Temporary use the charabia git repository to get the latest fixes

4053: Fix the stats of the documents deletion by filter r=Kerollmops a=irevoire

# Pull Request

The issue was that the operation « DocumentDeletionByFilter » was not declared as an index operation. That means the index stats were not reprocessed after the application of the operation.

## Related issue
Fixes #4018

## What does this PR do?
- Move the `DocumentDeletionByFilter` internal operation into the category of the `IndexOperation`. This means that the stats will automatically be re-processed after a batch is processed.
- Update a test to ensure that the stats are valid after each operation

## PR checklist
Please check if your PR fulfills the following requirements:
- [x] Does this PR fix an existing issue, or have you listed the changes applied in the PR description (and why they are needed)?
- [x] Have you read the contributing guidelines?
- [x] Have you made sure that the title is accurate and descriptive of the changes?

Thank you so much for contributing to Meilisearch!


Co-authored-by: Tamo <tamo@meilisearch.com>

.github/workflows/publish-apt-brew-pkg.yml

@@ -53,5 +53,6 @@ jobs:
         uses: mislav/bump-homebrew-formula-action@v2
         with:
           formula-name: meilisearch
+          formula-path: Formula/m/meilisearch.rb
         env:
           COMMITTER_TOKEN: ${{ secrets.HOMEBREW_COMMITTER_TOKEN }}

.github/workflows/test-suite.yml

@@ -37,7 +37,7 @@ jobs:
           toolchain: stable
           override: true
       - name: Setup test with Rust nightly
-        if: github.event_name == 'schedule'
+        if: github.event_name == 'schedule' || github.event_name == 'workflow_dispatch'
         uses: actions-rs/toolchain@v1
         with:
           toolchain: nightly
@@ -78,12 +78,12 @@ jobs:
           args: --locked --release --all
 
   test-all-features:
-    name: Tests all features on cron schedule only
+    name: Tests all features
     runs-on: ubuntu-latest
     container:
       # Use ubuntu-18.04 to compile with glibc 2.27, which are the production expectations
       image: ubuntu:18.04
-    if: github.event_name == 'schedule'
+    if: github.event_name == 'schedule' || github.event_name == 'workflow_dispatch'
     steps:
       - uses: actions/checkout@v3
       - name: Install needed dependencies
@@ -110,7 +110,7 @@ jobs:
     runs-on: ubuntu-latest
     container:
       image: ubuntu:18.04
-    if: github.event_name == 'schedule'
+    if: github.event_name == 'schedule' || github.event_name == 'workflow_dispatch'
     steps:
       - uses: actions/checkout@v3
       - name: Install needed dependencies
@@ -161,7 +161,7 @@ jobs:
       - uses: actions-rs/toolchain@v1
         with:
           profile: minimal
-          toolchain: 1.69.0
+          toolchain: 1.71.1
           override: true
           components: clippy
       - name: Cache dependencies

Cargo.toml

@@ -18,7 +18,7 @@ members = [
 ]
 
 [workspace.package]
-version = "1.3.0"
+version = "1.4.0"
 authors = ["Quentin de Quelen <quentin@dequelen.me>", "Clément Renault <clement@meilisearch.com>"]
 description = "Meilisearch HTTP server"
 homepage = "https://meilisearch.com"

dump/src/reader/snapshots/dump__reader__test__import_dump_v1-10.snap

@@ -0,0 +1,24 @@
+---
+source: dump/src/reader/mod.rs
+expression: spells.settings().unwrap()
+---
+{
+  "displayedAttributes": [
+    "*"
+  ],
+  "searchableAttributes": [
+    "*"
+  ],
+  "filterableAttributes": [],
+  "sortableAttributes": [],
+  "rankingRules": [
+    "typo",
+    "words",
+    "proximity",
+    "attribute",
+    "exactness"
+  ],
+  "stopWords": [],
+  "synonyms": {},
+  "distinctAttribute": null
+}

dump/src/reader/snapshots/dump__reader__test__import_dump_v1-4.snap

@@ -0,0 +1,38 @@
+---
+source: dump/src/reader/mod.rs
+expression: products.settings().unwrap()
+---
+{
+  "displayedAttributes": [
+    "*"
+  ],
+  "searchableAttributes": [
+    "*"
+  ],
+  "filterableAttributes": [],
+  "sortableAttributes": [],
+  "rankingRules": [
+    "typo",
+    "words",
+    "proximity",
+    "attribute",
+    "exactness"
+  ],
+  "stopWords": [],
+  "synonyms": {
+    "android": [
+      "phone",
+      "smartphone"
+    ],
+    "iphone": [
+      "phone",
+      "smartphone"
+    ],
+    "phone": [
+      "android",
+      "iphone",
+      "smartphone"
+    ]
+  },
+  "distinctAttribute": null
+}

dump/src/reader/snapshots/dump__reader__test__import_dump_v1-7.snap

@@ -0,0 +1,31 @@
+---
+source: dump/src/reader/mod.rs
+expression: movies.settings().unwrap()
+---
+{
+  "displayedAttributes": [
+    "*"
+  ],
+  "searchableAttributes": [
+    "*"
+  ],
+  "filterableAttributes": [
+    "genres",
+    "id"
+  ],
+  "sortableAttributes": [
+    "genres",
+    "id"
+  ],
+  "rankingRules": [
+    "typo",
+    "words",
+    "proximity",
+    "attribute",
+    "exactness",
+    "release_date:asc"
+  ],
+  "stopWords": [],
+  "synonyms": {},
+  "distinctAttribute": null
+}

filter-parser/Cargo.toml

@@ -14,6 +14,7 @@ license.workspace = true
 [dependencies]
 nom = "7.1.3"
 nom_locate = "4.1.0"
+unescaper = "0.1.2"
 
