.unwrap_or_default())
}
@ -550,7 +565,7 @@ impl Index {
wtxn: &mut RwTxn,
fields: &[&str],
) -> heed::Result<()> {
self.main.put::<_, Str, SerdeBincode<&[&str]>>(
self.main.remap_types::<Str, SerdeBincode<&[&str]>>().put(
wtxn,
main_key::DISPLAYED_FIELDS_KEY,
&fields,
@ -560,13 +575,15 @@ impl Index {
/// Deletes the displayed fields ids, this will make the engine to display
/// all the documents attributes in the order of the `FieldsIdsMap`.
pub(crate) fn delete_displayed_fields(&self, wtxn: &mut RwTxn) -> heed::Result<bool> {
self.main.delete::<_, Str>(wtxn, main_key::DISPLAYED_FIELDS_KEY)
self.main.remap_key_type::<Str>().delete(wtxn, main_key::DISPLAYED_FIELDS_KEY)
}
/// Returns the displayed fields in the order they were set by the user. If it returns
/// `None` it means that all the attributes are set as displayed in the order of the `FieldsIdsMap`.
pub fn displayed_fields<'t>(&self, rtxn: &'t RoTxn) -> heed::Result<Option<Vec<&'t str>>> {
self.main.get::<_, Str, SerdeBincode<Vec<&'t str>>>(rtxn, main_key::DISPLAYED_FIELDS_KEY)
self.main
.remap_types::<Str, SerdeBincode<Vec<&'t str>>>()
.get(rtxn, main_key::DISPLAYED_FIELDS_KEY)
}
/// Identical to `displayed_fields`, but returns the ids instead.
@ -646,7 +663,7 @@ impl Index {
/// Writes the searchable fields, when this list is specified, only these are indexed.
fn put_searchable_fields(&self, wtxn: &mut RwTxn, fields: &[&str]) -> heed::Result<()> {
self.main.put::<_, Str, SerdeBincode<&[&str]>>(
self.main.remap_types::<Str, SerdeBincode<&[&str]>>().put(
wtxn,
main_key::SEARCHABLE_FIELDS_KEY,
&fields,
@ -655,13 +672,15 @@ impl Index {
/// Deletes the searchable fields, when no fields are specified, all fields are indexed.
fn delete_searchable_fields(&self, wtxn: &mut RwTxn) -> heed::Result<bool> {
self.main.delete::<_, Str>(wtxn, main_key::SEARCHABLE_FIELDS_KEY)
self.main.remap_key_type::<Str>().delete(wtxn, main_key::SEARCHABLE_FIELDS_KEY)
}
/// Returns the searchable fields, those are the fields that are indexed,
/// if the searchable fields aren't there it means that **all** the fields are indexed.
pub fn searchable_fields<'t>(&self, rtxn: &'t RoTxn) -> heed::Result<Option<Vec<&'t str>>> {
self.main.get::<_, Str, SerdeBincode<Vec<&'t str>>>(rtxn, main_key::SEARCHABLE_FIELDS_KEY)
self.main
.remap_types::<Str, SerdeBincode<Vec<&'t str>>>()
.get(rtxn, main_key::SEARCHABLE_FIELDS_KEY)
}
/// Identical to `searchable_fields`, but returns the ids instead.
@ -687,7 +706,7 @@ impl Index {
wtxn: &mut RwTxn,
fields: &[&str],
) -> heed::Result<()> {
self.main.put::<_, Str, SerdeBincode<_>>(
self.main.remap_types::<Str, SerdeBincode<_>>().put(
wtxn,
main_key::USER_DEFINED_SEARCHABLE_FIELDS_KEY,
&fields,
@ -699,7 +718,7 @@ impl Index {
&self,
wtxn: &mut RwTxn,
) -> heed::Result<bool> {
self.main.delete::<_, Str>(wtxn, main_key::USER_DEFINED_SEARCHABLE_FIELDS_KEY)
self.main.remap_key_type::<Str>().delete(wtxn, main_key::USER_DEFINED_SEARCHABLE_FIELDS_KEY)
}
/// Returns the user defined searchable fields.
@ -708,7 +727,8 @@ impl Index {
rtxn: &'t RoTxn,
) -> heed::Result<Option<Vec<&'t str>>> {
self.main
.get::<_, Str, SerdeBincode<Vec<_>>>(rtxn, main_key::USER_DEFINED_SEARCHABLE_FIELDS_KEY)
.remap_types::<Str, SerdeBincode<Vec<_>>>()
.get(rtxn, main_key::USER_DEFINED_SEARCHABLE_FIELDS_KEY)
}
/* filterable fields */
@ -719,19 +739,24 @@ impl Index {
wtxn: &mut RwTxn,
fields: &HashSet<String>,
) -> heed::Result<()> {
self.main.put::<_, Str, SerdeJson<_>>(wtxn, main_key::FILTERABLE_FIELDS_KEY, fields)
self.main.remap_types::<Str, SerdeJson<_>>().put(
wtxn,
main_key::FILTERABLE_FIELDS_KEY,
fields,
)
}
/// Deletes the filterable fields ids in the database.
pub(crate) fn delete_filterable_fields(&self, wtxn: &mut RwTxn) -> heed::Result<bool> {
self.main.delete::<_, Str>(wtxn, main_key::FILTERABLE_FIELDS_KEY)
self.main.remap_key_type::<Str>().delete(wtxn, main_key::FILTERABLE_FIELDS_KEY)
}
/// Returns the filterable fields names.
pub fn filterable_fields(&self, rtxn: &RoTxn) -> heed::Result<HashSet<String>> {
Ok(self
.main
.get::<_, Str, SerdeJson<_>>(rtxn, main_key::FILTERABLE_FIELDS_KEY)?
.remap_types::<Str, SerdeJson<_>>()
.get(rtxn, main_key::FILTERABLE_FIELDS_KEY)?
.unwrap_or_default())
}
@ -758,19 +783,24 @@ impl Index {
wtxn: &mut RwTxn,
fields: &HashSet<String>,
) -> heed::Result<()> {
self.main.put::<_, Str, SerdeJson<_>>(wtxn, main_key::SORTABLE_FIELDS_KEY, fields)
self.main.remap_types::<Str, SerdeJson<_>>().put(
wtxn,
main_key::SORTABLE_FIELDS_KEY,
fields,
)
}
/// Deletes the sortable fields ids in the database.
pub(crate) fn delete_sortable_fields(&self, wtxn: &mut RwTxn) -> heed::Result<bool> {
self.main.delete::<_, Str>(wtxn, main_key::SORTABLE_FIELDS_KEY)
self.main.remap_key_type::<Str>().delete(wtxn, main_key::SORTABLE_FIELDS_KEY)
}
/// Returns the sortable fields names.
pub fn sortable_fields(&self, rtxn: &RoTxn) -> heed::Result<HashSet<String>> {
Ok(self
.main
.get::<_, Str, SerdeJson<_>>(rtxn, main_key::SORTABLE_FIELDS_KEY)?
.remap_types::<Str, SerdeJson<_>>()
.get(rtxn, main_key::SORTABLE_FIELDS_KEY)?
.unwrap_or_default())
}
@ -789,14 +819,19 @@ impl Index {
wtxn: &mut RwTxn,
fields: &HashSet<String>,
) -> heed::Result<()> {
self.main.put::<_, Str, SerdeJson<_>>(wtxn, main_key::HIDDEN_FACETED_FIELDS_KEY, fields)
self.main.remap_types::<Str, SerdeJson<_>>().put(
wtxn,
main_key::HIDDEN_FACETED_FIELDS_KEY,
fields,
)
}
/// Returns the faceted fields names.
pub fn faceted_fields(&self, rtxn: &RoTxn) -> heed::Result<HashSet<String>> {
Ok(self
.main
.get::<_, Str, SerdeJson<_>>(rtxn, main_key::HIDDEN_FACETED_FIELDS_KEY)?
.remap_types::<Str, SerdeJson<_>>()
.get(rtxn, main_key::HIDDEN_FACETED_FIELDS_KEY)?
.unwrap_or_default())
}
@ -863,7 +898,7 @@ impl Index {
rtxn: &RoTxn,
field_id: FieldId,
) -> heed::Result<RoaringBitmap> {
match self.facet_id_is_null_docids.get(rtxn, &BEU16::new(field_id))? {
match self.facet_id_is_null_docids.get(rtxn, &field_id)? {
Some(docids) => Ok(docids),
None => Ok(RoaringBitmap::new()),
}
@ -875,7 +910,7 @@ impl Index {
rtxn: &RoTxn,
field_id: FieldId,
) -> heed::Result<RoaringBitmap> {
match self.facet_id_is_empty_docids.get(rtxn, &BEU16::new(field_id))? {
match self.facet_id_is_empty_docids.get(rtxn, &field_id)? {
Some(docids) => Ok(docids),
None => Ok(RoaringBitmap::new()),
}
@ -887,7 +922,7 @@ impl Index {
rtxn: &RoTxn,
field_id: FieldId,
) -> heed::Result<RoaringBitmap> {
match self.facet_id_exists_docids.get(rtxn, &BEU16::new(field_id))? {
match self.facet_id_exists_docids.get(rtxn, &field_id)? {
Some(docids) => Ok(docids),
None => Ok(RoaringBitmap::new()),
}
@ -900,15 +935,15 @@ impl Index {
wtxn: &mut RwTxn,
distinct_field: &str,
) -> heed::Result<()> {
self.main.put::<_, Str, Str>(wtxn, main_key::DISTINCT_FIELD_KEY, distinct_field)
self.main.remap_types::<Str, Str>().put(wtxn, main_key::DISTINCT_FIELD_KEY, distinct_field)
}
pub fn distinct_field<'a>(&self, rtxn: &'a RoTxn) -> heed::Result<Option<&'a str>> {
self.main.get::<_, Str, Str>(rtxn, main_key::DISTINCT_FIELD_KEY)
self.main.remap_types::<Str, Str>().get(rtxn, main_key::DISTINCT_FIELD_KEY)
}
pub(crate) fn delete_distinct_field(&self, wtxn: &mut RwTxn) -> heed::Result<bool> {
self.main.delete::<_, Str>(wtxn, main_key::DISTINCT_FIELD_KEY)
self.main.remap_key_type::<Str>().delete(wtxn, main_key::DISTINCT_FIELD_KEY)
}
/* criteria */
@ -918,15 +953,23 @@ impl Index {
wtxn: &mut RwTxn,
criteria: &[Criterion],
) -> heed::Result<()> {
self.main.put::<_, Str, SerdeJson<&[Criterion]>>(wtxn, main_key::CRITERIA_KEY, &criteria)
self.main.remap_types::<Str, SerdeJson<&[Criterion]>>().put(
wtxn,
main_key::CRITERIA_KEY,
&criteria,
)
}
pub(crate) fn delete_criteria(&self, wtxn: &mut RwTxn) -> heed::Result<bool> {
self.main.delete::<_, Str>(wtxn, main_key::CRITERIA_KEY)
self.main.remap_key_type::<Str>().delete(wtxn, main_key::CRITERIA_KEY)
}
pub fn criteria(&self, rtxn: &RoTxn) -> heed::Result<Vec<Criterion>> {
match self.main.get::<_, Str, SerdeJson<Vec<Criterion>>>(rtxn, main_key::CRITERIA_KEY)? {
match self
.main
.remap_types::<Str, SerdeJson<Vec<Criterion>>>()
.get(rtxn, main_key::CRITERIA_KEY)?
{
Some(criteria) => Ok(criteria),
None => Ok(default_criteria()),
}
@ -940,12 +983,16 @@ impl Index {
wtxn: &mut RwTxn,
fst: &fst::Set<A>,
) -> heed::Result<()> {
self.main.put::<_, Str, ByteSlice>(wtxn, main_key::WORDS_FST_KEY, fst.as_fst().as_bytes())
self.main.remap_types::<Str, Bytes>().put(
wtxn,
main_key::WORDS_FST_KEY,
fst.as_fst().as_bytes(),
)
}
/// Returns the FST which is the words dictionary of the engine.
pub fn words_fst<'t>(&self, rtxn: &'t RoTxn) -> Result<fst::Set<Cow<'t, [u8]>>> {
match self.main.get::<_, Str, ByteSlice>(rtxn, main_key::WORDS_FST_KEY)? {
match self.main.remap_types::<Str, Bytes>().get(rtxn, main_key::WORDS_FST_KEY)? {
Some(bytes) => Ok(fst::Set::new(bytes)?.map_data(Cow::Borrowed)?),
None => Ok(fst::Set::default().map_data(Cow::Owned)?),
}
@ -958,15 +1005,19 @@ impl Index {
wtxn: &mut RwTxn,
fst: &fst::Set<A>,
) -> heed::Result<()> {
self.main.put::<_, Str, ByteSlice>(wtxn, main_key::STOP_WORDS_KEY, fst.as_fst().as_bytes())
self.main.remap_types::<Str, Bytes>().put(
wtxn,
main_key::STOP_WORDS_KEY,
fst.as_fst().as_bytes(),
)
}
pub(crate) fn delete_stop_words(&self, wtxn: &mut RwTxn) -> heed::Result<bool> {
self.main.delete::<_, Str>(wtxn, main_key::STOP_WORDS_KEY)
self.main.remap_key_type::<Str>().delete(wtxn, main_key::STOP_WORDS_KEY)
}
pub fn stop_words<'t>(&self, rtxn: &'t RoTxn) -> Result<Option<fst::Set<&'t [u8]>>> {
match self.main.get::<_, Str, ByteSlice>(rtxn, main_key::STOP_WORDS_KEY)? {
match self.main.remap_types::<Str, Bytes>().get(rtxn, main_key::STOP_WORDS_KEY)? {
Some(bytes) => Ok(Some(fst::Set::new(bytes)?)),
None => Ok(None),
}
@ -979,18 +1030,22 @@ impl Index {
wtxn: &mut RwTxn,
set: &BTreeSet<String>,
) -> heed::Result<()> {
self.main.put::<_, Str, SerdeBincode<_>>(wtxn, main_key::NON_SEPARATOR_TOKENS_KEY, set)
self.main.remap_types::<Str, SerdeBincode<_>>().put(
wtxn,
main_key::NON_SEPARATOR_TOKENS_KEY,
set,
)
}
pub(crate) fn delete_non_separator_tokens(&self, wtxn: &mut RwTxn) -> heed::Result<bool> {
self.main.delete::<_, Str>(wtxn, main_key::NON_SEPARATOR_TOKENS_KEY)
self.main.remap_key_type::<Str>().delete(wtxn, main_key::NON_SEPARATOR_TOKENS_KEY)
}
pub fn non_separator_tokens(&self, rtxn: &RoTxn) -> Result<Option<BTreeSet<String>>> {
Ok(self.main.get::<_, Str, SerdeBincode<BTreeSet<String>>>(
rtxn,
main_key::NON_SEPARATOR_TOKENS_KEY,
)?)
Ok(self
.main
.remap_types::<Str, SerdeBincode<BTreeSet<String>>>()
.get(rtxn, main_key::NON_SEPARATOR_TOKENS_KEY)?)
}
/* separator tokens */
@ -1000,17 +1055,22 @@ impl Index {
wtxn: &mut RwTxn,
set: &BTreeSet<String>,
) -> heed::Result<()> {
self.main.put::<_, Str, SerdeBincode<_>>(wtxn, main_key::SEPARATOR_TOKENS_KEY, set)
self.main.remap_types::<Str, SerdeBincode<_>>().put(
wtxn,
main_key::SEPARATOR_TOKENS_KEY,
set,
)
}
pub(crate) fn delete_separator_tokens(&self, wtxn: &mut RwTxn) -> heed::Result<bool> {
self.main.delete::<_, Str>(wtxn, main_key::SEPARATOR_TOKENS_KEY)
self.main.remap_key_type::<Str>().delete(wtxn, main_key::SEPARATOR_TOKENS_KEY)
}
pub fn separator_tokens(&self, rtxn: &RoTxn) -> Result<Option<BTreeSet<String>>> {
Ok(self
.main
.get::<_, Str, SerdeBincode<BTreeSet<String>>>(rtxn, main_key::SEPARATOR_TOKENS_KEY)?)
.remap_types::<Str, SerdeBincode<BTreeSet<String>>>()
.get(rtxn, main_key::SEPARATOR_TOKENS_KEY)?)
}
/* separators easing method */
@ -1040,17 +1100,18 @@ impl Index {
wtxn: &mut RwTxn,
set: &BTreeSet<String>,
) -> heed::Result<()> {
self.main.put::<_, Str, SerdeBincode<_>>(wtxn, main_key::DICTIONARY_KEY, set)