 [dev-dependencies]
 insta = "1.29.0"

filter-parser/src/error.rs

@@ -62,6 +62,7 @@ pub enum ErrorKind<'a> {
     MisusedGeoRadius,
     MisusedGeoBoundingBox,
     InvalidPrimary,
+    InvalidEscapedNumber,
     ExpectedEof,
     ExpectedValue(ExpectedValueKind),
     MalformedValue,
@@ -147,6 +148,9 @@ impl<'a> Display for Error<'a> {
                 let text = if input.trim().is_empty() { "but instead got nothing.".to_string() } else { format!("at `{}`.", escaped_input) };
                 writeln!(f, "Was expecting an operation `=`, `!=`, `>=`, `>`, `<=`, `<`, `IN`, `NOT IN`, `TO`, `EXISTS`, `NOT EXISTS`, `IS NULL`, `IS NOT NULL`, `IS EMPTY`, `IS NOT EMPTY`, `_geoRadius`, or `_geoBoundingBox` {}", text)?
             }
+            ErrorKind::InvalidEscapedNumber => {
+                writeln!(f, "Found an invalid escaped sequence number: `{}`.", escaped_input)?
+            }
             ErrorKind::ExpectedEof => {
                 writeln!(f, "Found unexpected characters at the end of the filter: `{}`. You probably forgot an `OR` or an `AND` rule.", escaped_input)?
             }

filter-parser/src/lib.rs

@@ -545,6 +545,8 @@ impl<'a> std::fmt::Display for Token<'a> {
 
 #[cfg(test)]
 pub mod tests {
+    use FilterCondition as Fc;
+
     use super::*;
 
     /// Create a raw [Token]. You must specify the string that appear BEFORE your element followed by your element
@@ -556,14 +558,22 @@ pub mod tests {
         unsafe { Span::new_from_raw_offset(offset, lines as u32, value, "") }.into()
     }
 
+    fn p(s: &str) -> impl std::fmt::Display + '_ {
+        Fc::parse(s).unwrap().unwrap()
+    }
+
+    #[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\\\\}"#);
+        // but it also works with other sequencies
+        insta::assert_display_snapshot!(p(r#"title = 'foo\x20\n\t\"\'"'"#), @"{title} = {foo \n\t\"\'\"}");
+    }
+
     #[test]
     fn parse() {
-        use FilterCondition as Fc;
-
-        fn p(s: &str) -> impl std::fmt::Display + '_ {
-            Fc::parse(s).unwrap().unwrap()
-        }
-
         // Test equal
         insta::assert_display_snapshot!(p("channel = Ponce"), @"{channel} = {Ponce}");
         insta::assert_display_snapshot!(p("subscribers = 12"), @"{subscribers} = {12}");

filter-parser/src/value.rs

@@ -171,7 +171,24 @@ pub fn parse_value(input: Span) -> IResult<Token> {
         })
     })?;
 
-    Ok((input, value))
+    match unescaper::unescape(value.value()) {
+        Ok(content) => {
+            if content.len() != value.value().len() {
+                Ok((input, Token::new(value.original_span(), Some(content))))
+            } else {
+                Ok((input, value))
+            }
+        }
+        Err(unescaper::Error::IncompleteStr(_)) => Err(nom::Err::Incomplete(nom::Needed::Unknown)),
+        Err(unescaper::Error::ParseIntError { .. }) => Err(nom::Err::Error(Error::new_from_kind(
+            value.original_span(),
+            ErrorKind::InvalidEscapedNumber,
+        ))),
+        Err(unescaper::Error::InvalidChar { .. }) => Err(nom::Err::Error(Error::new_from_kind(
+            value.original_span(),
+            ErrorKind::MalformedValue,
+        ))),
+    }
 }
 
 fn is_value_component(c: char) -> bool {
@@ -318,17 +335,17 @@ pub mod test {
             ("\"cha'nnel\"", "cha'nnel", false),
             ("I'm tamo", "I", false),
             // escaped thing but not quote
-            (r#""\\""#, r#"\\"#, false),
-            (r#""\\\\\\""#, r#"\\\\\\"#, false),
-            (r#""aa\\aa""#, r#"aa\\aa"#, false),
+            (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#""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#"'I\'m "super" tamo'"#, r#"I'm "super" tamo"#, true),
             (r#"'\'\''"#, r#"''"#, true),
         ];
@@ -350,7 +367,14 @@ pub mod test {
                 "Filter `{}` was not supposed to be escaped",
                 input
             );
-            assert_eq!(token.value(), expected, "Filter `{}` failed.", input);
+            assert_eq!(
+                token.value(),
+                expected,
+                "Filter `{}` failed by giving `{}` instead of `{}`.",
+                input,
+                token.value(),
+                expected
+            );
         }
     }
 

fuzzers/Cargo.toml

@@ -13,7 +13,7 @@ license.workspace = true
 [dependencies]
 arbitrary = { version = "1.3.0", features = ["derive"] }
 clap = { version = "4.3.0", features = ["derive"] }
-fastrand = "1.9.0"
+fastrand = "2.0.0"
 milli = { path = "../milli" }
 serde = { version = "1.0.160", features = ["derive"] }
 serde_json = { version = "1.0.95", features = ["preserve_order"] }