self.main.remap_types::<Str, SerdeBincode<_>>().put(wtxn, main_key::DICTIONARY_KEY, set)
}
pub(crate) fn delete_dictionary(&self, wtxn: &mut RwTxn) -> heed::Result<bool> {
self.main.delete::<_, Str>(wtxn, main_key::DICTIONARY_KEY)
self.main.remap_key_type::<Str>().delete(wtxn, main_key::DICTIONARY_KEY)
}
pub fn dictionary(&self, rtxn: &RoTxn) -> Result<Option<BTreeSet<String>>> {
Ok(self
.main
.get::<_, Str, SerdeBincode<BTreeSet<String>>>(rtxn, main_key::DICTIONARY_KEY)?)
.remap_types::<Str, SerdeBincode<BTreeSet<String>>>()
.get(rtxn, main_key::DICTIONARY_KEY)?)
}
/* synonyms */
@ -1061,8 +1122,12 @@ impl Index {
synonyms: &HashMap<Vec<String>, Vec<Vec<String>>>,
user_defined_synonyms: &BTreeMap<String, Vec<String>>,
) -> heed::Result<()> {
self.main.put::<_, Str, SerdeBincode<_>>(wtxn, main_key::SYNONYMS_KEY, synonyms)?;
self.main.put::<_, Str, SerdeBincode<_>>(
self.main.remap_types::<Str, SerdeBincode<_>>().put(
wtxn,
main_key::SYNONYMS_KEY,
synonyms,
)?;
self.main.remap_types::<Str, SerdeBincode<_>>().put(
wtxn,
main_key::USER_DEFINED_SYNONYMS_KEY,
user_defined_synonyms,
@ -1070,8 +1135,8 @@ impl Index {
}
pub(crate) fn delete_synonyms(&self, wtxn: &mut RwTxn) -> heed::Result<bool> {
self.main.delete::<_, Str>(wtxn, main_key::SYNONYMS_KEY)?;
self.main.delete::<_, Str>(wtxn, main_key::USER_DEFINED_SYNONYMS_KEY)
self.main.remap_key_type::<Str>().delete(wtxn, main_key::SYNONYMS_KEY)?;
self.main.remap_key_type::<Str>().delete(wtxn, main_key::USER_DEFINED_SYNONYMS_KEY)
}
pub fn user_defined_synonyms(
@ -1080,14 +1145,16 @@ impl Index {
) -> heed::Result<BTreeMap<String, Vec<String>>> {
Ok(self
.main
.get::<_, Str, SerdeBincode<_>>(rtxn, main_key::USER_DEFINED_SYNONYMS_KEY)?
.remap_types::<Str, SerdeBincode<_>>()
.get(rtxn, main_key::USER_DEFINED_SYNONYMS_KEY)?
.unwrap_or_default())
}
pub fn synonyms(&self, rtxn: &RoTxn) -> heed::Result<HashMap<Vec<String>, Vec<Vec<String>>>> {
Ok(self
.main
.get::<_, Str, SerdeBincode<_>>(rtxn, main_key::SYNONYMS_KEY)?
.remap_types::<Str, SerdeBincode<_>>()
.get(rtxn, main_key::SYNONYMS_KEY)?
.unwrap_or_default())
}
@ -1108,7 +1175,7 @@ impl Index {
wtxn: &mut RwTxn,
fst: &fst::Set<A>,
) -> heed::Result<()> {
self.main.put::<_, Str, ByteSlice>(
self.main.remap_types::<Str, Bytes>().put(
wtxn,
main_key::WORDS_PREFIXES_FST_KEY,
fst.as_fst().as_bytes(),
@ -1117,7 +1184,7 @@ impl Index {
/// Returns the FST which is the words prefixes dictionnary of the engine.
pub fn words_prefixes_fst<'t>(&self, rtxn: &'t RoTxn) -> Result<fst::Set<Cow<'t, [u8]>>> {
match self.main.get::<_, Str, ByteSlice>(rtxn, main_key::WORDS_PREFIXES_FST_KEY)? {
match self.main.remap_types::<Str, Bytes>().get(rtxn, main_key::WORDS_PREFIXES_FST_KEY)? {
Some(bytes) => Ok(fst::Set::new(bytes)?.map_data(Cow::Borrowed)?),
None => Ok(fst::Set::default().map_data(Cow::Owned)?),
}
@ -1142,7 +1209,7 @@ impl Index {
Ok(ids.into_iter().map(move |id| {
let kv = self
.documents
.get(rtxn, &BEU32::new(id))?
.get(rtxn, &id)?
.ok_or(UserError::UnknownInternalDocumentId { document_id: id })?;
Ok((id, kv))
}))
@ -1207,7 +1274,8 @@ impl Index {
pub fn created_at(&self, rtxn: &RoTxn) -> Result<OffsetDateTime> {
Ok(self
.main
.get::<_, Str, SerdeJson<OffsetDateTime>>(rtxn, main_key::CREATED_AT_KEY)?
.remap_types::<Str, SerdeJson<OffsetDateTime>>()
.get(rtxn, main_key::CREATED_AT_KEY)?
.ok_or(InternalError::DatabaseMissingEntry {
db_name: db_name::MAIN,
key: Some(main_key::CREATED_AT_KEY),
@ -1218,7 +1286,8 @@ impl Index {
pub fn updated_at(&self, rtxn: &RoTxn) -> Result<OffsetDateTime> {
Ok(self
.main
.get::<_, Str, SerdeJson<OffsetDateTime>>(rtxn, main_key::UPDATED_AT_KEY)?
.remap_types::<Str, SerdeJson<OffsetDateTime>>()
.get(rtxn, main_key::UPDATED_AT_KEY)?
.ok_or(InternalError::DatabaseMissingEntry {
db_name: db_name::MAIN,
key: Some(main_key::UPDATED_AT_KEY),
@ -1230,14 +1299,18 @@ impl Index {
wtxn: &mut RwTxn,
time: &OffsetDateTime,
) -> heed::Result<()> {
self.main.put::<_, Str, SerdeJson<OffsetDateTime>>(wtxn, main_key::UPDATED_AT_KEY, time)
self.main.remap_types::<Str, SerdeJson<OffsetDateTime>>().put(
wtxn,
main_key::UPDATED_AT_KEY,
time,
)
}
pub fn authorize_typos(&self, txn: &RoTxn) -> heed::Result<bool> {
// It is not possible to put a bool in heed with OwnedType, so we put a u8 instead. We
// identify 0 as being false, and anything else as true. The absence of a value is true,
// because by default, we authorize typos.
match self.main.get::<_, Str, OwnedType<u8>>(txn, main_key::AUTHORIZE_TYPOS)? {
match self.main.remap_types::<Str, U8>().get(txn, main_key::AUTHORIZE_TYPOS)? {
Some(0) => Ok(false),
_ => Ok(true),
}
@ -1247,7 +1320,7 @@ impl Index {
// It is not possible to put a bool in heed with OwnedType, so we put a u8 instead. We
// identify 0 as being false, and anything else as true. The absence of a value is true,
// because by default, we authorize typos.
self.main.put::<_, Str, OwnedType<u8>>(txn, main_key::AUTHORIZE_TYPOS, &(flag as u8))?;
self.main.remap_types::<Str, U8>().put(txn, main_key::AUTHORIZE_TYPOS, &(flag as u8))?;
Ok(())
}
@ -1258,7 +1331,8 @@ impl Index {
// because by default, we authorize typos.
Ok(self
.main
.get::<_, Str, OwnedType<u8>>(txn, main_key::ONE_TYPO_WORD_LEN)?
.remap_types::<Str, U8>()
.get(txn, main_key::ONE_TYPO_WORD_LEN)?
.unwrap_or(DEFAULT_MIN_WORD_LEN_ONE_TYPO))
}
@ -1266,7 +1340,7 @@ impl Index {
// It is not possible to put a bool in heed with OwnedType, so we put a u8 instead. We
// identify 0 as being false, and anything else as true. The absence of a value is true,
// because by default, we authorize typos.
self.main.put::<_, Str, OwnedType<u8>>(txn, main_key::ONE_TYPO_WORD_LEN, &val)?;
self.main.remap_types::<Str, U8>().put(txn, main_key::ONE_TYPO_WORD_LEN, &val)?;
Ok(())
}
@ -1276,7 +1350,8 @@ impl Index {
// because by default, we authorize typos.
Ok(self
.main
.get::<_, Str, OwnedType<u8>>(txn, main_key::TWO_TYPOS_WORD_LEN)?
.remap_types::<Str, U8>()
.get(txn, main_key::TWO_TYPOS_WORD_LEN)?
.unwrap_or(DEFAULT_MIN_WORD_LEN_TWO_TYPOS))
}
@ -1284,13 +1359,13 @@ impl Index {
// It is not possible to put a bool in heed with OwnedType, so we put a u8 instead. We
// identify 0 as being false, and anything else as true. The absence of a value is true,
// because by default, we authorize typos.
self.main.put::<_, Str, OwnedType<u8>>(txn, main_key::TWO_TYPOS_WORD_LEN, &val)?;
self.main.remap_types::<Str, U8>().put(txn, main_key::TWO_TYPOS_WORD_LEN, &val)?;
Ok(())
}
/// List the words on which typo are not allowed
pub fn exact_words<'t>(&self, txn: &'t RoTxn) -> Result<Option<fst::Set<Cow<'t, [u8]>>>> {
match self.main.get::<_, Str, ByteSlice>(txn, main_key::EXACT_WORDS)? {
match self.main.remap_types::<Str, Bytes>().get(txn, main_key::EXACT_WORDS)? {
Some(bytes) => Ok(Some(fst::Set::new(bytes)?.map_data(Cow::Borrowed)?)),
None => Ok(None),
}
@ -1301,7 +1376,7 @@ impl Index {
txn: &mut RwTxn,
words: &fst::Set<A>,
) -> Result<()> {
self.main.put::<_, Str, ByteSlice>(
self.main.remap_types::<Str, Bytes>().put(
txn,
main_key::EXACT_WORDS,
words.as_fst().as_bytes(),
@ -1313,7 +1388,8 @@ impl Index {
pub fn exact_attributes<'t>(&self, txn: &'t RoTxn) -> Result<Vec<&'t str>> {
Ok(self
.main
.get::<_, Str, SerdeBincode<Vec<&str>>>(txn, main_key::EXACT_ATTRIBUTES)?
.remap_types::<Str, SerdeBincode<Vec<&str>>>()
.get(txn, main_key::EXACT_ATTRIBUTES)?
.unwrap_or_default())
}
@ -1326,34 +1402,36 @@ impl Index {
/// Writes the exact attributes to the database.
pub(crate) fn put_exact_attributes(&self, txn: &mut RwTxn, attrs: &[&str]) -> Result<()> {
self.main.put::<_, Str, SerdeBincode<&[&str]>>(txn, main_key::EXACT_ATTRIBUTES, &attrs)?;
self.main.remap_types::<Str, SerdeBincode<&[&str]>>().put(
txn,
main_key::EXACT_ATTRIBUTES,
&attrs,
)?;
Ok(())
}
/// Clears the exact attributes from the store.
pub(crate) fn delete_exact_attributes(&self, txn: &mut RwTxn) -> heed::Result<bool> {
self.main.delete::<_, Str>(txn, main_key::EXACT_ATTRIBUTES)
self.main.remap_key_type::<Str>().delete(txn, main_key::EXACT_ATTRIBUTES)
}
pub fn max_values_per_facet(&self, txn: &RoTxn) -> heed::Result<Option<usize>> {
self.main.get::<_, Str, OwnedType<usize>>(txn, main_key::MAX_VALUES_PER_FACET)
pub fn max_values_per_facet(&self, txn: &RoTxn) -> heed::Result<Option<u64>> {
self.main.remap_types::<Str, BEU64>().get(txn, main_key::MAX_VALUES_PER_FACET)
}
pub(crate) fn put_max_values_per_facet(&self, txn: &mut RwTxn, val: usize) -> heed::Result<()> {
self.main.put::<_, Str, OwnedType<usize>>(txn, main_key::MAX_VALUES_PER_FACET, &val)
pub(crate) fn put_max_values_per_facet(&self, txn: &mut RwTxn, val: u64) -> heed::Result<()> {
self.main.remap_types::<Str, BEU64>().put(txn, main_key::MAX_VALUES_PER_FACET, &val)
}
pub(crate) fn delete_max_values_per_facet(&self, txn: &mut RwTxn) -> heed::Result<bool> {
self.main.delete::<_, Str>(txn, main_key::MAX_VALUES_PER_FACET)
self.main.remap_key_type::<Str>().delete(txn, main_key::MAX_VALUES_PER_FACET)
}
pub fn sort_facet_values_by(&self, txn: &RoTxn) -> heed::Result<HashMap<String, OrderBy>> {
let mut orders = self
.main
.get::<_, Str, SerdeJson<HashMap<String, OrderBy>>>(
txn,
main_key::SORT_FACET_VALUES_BY,
)?
.remap_types::<Str, SerdeJson<HashMap<String, OrderBy>>>()
.get(txn, main_key::SORT_FACET_VALUES_BY)?
.unwrap_or_default();
// Insert the default ordering if it is not already overwritten by the user.
orders.entry("*".to_string()).or_insert(OrderBy::Lexicographic);
@ -1365,27 +1443,27 @@ impl Index {
txn: &mut RwTxn,
val: &HashMap<String, OrderBy>,
) -> heed::Result<()> {
self.main.put::<_, Str, SerdeJson<_>>(txn, main_key::SORT_FACET_VALUES_BY, &val)
self.main.remap_types::<Str, SerdeJson<_>>().put(txn, main_key::SORT_FACET_VALUES_BY, &val)
}
pub(crate) fn delete_sort_facet_values_by(&self, txn: &mut RwTxn) -> heed::Result<bool> {
self.main.delete::<_, Str>(txn, main_key::SORT_FACET_VALUES_BY)
self.main.remap_key_type::<Str>().delete(txn, main_key::SORT_FACET_VALUES_BY)
}
pub fn pagination_max_total_hits(&self, txn: &RoTxn) -> heed::Result<Option<usize>> {
self.main.get::<_, Str, OwnedType<usize>>(txn, main_key::PAGINATION_MAX_TOTAL_HITS)
pub fn pagination_max_total_hits(&self, txn: &RoTxn) -> heed::Result<Option<u64>> {
self.main.remap_types::<Str, BEU64>().get(txn, main_key::PAGINATION_MAX_TOTAL_HITS)
}
pub(crate) fn put_pagination_max_total_hits(
&self,
txn: &mut RwTxn,
val: usize,
val: u64,
) -> heed::Result<()> {
self.main.put::<_, Str, OwnedType<usize>>(txn, main_key::PAGINATION_MAX_TOTAL_HITS, &val)
self.main.remap_types::<Str, BEU64>().put(txn, main_key::PAGINATION_MAX_TOTAL_HITS, &val)
}
pub(crate) fn delete_pagination_max_total_hits(&self, txn: &mut RwTxn) -> heed::Result<bool> {
self.main.delete::<_, Str>(txn, main_key::PAGINATION_MAX_TOTAL_HITS)
self.main.remap_key_type::<Str>().delete(txn, main_key::PAGINATION_MAX_TOTAL_HITS)
}
/* script language docids */
@ -1479,7 +1557,7 @@ pub(crate) mod tests {
}
pub fn add_documents_using_wtxn<'t, R>(
&'t self,
wtxn: &mut RwTxn<'t, '_>,
wtxn: &mut RwTxn<'t>,
documents: DocumentsBatchReader<R>,
) -> Result<(), crate::error::Error>
where
@ -1523,7 +1601,7 @@ pub(crate) mod tests {
}
pub fn update_settings_using_wtxn<'t>(
&'t self,
wtxn: &mut RwTxn<'t, '_>,
wtxn: &mut RwTxn<'t>,
update: impl Fn(&mut Settings),
) -> Result<(), crate::error::Error> {
let mut builder = update::Settings::new(wtxn, &self.inner, &self.indexer_config);
@ -1534,7 +1612,7 @@ pub(crate) mod tests {
pub fn delete_documents_using_wtxn<'t>(
&'t self,
wtxn: &mut RwTxn<'t, '_>,
wtxn: &mut RwTxn<'t>,
external_document_ids: Vec<String>,
) {
let builder = IndexDocuments::new(

6
milli/src/lib.rs
View File

@ -66,9 +66,9 @@ pub use self::search::{
pub type Result<T> = std::result::Result<T, error::Error>;
pub type Attribute = u32;
pub type BEU16 = heed::zerocopy::U16<heed::byteorder::BE>;
pub type BEU32 = heed::zerocopy::U32<heed::byteorder::BE>;
pub type BEU64 = heed::zerocopy::U64<heed::byteorder::BE>;
pub type BEU16 = heed::types::U16<heed::byteorder::BE>;
pub type BEU32 = heed::types::U32<heed::byteorder::BE>;
pub type BEU64 = heed::types::U64<heed::byteorder::BE>;
pub type DocumentId = u32;
pub type FastMap4<K, V> = HashMap<K, V, BuildHasherDefault<FxHasher32>>;
pub type FastMap8<K, V> = HashMap<K, V, BuildHasherDefault<FxHasher64>>;

16
milli/src/search/facet/facet_distribution.rs
View File

@ -2,7 +2,7 @@ use std::collections::{BTreeMap, HashMap, HashSet};
use std::ops::ControlFlow;
use std::{fmt, mem};
use heed::types::ByteSlice;
use heed::types::Bytes;
use heed::BytesDecode;
use indexmap::IndexMap;
use roaring::RoaringBitmap;
@ -13,7 +13,7 @@ use crate::facet::FacetType;
use crate::heed_codec::facet::{
FacetGroupKeyCodec, FieldDocIdFacetF64Codec, FieldDocIdFacetStringCodec, OrderedF64Codec,
};
use crate::heed_codec::{ByteSliceRefCodec, StrRefCodec};
use crate::heed_codec::{BytesRefCodec, StrRefCodec};
use crate::search::facet::facet_distribution_iter::{
count_iterate_over_facet_distribution, lexicographically_iterate_over_facet_distribution,
};
@ -105,7 +105,7 @@ impl<'a> FacetDistribution<'a> {
key_buffer.truncate(mem::size_of::<FieldId>());
key_buffer.extend_from_slice(&docid.to_be_bytes());
let iter = db
.remap_key_type::<ByteSlice>()
.remap_key_type::<Bytes>()
.prefix_iter(self.rtxn, &key_buffer)?
.remap_key_type::<FieldDocIdFacetF64Codec>();
@ -129,7 +129,7 @@ impl<'a> FacetDistribution<'a> {
key_buffer.truncate(mem::size_of::<FieldId>());
key_buffer.extend_from_slice(&docid.to_be_bytes());
let iter = db
.remap_key_type::<ByteSlice>()
.remap_key_type::<Bytes>()
.prefix_iter(self.rtxn, &key_buffer)?
.remap_key_type::<FieldDocIdFacetStringCodec>();
@ -172,9 +172,7 @@ impl<'a> FacetDistribution<'a> {
search_function(
self.rtxn,
self.index
.facet_id_f64_docids
.remap_key_type::<FacetGroupKeyCodec<ByteSliceRefCodec>>(),
self.index.facet_id_f64_docids.remap_key_type::<FacetGroupKeyCodec<BytesRefCodec>>(),
field_id,
candidates,
|facet_key, nbr_docids, _| {
@ -203,9 +201,7 @@ impl<'a> FacetDistribution<'a> {
search_function(
self.rtxn,
self.index
.facet_id_string_docids
.remap_key_type::<FacetGroupKeyCodec<ByteSliceRefCodec>>(),
self.index.facet_id_string_docids.remap_key_type::<FacetGroupKeyCodec<BytesRefCodec>>(),
field_id,
candidates,
|facet_key, nbr_docids, any_docid| {

16
milli/src/search/facet/facet_distribution_iter.rs
View File

@ -7,7 +7,7 @@ use roaring::RoaringBitmap;
use super::{get_first_facet_value, get_highest_level};
use crate::heed_codec::facet::{FacetGroupKey, FacetGroupKeyCodec, FacetGroupValueCodec};
use crate::heed_codec::ByteSliceRefCodec;
use crate::heed_codec::BytesRefCodec;
use crate::DocumentId;
/// Call the given closure on the facet distribution of the candidate documents.
@ -23,7 +23,7 @@ use crate::DocumentId;
/// keep iterating over the different facet values or stop.
pub fn lexicographically_iterate_over_facet_distribution<'t, CB>(
rtxn: &'t heed::RoTxn<'t>,
db: heed::Database<FacetGroupKeyCodec<ByteSliceRefCodec>, FacetGroupValueCodec>,
db: heed::Database<FacetGroupKeyCodec<BytesRefCodec>, FacetGroupValueCodec>,
field_id: u16,
candidates: &RoaringBitmap,
callback: CB,
@ -34,11 +34,11 @@ where
let mut fd = LexicographicFacetDistribution { rtxn, db, field_id, callback };
let highest_level = get_highest_level(
rtxn,
db.remap_key_type::<FacetGroupKeyCodec<ByteSliceRefCodec>>(),
db.remap_key_type::<FacetGroupKeyCodec<BytesRefCodec>>(),
field_id,
)?;
if let Some(first_bound) = get_first_facet_value::<ByteSliceRefCodec>(rtxn, db, field_id)? {
if let Some(first_bound) = get_first_facet_value::<BytesRefCodec>(rtxn, db, field_id)? {
fd.iterate(candidates, highest_level, first_bound, usize::MAX)?;
Ok(())
} else {
@ -48,7 +48,7 @@ where
pub fn count_iterate_over_facet_distribution<'t, CB>(
rtxn: &'t heed::RoTxn<'t>,
db: heed::Database<FacetGroupKeyCodec<ByteSliceRefCodec>, FacetGroupValueCodec>,
db: heed::Database<FacetGroupKeyCodec<BytesRefCodec>, FacetGroupValueCodec>,
field_id: u16,
candidates: &RoaringBitmap,
mut callback: CB,
@ -77,11 +77,11 @@ where
let mut heap = BinaryHeap::new();
let highest_level = get_highest_level(
rtxn,
db.remap_key_type::<FacetGroupKeyCodec<ByteSliceRefCodec>>(),
db.remap_key_type::<FacetGroupKeyCodec<BytesRefCodec>>(),
field_id,
)?;
if let Some(first_bound) = get_first_facet_value::<ByteSliceRefCodec>(rtxn, db, field_id)? {
if let Some(first_bound) = get_first_facet_value::<BytesRefCodec>(rtxn, db, field_id)? {
// We first fill the heap with values from the highest level
let starting_key =
FacetGroupKey { field_id, level: highest_level, left_bound: first_bound };
@ -146,7 +146,7 @@ where
CB: FnMut(&'t [u8], u64, DocumentId) -> Result<ControlFlow<()>>,
{
rtxn: &'t heed::RoTxn<'t>,
db: heed::Database<FacetGroupKeyCodec<ByteSliceRefCodec>, FacetGroupValueCodec>,
db: heed::Database<FacetGroupKeyCodec<BytesRefCodec>, FacetGroupValueCodec>,
field_id: u16,
callback: CB,
}

23
milli/src/search/facet/facet_range_search.rs
View File

@ -5,7 +5,7 @@ use roaring::RoaringBitmap;
use super::{get_first_facet_value, get_highest_level, get_last_facet_value};
use crate::heed_codec::facet::{FacetGroupKey, FacetGroupKeyCodec, FacetGroupValueCodec};
use crate::heed_codec::ByteSliceRefCodec;
use crate::heed_codec::BytesRefCodec;
use crate::Result;
/// Find all the document ids for which the given field contains a value contained within
@ -25,11 +25,11 @@ where
let inner;
let left = match left {
Bound::Included(left) => {
inner = BoundCodec::bytes_encode(left).ok_or(heed::Error::Encoding)?;
inner = BoundCodec::bytes_encode(left).map_err(heed::Error::Encoding)?;
Bound::Included(inner.as_ref())
}
Bound::Excluded(left) => {
inner = BoundCodec::bytes_encode(left).ok_or(heed::Error::Encoding)?;
inner = BoundCodec::bytes_encode(left).map_err(heed::Error::Encoding)?;
Bound::Excluded(inner.as_ref())
}
Bound::Unbounded => Bound::Unbounded,
@ -37,25 +37,22 @@ where
let inner;
let right = match right {
Bound::Included(right) => {
inner = BoundCodec::bytes_encode(right).ok_or(heed::Error::Encoding)?;
inner = BoundCodec::bytes_encode(right).map_err(heed::Error::Encoding)?;
Bound::Included(inner.as_ref())
}
Bound::Excluded(right) => {
inner = BoundCodec::bytes_encode(right).ok_or(heed::Error::Encoding)?;
inner = BoundCodec::bytes_encode(right).map_err(heed::Error::Encoding)?;
Bound::Excluded(inner.as_ref())
}
Bound::Unbounded => Bound::Unbounded,
};
let db = db.remap_key_type::<FacetGroupKeyCodec<ByteSliceRefCodec>>();
let db = db.remap_key_type::<FacetGroupKeyCodec<BytesRefCodec>>();
let mut f = FacetRangeSearch { rtxn, db, field_id, left, right, docids };
let highest_level = get_highest_level(rtxn, db, field_id)?;
if let Some(starting_left_bound) =
get_first_facet_value::<ByteSliceRefCodec>(rtxn, db, field_id)?
{
let rightmost_bound = Bound::Included(
get_last_facet_value::<ByteSliceRefCodec>(rtxn, db, field_id)?.unwrap(),
); // will not fail because get_first_facet_value succeeded
if let Some(starting_left_bound) = get_first_facet_value::<BytesRefCodec>(rtxn, db, field_id)? {
let rightmost_bound =
Bound::Included(get_last_facet_value::<BytesRefCodec>(rtxn, db, field_id)?.unwrap()); // will not fail because get_first_facet_value succeeded
let group_size = usize::MAX;
f.run(highest_level, starting_left_bound, rightmost_bound, group_size)?;
Ok(())
@ -67,7 +64,7 @@ where
/// Fetch the document ids that have a facet with a value between the two given bounds
struct FacetRangeSearch<'t, 'b, 'bitmap> {
rtxn: &'t heed::RoTxn<'t>,
db: heed::Database<FacetGroupKeyCodec<ByteSliceRefCodec>, FacetGroupValueCodec>,
db: heed::Database<FacetGroupKeyCodec<BytesRefCodec>, FacetGroupValueCodec>,
field_id: u16,
left: Bound<&'b [u8]>,
right: Bound<&'b [u8]>,

12
milli/src/search/facet/facet_sort_ascending.rs
View File

@ -5,7 +5,7 @@ use super::{get_first_facet_value, get_highest_level};
use crate::heed_codec::facet::{
FacetGroupKey, FacetGroupKeyCodec, FacetGroupValue, FacetGroupValueCodec,
};
use crate::heed_codec::ByteSliceRefCodec;
use crate::heed_codec::BytesRefCodec;
/// Return an iterator which iterates over the given candidate documents in
/// ascending order of their facet value for the given field id.
@ -31,12 +31,12 @@ use crate::heed_codec::ByteSliceRefCodec;
/// Note that once a document id is returned by the iterator, it is never returned again.
pub fn ascending_facet_sort<'t>(
rtxn: &'t heed::RoTxn<'t>,
db: heed::Database<FacetGroupKeyCodec<ByteSliceRefCodec>, FacetGroupValueCodec>,
db: heed::Database<FacetGroupKeyCodec<BytesRefCodec>, FacetGroupValueCodec>,
field_id: u16,
candidates: RoaringBitmap,
) -> Result<impl Iterator<Item = Result<(RoaringBitmap, &'t [u8])>> + 't> {
let highest_level = get_highest_level(rtxn, db, field_id)?;
if let Some(first_bound) = get_first_facet_value::<ByteSliceRefCodec>(rtxn, db, field_id)? {
if let Some(first_bound) = get_first_facet_value::<BytesRefCodec>(rtxn, db, field_id)? {
let first_key = FacetGroupKey { field_id, level: highest_level, left_bound: first_bound };
let iter = db.range(rtxn, &(first_key..)).unwrap().take(usize::MAX);
@ -53,14 +53,12 @@ pub fn ascending_facet_sort<'t>(
struct AscendingFacetSort<'t, 'e> {
rtxn: &'t heed::RoTxn<'e>,
db: heed::Database<FacetGroupKeyCodec<ByteSliceRefCodec>, FacetGroupValueCodec>,
db: heed::Database<FacetGroupKeyCodec<BytesRefCodec>, FacetGroupValueCodec>,
field_id: u16,
#[allow(clippy::type_complexity)]
stack: Vec<(
RoaringBitmap,
std::iter::Take<
heed::RoRange<'t, FacetGroupKeyCodec<ByteSliceRefCodec>, FacetGroupValueCodec>,
>,
std::iter::Take<heed::RoRange<'t, FacetGroupKeyCodec<BytesRefCodec>, FacetGroupValueCodec>>,
)>,
}

20
milli/src/search/facet/facet_sort_descending.rs
View File

@ -7,21 +7,21 @@ use super::{get_first_facet_value, get_highest_level, get_last_facet_value};
use crate::heed_codec::facet::{
FacetGroupKey, FacetGroupKeyCodec, FacetGroupValue, FacetGroupValueCodec,
};
use crate::heed_codec::ByteSliceRefCodec;
use crate::heed_codec::BytesRefCodec;
/// See documentationg for [`ascending_facet_sort`](super::ascending_facet_sort).
///
/// This function does the same thing, but in the opposite order.
pub fn descending_facet_sort<'t>(
rtxn: &'t heed::RoTxn<'t>,
db: heed::Database<FacetGroupKeyCodec<ByteSliceRefCodec>, FacetGroupValueCodec>,
db: heed::Database<FacetGroupKeyCodec<BytesRefCodec>, FacetGroupValueCodec>,
field_id: u16,
candidates: RoaringBitmap,
) -> Result<impl Iterator<Item = Result<(RoaringBitmap, &'t [u8])>> + 't> {
let highest_level = get_highest_level(rtxn, db, field_id)?;
if let Some(first_bound) = get_first_facet_value::<ByteSliceRefCodec>(rtxn, db, field_id)? {
if let Some(first_bound) = get_first_facet_value::<BytesRefCodec>(rtxn, db, field_id)? {
let first_key = FacetGroupKey { field_id, level: highest_level, left_bound: first_bound };
let last_bound = get_last_facet_value::<ByteSliceRefCodec>(rtxn, db, field_id)?.unwrap();
let last_bound = get_last_facet_value::<BytesRefCodec>(rtxn, db, field_id)?.unwrap();
let last_key = FacetGroupKey { field_id, level: highest_level, left_bound: last_bound };
let iter = db.rev_range(rtxn, &(first_key..=last_key))?.take(usize::MAX);
Ok(itertools::Either::Left(DescendingFacetSort {
@ -37,13 +37,13 @@ pub fn descending_facet_sort<'t>(
struct DescendingFacetSort<'t> {
rtxn: &'t heed::RoTxn<'t>,
db: heed::Database<FacetGroupKeyCodec<ByteSliceRefCodec>, FacetGroupValueCodec>,
db: heed::Database<FacetGroupKeyCodec<BytesRefCodec>, FacetGroupValueCodec>,
field_id: u16,
#[allow(clippy::type_complexity)]
stack: Vec<(
RoaringBitmap,
std::iter::Take<
heed::RoRevRange<'t, FacetGroupKeyCodec<ByteSliceRefCodec>, FacetGroupValueCodec>,
heed::RoRevRange<'t, FacetGroupKeyCodec<BytesRefCodec>, FacetGroupValueCodec>,
>,
Bound<&'t [u8]>,
)>,
@ -100,7 +100,7 @@ impl<'t> Iterator for DescendingFacetSort<'t> {
*right_bound = Bound::Excluded(left_bound);
let iter = match self
.db
.remap_key_type::<FacetGroupKeyCodec<ByteSliceRefCodec>>()
.remap_key_type::<FacetGroupKeyCodec<BytesRefCodec>>()
.rev_range(self.rtxn, &(Bound::Included(starting_key_below), end_key_kelow))
{
Ok(iter) => iter,
@ -123,7 +123,7 @@ mod tests {
use roaring::RoaringBitmap;
use crate::heed_codec::facet::FacetGroupKeyCodec;
use crate::heed_codec::ByteSliceRefCodec;
use crate::heed_codec::BytesRefCodec;
use crate::milli_snap;
use crate::search::facet::facet_sort_descending::descending_facet_sort;
use crate::search::facet::tests::{
@ -144,7 +144,7 @@ mod tests {
let txn = index.env.read_txn().unwrap();
let candidates = (200..=300).collect::<RoaringBitmap>();
let mut results = String::new();
let db = index.content.remap_key_type::<FacetGroupKeyCodec<ByteSliceRefCodec>>();
let db = index.content.remap_key_type::<FacetGroupKeyCodec<BytesRefCodec>>();
let iter = descending_facet_sort(&txn, db, 0, candidates).unwrap();
for el in iter {
let (docids, _) = el.unwrap();
@ -167,7 +167,7 @@ mod tests {
let txn = index.env.read_txn().unwrap();
let candidates = (200..=300).collect::<RoaringBitmap>();
let mut results = String::new();
let db = index.content.remap_key_type::<FacetGroupKeyCodec<ByteSliceRefCodec>>();
let db = index.content.remap_key_type::<FacetGroupKeyCodec<BytesRefCodec>>();
let iter = descending_facet_sort(&txn, db, 0, candidates.clone()).unwrap();
for el in iter {
let (docids, _) = el.unwrap();