index-scheduler/src/batch.rs

@@ -67,10 +67,6 @@ pub(crate) enum Batch {
         op: IndexOperation,
         must_create_index: bool,
     },
-    IndexDocumentDeletionByFilter {
-        index_uid: String,
-        task: Task,
-    },
     IndexCreation {
         index_uid: String,
         primary_key: Option<String>,
@@ -114,6 +110,10 @@ pub(crate) enum IndexOperation {
         documents: Vec<Vec<String>>,
         tasks: Vec<Task>,
     },
+    IndexDocumentDeletionByFilter {
+        index_uid: String,
+        task: Task,
+    },
     DocumentClear {
         index_uid: String,
         tasks: Vec<Task>,
@@ -155,7 +155,6 @@ impl Batch {
             | Batch::TaskDeletion(task)
             | Batch::Dump(task)
             | Batch::IndexCreation { task, .. }
-            | Batch::IndexDocumentDeletionByFilter { task, .. }
             | Batch::IndexUpdate { task, .. } => vec![task.uid],
             Batch::SnapshotCreation(tasks) | Batch::IndexDeletion { tasks, .. } => {
                 tasks.iter().map(|task| task.uid).collect()
@@ -167,6 +166,7 @@ impl Batch {
                 | IndexOperation::DocumentClear { tasks, .. } => {
                     tasks.iter().map(|task| task.uid).collect()
                 }
+                IndexOperation::IndexDocumentDeletionByFilter { task, .. } => vec![task.uid],
                 IndexOperation::SettingsAndDocumentOperation {
                     document_import_tasks: tasks,
                     settings_tasks: other,
@@ -194,8 +194,7 @@ impl Batch {
             IndexOperation { op, .. } => Some(op.index_uid()),
             IndexCreation { index_uid, .. }
             | IndexUpdate { index_uid, .. }
-            | IndexDeletion { index_uid, .. }
-            | IndexDocumentDeletionByFilter { index_uid, .. } => Some(index_uid),
+            | IndexDeletion { index_uid, .. } => Some(index_uid),
         }
     }
 }
@@ -205,6 +204,7 @@ impl IndexOperation {
         match self {
             IndexOperation::DocumentOperation { index_uid, .. }
             | IndexOperation::DocumentDeletion { index_uid, .. }
+            | IndexOperation::IndexDocumentDeletionByFilter { index_uid, .. }
             | IndexOperation::DocumentClear { index_uid, .. }
             | IndexOperation::Settings { index_uid, .. }
             | IndexOperation::DocumentClearAndSetting { index_uid, .. }
@@ -239,9 +239,12 @@ impl IndexScheduler {
                 let task = self.get_task(rtxn, id)?.ok_or(Error::CorruptedTaskQueue)?;
                 match &task.kind {
                     KindWithContent::DocumentDeletionByFilter { index_uid, .. } => {
-                        Ok(Some(Batch::IndexDocumentDeletionByFilter {
-                            index_uid: index_uid.clone(),
-                            task,
+                        Ok(Some(Batch::IndexOperation {
+                            op: IndexOperation::IndexDocumentDeletionByFilter {
+                                index_uid: index_uid.clone(),
+                                task,
+                            },
+                            must_create_index: false,
                         }))
                     }
                     _ => unreachable!(),
@@ -536,7 +539,9 @@ impl IndexScheduler {
         let index_tasks = self.index_tasks(rtxn, index_name)? & enqueued;
 
         // If autobatching is disabled we only take one task at a time.
-        let tasks_limit = if self.autobatching_enabled { usize::MAX } else { 1 };
+        // Otherwise, we take only a maximum of tasks to create batches.
+        let tasks_limit =
+            if self.autobatching_enabled { self.maximum_number_of_batched_tasks } else { 1 };
 
         let enqueued = index_tasks
             .into_iter()
@@ -896,51 +901,6 @@ impl IndexScheduler {
 
                 Ok(tasks)
             }
-            Batch::IndexDocumentDeletionByFilter { mut task, index_uid: _ } => {
-                let (index_uid, filter) =
-                    if let KindWithContent::DocumentDeletionByFilter { index_uid, filter_expr } =
-                        &task.kind
-                    {
-                        (index_uid, filter_expr)
-                    } else {
-                        unreachable!()
-                    };
-                let index = {
-                    let rtxn = self.env.read_txn()?;
-                    self.index_mapper.index(&rtxn, index_uid)?
-                };
-                let deleted_documents = delete_document_by_filter(filter, index);
-                let original_filter = if let Some(Details::DocumentDeletionByFilter {
-                    original_filter,
-                    deleted_documents: _,
-                }) = task.details
-                {
-                    original_filter
-                } else {
-                    // In the case of a `documentDeleteByFilter` the details MUST be set
-                    unreachable!();
-                };
-
-                match deleted_documents {
-                    Ok(deleted_documents) => {
-                        task.status = Status::Succeeded;
-                        task.details = Some(Details::DocumentDeletionByFilter {
-                            original_filter,
-                            deleted_documents: Some(deleted_documents),
-                        });
-                    }
-                    Err(e) => {
-                        task.status = Status::Failed;
-                        task.details = Some(Details::DocumentDeletionByFilter {
-                            original_filter,
-                            deleted_documents: Some(0),
-                        });
-                        task.error = Some(e.into());
-                    }
-                }
-
-                Ok(vec![task])
-            }
             Batch::IndexCreation { index_uid, primary_key, task } => {
                 let wtxn = self.env.write_txn()?;
                 if self.index_mapper.exists(&wtxn, &index_uid)? {
@@ -1299,6 +1259,47 @@ impl IndexScheduler {
 