40
milli/src/search/facet/mod.rs
View File

@ -1,13 +1,13 @@
pub use facet_sort_ascending::ascending_facet_sort;
pub use facet_sort_descending::descending_facet_sort;
use heed::types::{ByteSlice, DecodeIgnore};
use heed::types::{Bytes, DecodeIgnore};
use heed::{BytesDecode, RoTxn};
use roaring::RoaringBitmap;
pub use self::facet_distribution::{FacetDistribution, OrderBy, DEFAULT_VALUES_PER_FACET};
pub use self::filter::{BadGeoError, Filter};
use crate::heed_codec::facet::{FacetGroupKeyCodec, FacetGroupValueCodec, OrderedF64Codec};
use crate::heed_codec::ByteSliceRefCodec;
use crate::heed_codec::BytesRefCodec;
use crate::{Index, Result};
mod facet_distribution;
mod facet_distribution_iter;
@ -22,8 +22,10 @@ fn facet_extreme_value<'t>(
let extreme_value =
if let Some(extreme_value) = extreme_it.next() { extreme_value } else { return Ok(None) };
let (_, extreme_value) = extreme_value?;
Ok(OrderedF64Codec::bytes_decode(extreme_value))
OrderedF64Codec::bytes_decode(extreme_value)
.map(Some)
.map_err(heed::Error::Decoding)
.map_err(Into::into)
}
pub fn facet_min_value<'t>(
@ -32,7 +34,7 @@ pub fn facet_min_value<'t>(
field_id: u16,
candidates: RoaringBitmap,
) -> Result<Option<f64>> {
let db = index.facet_id_f64_docids.remap_key_type::<FacetGroupKeyCodec<ByteSliceRefCodec>>();
let db = index.facet_id_f64_docids.remap_key_type::<FacetGroupKeyCodec<BytesRefCodec>>();
let it = ascending_facet_sort(rtxn, db, field_id, candidates)?;
facet_extreme_value(it)
}
@ -43,7 +45,7 @@ pub fn facet_max_value<'t>(
field_id: u16,
candidates: RoaringBitmap,
) -> Result<Option<f64>> {
let db = index.facet_id_f64_docids.remap_key_type::<FacetGroupKeyCodec<ByteSliceRefCodec>>();
let db = index.facet_id_f64_docids.remap_key_type::<FacetGroupKeyCodec<BytesRefCodec>>();
let it = descending_facet_sort(rtxn, db, field_id, candidates)?;
facet_extreme_value(it)
}
@ -51,7 +53,7 @@ pub fn facet_max_value<'t>(
/// Get the first facet value in the facet database
pub(crate) fn get_first_facet_value<'t, BoundCodec>(
txn: &'t RoTxn,
db: heed::Database<FacetGroupKeyCodec<ByteSliceRefCodec>, FacetGroupValueCodec>,
db: heed::Database<FacetGroupKeyCodec<BytesRefCodec>, FacetGroupValueCodec>,
field_id: u16,
) -> heed::Result<Option<BoundCodec::DItem>>
where
@ -60,13 +62,12 @@ where
let mut level0prefix = vec![];
level0prefix.extend_from_slice(&field_id.to_be_bytes());
level0prefix.push(0);
let mut level0_iter_forward = db
.as_polymorph()
.prefix_iter::<_, ByteSlice, DecodeIgnore>(txn, level0prefix.as_slice())?;
let mut level0_iter_forward =
db.remap_types::<Bytes, DecodeIgnore>().prefix_iter(txn, level0prefix.as_slice())?;
if let Some(first) = level0_iter_forward.next() {
let (first_key, _) = first?;
let first_key = FacetGroupKeyCodec::<BoundCodec>::bytes_decode(first_key)
.ok_or(heed::Error::Encoding)?;
.map_err(heed::Error::Decoding)?;
Ok(Some(first_key.left_bound))
} else {
Ok(None)
@ -76,7 +77,7 @@ where
/// Get the last facet value in the facet database
pub(crate) fn get_last_facet_value<'t, BoundCodec>(
txn: &'t RoTxn,
db: heed::Database<FacetGroupKeyCodec<ByteSliceRefCodec>, FacetGroupValueCodec>,
db: heed::Database<FacetGroupKeyCodec<BytesRefCodec>, FacetGroupValueCodec>,
field_id: u16,
) -> heed::Result<Option<BoundCodec::DItem>>
where
@ -85,13 +86,12 @@ where
let mut level0prefix = vec![];
level0prefix.extend_from_slice(&field_id.to_be_bytes());
level0prefix.push(0);
let mut level0_iter_backward = db
.as_polymorph()
.rev_prefix_iter::<_, ByteSlice, DecodeIgnore>(txn, level0prefix.as_slice())?;
let mut level0_iter_backward =
db.remap_types::<Bytes, DecodeIgnore>().rev_prefix_iter(txn, level0prefix.as_slice())?;
if let Some(last) = level0_iter_backward.next() {
let (last_key, _) = last?;
let last_key = FacetGroupKeyCodec::<BoundCodec>::bytes_decode(last_key)
.ok_or(heed::Error::Encoding)?;
.map_err(heed::Error::Decoding)?;
Ok(Some(last_key.left_bound))
} else {
Ok(None)
@ -101,17 +101,17 @@ where
/// Get the height of the highest level in the facet database
pub(crate) fn get_highest_level<'t>(
txn: &'t RoTxn<'t>,
db: heed::Database<FacetGroupKeyCodec<ByteSliceRefCodec>, FacetGroupValueCodec>,
db: heed::Database<FacetGroupKeyCodec<BytesRefCodec>, FacetGroupValueCodec>,
field_id: u16,
) -> heed::Result<u8> {
let field_id_prefix = &field_id.to_be_bytes();
Ok(db
.as_polymorph()
.rev_prefix_iter::<_, ByteSlice, DecodeIgnore>(txn, field_id_prefix)?
.remap_types::<Bytes, DecodeIgnore>()
.rev_prefix_iter(txn, field_id_prefix)?
.next()
.map(|el| {
let (key, _) = el.unwrap();
let key = FacetGroupKeyCodec::<ByteSliceRefCodec>::bytes_decode(key).unwrap();
let key = FacetGroupKeyCodec::<BytesRefCodec>::bytes_decode(key).unwrap();
key.level
})
.unwrap_or(0))

5
milli/src/search/mod.rs
View File

@ -17,8 +17,7 @@ use crate::error::UserError;
use crate::heed_codec::facet::{FacetGroupKey, FacetGroupValue};
use crate::score_details::{ScoreDetails, ScoringStrategy};
use crate::{
execute_search, AscDesc, DefaultSearchLogger, DocumentId, FieldId, Index, Result,
SearchContext, BEU16,
execute_search, AscDesc, DefaultSearchLogger, DocumentId, FieldId, Index, Result, SearchContext,
};
// Building these factories is not free.
@ -299,7 +298,7 @@ impl<'a> SearchForFacetValues<'a> {
None => return Ok(Vec::new()),
};
let fst = match self.search_query.index.facet_id_string_fst.get(rtxn, &BEU16::new(fid))? {
let fst = match self.search_query.index.facet_id_string_fst.get(rtxn, &fid)? {
Some(fst) => fst,
None => return Ok(vec![]),
};

68
milli/src/search/new/db_cache.rs
View File

@ -3,7 +3,7 @@ use std::collections::hash_map::Entry;
use std::hash::Hash;
use fxhash::FxHashMap;
use heed::types::ByteSlice;
use heed::types::Bytes;
use heed::{BytesEncode, Database, RoTxn};
use roaring::RoaringBitmap;
@ -50,7 +50,7 @@ impl<'ctx> DatabaseCache<'ctx> {
cache_key: K1,
db_key: &'v KC::EItem,
cache: &mut FxHashMap<K1, Option<Cow<'ctx, [u8]>>>,
db: Database<KC, ByteSlice>,
db: Database<KC, Bytes>,
) -> Result<Option<DC::DItem>>
where
K1: Copy + Eq + Hash,
@ -63,12 +63,14 @@ impl<'ctx> DatabaseCache<'ctx> {
}
match cache.get(&cache_key).unwrap() {
Some(Cow::Borrowed(bytes)) => {
DC::bytes_decode_owned(bytes).ok_or(heed::Error::Decoding.into()).map(Some)
}
Some(Cow::Owned(bytes)) => {
DC::bytes_decode_owned(bytes).ok_or(heed::Error::Decoding.into()).map(Some)
}
Some(Cow::Borrowed(bytes)) => DC::bytes_decode_owned(bytes)
.map(Some)
.map_err(heed::Error::Decoding)
.map_err(Into::into),
Some(Cow::Owned(bytes)) => DC::bytes_decode_owned(bytes)
.map(Some)
.map_err(heed::Error::Decoding)
.map_err(Into::into),
None => Ok(None),
}
}
@ -78,7 +80,7 @@ impl<'ctx> DatabaseCache<'ctx> {
cache_key: K1,
db_keys: &'v [KC::EItem],
cache: &mut FxHashMap<K1, Option<Cow<'ctx, [u8]>>>,
db: Database<KC, ByteSlice>,
db: Database<KC, Bytes>,
merger: MergeFn,
) -> Result<Option<DC::DItem>>
where
@ -110,12 +112,14 @@ impl<'ctx> DatabaseCache<'ctx> {
}
match cache.get(&cache_key).unwrap() {
Some(Cow::Borrowed(bytes)) => {
DC::bytes_decode_owned(bytes).ok_or(heed::Error::Decoding.into()).map(Some)
}
Some(Cow::Owned(bytes)) => {
DC::bytes_decode_owned(bytes).ok_or(heed::Error::Decoding.into()).map(Some)
}
Some(Cow::Borrowed(bytes)) => DC::bytes_decode_owned(bytes)
.map(Some)
.map_err(heed::Error::Decoding)
.map_err(Into::into),
Some(Cow::Owned(bytes)) => DC::bytes_decode_owned(bytes)
.map(Some)
.map_err(heed::Error::Decoding)
.map_err(Into::into),
None => Ok(None),
}
}
@ -164,7 +168,7 @@ impl<'ctx> SearchContext<'ctx> {
word,
&keys[..],
&mut self.db_cache.word_docids,
self.index.word_fid_docids.remap_data_type::<ByteSlice>(),
self.index.word_fid_docids.remap_data_type::<Bytes>(),
merge_cbo_roaring_bitmaps,
)
}
@ -173,7 +177,7 @@ impl<'ctx> SearchContext<'ctx> {
word,
self.word_interner.get(word).as_str(),
&mut self.db_cache.word_docids,
self.index.word_docids.remap_data_type::<ByteSlice>(),
self.index.word_docids.remap_data_type::<Bytes>(),
),
}
}
@ -187,7 +191,7 @@ impl<'ctx> SearchContext<'ctx> {
word,
self.word_interner.get(word).as_str(),
&mut self.db_cache.exact_word_docids,
self.index.exact_word_docids.remap_data_type::<ByteSlice>(),
self.index.exact_word_docids.remap_data_type::<Bytes>(),
)
}
@ -226,7 +230,7 @@ impl<'ctx> SearchContext<'ctx> {
prefix,
&keys[..],
&mut self.db_cache.word_prefix_docids,
self.index.word_prefix_fid_docids.remap_data_type::<ByteSlice>(),
self.index.word_prefix_fid_docids.remap_data_type::<Bytes>(),
merge_cbo_roaring_bitmaps,
)
}
@ -235,7 +239,7 @@ impl<'ctx> SearchContext<'ctx> {
prefix,
self.word_interner.get(prefix).as_str(),
&mut self.db_cache.word_prefix_docids,
self.index.word_prefix_docids.remap_data_type::<ByteSlice>(),
self.index.word_prefix_docids.remap_data_type::<Bytes>(),
),
}
}
@ -249,7 +253,7 @@ impl<'ctx> SearchContext<'ctx> {
prefix,
self.word_interner.get(prefix).as_str(),
&mut self.db_cache.exact_word_prefix_docids,
self.index.exact_word_prefix_docids.remap_data_type::<ByteSlice>(),
self.index.exact_word_prefix_docids.remap_data_type::<Bytes>(),
)
}
@ -268,7 +272,7 @@ impl<'ctx> SearchContext<'ctx> {
self.word_interner.get(word2).as_str(),
),
&mut self.db_cache.word_pair_proximity_docids,
self.index.word_pair_proximity_docids.remap_data_type::<ByteSlice>(),
self.index.word_pair_proximity_docids.remap_data_type::<Bytes>(),
)
}
@ -287,7 +291,7 @@ impl<'ctx> SearchContext<'ctx> {
self.word_interner.get(word2).as_str(),
),
&mut self.db_cache.word_pair_proximity_docids,
self.index.word_pair_proximity_docids.remap_data_type::<ByteSlice>(),
self.index.word_pair_proximity_docids.remap_data_type::<Bytes>(),
)
}
@ -316,7 +320,7 @@ impl<'ctx> SearchContext<'ctx> {
let remap_key_type = self
.index
.word_pair_proximity_docids
.remap_key_type::<ByteSlice>()
.remap_key_type::<Bytes>()
.prefix_iter(self.txn, &key)?;
for result in remap_key_type {
let (_, docids) = result?;
@ -355,7 +359,7 @@ impl<'ctx> SearchContext<'ctx> {
(word, fid),
&(self.word_interner.get(word).as_str(), fid),
&mut self.db_cache.word_fid_docids,
self.index.word_fid_docids.remap_data_type::<ByteSlice>(),
self.index.word_fid_docids.remap_data_type::<Bytes>(),
)
}
@ -374,7 +378,7 @@ impl<'ctx> SearchContext<'ctx> {
(word_prefix, fid),
&(self.word_interner.get(word_prefix).as_str(), fid),
&mut self.db_cache.word_prefix_fid_docids,
self.index.word_prefix_fid_docids.remap_data_type::<ByteSlice>(),
self.index.word_prefix_fid_docids.remap_data_type::<Bytes>(),
)
}
@ -388,7 +392,7 @@ impl<'ctx> SearchContext<'ctx> {
let remap_key_type = self
.index
.word_fid_docids
.remap_types::<ByteSlice, ByteSlice>()
.remap_types::<Bytes, Bytes>()
.prefix_iter(self.txn, &key)?
.remap_key_type::<StrBEU16Codec>();
for result in remap_key_type {
@ -414,7 +418,7 @@ impl<'ctx> SearchContext<'ctx> {
let remap_key_type = self
.index
.word_prefix_fid_docids
.remap_types::<ByteSlice, ByteSlice>()
.remap_types::<Bytes, Bytes>()
.prefix_iter(self.txn, &key)?
.remap_key_type::<StrBEU16Codec>();
for result in remap_key_type {
@ -442,7 +446,7 @@ impl<'ctx> SearchContext<'ctx> {
(word, position),
&(self.word_interner.get(word).as_str(), position),
&mut self.db_cache.word_position_docids,
self.index.word_position_docids.remap_data_type::<ByteSlice>(),
self.index.word_position_docids.remap_data_type::<Bytes>(),
)
}
@ -456,7 +460,7 @@ impl<'ctx> SearchContext<'ctx> {
(word_prefix, position),
&(self.word_interner.get(word_prefix).as_str(), position),
&mut self.db_cache.word_prefix_position_docids,
self.index.word_prefix_position_docids.remap_data_type::<ByteSlice>(),
self.index.word_prefix_position_docids.remap_data_type::<Bytes>(),
)
}
@ -470,7 +474,7 @@ impl<'ctx> SearchContext<'ctx> {
let remap_key_type = self
.index
.word_position_docids
.remap_types::<ByteSlice, ByteSlice>()
.remap_types::<Bytes, Bytes>()
.prefix_iter(self.txn, &key)?
.remap_key_type::<StrBEU16Codec>();
for result in remap_key_type {
@ -501,7 +505,7 @@ impl<'ctx> SearchContext<'ctx> {
let remap_key_type = self
.index
.word_prefix_position_docids
.remap_types::<ByteSlice, ByteSlice>()
.remap_types::<Bytes, Bytes>()
.prefix_iter(self.txn, &key)?
.remap_key_type::<StrBEU16Codec>();
for result in remap_key_type {

14
milli/src/search/new/distinct.rs
View File

@ -1,4 +1,4 @@
use heed::types::{ByteSlice, Str, Unit};
use heed::types::{Bytes, Str, Unit};
use heed::{Database, RoPrefix, RoTxn};
use roaring::RoaringBitmap;
@ -8,7 +8,7 @@ const DOCID_SIZE: usize = 4;
use crate::heed_codec::facet::{
FacetGroupKey, FacetGroupKeyCodec, FacetGroupValueCodec, FieldDocIdFacetCodec,
};
use crate::heed_codec::ByteSliceRefCodec;
use crate::heed_codec::BytesRefCodec;
use crate::{Index, Result, SearchContext};
pub struct DistinctOutput {
@ -71,7 +71,7 @@ pub fn distinct_single_docid(
/// Return all the docids containing the given value in the given field
fn facet_value_docids(
database: Database<FacetGroupKeyCodec<ByteSliceRefCodec>, FacetGroupValueCodec>,
database: Database<FacetGroupKeyCodec<BytesRefCodec>, FacetGroupValueCodec>,
txn: &RoTxn,
field_id: u16,
facet_value: &[u8],
@ -87,12 +87,12 @@ fn facet_number_values<'a>(
field_id: u16,
index: &Index,
txn: &'a RoTxn,
) -> Result<RoPrefix<'a, FieldDocIdFacetCodec<ByteSliceRefCodec>, Unit>> {
) -> Result<RoPrefix<'a, FieldDocIdFacetCodec<BytesRefCodec>, Unit>> {
let key = facet_values_prefix_key(field_id, docid);
let iter = index
.field_id_docid_facet_f64s
.remap_key_type::<ByteSlice>()
.remap_key_type::<Bytes>()
.prefix_iter(txn, &key)?
.remap_key_type();
@ -105,12 +105,12 @@ pub fn facet_string_values<'a>(
field_id: u16,
index: &Index,
txn: &'a RoTxn,
) -> Result<RoPrefix<'a, FieldDocIdFacetCodec<ByteSliceRefCodec>, Str>> {
) -> Result<RoPrefix<'a, FieldDocIdFacetCodec<BytesRefCodec>, Str>> {
let key = facet_values_prefix_key(field_id, docid);
let iter = index
.field_id_docid_facet_strings
.remap_key_type::<ByteSlice>()
.remap_key_type::<Bytes>()
.prefix_iter(txn, &key)?
.remap_types();

6
milli/src/search/new/geo_sort.rs
View File

@ -1,7 +1,7 @@
use std::collections::VecDeque;
use std::iter::FromIterator;
use heed::types::{ByteSlice, Unit};
use heed::types::{Bytes, Unit};
use heed::{RoPrefix, RoTxn};
use roaring::RoaringBitmap;
use rstar::RTree;
@ -34,7 +34,7 @@ fn facet_number_values<'a>(
let iter = index
.field_id_docid_facet_f64s
.remap_key_type::<ByteSlice>()
.remap_key_type::<Bytes>()
.prefix_iter(txn, &key)?
.remap_key_type();
@ -163,7 +163,7 @@ impl<Q: RankingRuleQueryTrait> GeoSort<Q> {
// computing the distance between two points is expensive thus we cache the result
documents
.sort_by_cached_key(|(_, p)| distance_between_two_points(&self.point, p) as usize);
self.cached_sorted_docids.extend(documents.into_iter());
self.cached_sorted_docids.extend(documents);
};
Ok(())

4
milli/src/search/new/interner.rs
View File

@ -228,7 +228,7 @@ impl<T> Ord for Interned<T> {
impl<T> PartialOrd for Interned<T> {
fn partial_cmp(&self, other: &Self) -> Option<std::cmp::Ordering> {
self.idx.partial_cmp(&other.idx)
Some(self.cmp(other))
}
}
@ -241,7 +241,7 @@ impl<T> PartialEq for Interned<T> {
}
impl<T> Clone for Interned<T> {
fn clone(&self) -> Self {
Self { idx: self.idx, _phantom: PhantomData }
*self
}
}

14
milli/src/search/new/mod.rs
View File

@ -50,9 +50,7 @@ use crate::distance::NDotProductPoint;
use crate::error::FieldIdMapMissingEntry;
use crate::score_details::{ScoreDetails, ScoringStrategy};
use crate::search::new::distinct::apply_distinct_rule;
use crate::{
AscDesc, DocumentId, Filter, Index, Member, Result, TermsMatchingStrategy, UserError, BEU32,
};
use crate::{AscDesc, DocumentId, Filter, Index, Member, Result, TermsMatchingStrategy, UserError};
/// A structure used throughout the execution of a search query.
pub struct SearchContext<'ctx> {
@ -451,8 +449,8 @@ pub fn execute_search(
let mut docids = Vec::new();
let mut uniq_docids = RoaringBitmap::new();
for instant_distance::Item { distance: _, pid, point: _ } in neighbors {
let index = BEU32::new(pid.into_inner());
let docid = ctx.index.vector_id_docid.get(ctx.txn, &index)?.unwrap().get();
let index = pid.into_inner();
let docid = ctx.index.vector_id_docid.get(ctx.txn, &index)?.unwrap();
if universe.contains(docid) && uniq_docids.insert(docid) {
docids.push(docid);
if docids.len() == (from + length) {
@ -609,7 +607,8 @@ fn check_sort_criteria(ctx: &SearchContext, sort_criteria: Option<&Vec<AscDesc>>
field: field.to_string(),
valid_fields,
hidden_fields,
})?;
}
.into());
}
Member::Geo(_) if !sortable_fields.contains("_geo") => {
let (valid_fields, hidden_fields) =
@ -619,7 +618,8 @@ fn check_sort_criteria(ctx: &SearchContext, sort_criteria: Option<&Vec<AscDesc>>
field: "_geo".to_string(),
valid_fields,
hidden_fields,
})?;
}
.into());
}
_ => (),
}

2
milli/src/search/new/query_term/mod.rs
View File

@ -175,7 +175,7 @@ impl QueryTermSubset {
pub fn use_prefix_db(&self, ctx: &SearchContext) -> Option<Word> {
let original = ctx.term_interner.get(self.original);
let Some(use_prefix_db) = original.zero_typo.use_prefix_db else { return None };
let use_prefix_db = original.zero_typo.use_prefix_db?;
let word = match &self.zero_typo_subset {
NTypoTermSubset::All => Some(use_prefix_db),
NTypoTermSubset::Subset { words, phrases: _ } => {

6
milli/src/search/new/sort.rs
View File

@ -4,7 +4,7 @@ use roaring::RoaringBitmap;
use super::logger::SearchLogger;
use super::{RankingRule, RankingRuleOutput, RankingRuleQueryTrait, SearchContext};
use crate::heed_codec::facet::{FacetGroupKeyCodec, OrderedF64Codec};
use crate::heed_codec::{ByteSliceRefCodec, StrRefCodec};
use crate::heed_codec::{BytesRefCodec, StrRefCodec};
use crate::score_details::{self, ScoreDetails};
use crate::search::facet::{ascending_facet_sort, descending_facet_sort};
use crate::{FieldId, Index, Result};
@ -100,11 +100,11 @@ impl<'ctx, Query: RankingRuleQueryTrait> RankingRule<'ctx, Query> for Sort<'ctx,
let number_db = ctx
.index
.facet_id_f64_docids
.remap_key_type::<FacetGroupKeyCodec<ByteSliceRefCodec>>();
.remap_key_type::<FacetGroupKeyCodec<BytesRefCodec>>();
let string_db = ctx
.index
.facet_id_string_docids
.remap_key_type::<FacetGroupKeyCodec<ByteSliceRefCodec>>();
.remap_key_type::<FacetGroupKeyCodec<BytesRefCodec>>();
let (number_iter, string_iter) = if self.is_ascending {
let number_iter = ascending_facet_sort(

6
milli/src/search/new/tests/attribute_fid.rs
View File

@ -124,8 +124,7 @@ fn test_attribute_fid_simple() {
s.query("the quick brown fox jumps over the lazy dog");
s.scoring_strategy(crate::score_details::ScoringStrategy::Detailed);
let SearchResult { documents_ids, document_scores, .. } = s.execute().unwrap();
let document_ids_scores: Vec<_> =
documents_ids.iter().zip(document_scores.into_iter()).collect();
let document_ids_scores: Vec<_> = documents_ids.iter().zip(document_scores).collect();
insta::assert_snapshot!(format!("{document_ids_scores:#?}"));
}
@ -142,7 +141,6 @@ fn test_attribute_fid_ngrams() {
let SearchResult { documents_ids, document_scores, .. } = s.execute().unwrap();
let document_ids_scores: Vec<_> =
documents_ids.iter().zip(document_scores.into_iter()).collect();
let document_ids_scores: Vec<_> = documents_ids.iter().zip(document_scores).collect();
insta::assert_snapshot!(format!("{document_ids_scores:#?}"));
}

12
milli/src/search/new/tests/attribute_position.rs
View File

@ -141,8 +141,7 @@ fn test_attribute_position_simple() {
let SearchResult { documents_ids, document_scores, .. } = s.execute().unwrap();
let document_ids_scores: Vec<_> =
documents_ids.iter().zip(document_scores.into_iter()).collect();
let document_ids_scores: Vec<_> = documents_ids.iter().zip(document_scores).collect();
insta::assert_snapshot!(format!("{document_ids_scores:#?}"));
}
#[test]
@ -158,8 +157,7 @@ fn test_attribute_position_repeated() {
let SearchResult { documents_ids, document_scores, .. } = s.execute().unwrap();
let document_ids_scores: Vec<_> =
documents_ids.iter().zip(document_scores.into_iter()).collect();
let document_ids_scores: Vec<_> = documents_ids.iter().zip(document_scores).collect();
insta::assert_snapshot!(format!("{document_ids_scores:#?}"));
}
@ -176,8 +174,7 @@ fn test_attribute_position_different_fields() {
let SearchResult { documents_ids, document_scores, .. } = s.execute().unwrap();
let document_ids_scores: Vec<_> =
documents_ids.iter().zip(document_scores.into_iter()).collect();
let document_ids_scores: Vec<_> = documents_ids.iter().zip(document_scores).collect();
insta::assert_snapshot!(format!("{document_ids_scores:#?}"));
}
@ -194,7 +191,6 @@ fn test_attribute_position_ngrams() {
let SearchResult { documents_ids, document_scores, .. } = s.execute().unwrap();
let document_ids_scores: Vec<_> =
documents_ids.iter().zip(document_scores.into_iter()).collect();
let document_ids_scores: Vec<_> = documents_ids.iter().zip(document_scores).collect();
insta::assert_snapshot!(format!("{document_ids_scores:#?}"));
}

42
milli/src/search/new/tests/exactness.rs
View File

@ -478,8 +478,7 @@ fn test_exactness_simple_ordered() {
let SearchResult { documents_ids, document_scores, .. } = s.execute().unwrap();
let document_ids_scores: Vec<_> =
documents_ids.iter().zip(document_scores.into_iter()).collect();
let document_ids_scores: Vec<_> = documents_ids.iter().zip(document_scores).collect();
insta::assert_snapshot!(format!("{document_ids_scores:#?}"));
let texts = collect_field_values(&index, &txn, "text", &documents_ids);
@ -511,8 +510,7 @@ fn test_exactness_simple_reversed() {
let SearchResult { documents_ids, document_scores, .. } = s.execute().unwrap();
let document_ids_scores: Vec<_> =
documents_ids.iter().zip(document_scores.into_iter()).collect();
let document_ids_scores: Vec<_> = documents_ids.iter().zip(document_scores).collect();
insta::assert_snapshot!(format!("{document_ids_scores:#?}"));
let texts = collect_field_values(&index, &txn, "text", &documents_ids);
@ -535,8 +533,7 @@ fn test_exactness_simple_reversed() {
let SearchResult { documents_ids, document_scores, .. } = s.execute().unwrap();
let document_ids_scores: Vec<_> =
documents_ids.iter().zip(document_scores.into_iter()).collect();
let document_ids_scores: Vec<_> = documents_ids.iter().zip(document_scores).collect();
insta::assert_snapshot!(format!("{document_ids_scores:#?}"));
let texts = collect_field_values(&index, &txn, "text", &documents_ids);
@ -566,8 +563,7 @@ fn test_exactness_simple_random() {
let SearchResult { documents_ids, document_scores, .. } = s.execute().unwrap();
let document_ids_scores: Vec<_> =
documents_ids.iter().zip(document_scores.into_iter()).collect();
let document_ids_scores: Vec<_> = documents_ids.iter().zip(document_scores).collect();
insta::assert_snapshot!(format!("{document_ids_scores:#?}"));
let texts = collect_field_values(&index, &txn, "text", &documents_ids);
@ -596,8 +592,7 @@ fn test_exactness_attribute_starts_with_simple() {
let SearchResult { documents_ids, document_scores, .. } = s.execute().unwrap();
let document_ids_scores: Vec<_> =
documents_ids.iter().zip(document_scores.into_iter()).collect();
let document_ids_scores: Vec<_> = documents_ids.iter().zip(document_scores).collect();
insta::assert_snapshot!(format!("{document_ids_scores:#?}"));
let texts = collect_field_values(&index, &txn, "text", &documents_ids);
@ -623,8 +618,7 @@ fn test_exactness_attribute_starts_with_phrase() {
let SearchResult { documents_ids, document_scores, .. } = s.execute().unwrap();
let document_ids_scores: Vec<_> =
documents_ids.iter().zip(document_scores.into_iter()).collect();
let document_ids_scores: Vec<_> = documents_ids.iter().zip(document_scores).collect();
insta::assert_snapshot!(format!("{document_ids_scores:#?}"));
let texts = collect_field_values(&index, &txn, "text", &documents_ids);
@ -644,8 +638,7 @@ fn test_exactness_attribute_starts_with_phrase() {
let SearchResult { documents_ids, document_scores, .. } = s.execute().unwrap();
let document_ids_scores: Vec<_> =
documents_ids.iter().zip(document_scores.into_iter()).collect();
let document_ids_scores: Vec<_> = documents_ids.iter().zip(document_scores).collect();
insta::assert_snapshot!(format!("{document_ids_scores:#?}"));
let texts = collect_field_values(&index, &txn, "text", &documents_ids);
@ -674,8 +667,7 @@ fn test_exactness_all_candidates_with_typo() {
let SearchResult { documents_ids, document_scores, .. } = s.execute().unwrap();
let document_ids_scores: Vec<_> =
documents_ids.iter().zip(document_scores.into_iter()).collect();
let document_ids_scores: Vec<_> = documents_ids.iter().zip(document_scores).collect();
insta::assert_snapshot!(format!("{document_ids_scores:#?}"));
let texts = collect_field_values(&index, &txn, "text", &documents_ids);
@ -711,8 +703,7 @@ fn test_exactness_after_words() {
let SearchResult { documents_ids, document_scores, .. } = s.execute().unwrap();
let document_ids_scores: Vec<_> =
documents_ids.iter().zip(document_scores.into_iter()).collect();
let document_ids_scores: Vec<_> = documents_ids.iter().zip(document_scores).collect();
insta::assert_snapshot!(format!("{document_ids_scores:#?}"));
let texts = collect_field_values(&index, &txn, "text", &documents_ids);
@ -760,8 +751,7 @@ fn test_words_after_exactness() {
let SearchResult { documents_ids, document_scores, .. } = s.execute().unwrap();
let document_ids_scores: Vec<_> =
documents_ids.iter().zip(document_scores.into_iter()).collect();
let document_ids_scores: Vec<_> = documents_ids.iter().zip(document_scores).collect();
insta::assert_snapshot!(format!("{document_ids_scores:#?}"));
insta::assert_snapshot!(format!("{documents_ids:?}"), @"[19, 9, 18, 8, 17, 16, 6, 7, 15, 5, 14, 4, 13, 3, 12, 2, 1, 11]");
let texts = collect_field_values(&index, &txn, "text", &documents_ids);
@ -809,8 +799,7 @@ fn test_proximity_after_exactness() {
let SearchResult { documents_ids, document_scores, .. } = s.execute().unwrap();
let document_ids_scores: Vec<_> =
documents_ids.iter().zip(document_scores.into_iter()).collect();
let document_ids_scores: Vec<_> = documents_ids.iter().zip(document_scores).collect();
insta::assert_snapshot!(format!("{document_ids_scores:#?}"));
insta::assert_snapshot!(format!("{documents_ids:?}"), @"[2, 1, 0, 4, 5, 8, 7, 3, 6]");
@ -847,8 +836,7 @@ fn test_proximity_after_exactness() {
let SearchResult { documents_ids, document_scores, .. } = s.execute().unwrap();
let document_ids_scores: Vec<_> =
documents_ids.iter().zip(document_scores.into_iter()).collect();
let document_ids_scores: Vec<_> = documents_ids.iter().zip(document_scores).collect();
insta::assert_snapshot!(format!("{document_ids_scores:#?}"));
insta::assert_snapshot!(format!("{documents_ids:?}"), @"[0, 1, 2]");
let texts = collect_field_values(&index, &txn, "text", &documents_ids);
@ -881,8 +869,7 @@ fn test_exactness_followed_by_typo_prefer_no_typo_prefix() {
let SearchResult { documents_ids, document_scores, .. } = s.execute().unwrap();
let document_ids_scores: Vec<_> =
documents_ids.iter().zip(document_scores.into_iter()).collect();
let document_ids_scores: Vec<_> = documents_ids.iter().zip(document_scores).collect();
insta::assert_snapshot!(format!("{document_ids_scores:#?}"));
insta::assert_snapshot!(format!("{documents_ids:?}"), @"[2, 1, 0, 4, 3]");
let texts = collect_field_values(&index, &txn, "text", &documents_ids);
@ -917,8 +904,7 @@ fn test_typo_followed_by_exactness() {
let SearchResult { documents_ids, document_scores, .. } = s.execute().unwrap();
let document_ids_scores: Vec<_> =
documents_ids.iter().zip(document_scores.into_iter()).collect();
let document_ids_scores: Vec<_> = documents_ids.iter().zip(document_scores).collect();
insta::assert_snapshot!(format!("{document_ids_scores:#?}"));
insta::assert_snapshot!(format!("{documents_ids:?}"), @"[1, 0, 4, 3]");
let texts = collect_field_values(&index, &txn, "text", &documents_ids);

9
milli/src/update/clear_documents.rs
View File

@ -1,15 +1,16 @@
use heed::RwTxn;
use roaring::RoaringBitmap;
use time::OffsetDateTime;
use crate::{FieldDistribution, Index, Result};
pub struct ClearDocuments<'t, 'u, 'i> {
wtxn: &'t mut heed::RwTxn<'i, 'u>,
pub struct ClearDocuments<'t, 'i> {
wtxn: &'t mut RwTxn<'i>,
index: &'i Index,
}
impl<'t, 'u, 'i> ClearDocuments<'t, 'u, 'i> {
pub fn new(wtxn: &'t mut heed::RwTxn<'i, 'u>, index: &'i Index) -> ClearDocuments<'t, 'u, 'i> {
impl<'t, 'i> ClearDocuments<'t, 'i> {
pub fn new(wtxn: &'t mut RwTxn<'i>, index: &'i Index) -> ClearDocuments<'t, 'i> {
ClearDocuments { wtxn, index }
}