                 Ok(tasks)
             }
+            IndexOperation::IndexDocumentDeletionByFilter { mut task, index_uid: _ } => {
+                let filter =
+                    if let KindWithContent::DocumentDeletionByFilter { filter_expr, .. } =
+                        &task.kind
+                    {
+                        filter_expr
+                    } else {
+                        unreachable!()
+                    };
+                let deleted_documents = delete_document_by_filter(index_wtxn, filter, index);
+                let original_filter = if let Some(Details::DocumentDeletionByFilter {
+                    original_filter,
+                    deleted_documents: _,
+                }) = task.details
+                {
+                    original_filter
+                } else {
+                    // In the case of a `documentDeleteByFilter` the details MUST be set
+                    unreachable!();
+                };
+
+                match deleted_documents {
+                    Ok(deleted_documents) => {
+                        task.status = Status::Succeeded;
+                        task.details = Some(Details::DocumentDeletionByFilter {
+                            original_filter,
+                            deleted_documents: Some(deleted_documents),
+                        });
+                    }
+                    Err(e) => {
+                        task.status = Status::Failed;
+                        task.details = Some(Details::DocumentDeletionByFilter {
+                            original_filter,
+                            deleted_documents: Some(0),
+                        });
+                        task.error = Some(e.into());
+                    }
+                }
+
+                Ok(vec![task])
+            }
             IndexOperation::Settings { index_uid: _, settings, mut tasks } => {
                 let indexer_config = self.index_mapper.indexer_config();
                 let mut builder = milli::update::Settings::new(index_wtxn, index, indexer_config);
@@ -1498,23 +1499,22 @@ impl IndexScheduler {
     }
 }
 
-fn delete_document_by_filter(filter: &serde_json::Value, index: Index) -> Result<u64> {
+fn delete_document_by_filter<'a>(
+    wtxn: &mut RwTxn<'a, '_>,
+    filter: &serde_json::Value,
+    index: &'a Index,
+) -> Result<u64> {
     let filter = Filter::from_json(filter)?;
     Ok(if let Some(filter) = filter {
-        let mut wtxn = index.write_txn()?;
-
-        let candidates = filter.evaluate(&wtxn, &index).map_err(|err| match err {
+        let candidates = filter.evaluate(wtxn, index).map_err(|err| match err {
             milli::Error::UserError(milli::UserError::InvalidFilter(_)) => {
                 Error::from(err).with_custom_error_code(Code::InvalidDocumentFilter)
             }
             e => e.into(),
         })?;
-        let mut delete_operation = DeleteDocuments::new(&mut wtxn, &index)?;
+        let mut delete_operation = DeleteDocuments::new(wtxn, index)?;
         delete_operation.delete_documents(&candidates);
-        let deleted_documents =
-            delete_operation.execute().map(|result| result.deleted_documents)?;
-        wtxn.commit()?;
-        deleted_documents
+        delete_operation.execute().map(|result| result.deleted_documents)?
     } else {
         0
     })

index-scheduler/src/insta_snapshot.rs

@@ -15,6 +15,7 @@ pub fn snapshot_index_scheduler(scheduler: &IndexScheduler) -> String {
 
     let IndexScheduler {
         autobatching_enabled,
+        maximum_number_of_batched_tasks: _,
         must_stop_processing: _,
         processing_tasks,
         file_store,

index-scheduler/src/lib.rs

@@ -253,6 +253,9 @@ pub struct IndexSchedulerOptions {
     /// Set to `true` iff the index scheduler is allowed to automatically
     /// batch tasks together, to process multiple tasks at once.
     pub autobatching_enabled: bool,
+    /// If the autobatcher is allowed to automatically batch tasks
+    /// it will only batch this defined number of tasks at once.
+    pub maximum_number_of_batched_tasks: usize,
     /// The maximum number of tasks stored in the task queue before starting
     /// to auto schedule task deletions.
     pub max_number_of_tasks: usize,
@@ -310,6 +313,9 @@ pub struct IndexScheduler {
     /// Whether auto-batching is enabled or not.
     pub(crate) autobatching_enabled: bool,
 
+    /// The maximum number of tasks that will be batched together.
+    pub(crate) maximum_number_of_batched_tasks: usize,
+
     /// The max number of tasks allowed before the scheduler starts to delete
     /// the finished tasks automatically.
     pub(crate) max_number_of_tasks: usize,
@@ -363,6 +369,7 @@ impl IndexScheduler {
             index_mapper: self.index_mapper.clone(),
             wake_up: self.wake_up.clone(),
             autobatching_enabled: self.autobatching_enabled,
+            maximum_number_of_batched_tasks: self.maximum_number_of_batched_tasks,
             max_number_of_tasks: self.max_number_of_tasks,
             snapshots_path: self.snapshots_path.clone(),
             dumps_path: self.dumps_path.clone(),
@@ -458,6 +465,7 @@ impl IndexScheduler {
             // we want to start the loop right away in case meilisearch was ctrl+Ced while processing things
             wake_up: Arc::new(SignalEvent::auto(true)),
             autobatching_enabled: options.autobatching_enabled,
+            maximum_number_of_batched_tasks: options.maximum_number_of_batched_tasks,
             max_number_of_tasks: options.max_number_of_tasks,
             dumps_path: options.dumps_path,
             snapshots_path: options.snapshots_path,
@@ -790,10 +798,19 @@ impl IndexScheduler {
 
         let mut res = BTreeMap::new();
 