44
milli/src/update/facet/bulk.rs
View File

@ -2,8 +2,8 @@ use std::fs::File;
use std::io::BufReader;
use grenad::CompressionType;
use heed::types::ByteSlice;
use heed::{BytesDecode, BytesEncode, Error, RoTxn, RwTxn};
use heed::types::Bytes;
use heed::{BytesDecode, BytesEncode, Error, PutFlags, RoTxn, RwTxn};
use roaring::RoaringBitmap;
use super::{FACET_GROUP_SIZE, FACET_MIN_LEVEL_SIZE};
@ -11,7 +11,7 @@ use crate::facet::FacetType;
use crate::heed_codec::facet::{
FacetGroupKey, FacetGroupKeyCodec, FacetGroupValue, FacetGroupValueCodec,
};
use crate::heed_codec::ByteSliceRefCodec;
use crate::heed_codec::BytesRefCodec;
use crate::update::del_add::{DelAdd, KvReaderDelAdd};
use crate::update::index_documents::{create_writer, valid_lmdb_key, writer_into_reader};
use crate::{CboRoaringBitmapCodec, CboRoaringBitmapLenCodec, FieldId, Index, Result};
@ -70,11 +70,11 @@ impl<'i> FacetsUpdateBulk<'i> {
let Self { index, field_ids, group_size, min_level_size, facet_type, delta_data } = self;
let db = match facet_type {
FacetType::String => index
.facet_id_string_docids
.remap_key_type::<FacetGroupKeyCodec<ByteSliceRefCodec>>(),
FacetType::String => {
index.facet_id_string_docids.remap_key_type::<FacetGroupKeyCodec<BytesRefCodec>>()
}
FacetType::Number => {
index.facet_id_f64_docids.remap_key_type::<FacetGroupKeyCodec<ByteSliceRefCodec>>()
index.facet_id_f64_docids.remap_key_type::<FacetGroupKeyCodec<BytesRefCodec>>()
}
};
@ -88,7 +88,7 @@ impl<'i> FacetsUpdateBulk<'i> {
/// Implementation of `FacetsUpdateBulk` that is independent of milli's `Index` type
pub(crate) struct FacetsUpdateBulkInner<R: std::io::Read + std::io::Seek> {
pub db: heed::Database<FacetGroupKeyCodec<ByteSliceRefCodec>, FacetGroupValueCodec>,
pub db: heed::Database<FacetGroupKeyCodec<BytesRefCodec>, FacetGroupValueCodec>,
pub delta_data: Option<grenad::Reader<R>>,
pub group_size: u8,
pub min_level_size: u8,
@ -106,7 +106,7 @@ impl<R: std::io::Read + std::io::Seek> FacetsUpdateBulkInner<R> {
for level_reader in level_readers {
let mut cursor = level_reader.into_cursor()?;
while let Some((k, v)) = cursor.move_on_next()? {
self.db.remap_types::<ByteSlice, ByteSlice>().put(wtxn, k, v)?;
self.db.remap_types::<Bytes, Bytes>().put(wtxn, k, v)?;
}
}
}
@ -128,7 +128,7 @@ impl<R: std::io::Read + std::io::Seek> FacetsUpdateBulkInner<R> {
};
if self.db.is_empty(wtxn)? {
let mut buffer = Vec::new();
let mut database = self.db.iter_mut(wtxn)?.remap_types::<ByteSlice, ByteSlice>();
let mut database = self.db.iter_mut(wtxn)?.remap_types::<Bytes, Bytes>();
let mut cursor = delta_data.into_cursor()?;
while let Some((key, value)) = cursor.move_on_next()? {
if !valid_lmdb_key(key) {
@ -146,11 +146,13 @@ impl<R: std::io::Read + std::io::Seek> FacetsUpdateBulkInner<R> {
buffer.push(1);
// then we extend the buffer with the docids bitmap
buffer.extend_from_slice(value);
unsafe { database.append(key, &buffer)? };
unsafe {
database.put_current_with_options::<Bytes>(PutFlags::APPEND, key, &buffer)?
};
}
} else {
let mut buffer = Vec::new();
let database = self.db.remap_types::<ByteSlice, ByteSlice>();
let database = self.db.remap_types::<Bytes, Bytes>();
let mut cursor = delta_data.into_cursor()?;
while let Some((key, value)) = cursor.move_on_next()? {
@ -219,9 +221,9 @@ impl<R: std::io::Read + std::io::Seek> FacetsUpdateBulkInner<R> {
let level_0_iter = self
.db
.as_polymorph()
.prefix_iter::<_, ByteSlice, ByteSlice>(rtxn, level_0_prefix.as_slice())?
.remap_types::<FacetGroupKeyCodec<ByteSliceRefCodec>, FacetGroupValueCodec>();
.remap_types::<Bytes, Bytes>()
.prefix_iter(rtxn, level_0_prefix.as_slice())?
.remap_types::<FacetGroupKeyCodec<BytesRefCodec>, FacetGroupValueCodec>();
let mut left_bound: &[u8] = &[];
let mut first_iteration_for_new_group = true;
@ -307,11 +309,11 @@ impl<R: std::io::Read + std::io::Seek> FacetsUpdateBulkInner<R> {
bitmaps.drain(..).zip(left_bounds.drain(..)).zip(group_sizes.drain(..))
{
let key = FacetGroupKey { field_id, level, left_bound };
let key = FacetGroupKeyCodec::<ByteSliceRefCodec>::bytes_encode(&key)
.ok_or(Error::Encoding)?;
let key = FacetGroupKeyCodec::<BytesRefCodec>::bytes_encode(&key)
.map_err(Error::Encoding)?;
let value = FacetGroupValue { size: group_size, bitmap };
let value =
FacetGroupValueCodec::bytes_encode(&value).ok_or(Error::Encoding)?;
FacetGroupValueCodec::bytes_encode(&value).map_err(Error::Encoding)?;
cur_writer.insert(key, value)?;
cur_writer_len += 1;
}
@ -336,10 +338,10 @@ impl<R: std::io::Read + std::io::Seek> FacetsUpdateBulkInner<R> {
bitmaps.drain(..).zip(left_bounds.drain(..)).zip(group_sizes.drain(..))
{
let key = FacetGroupKey { field_id, level, left_bound };
let key = FacetGroupKeyCodec::<ByteSliceRefCodec>::bytes_encode(&key)
.ok_or(Error::Encoding)?;
let key = FacetGroupKeyCodec::<BytesRefCodec>::bytes_encode(&key)
.map_err(Error::Encoding)?;
let value = FacetGroupValue { size: group_size, bitmap };
let value = FacetGroupValueCodec::bytes_encode(&value).ok_or(Error::Encoding)?;
let value = FacetGroupValueCodec::bytes_encode(&value).map_err(Error::Encoding)?;
cur_writer.insert(key, value)?;
cur_writer_len += 1;
}

83
milli/src/update/facet/incremental.rs
View File

@ -1,7 +1,7 @@
use std::fs::File;
use std::io::BufReader;
use heed::types::{ByteSlice, DecodeIgnore};
use heed::types::{Bytes, DecodeIgnore};
use heed::{BytesDecode, Error, RoTxn, RwTxn};
use obkv::KvReader;
use roaring::RoaringBitmap;
@ -10,7 +10,7 @@ use crate::facet::FacetType;
use crate::heed_codec::facet::{
FacetGroupKey, FacetGroupKeyCodec, FacetGroupValue, FacetGroupValueCodec,
};
use crate::heed_codec::ByteSliceRefCodec;
use crate::heed_codec::BytesRefCodec;
use crate::search::facet::get_highest_level;
use crate::update::del_add::DelAdd;
use crate::update::index_documents::valid_lmdb_key;
@ -48,10 +48,10 @@ impl FacetsUpdateIncremental {
db: match facet_type {
FacetType::String => index
.facet_id_string_docids
.remap_key_type::<FacetGroupKeyCodec<ByteSliceRefCodec>>(),
.remap_key_type::<FacetGroupKeyCodec<BytesRefCodec>>(),
FacetType::Number => index
.facet_id_f64_docids
.remap_key_type::<FacetGroupKeyCodec<ByteSliceRefCodec>>(),
.remap_key_type::<FacetGroupKeyCodec<BytesRefCodec>>(),
},
group_size,
max_group_size,
@ -67,19 +67,19 @@ impl FacetsUpdateIncremental {
if !valid_lmdb_key(key) {
continue;
}
let key = FacetGroupKeyCodec::<ByteSliceRefCodec>::bytes_decode(key)
.ok_or(heed::Error::Encoding)?;
let key = FacetGroupKeyCodec::<BytesRefCodec>::bytes_decode(key)
.map_err(heed::Error::Encoding)?;
let value = KvReader::new(value);
let docids_to_delete = value
.get(DelAdd::Deletion)
.map(CboRoaringBitmapCodec::bytes_decode)
.map(|o| o.ok_or(heed::Error::Encoding));
.map(|o| o.map_err(heed::Error::Encoding));
let docids_to_add = value
.get(DelAdd::Addition)
.map(CboRoaringBitmapCodec::bytes_decode)
.map(|o| o.ok_or(heed::Error::Encoding));
.map(|o| o.map_err(heed::Error::Encoding));
if let Some(docids_to_delete) = docids_to_delete {
let docids_to_delete = docids_to_delete?;
@ -98,7 +98,7 @@ impl FacetsUpdateIncremental {
/// Implementation of `FacetsUpdateIncremental` that is independent of milli's `Index` type
pub struct FacetsUpdateIncrementalInner {
pub db: heed::Database<FacetGroupKeyCodec<ByteSliceRefCodec>, FacetGroupValueCodec>,
pub db: heed::Database<FacetGroupKeyCodec<BytesRefCodec>, FacetGroupValueCodec>,
pub group_size: u8,
pub min_level_size: u8,
pub max_group_size: u8,
@ -134,15 +134,14 @@ impl FacetsUpdateIncrementalInner {
prefix.extend_from_slice(&field_id.to_be_bytes());
prefix.push(level);
let mut iter =
self.db.as_polymorph().prefix_iter::<_, ByteSlice, FacetGroupValueCodec>(
txn,
prefix.as_slice(),
)?;
let mut iter = self
.db
.remap_types::<Bytes, FacetGroupValueCodec>()
.prefix_iter(txn, prefix.as_slice())?;
let (key_bytes, value) = iter.next().unwrap()?;
Ok((
FacetGroupKeyCodec::<ByteSliceRefCodec>::bytes_decode(key_bytes)
.ok_or(Error::Encoding)?
FacetGroupKeyCodec::<BytesRefCodec>::bytes_decode(key_bytes)
.map_err(Error::Encoding)?
.into_owned(),
value,
))
@ -177,10 +176,8 @@ impl FacetsUpdateIncrementalInner {
level0_prefix.extend_from_slice(&field_id.to_be_bytes());
level0_prefix.push(0);
let mut iter = self
.db
.as_polymorph()
.prefix_iter::<_, ByteSlice, DecodeIgnore>(txn, &level0_prefix)?;
let mut iter =
self.db.remap_types::<Bytes, DecodeIgnore>().prefix_iter(txn, &level0_prefix)?;
if iter.next().is_none() {
drop(iter);
@ -382,11 +379,8 @@ impl FacetsUpdateIncrementalInner {
highest_level_prefix.extend_from_slice(&field_id.to_be_bytes());
highest_level_prefix.push(highest_level);
let size_highest_level = self
.db
.as_polymorph()
.prefix_iter::<_, ByteSlice, ByteSlice>(txn, &highest_level_prefix)?
.count();
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 {
return Ok(());
@ -394,8 +388,8 @@ impl FacetsUpdateIncrementalInner {
let mut groups_iter = self
.db
.as_polymorph()
.prefix_iter::<_, ByteSlice, FacetGroupValueCodec>(txn, &highest_level_prefix)?;
.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;
@ -406,8 +400,8 @@ impl FacetsUpdateIncrementalInner {
let mut values = RoaringBitmap::new();
for _ in 0..group_size {
let (key_bytes, value_i) = groups_iter.next().unwrap()?;
let key_i = FacetGroupKeyCodec::<ByteSliceRefCodec>::bytes_decode(key_bytes)
.ok_or(Error::Encoding)?;
let key_i = FacetGroupKeyCodec::<BytesRefCodec>::bytes_decode(key_bytes)
.map_err(Error::Encoding)?;
if first_key.is_none() {
first_key = Some(key_i);
@ -429,8 +423,8 @@ impl FacetsUpdateIncrementalInner {
let mut values = RoaringBitmap::new();
for _ in 0..nbr_leftover_elements {
let (key_bytes, value_i) = groups_iter.next().unwrap()?;
let key_i = FacetGroupKeyCodec::<ByteSliceRefCodec>::bytes_decode(key_bytes)
.ok_or(Error::Encoding)?;
let key_i = FacetGroupKeyCodec::<BytesRefCodec>::bytes_decode(key_bytes)
.map_err(Error::Encoding)?;
if first_key.is_none() {
first_key = Some(key_i);
@ -597,23 +591,21 @@ impl FacetsUpdateIncrementalInner {
if highest_level == 0
|| self
.db
.as_polymorph()
.prefix_iter::<_, ByteSlice, ByteSlice>(txn, &highest_level_prefix)?
.remap_types::<Bytes, Bytes>()
.prefix_iter(txn, &highest_level_prefix)?
.count()
>= self.min_level_size as usize
{
return Ok(());
}
let mut to_delete = vec![];
let mut iter = self
.db
.as_polymorph()
.prefix_iter::<_, ByteSlice, ByteSlice>(txn, &highest_level_prefix)?;
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::<ByteSliceRefCodec>::bytes_decode(k)
.ok_or(Error::Encoding)?
FacetGroupKeyCodec::<BytesRefCodec>::bytes_decode(k)
.map_err(Error::Encoding)?
.into_owned(),
);
}
@ -1121,7 +1113,7 @@ mod fuzz {
#[no_coverage]
fn compare_with_trivial_database(tempdir: Rc<TempDir>, operations: &[Operation]) {
let index = FacetIndex::<ByteSliceRefCodec>::open_from_tempdir(tempdir, 4, 8, 5); // dummy params, they'll be overwritten
let index = FacetIndex::<BytesRefCodec>::open_from_tempdir(tempdir, 4, 8, 5); // dummy params, they'll be overwritten
let mut txn = index.env.write_txn().unwrap();
let mut trivial_db = TrivialDatabase::<Vec<u8>>::default();
@ -1167,16 +1159,13 @@ mod fuzz {
let level0iter = index
.content
.as_polymorph()
.prefix_iter::<_, ByteSlice, FacetGroupValueCodec>(
&mut txn,
&field_id.to_be_bytes(),
)
.prefix_iter::<_, Bytes, FacetGroupValueCodec>(&mut txn, &field_id.to_be_bytes())
.unwrap();
for ((key, values), group) in values_field_id.iter().zip(level0iter) {
let (group_key, group_values) = group.unwrap();
let group_key =
FacetGroupKeyCodec::<ByteSliceRefCodec>::bytes_decode(group_key).unwrap();
FacetGroupKeyCodec::<BytesRefCodec>::bytes_decode(group_key).unwrap();
assert_eq!(key, &group_key.left_bound);
assert_eq!(values, &group_values.bitmap);
}
@ -1186,13 +1175,13 @@ mod fuzz {
let level0iter = index
.content
.as_polymorph()
.prefix_iter::<_, ByteSlice, FacetGroupValueCodec>(&txn, &field_id.to_be_bytes())
.prefix_iter::<_, Bytes, FacetGroupValueCodec>(&txn, &field_id.to_be_bytes())
.unwrap();
for ((key, values), group) in values_field_id.iter().zip(level0iter) {
let (group_key, group_values) = group.unwrap();
let group_key =
FacetGroupKeyCodec::<ByteSliceRefCodec>::bytes_decode(group_key).unwrap();
FacetGroupKeyCodec::<BytesRefCodec>::bytes_decode(group_key).unwrap();
assert_eq!(key, &group_key.left_bound);
assert_eq!(values, &group_values.bitmap);
}