+        let processing_tasks = { self.processing_tasks.read().unwrap().processing.len() };
+
         res.insert(
             "statuses".to_string(),
             enum_iterator::all::<Status>()
-                .map(|s| Ok((s.to_string(), self.get_status(&rtxn, s)?.len())))
+                .map(|s| {
+                    let tasks = self.get_status(&rtxn, s)?.len();
+                    match s {
+                        Status::Enqueued => Ok((s.to_string(), tasks - processing_tasks)),
+                        Status::Processing => Ok((s.to_string(), processing_tasks)),
+                        s => Ok((s.to_string(), tasks)),
+                    }
+                })
                 .collect::<Result<BTreeMap<String, u64>>>()?,
         );
         res.insert(
@@ -1580,6 +1597,7 @@ mod tests {
                 index_count: 5,
                 indexer_config,
                 autobatching_enabled: true,
+                maximum_number_of_batched_tasks: usize::MAX,
                 max_number_of_tasks: 1_000_000,
                 instance_features: Default::default(),
             };
@@ -4131,4 +4149,154 @@ mod tests {
         snapshot!(json_string!(tasks, { "[].enqueuedAt" => "[date]", "[].startedAt" => "[date]", "[].finishedAt" => "[date]", ".**.original_filter" => "[filter]", ".**.query" => "[query]" }), name: "everything_has_been_processed");
         drop(rtxn);
     }
+
+    #[test]
+    fn basic_get_stats() {
+        let (index_scheduler, mut handle) = IndexScheduler::test(true, vec![]);
+
+        let kind = index_creation_task("catto", "mouse");
+        let _task = index_scheduler.register(kind).unwrap();
+        let kind = index_creation_task("doggo", "sheep");
+        let _task = index_scheduler.register(kind).unwrap();
+        let kind = index_creation_task("whalo", "fish");
+        let _task = index_scheduler.register(kind).unwrap();
+
+        snapshot!(json_string!(index_scheduler.get_stats().unwrap()), @r###"
+        {
+          "indexes": {
+            "catto": 1,
+            "doggo": 1,
+            "whalo": 1
+          },
+          "statuses": {
+            "canceled": 0,
+            "enqueued": 3,
+            "failed": 0,
+            "processing": 0,
+            "succeeded": 0
+          },
+          "types": {
+            "documentAdditionOrUpdate": 0,
+            "documentDeletion": 0,
+            "dumpCreation": 0,
+            "indexCreation": 3,
+            "indexDeletion": 0,
+            "indexSwap": 0,
+            "indexUpdate": 0,
+            "settingsUpdate": 0,
+            "snapshotCreation": 0,
+            "taskCancelation": 0,
+            "taskDeletion": 0
+          }
+        }
+        "###);
+
+        handle.advance_till([Start, BatchCreated]);
+        snapshot!(json_string!(index_scheduler.get_stats().unwrap()), @r###"
+        {
+          "indexes": {
+            "catto": 1,
+            "doggo": 1,
+            "whalo": 1
+          },
+          "statuses": {
+            "canceled": 0,
+            "enqueued": 2,
+            "failed": 0,
+            "processing": 1,
+            "succeeded": 0
+          },
+          "types": {
+            "documentAdditionOrUpdate": 0,
+            "documentDeletion": 0,
+            "dumpCreation": 0,
+            "indexCreation": 3,
+            "indexDeletion": 0,
+            "indexSwap": 0,
+            "indexUpdate": 0,
+            "settingsUpdate": 0,
+            "snapshotCreation": 0,
+            "taskCancelation": 0,
+            "taskDeletion": 0
+          }
+        }
+        "###);
+
+        handle.advance_till([
+            InsideProcessBatch,
+            InsideProcessBatch,
+            ProcessBatchSucceeded,
+            AfterProcessing,
+            Start,
+            BatchCreated,
+        ]);
+        snapshot!(json_string!(index_scheduler.get_stats().unwrap()), @r###"
+        {
+          "indexes": {
+            "catto": 1,
+            "doggo": 1,
+            "whalo": 1
+          },
+          "statuses": {
+            "canceled": 0,
+            "enqueued": 1,
+            "failed": 0,
+            "processing": 1,
+            "succeeded": 1
+          },
+          "types": {
+            "documentAdditionOrUpdate": 0,
+            "documentDeletion": 0,
+            "dumpCreation": 0,
+            "indexCreation": 3,
+            "indexDeletion": 0,
+            "indexSwap": 0,
+            "indexUpdate": 0,
+            "settingsUpdate": 0,
+            "snapshotCreation": 0,
+            "taskCancelation": 0,
+            "taskDeletion": 0
+          }
+        }
+        "###);
+
+        // now we make one more batch, the started_at field of the new tasks will be past `second_start_time`
+        handle.advance_till([
+            InsideProcessBatch,
+            InsideProcessBatch,
+            ProcessBatchSucceeded,
+            AfterProcessing,
+            Start,
+            BatchCreated,
+        ]);
+        snapshot!(json_string!(index_scheduler.get_stats().unwrap()), @r###"
+        {
+          "indexes": {
+            "catto": 1,
+            "doggo": 1,
+            "whalo": 1
+          },
+          "statuses": {
+            "canceled": 0,
+            "enqueued": 0,
+            "failed": 0,
+            "processing": 1,
+            "succeeded": 2
+          },
+          "types": {
+            "documentAdditionOrUpdate": 0,
+            "documentDeletion": 0,
+            "dumpCreation": 0,
+            "indexCreation": 3,
+            "indexDeletion": 0,
+            "indexSwap": 0,
+            "indexUpdate": 0,
+            "settingsUpdate": 0,
+            "snapshotCreation": 0,
+            "taskCancelation": 0,
+            "taskDeletion": 0
+          }
+        }
+        "###);
+    }
 }

meili-snap/src/lib.rs

@@ -167,7 +167,9 @@ macro_rules! snapshot {
         let (settings, snap_name, _) = $crate::default_snapshot_settings_for_test(test_name, Some(&snap_name));
         settings.bind(|| {
             let snap = format!("{}", $value);
-            meili_snap::insta::assert_snapshot!(format!("{}", snap_name), snap);
+            insta::allow_duplicates! {
+                meili_snap::insta::assert_snapshot!(format!("{}", snap_name), snap);
+            }
         });
     };
     ($value:expr, @$inline:literal) => {
@@ -176,7 +178,9 @@ macro_rules! snapshot {
         let (settings, _, _) = $crate::default_snapshot_settings_for_test("", Some("_dummy_argument"));
         settings.bind(|| {
             let snap = format!("{}", $value);
-            meili_snap::insta::assert_snapshot!(snap, @$inline);
+            insta::allow_duplicates! {
+                meili_snap::insta::assert_snapshot!(snap, @$inline);
+            }
         });
     };
     ($value:expr) => {
@@ -194,7 +198,9 @@ macro_rules! snapshot {
         let (settings, snap_name, _) = $crate::default_snapshot_settings_for_test(test_name, None);
         settings.bind(|| {
             let snap = format!("{}", $value);
-            meili_snap::insta::assert_snapshot!(format!("{}", snap_name), snap);
+            insta::allow_duplicates! {
+                meili_snap::insta::assert_snapshot!(format!("{}", snap_name), snap);
+            }
         });
     };
 }

meilisearch-types/Cargo.toml

@@ -15,13 +15,13 @@ actix-web = { version = "4.3.1", default-features = false }
 anyhow = "1.0.70"
 convert_case = "0.6.0"
 csv = "1.2.1"
-deserr = "0.5.0"
+deserr = { version = "0.6.0", features = ["actix-web"]}
 either = { version = "1.8.1", features = ["serde"] }
 enum-iterator = "1.4.0"
 file-store = { path = "../file-store" }
 flate2 = "1.0.25"
 fst = "0.4.7"
-memmap2 = "0.5.10"
+memmap2 = "0.7.1"
 milli = { path = "../milli" }
 roaring = { version = "0.10.1", features = ["serde"] }
 serde = { version = "1.0.160", features = ["derive"] }

meilisearch-types/src/document_formats.rs

@@ -1,4 +1,3 @@
-use std::borrow::Borrow;
 use std::fmt::{self, Debug, Display};
 use std::fs::File;
 use std::io::{self, Seek, Write};
@@ -42,7 +41,7 @@ impl Display for DocumentFormatError {
     fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
         match self {
             Self::Io(e) => write!(f, "{e}"),
-            Self::MalformedPayload(me, b) => match me.borrow() {
+            Self::MalformedPayload(me, b) => match me {
                 Error::Json(se) => {
                     let mut message = match se.classify() {
                         Category::Data => {

meilisearch/Cargo.toml

@@ -39,7 +39,7 @@ byte-unit = { version = "4.0.19", default-features = false, features = [
 bytes = "1.4.0"
 clap = { version = "4.2.1", features = ["derive", "env"] }
 crossbeam-channel = "0.5.8"
-deserr = "0.5.0"
+deserr = { version = "0.6.0", features = ["actix-web"]}
 dump = { path = "../dump" }
 either = "1.8.1"
 env_logger = "0.10.0"
@@ -50,9 +50,9 @@ futures = "0.3.28"
 futures-util = "0.3.28"
 http = "0.2.9"
 index-scheduler = { path = "../index-scheduler" }
-indexmap = { version = "1.9.3", features = ["serde-1"] }
+indexmap = { version = "2.0.0", features = ["serde"] }
 is-terminal = "0.4.8"
-itertools = "0.10.5"
+itertools = "0.11.0"
 jsonwebtoken = "8.3.0"
 lazy_static = "1.4.0"
 log = "0.4.17"
@@ -87,7 +87,7 @@ sha2 = "0.10.6"
 siphasher = "0.3.10"
 slice-group-by = "0.3.0"
 static-files = { version = "0.2.3", optional = true }
-sysinfo = "0.28.4"
+sysinfo = "0.29.7"
 tar = "0.4.38"
 tempfile = "3.5.0"
 thiserror = "1.0.40"
@@ -133,7 +133,7 @@ vergen = { version = "7.5.1", default-features = false, features = ["git"] }
 zip = { version = "0.6.4", optional = true }
 
 [features]
-default = ["analytics", "meilisearch-types/all-tokenizations", "mini-dashboard"]
+default = ["analytics", "meilisearch-types/all-tokenizations", "mini-dashboard", "profile-with-puffin"]
 analytics = ["segment"]
 profile-with-puffin = ["dep:puffin_http"]
 mini-dashboard = [

meilisearch/src/analytics/mock_analytics.rs

@@ -20,7 +20,7 @@ pub struct SearchAggregator;
 #[allow(dead_code)]
 impl SearchAggregator {
     pub fn from_query(_: &dyn Any, _: &dyn Any) -> Self {
-        Self::default()
+        Self
     }
 
     pub fn succeed(&mut self, _: &dyn Any) {}
@@ -32,7 +32,7 @@ pub struct MultiSearchAggregator;
 #[allow(dead_code)]
 impl MultiSearchAggregator {
     pub fn from_queries(_: &dyn Any, _: &dyn Any) -> Self {
-        Self::default()
+        Self
     }
 
     pub fn succeed(&mut self) {}
@@ -44,7 +44,7 @@ pub struct FacetSearchAggregator;
 #[allow(dead_code)]
 impl FacetSearchAggregator {
     pub fn from_query(_: &dyn Any, _: &dyn Any) -> Self {
-        Self::default()
+        Self
     }
 
     pub fn succeed(&mut self, _: &dyn Any) {}

meilisearch/src/analytics/segment_analytics.rs

@@ -285,6 +285,7 @@ impl From<Opt> for Infos {
             db_path,
             experimental_enable_metrics,
             experimental_reduce_indexing_memory_usage,
+            experimental_limit_batched_tasks: _,
             http_addr,
             master_key: _,
             env,

meilisearch/src/lib.rs

@@ -236,6 +236,7 @@ fn open_or_create_database_unchecked(
             enable_mdb_writemap: opt.experimental_reduce_indexing_memory_usage,
             indexer_config: (&opt.indexer_options).try_into()?,
             autobatching_enabled: true,
+            maximum_number_of_batched_tasks: opt.experimental_limit_batched_tasks,
             max_number_of_tasks: 1_000_000,
             index_growth_amount: byte_unit::Byte::from_str("10GiB").unwrap().get_bytes() as usize,
             index_count: DEFAULT_INDEX_COUNT,