58
milli/src/update/facet/mod.rs
View File

@ -83,7 +83,7 @@ use std::iter::FromIterator;
use charabia::normalizer::{Normalize, NormalizerOption};
use grenad::{CompressionType, SortAlgorithm};
use heed::types::{ByteSlice, DecodeIgnore, SerdeJson};
use heed::types::{Bytes, DecodeIgnore, SerdeJson};
use heed::BytesEncode;
use log::debug;
use time::OffsetDateTime;
@ -92,10 +92,10 @@ use self::incremental::FacetsUpdateIncremental;
use super::FacetsUpdateBulk;
use crate::facet::FacetType;
use crate::heed_codec::facet::{FacetGroupKey, FacetGroupKeyCodec, FacetGroupValueCodec};
use crate::heed_codec::ByteSliceRefCodec;
use crate::heed_codec::BytesRefCodec;
use crate::update::index_documents::create_sorter;
use crate::update::merge_btreeset_string;
use crate::{BEU16StrCodec, Index, Result, BEU16, MAX_FACET_VALUE_LENGTH};
use crate::{BEU16StrCodec, Index, Result, MAX_FACET_VALUE_LENGTH};
pub mod bulk;
pub mod incremental;
@ -106,7 +106,7 @@ pub mod incremental;
/// a bulk update method or an incremental update method.
pub struct FacetsUpdate<'i> {
index: &'i Index,
database: heed::Database<FacetGroupKeyCodec<ByteSliceRefCodec>, FacetGroupValueCodec>,
database: heed::Database<FacetGroupKeyCodec<BytesRefCodec>, FacetGroupValueCodec>,
facet_type: FacetType,
delta_data: grenad::Reader<BufReader<File>>,
group_size: u8,
@ -120,11 +120,11 @@ impl<'i> FacetsUpdate<'i> {
delta_data: grenad::Reader<BufReader<File>>,
) -> Self {
let database = match facet_type {
FacetType::String => index
.facet_id_string_docids
.remap_key_type::<FacetGroupKeyCodec<ByteSliceRefCodec>>(),
FacetType::String => {
index.facet_id_string_docids.remap_key_type::<FacetGroupKeyCodec<BytesRefCodec>>()
}
FacetType::Number => {
index.facet_id_f64_docids.remap_key_type::<FacetGroupKeyCodec<ByteSliceRefCodec>>()
index.facet_id_f64_docids.remap_key_type::<FacetGroupKeyCodec<BytesRefCodec>>()
}
};
Self {
@ -146,7 +146,7 @@ impl<'i> FacetsUpdate<'i> {
self.index.set_updated_at(wtxn, &OffsetDateTime::now_utc())?;
// See self::comparison_bench::benchmark_facet_indexing
if self.delta_data.len() >= (self.database.len(wtxn)? as u64 / 50) {
if self.delta_data.len() >= (self.database.len(wtxn)? / 50) {
let field_ids =
self.index.faceted_fields_ids(wtxn)?.iter().copied().collect::<Vec<_>>();
let bulk_update = FacetsUpdateBulk::new(
@ -207,8 +207,8 @@ impl<'i> FacetsUpdate<'i> {
}
let set = BTreeSet::from_iter(std::iter::once(left_bound));
let key = (field_id, normalized_facet.as_ref());
let key = BEU16StrCodec::bytes_encode(&key).ok_or(heed::Error::Encoding)?;
let val = SerdeJson::bytes_encode(&set).ok_or(heed::Error::Encoding)?;
let key = BEU16StrCodec::bytes_encode(&key).map_err(heed::Error::Encoding)?;
let val = SerdeJson::bytes_encode(&set).map_err(heed::Error::Encoding)?;
sorter.insert(key, val)?;
}
}
@ -217,10 +217,11 @@ impl<'i> FacetsUpdate<'i> {
// as the grenad sorter already merged them for us.
let mut merger_iter = sorter.into_stream_merger_iter()?;
while let Some((key_bytes, btreeset_bytes)) = merger_iter.next()? {
self.index
.facet_id_normalized_string_strings
.remap_types::<ByteSlice, ByteSlice>()
.put(wtxn, key_bytes, btreeset_bytes)?;
self.index.facet_id_normalized_string_strings.remap_types::<Bytes, Bytes>().put(
wtxn,
key_bytes,
btreeset_bytes,
)?;
}
// We compute one FST by string facet
@ -252,7 +253,7 @@ impl<'i> FacetsUpdate<'i> {
// We write those FSTs in LMDB now
for (field_id, fst) in text_fsts {
self.index.facet_id_string_fst.put(wtxn, &BEU16::new(field_id), &fst)?;
self.index.facet_id_string_fst.put(wtxn, &field_id, &fst)?;
}
Ok(())
@ -267,7 +268,7 @@ pub(crate) mod test_helpers {
use std::marker::PhantomData;
use std::rc::Rc;
use heed::types::ByteSlice;
use heed::types::Bytes;
use heed::{BytesDecode, BytesEncode, Env, RoTxn, RwTxn};
use roaring::RoaringBitmap;
@ -275,7 +276,7 @@ pub(crate) mod test_helpers {
use crate::heed_codec::facet::{
FacetGroupKey, FacetGroupKeyCodec, FacetGroupValue, FacetGroupValueCodec,
};
use crate::heed_codec::ByteSliceRefCodec;
use crate::heed_codec::BytesRefCodec;
use crate::search::facet::get_highest_level;
use crate::snapshot_tests::display_bitmap;
use crate::update::del_add::{DelAdd, KvWriterDelAdd};
@ -306,7 +307,7 @@ pub(crate) mod test_helpers {
BytesEncode<'a> + BytesDecode<'a, DItem = <BoundCodec as BytesEncode<'a>>::EItem>,
{
pub env: Env,
pub content: heed::Database<FacetGroupKeyCodec<ByteSliceRefCodec>, FacetGroupValueCodec>,
pub content: heed::Database<FacetGroupKeyCodec<BytesRefCodec>, FacetGroupValueCodec>,
pub group_size: Cell<u8>,
pub min_level_size: Cell<u8>,
pub max_group_size: Cell<u8>,
@ -454,7 +455,7 @@ pub(crate) mod test_helpers {
let left_bound_bytes = BoundCodec::bytes_encode(left_bound).unwrap().into_owned();
let key: FacetGroupKey<&[u8]> =
FacetGroupKey { field_id: *field_id, level: 0, left_bound: &left_bound_bytes };
let key = FacetGroupKeyCodec::<ByteSliceRefCodec>::bytes_encode(&key).unwrap();
let key = FacetGroupKeyCodec::<BytesRefCodec>::bytes_encode(&key).unwrap();
let mut inner_writer = KvWriterDelAdd::memory();
let value = CboRoaringBitmapCodec::bytes_encode(docids).unwrap();
inner_writer.insert(DelAdd::Addition, value).unwrap();
@ -486,12 +487,12 @@ pub(crate) mod test_helpers {
let iter = self
.content
.as_polymorph()
.prefix_iter::<_, ByteSlice, FacetGroupValueCodec>(txn, &level_no_prefix)
.remap_types::<Bytes, FacetGroupValueCodec>()
.prefix_iter(txn, &level_no_prefix)
.unwrap();
for el in iter {
let (key, value) = el.unwrap();
let key = FacetGroupKeyCodec::<ByteSliceRefCodec>::bytes_decode(key).unwrap();
let key = FacetGroupKeyCodec::<BytesRefCodec>::bytes_decode(key).unwrap();
let mut prefix_start_below = vec![];
prefix_start_below.extend_from_slice(&field_id.to_be_bytes());
@ -501,14 +502,11 @@ pub(crate) mod test_helpers {
let start_below = {
let mut start_below_iter = self
.content
.as_polymorph()
.prefix_iter::<_, ByteSlice, FacetGroupValueCodec>(
txn,
&prefix_start_below,
)
.remap_types::<Bytes, FacetGroupValueCodec>()
.prefix_iter(txn, &prefix_start_below)
.unwrap();
let (key_bytes, _) = start_below_iter.next().unwrap().unwrap();
FacetGroupKeyCodec::<ByteSliceRefCodec>::bytes_decode(key_bytes).unwrap()
FacetGroupKeyCodec::<BytesRefCodec>::bytes_decode(key_bytes).unwrap()
};
assert!(value.size > 0);
@ -612,7 +610,7 @@ mod comparison_bench {
}
let time_spent = timer.elapsed().as_millis();
println!(" add {nbr_doc} : {time_spent}ms");
txn.abort().unwrap();
txn.abort();
}
}
}

3
milli/src/update/index_documents/extract/extract_docid_word_positions.rs
View File

@ -309,8 +309,7 @@ fn tokens_from_document<'a>(
// if a language has been detected for the token, we update the counter.
if let Some(language) = token.language {
let script = token.script;
let entry =
script_language_word_count.entry(script).or_insert_with(Vec::new);
let entry = script_language_word_count.entry(script).or_default();
match entry.iter_mut().find(|(l, _)| *l == language) {
Some((_, n)) => *n += 1,
None => entry.push((language, 1)),

12
milli/src/update/index_documents/extract/extract_fid_docid_facet_values.rs
View File

@ -6,8 +6,8 @@ use std::io::{self, BufReader};
use std::mem::size_of;
use std::result::Result as StdResult;
use bytemuck::bytes_of;
use grenad::Sorter;
use heed::zerocopy::AsBytes;
use heed::BytesEncode;
use itertools::EitherOrBoth;
use ordered_float::OrderedFloat;
@ -20,9 +20,7 @@ use crate::error::InternalError;
use crate::facet::value_encoding::f64_into_bytes;
use crate::update::del_add::{DelAdd, KvWriterDelAdd};
use crate::update::index_documents::{create_writer, writer_into_reader};
use crate::{
CboRoaringBitmapCodec, DocumentId, Error, FieldId, Result, BEU32, MAX_FACET_VALUE_LENGTH,
};
use crate::{CboRoaringBitmapCodec, DocumentId, Error, FieldId, Result, MAX_FACET_VALUE_LENGTH};
/// The length of the elements that are always in the buffer when inserting new values.
const TRUNCATE_SIZE: usize = size_of::<FieldId>() + size_of::<DocumentId>();
@ -94,7 +92,7 @@ pub fn extract_fid_docid_facet_values<R: io::Read + io::Seek>(
strings_key_buffer.extend_from_slice(&field_id.to_be_bytes());
let document: [u8; 4] = docid_bytes[..4].try_into().ok().unwrap();
let document = BEU32::from(document).get();
let document = DocumentId::from_be_bytes(document);
// For the other extraction tasks, prefix the key with the field_id and the document_id
numbers_key_buffer.extend_from_slice(docid_bytes);
@ -323,7 +321,7 @@ where
// We insert only the Del part of the Obkv to inform
// that we only want to remove all those numbers.
let mut obkv = KvWriterDelAdd::memory();
obkv.insert(DelAdd::Deletion, ().as_bytes())?;
obkv.insert(DelAdd::Deletion, bytes_of(&()))?;
let bytes = obkv.into_inner()?;
fid_docid_facet_numbers_sorter.insert(&key_buffer, bytes)?;
}
@ -336,7 +334,7 @@ where
// We insert only the Add part of the Obkv to inform
// that we only want to remove all those numbers.
let mut obkv = KvWriterDelAdd::memory();
obkv.insert(DelAdd::Addition, ().as_bytes())?;
obkv.insert(DelAdd::Addition, bytes_of(&()))?;
let bytes = obkv.into_inner()?;
fid_docid_facet_numbers_sorter.insert(&key_buffer, bytes)?;
}

2
milli/src/update/index_documents/extract/extract_word_docids.rs
View File

@ -118,7 +118,7 @@ pub fn extract_word_docids<R: io::Read + io::Seek>(
}
let (word, fid) = StrBEU16Codec::bytes_decode(key)
.ok_or(SerializationError::Decoding { db_name: Some(DOCID_WORD_POSITIONS) })?;
.map_err(|_| SerializationError::Decoding { db_name: Some(DOCID_WORD_POSITIONS) })?;
// every words contained in an attribute set to exact must be pushed in the exact_words list.
if exact_attributes.contains(&fid) {

4
milli/src/update/index_documents/helpers/grenad_helpers.rs
View File

@ -3,7 +3,7 @@ use std::fs::File;
use std::io::{self, BufReader, BufWriter, Seek};
use grenad::{CompressionType, Sorter};
use heed::types::ByteSlice;
use heed::types::Bytes;
use super::{ClonableMmap, MergeFn};
use crate::update::index_documents::valid_lmdb_key;
@ -255,7 +255,7 @@ where
puffin::profile_function!();
let mut buffer = Vec::new();
let database = database.remap_types::<ByteSlice, ByteSlice>();
let database = database.remap_types::<Bytes, Bytes>();
let mut merger_iter = sorter.into_stream_merger_iter()?;
while let Some((key, value)) = merger_iter.next()? {

55
milli/src/update/index_documents/mod.rs
View File

@ -68,8 +68,8 @@ impl Default for IndexDocumentsMethod {
}
}
pub struct IndexDocuments<'t, 'u, 'i, 'a, FP, FA> {
wtxn: &'t mut heed::RwTxn<'i, 'u>,
pub struct IndexDocuments<'t, 'i, 'a, FP, FA> {
wtxn: &'t mut heed::RwTxn<'i>,
index: &'i Index,
config: IndexDocumentsConfig,
indexer_config: &'a IndexerConfig,
@ -90,19 +90,19 @@ pub struct IndexDocumentsConfig {
pub autogenerate_docids: bool,
}
impl<'t, 'u, 'i, 'a, FP, FA> IndexDocuments<'t, 'u, 'i, 'a, FP, FA>
impl<'t, 'i, 'a, FP, FA> IndexDocuments<'t, 'i, 'a, FP, FA>
where
FP: Fn(UpdateIndexingStep) + Sync,
FA: Fn() -> bool + Sync,
{
pub fn new(
wtxn: &'t mut heed::RwTxn<'i, 'u>,
wtxn: &'t mut heed::RwTxn<'i>,
index: &'i Index,
indexer_config: &'a IndexerConfig,
config: IndexDocumentsConfig,
progress: FP,
should_abort: FA,
) -> Result<IndexDocuments<'t, 'u, 'i, 'a, FP, FA>> {
) -> Result<IndexDocuments<'t, 'i, 'a, FP, FA>> {
let transform = Some(Transform::new(
wtxn,
index,
@ -701,7 +701,7 @@ mod tests {
use crate::documents::documents_batch_reader_from_objects;
use crate::index::tests::TempIndex;
use crate::search::TermsMatchingStrategy;
use crate::{db_snap, Filter, Search, BEU16};
use crate::{db_snap, Filter, Search};
#[test]
fn simple_document_replacement() {
@ -1743,14 +1743,11 @@ mod tests {
let colour_green_id = index.fields_ids_map(&rtxn).unwrap().id("colour.green").unwrap();
let bitmap_colour =
index.facet_id_exists_docids.get(&rtxn, &BEU16::new(colour_id)).unwrap().unwrap();
index.facet_id_exists_docids.get(&rtxn, &colour_id).unwrap().unwrap();
assert_eq!(bitmap_colour.into_iter().collect::<Vec<_>>(), vec![0, 1, 2, 3, 4, 6, 7]);
let bitmap_colour_green = index
.facet_id_exists_docids
.get(&rtxn, &BEU16::new(colour_green_id))
.unwrap()
.unwrap();
let bitmap_colour_green =
index.facet_id_exists_docids.get(&rtxn, &colour_green_id).unwrap().unwrap();
assert_eq!(bitmap_colour_green.into_iter().collect::<Vec<_>>(), vec![6, 7]);
};
@ -1848,21 +1845,15 @@ mod tests {
index.fields_ids_map(&rtxn).unwrap().id("colour.green.blue").unwrap();
let bitmap_null_colour =
index.facet_id_is_null_docids.get(&rtxn, &BEU16::new(colour_id)).unwrap().unwrap();
index.facet_id_is_null_docids.get(&rtxn, &colour_id).unwrap().unwrap();
assert_eq!(bitmap_null_colour.into_iter().collect::<Vec<_>>(), vec![0]);
let bitmap_colour_green = index
.facet_id_is_null_docids
.get(&rtxn, &BEU16::new(colour_green_id))
.unwrap()
.unwrap();
let bitmap_colour_green =
index.facet_id_is_null_docids.get(&rtxn, &colour_green_id).unwrap().unwrap();
assert_eq!(bitmap_colour_green.into_iter().collect::<Vec<_>>(), vec![2]);
let bitmap_colour_blue = index
.facet_id_is_null_docids
.get(&rtxn, &BEU16::new(colour_blue_id))
.unwrap()
.unwrap();
let bitmap_colour_blue =
index.facet_id_is_null_docids.get(&rtxn, &colour_blue_id).unwrap().unwrap();
assert_eq!(bitmap_colour_blue.into_iter().collect::<Vec<_>>(), vec![3]);
};
@ -1917,21 +1908,15 @@ mod tests {
let tags_blue_id = index.fields_ids_map(&rtxn).unwrap().id("tags.green.blue").unwrap();
let bitmap_empty_tags =
index.facet_id_is_empty_docids.get(&rtxn, &BEU16::new(tags_id)).unwrap().unwrap();
index.facet_id_is_empty_docids.get(&rtxn, &tags_id).unwrap().unwrap();
assert_eq!(bitmap_empty_tags.into_iter().collect::<Vec<_>>(), vec![2, 6, 9]);
let bitmap_tags_green = index
.facet_id_is_empty_docids
.get(&rtxn, &BEU16::new(tags_green_id))
.unwrap()
.unwrap();
let bitmap_tags_green =
index.facet_id_is_empty_docids.get(&rtxn, &tags_green_id).unwrap().unwrap();
assert_eq!(bitmap_tags_green.into_iter().collect::<Vec<_>>(), vec![8]);
let bitmap_tags_blue = index
.facet_id_is_empty_docids
.get(&rtxn, &BEU16::new(tags_blue_id))
.unwrap()
.unwrap();
let bitmap_tags_blue =
index.facet_id_is_empty_docids.get(&rtxn, &tags_blue_id).unwrap().unwrap();
assert_eq!(bitmap_tags_blue.into_iter().collect::<Vec<_>>(), vec![12]);
};
@ -2684,7 +2669,7 @@ mod tests {
}
fn delete_documents<'t>(
wtxn: &mut RwTxn<'t, '_>,
wtxn: &mut RwTxn<'t>,
index: &'t TempIndex,
external_ids: &[&str],
) -> Vec<u32> {