meilisearch/src/option.rs

@@ -51,6 +51,7 @@ const MEILI_LOG_LEVEL: &str = "MEILI_LOG_LEVEL";
 const MEILI_EXPERIMENTAL_ENABLE_METRICS: &str = "MEILI_EXPERIMENTAL_ENABLE_METRICS";
 const MEILI_EXPERIMENTAL_REDUCE_INDEXING_MEMORY_USAGE: &str =
     "MEILI_EXPERIMENTAL_REDUCE_INDEXING_MEMORY_USAGE";
+const MEILI_EXPERIMENTAL_LIMIT_BATCHED_TASKS: &str = "MEILI_EXPERIMENTAL_LIMIT_BATCHED_TASKS";
 
 const DEFAULT_CONFIG_FILE_PATH: &str = "./config.toml";
 const DEFAULT_DB_PATH: &str = "./data.ms";
@@ -301,6 +302,11 @@ pub struct Opt {
     #[serde(default)]
     pub experimental_reduce_indexing_memory_usage: bool,
 
+    /// Experimental limit to the number of tasks per batch
+    #[clap(long, env = MEILI_EXPERIMENTAL_LIMIT_BATCHED_TASKS, default_value_t = default_limit_batched_tasks())]
+    #[serde(default = "default_limit_batched_tasks")]
+    pub experimental_limit_batched_tasks: usize,
+
     #[serde(flatten)]
     #[clap(flatten)]
     pub indexer_options: IndexerOpts,
@@ -393,7 +399,8 @@ impl Opt {
             #[cfg(all(not(debug_assertions), feature = "analytics"))]
             no_analytics,
             experimental_enable_metrics: enable_metrics_route,
-            experimental_reduce_indexing_memory_usage: reduce_indexing_memory_usage,
+            experimental_reduce_indexing_memory_usage,
+            experimental_limit_batched_tasks,
         } = self;
         export_to_env_if_not_present(MEILI_DB_PATH, db_path);
         export_to_env_if_not_present(MEILI_HTTP_ADDR, http_addr);
@@ -437,7 +444,11 @@ impl Opt {
         );
         export_to_env_if_not_present(
             MEILI_EXPERIMENTAL_REDUCE_INDEXING_MEMORY_USAGE,
-            reduce_indexing_memory_usage.to_string(),
+            experimental_reduce_indexing_memory_usage.to_string(),
+        );
+        export_to_env_if_not_present(
+            MEILI_EXPERIMENTAL_LIMIT_BATCHED_TASKS,
+            experimental_limit_batched_tasks.to_string(),
         );
         indexer_options.export_to_env();
     }
@@ -739,6 +750,10 @@ fn default_dump_dir() -> PathBuf {
     PathBuf::from(DEFAULT_DUMP_DIR)
 }
 
+fn default_limit_batched_tasks() -> usize {
+    usize::MAX
+}
+
 /// Indicates if a snapshot was scheduled, and if yes with which interval.
 #[derive(Debug, Default, Copy, Clone, Deserialize, Serialize)]
 pub enum ScheduleSnapshot {

meilisearch/src/routes/swap_indexes.rs

@@ -60,8 +60,7 @@ pub async fn swap_indexes(
     }
 
     let task = KindWithContent::IndexSwap { swaps };
-
-    let task = index_scheduler.register(task)?;
-    let task: SummarizedTaskView = task.into();
+    let task: SummarizedTaskView =
+        tokio::task::spawn_blocking(move || index_scheduler.register(task)).await??.into();
     Ok(HttpResponse::Accepted().json(task))
 }

meilisearch/src/search.rs

@@ -680,6 +680,7 @@ fn compute_semantic_score(query: &[f32], vectors: Value) -> milli::Result<Option
         .map_err(InternalError::SerdeJson)?;
     Ok(vectors
         .into_iter()
+        .flatten()
         .map(|v| OrderedFloat(dot_product_similarity(query, &v)))
         .max()
         .map(OrderedFloat::into_inner))

meilisearch/tests/documents/delete_documents.rs

@@ -154,6 +154,19 @@ async fn delete_document_by_filter() {
         )
         .await;
     index.wait_task(1).await;
+
+    let (stats, _) = index.stats().await;
+    snapshot!(json_string!(stats), @r###"
+    {
+      "numberOfDocuments": 4,
+      "isIndexing": false,
+      "fieldDistribution": {
+        "color": 3,
+        "id": 4
+      }
+    }
+    "###);
+
     let (response, code) =
         index.delete_document_by_filter(json!({ "filter": "color = blue"})).await;
     snapshot!(code, @"202 Accepted");
@@ -188,6 +201,18 @@ async fn delete_document_by_filter() {
     }
     "###);
 
+    let (stats, _) = index.stats().await;
+    snapshot!(json_string!(stats), @r###"
+    {
+      "numberOfDocuments": 2,
+      "isIndexing": false,
+      "fieldDistribution": {
+        "color": 1,
+        "id": 2
+      }
+    }
+    "###);
+
     let (documents, code) = index.get_all_documents(GetAllDocumentsOptions::default()).await;
     snapshot!(code, @"200 OK");
     snapshot!(json_string!(documents), @r###"
@@ -241,6 +266,18 @@ async fn delete_document_by_filter() {
     }
     "###);
 
+    let (stats, _) = index.stats().await;
+    snapshot!(json_string!(stats), @r###"
+    {
+      "numberOfDocuments": 1,
+      "isIndexing": false,
+      "fieldDistribution": {
+        "color": 1,
+        "id": 1
+      }
+    }
+    "###);
+
     let (documents, code) = index.get_all_documents(GetAllDocumentsOptions::default()).await;
     snapshot!(code, @"200 OK");
     snapshot!(json_string!(documents), @r###"

meilisearch/tests/search/geo.rs

@@ -1,3 +1,4 @@
+use meili_snap::{json_string, snapshot};
 use once_cell::sync::Lazy;
 use serde_json::{json, Value};
 