15
milli/src/update/index_documents/transform.rs
View File

@ -24,9 +24,7 @@ use crate::index::{db_name, main_key};
use crate::update::del_add::{into_del_add_obkv, DelAdd, DelAddOperation, KvReaderDelAdd};
use crate::update::index_documents::GrenadParameters;
use crate::update::{AvailableDocumentsIds, ClearDocuments, UpdateIndexingStep};
use crate::{
FieldDistribution, FieldId, FieldIdMapMissingEntry, FieldsIdsMap, Index, Result, BEU32,
};
use crate::{FieldDistribution, FieldId, FieldIdMapMissingEntry, FieldsIdsMap, Index, Result};
pub struct TransformOutput {
pub primary_key: String,
@ -245,11 +243,11 @@ impl<'a, 'i> Transform<'a, 'i> {
let mut skip_insertion = false;
if let Some(original_docid) = original_docid {
let original_key = BEU32::new(original_docid);
let original_key = original_docid;
let base_obkv = self
.index
.documents
.remap_data_type::<heed::types::ByteSlice>()
.remap_data_type::<heed::types::Bytes>()
.get(wtxn, &original_key)?
.ok_or(InternalError::DatabaseMissingEntry {
db_name: db_name::DOCUMENTS,
@ -499,11 +497,11 @@ impl<'a, 'i> Transform<'a, 'i> {
self.replaced_documents_ids.insert(internal_docid);
// fetch the obkv document
let original_key = BEU32::new(internal_docid);
let original_key = internal_docid;
let base_obkv = self
.index
.documents
.remap_data_type::<heed::types::ByteSlice>()
.remap_data_type::<heed::types::Bytes>()
.get(txn, &original_key)?
.ok_or(InternalError::DatabaseMissingEntry {
db_name: db_name::DOCUMENTS,
@ -811,7 +809,7 @@ impl<'a, 'i> Transform<'a, 'i> {
// TODO this can be done in parallel by using the rayon `ThreadPool`.
pub fn prepare_for_documents_reindexing(
self,
wtxn: &mut heed::RwTxn<'i, '_>,
wtxn: &mut heed::RwTxn<'i>,
old_fields_ids_map: FieldsIdsMap,
mut new_fields_ids_map: FieldsIdsMap,
) -> Result<TransformOutput> {
@ -857,7 +855,6 @@ impl<'a, 'i> Transform<'a, 'i> {
let obkv = self.index.documents.get(wtxn, &docid)?.ok_or(
InternalError::DatabaseMissingEntry { db_name: db_name::DOCUMENTS, key: None },
)?;
let docid = docid.get();
obkv_buffer.clear();
let mut obkv_writer = KvWriter::<_, FieldId>::new(&mut obkv_buffer);

39
milli/src/update/index_documents/typed_chunk.rs
View File

@ -6,8 +6,8 @@ use std::io::{self, BufReader};
use bytemuck::allocation::pod_collect_to_vec;
use charabia::{Language, Script};
use grenad::MergerBuilder;
use heed::types::ByteSlice;
use heed::RwTxn;
use heed::types::Bytes;
use heed::{PutFlags, RwTxn};
use log::error;
use obkv::{KvReader, KvWriter};
use ordered_float::OrderedFloat;
@ -27,9 +27,7 @@ use crate::index::Hnsw;
use crate::update::del_add::{deladd_serialize_add_side, DelAdd, KvReaderDelAdd};
use crate::update::facet::FacetsUpdate;
use crate::update::index_documents::helpers::{as_cloneable_grenad, try_split_array_at};
use crate::{
lat_lng_to_xyz, DocumentId, FieldId, GeoPoint, Index, Result, SerializationError, BEU32,
};
use crate::{lat_lng_to_xyz, DocumentId, FieldId, GeoPoint, Index, Result, SerializationError};
pub(crate) enum TypedChunk {
FieldIdDocidFacetStrings(grenad::Reader<CursorClonableMmap>),
@ -146,10 +144,10 @@ pub(crate) fn write_typed_chunk_into_index(
}
}
let db = index.documents.remap_data_type::<ByteSlice>();
let db = index.documents.remap_data_type::<Bytes>();
if !writer.is_empty() {
db.put(wtxn, &BEU32::new(docid), &writer.into_inner().unwrap())?;
db.put(wtxn, &docid, &writer.into_inner().unwrap())?;
operations.push(DocumentOperation {
external_id: external_id.to_string(),
internal_id: docid,
@ -157,7 +155,7 @@ pub(crate) fn write_typed_chunk_into_index(
});
docids.insert(docid);
} else {
db.delete(wtxn, &BEU32::new(docid))?;
db.delete(wtxn, &docid)?;
operations.push(DocumentOperation {
external_id: external_id.to_string(),
internal_id: docid,
@ -295,7 +293,7 @@ pub(crate) fn write_typed_chunk_into_index(
}
TypedChunk::FieldIdDocidFacetNumbers(fid_docid_facet_number) => {
let index_fid_docid_facet_numbers =
index.field_id_docid_facet_f64s.remap_types::<ByteSlice, ByteSlice>();
index.field_id_docid_facet_f64s.remap_types::<Bytes, Bytes>();
let mut cursor = fid_docid_facet_number.into_cursor()?;
while let Some((key, value)) = cursor.move_on_next()? {
let reader = KvReaderDelAdd::new(value);
@ -315,7 +313,7 @@ pub(crate) fn write_typed_chunk_into_index(
}
TypedChunk::FieldIdDocidFacetStrings(fid_docid_facet_string) => {
let index_fid_docid_facet_strings =
index.field_id_docid_facet_strings.remap_types::<ByteSlice, ByteSlice>();
index.field_id_docid_facet_strings.remap_types::<Bytes, Bytes>();
let mut cursor = fid_docid_facet_string.into_cursor()?;
while let Some((key, value)) = cursor.move_on_next()? {
let reader = KvReaderDelAdd::new(value);
@ -362,8 +360,8 @@ pub(crate) fn write_typed_chunk_into_index(
// We extract and store the previous vectors
if let Some(hnsw) = index.vector_hnsw(wtxn)? {
for (pid, point) in hnsw.iter() {
let pid_key = BEU32::new(pid.into_inner());
let docid = index.vector_id_docid.get(wtxn, &pid_key)?.unwrap().get();
let pid_key = pid.into_inner();
let docid = index.vector_id_docid.get(wtxn, &pid_key)?.unwrap();
let vector: Vec<_> = point.iter().copied().map(OrderedFloat).collect();
vectors_set.insert((docid, vector));
}
@ -424,11 +422,7 @@ pub(crate) fn write_typed_chunk_into_index(
// Store the vectors in the point-docid relation database
index.vector_id_docid.clear(wtxn)?;
for (docid, pid) in docids.into_iter().zip(pids) {
index.vector_id_docid.put(
wtxn,
&BEU32::new(pid.into_inner()),
&BEU32::new(docid),
)?;
index.vector_id_docid.put(wtxn, &pid.into_inner(), &docid)?;
}
log::debug!("There are {} entries in the HNSW so far", hnsw_length);
@ -504,7 +498,7 @@ where
puffin::profile_function!(format!("number of entries: {}", data.len()));
let mut buffer = Vec::new();
let database = database.remap_types::<ByteSlice, ByteSlice>();
let database = database.remap_types::<Bytes, Bytes>();
let mut cursor = data.into_cursor()?;
while let Some((key, value)) = cursor.move_on_next()? {
@ -562,20 +556,23 @@ where
}
let mut buffer = Vec::new();
let mut database = database.iter_mut(wtxn)?.remap_types::<ByteSlice, ByteSlice>();
let mut database = database.iter_mut(wtxn)?.remap_types::<Bytes, Bytes>();
let mut cursor = data.into_cursor()?;
while let Some((key, value)) = cursor.move_on_next()? {
if valid_lmdb_key(key) {
debug_assert!(
K::bytes_decode(key).is_some(),
K::bytes_decode(key).is_ok(),
"Couldn't decode key with the database decoder, key length: {} - key bytes: {:x?}",
key.len(),
&key
);
buffer.clear();
let value = serialize_value(value, &mut buffer)?;
unsafe { database.append(key, value)? };
unsafe {
// safety: We do not keep a reference to anything that lives inside the database
database.put_current_with_options::<Bytes>(PutFlags::APPEND, key, value)?
};
}
}

22
milli/src/update/settings.rs
View File

@ -100,8 +100,8 @@ impl<'de, T: Deserialize<'de>> Deserialize<'de> for Setting<T> {
}
}
pub struct Settings<'a, 't, 'u, 'i> {
wtxn: &'t mut heed::RwTxn<'i, 'u>,
pub struct Settings<'a, 't, 'i> {
wtxn: &'t mut heed::RwTxn<'i>,
index: &'i Index,
indexer_config: &'a IndexerConfig,
@ -129,12 +129,12 @@ pub struct Settings<'a, 't, 'u, 'i> {
pagination_max_total_hits: Setting<usize>,
}
impl<'a, 't, 'u, 'i> Settings<'a, 't, 'u, 'i> {
impl<'a, 't, 'i> Settings<'a, 't, 'i> {
pub fn new(
wtxn: &'t mut heed::RwTxn<'i, 'u>,
wtxn: &'t mut heed::RwTxn<'i>,
index: &'i Index,
indexer_config: &'a IndexerConfig,
) -> Settings<'a, 't, 'u, 'i> {
) -> Settings<'a, 't, 'i> {
Settings {
wtxn,
index,
@ -822,7 +822,7 @@ impl<'a, 't, 'u, 'i> Settings<'a, 't, 'u, 'i> {
fn update_max_values_per_facet(&mut self) -> Result<()> {
match self.max_values_per_facet {
Setting::Set(max) => {
self.index.put_max_values_per_facet(self.wtxn, max)?;
self.index.put_max_values_per_facet(self.wtxn, max as u64)?;
}
Setting::Reset => {
self.index.delete_max_values_per_facet(self.wtxn)?;
@ -850,7 +850,7 @@ impl<'a, 't, 'u, 'i> Settings<'a, 't, 'u, 'i> {
fn update_pagination_max_total_hits(&mut self) -> Result<()> {
match self.pagination_max_total_hits {
Setting::Set(max) => {
self.index.put_pagination_max_total_hits(self.wtxn, max)?;
self.index.put_pagination_max_total_hits(self.wtxn, max as u64)?;
}
Setting::Reset => {
self.index.delete_pagination_max_total_hits(self.wtxn)?;
@ -917,7 +917,7 @@ impl<'a, 't, 'u, 'i> Settings<'a, 't, 'u, 'i> {
#[cfg(test)]
mod tests {
use big_s::S;
use heed::types::ByteSlice;
use heed::types::Bytes;
use maplit::{btreemap, btreeset, hashset};
use super::*;
@ -1130,7 +1130,7 @@ mod tests {
}
let count = index
.facet_id_f64_docids
.remap_key_type::<ByteSlice>()
.remap_key_type::<Bytes>()
// The faceted field id is 1u16
.prefix_iter(&rtxn, &[0, 1, 0])
.unwrap()
@ -1151,7 +1151,7 @@ mod tests {
// Only count the field_id 0 and level 0 facet values.
let count = index
.facet_id_f64_docids
.remap_key_type::<ByteSlice>()
.remap_key_type::<Bytes>()
.prefix_iter(&rtxn, &[0, 1, 0])
.unwrap()
.count();
@ -1565,7 +1565,7 @@ mod tests {
})
.unwrap_err();
assert!(matches!(error, Error::UserError(UserError::PrimaryKeyCannotBeChanged(_))));
wtxn.abort().unwrap();
wtxn.abort();
// But if we clear the database...
let mut wtxn = index.write_txn().unwrap();

14
milli/src/update/word_prefix_docids.rs
View File

@ -1,7 +1,7 @@
use std::collections::{HashMap, HashSet};
use grenad::CompressionType;
use heed::types::{ByteSlice, Str};
use heed::types::{Bytes, Str};
use heed::Database;
use crate::update::del_add::{deladd_serialize_add_side, DelAdd, KvWriterDelAdd};
@ -12,8 +12,8 @@ use crate::update::index_documents::{
};
use crate::{CboRoaringBitmapCodec, Result};
pub struct WordPrefixDocids<'t, 'u, 'i> {
wtxn: &'t mut heed::RwTxn<'i, 'u>,
pub struct WordPrefixDocids<'t, 'i> {
wtxn: &'t mut heed::RwTxn<'i>,
word_docids: Database<Str, CboRoaringBitmapCodec>,
word_prefix_docids: Database<Str, CboRoaringBitmapCodec>,
pub(crate) chunk_compression_type: CompressionType,
@ -22,12 +22,12 @@ pub struct WordPrefixDocids<'t, 'u, 'i> {
pub(crate) max_memory: Option<usize>,
}
impl<'t, 'u, 'i> WordPrefixDocids<'t, 'u, 'i> {
impl<'t, 'i> WordPrefixDocids<'t, 'i> {
pub fn new(
wtxn: &'t mut heed::RwTxn<'i, 'u>,
wtxn: &'t mut heed::RwTxn<'i>,
word_docids: Database<Str, CboRoaringBitmapCodec>,
word_prefix_docids: Database<Str, CboRoaringBitmapCodec>,
) -> WordPrefixDocids<'t, 'u, 'i> {
) -> WordPrefixDocids<'t, 'i> {
WordPrefixDocids {
wtxn,
word_docids,
@ -93,7 +93,7 @@ impl<'t, 'u, 'i> WordPrefixDocids<'t, 'u, 'i> {
}
// We fetch the docids associated to the newly added word prefix fst only.
let db = self.word_docids.remap_data_type::<ByteSlice>();
let db = self.word_docids.remap_data_type::<Bytes>();
let mut buffer = Vec::new();
for prefix in new_prefix_fst_words {
let prefix = std::str::from_utf8(prefix.as_bytes())?;

19
milli/src/update/words_prefix_integer_docids.rs
View File

@ -2,7 +2,7 @@ use std::collections::{HashMap, HashSet};
use std::str;
use grenad::CompressionType;
use heed::types::ByteSlice;
use heed::types::Bytes;
use heed::{BytesDecode, BytesEncode, Database};
use log::debug;
@ -17,8 +17,8 @@ use crate::update::index_documents::{
};
use crate::{CboRoaringBitmapCodec, Result};
pub struct WordPrefixIntegerDocids<'t, 'u, 'i> {
wtxn: &'t mut heed::RwTxn<'i, 'u>,
pub struct WordPrefixIntegerDocids<'t, 'i> {
wtxn: &'t mut heed::RwTxn<'i>,
prefix_database: Database<StrBEU16Codec, CboRoaringBitmapCodec>,
word_database: Database<StrBEU16Codec, CboRoaringBitmapCodec>,
pub(crate) chunk_compression_type: CompressionType,
@ -27,12 +27,12 @@ pub struct WordPrefixIntegerDocids<'t, 'u, 'i> {
pub(crate) max_memory: Option<usize>,
}
impl<'t, 'u, 'i> WordPrefixIntegerDocids<'t, 'u, 'i> {
impl<'t, 'i> WordPrefixIntegerDocids<'t, 'i> {
pub fn new(
wtxn: &'t mut heed::RwTxn<'i, 'u>,
wtxn: &'t mut heed::RwTxn<'i>,
prefix_database: Database<StrBEU16Codec, CboRoaringBitmapCodec>,
word_database: Database<StrBEU16Codec, CboRoaringBitmapCodec>,
) -> WordPrefixIntegerDocids<'t, 'u, 'i> {
) -> WordPrefixIntegerDocids<'t, 'i> {
WordPrefixIntegerDocids {
wtxn,
prefix_database,
@ -72,7 +72,8 @@ impl<'t, 'u, 'i> WordPrefixIntegerDocids<'t, 'u, 'i> {
let mut current_prefixes: Option<&&[String]> = None;
let mut prefixes_cache = HashMap::new();
while let Some((key, data)) = new_word_integer_docids_iter.move_on_next()? {
let (word, pos) = StrBEU16Codec::bytes_decode(key).ok_or(heed::Error::Decoding)?;
let (word, pos) =
StrBEU16Codec::bytes_decode(key).map_err(heed::Error::Decoding)?;
current_prefixes = match current_prefixes.take() {
Some(prefixes) if word.starts_with(&prefixes[0]) => Some(prefixes),
@ -109,7 +110,7 @@ impl<'t, 'u, 'i> WordPrefixIntegerDocids<'t, 'u, 'i> {
}
// We fetch the docids associated to the newly added word prefix fst only.
let db = self.word_database.remap_data_type::<ByteSlice>();
let db = self.word_database.remap_data_type::<Bytes>();
let mut buffer = Vec::new();
for prefix_bytes in new_prefix_fst_words {
let prefix = str::from_utf8(prefix_bytes.as_bytes()).map_err(|_| {
@ -118,7 +119,7 @@ impl<'t, 'u, 'i> WordPrefixIntegerDocids<'t, 'u, 'i> {
// iter over all lines of the DB where the key is prefixed by the current prefix.
let iter = db
.remap_key_type::<ByteSlice>()
.remap_key_type::<Bytes>()
.prefix_iter(self.wtxn, prefix_bytes.as_bytes())?
.remap_key_type::<StrBEU16Codec>();
for result in iter {

12
milli/src/update/words_prefixes_fst.rs
View File

@ -2,21 +2,19 @@ use std::iter::{repeat_with, FromIterator};
use std::str;
use fst::{SetBuilder, Streamer};
use heed::RwTxn;
use crate::{Index, Result, SmallString32};
pub struct WordsPrefixesFst<'t, 'u, 'i> {
wtxn: &'t mut heed::RwTxn<'i, 'u>,
pub struct WordsPrefixesFst<'t, 'i> {
wtxn: &'t mut RwTxn<'i>,
index: &'i Index,
threshold: u32,
max_prefix_length: usize,
}
impl<'t, 'u, 'i> WordsPrefixesFst<'t, 'u, 'i> {
pub fn new(
wtxn: &'t mut heed::RwTxn<'i, 'u>,
index: &'i Index,
) -> WordsPrefixesFst<'t, 'u, 'i> {
impl<'t, 'i> WordsPrefixesFst<'t, 'i> {
pub fn new(wtxn: &'t mut RwTxn<'i>, index: &'i Index) -> WordsPrefixesFst<'t, 'i> {
WordsPrefixesFst { wtxn, index, threshold: 100, max_prefix_length: 4 }
}