@@ -60,3 +61,59 @@ async fn geo_sort_with_geo_strings() {
         )
         .await;
 }
+
+#[actix_rt::test]
+async fn geo_bounding_box_with_string_and_number() {
+    let server = Server::new().await;
+    let index = server.index("test");
+
+    let documents = DOCUMENTS.clone();
+    index.update_settings_filterable_attributes(json!(["_geo"])).await;
+    index.update_settings_sortable_attributes(json!(["_geo"])).await;
+    index.add_documents(documents, None).await;
+    index.wait_task(2).await;
+
+    index
+        .search(
+            json!({
+                "filter": "_geoBoundingBox([89, 179], [-89, -179])",
+            }),
+            |response, code| {
+                assert_eq!(code, 200, "{}", response);
+                snapshot!(json_string!(response, { ".processingTimeMs" => "[time]" }), @r###"
+                {
+                  "hits": [
+                    {
+                      "id": 1,
+                      "name": "Taco Truck",
+                      "address": "444 Salsa Street, Burritoville",
+                      "type": "Mexican",
+                      "rating": 9,
+                      "_geo": {
+                        "lat": 34.0522,
+                        "lng": -118.2437
+                      }
+                    },
+                    {
+                      "id": 2,
+                      "name": "La Bella Italia",
+                      "address": "456 Elm Street, Townsville",
+                      "type": "Italian",
+                      "rating": 9,
+                      "_geo": {
+                        "lat": "45.4777599",
+                        "lng": "9.1967508"
+                      }
+                    }
+                  ],
+                  "query": "",
+                  "processingTimeMs": "[time]",
+                  "limit": 20,
+                  "offset": 0,
+                  "estimatedTotalHits": 2
+                }
+                "###);
+            },
+        )
+        .await;
+}

meilisearch/tests/search/mod.rs

@@ -1104,3 +1104,59 @@ async fn camelcased_words() {
         })
         .await;
 }
+
+#[actix_rt::test]
+async fn simple_search_with_strange_synonyms() {
+    let server = Server::new().await;
+    let index = server.index("test");
+
+    index.update_settings(json!({ "synonyms": {"&": ["to"], "to": ["&"]} })).await;
+    let r = index.wait_task(0).await;
+    meili_snap::snapshot!(r["status"], @r###""succeeded""###);
+
+    let documents = DOCUMENTS.clone();
+    index.add_documents(documents, None).await;
+    index.wait_task(1).await;
+
+    index
+        .search(json!({"q": "How to train"}), |response, code| {
+            meili_snap::snapshot!(code, @"200 OK");
+            meili_snap::snapshot!(meili_snap::json_string!(response["hits"]), @r###"
+            [
+              {
+                "title": "How to Train Your Dragon: The Hidden World",
+                "id": "166428"
+              }
+            ]
+            "###);
+        })
+        .await;
+
+    index
+        .search(json!({"q": "How & train"}), |response, code| {
+            meili_snap::snapshot!(code, @"200 OK");
+            meili_snap::snapshot!(meili_snap::json_string!(response["hits"]), @r###"
+            [
+              {
+                "title": "How to Train Your Dragon: The Hidden World",
+                "id": "166428"
+              }
+            ]
+            "###);
+        })
+        .await;
+
+    index
+        .search(json!({"q": "to"}), |response, code| {
+            meili_snap::snapshot!(code, @"200 OK");
+            meili_snap::snapshot!(meili_snap::json_string!(response["hits"]), @r###"
+            [
+              {
+                "title": "How to Train Your Dragon: The Hidden World",
+                "id": "166428"
+              }
+            ]
+            "###);
+        })
+        .await;
+}

milli/Cargo.toml

@@ -17,10 +17,11 @@ bincode = "1.3.3"
 bstr = "1.4.0"
 bytemuck = { version = "1.13.1", features = ["extern_crate_alloc"] }
 byteorder = "1.4.3"
-charabia = { version = "0.8.2", default-features = false }
+# charabia = { version = "0.8.3", default-features = false }
+charabia = { git = "https://github.com/meilisearch/charabia", branch = "main", default-features = false }
 concat-arrays = "0.1.2"
 crossbeam-channel = "0.5.8"
-deserr = "0.5.0"
+deserr = { version = "0.6.0", features = ["actix-web"]}
 either = { version = "1.8.1", features = ["serde"] }
 flatten-serde-json = { path = "../flatten-serde-json" }
 fst = "0.4.7"
@@ -32,18 +33,18 @@ grenad = { version = "0.4.4", default-features = false, features = [
 heed = { git = "https://github.com/meilisearch/heed", tag = "v0.12.7", default-features = false, features = [
     "lmdb", "read-txn-no-tls"
 ] }
-indexmap = { version = "1.9.3", features = ["serde"] }
+indexmap = { version = "2.0.0", features = ["serde"] }
 instant-distance = { version = "0.6.1", features = ["with-serde"] }
 json-depth-checker = { path = "../json-depth-checker" }
 levenshtein_automata = { version = "0.2.1", features = ["fst_automaton"] }
-memmap2 = "0.5.10"
+memmap2 = "0.7.1"
 obkv = "0.2.0"
 once_cell = "1.17.1"
 ordered-float = "3.6.0"
 rand_pcg = { version = "0.3.1", features = ["serde1"] }
 rayon = "1.7.0"
 roaring = "0.10.1"
-rstar = { version = "0.10.0", features = ["serde"] }
+rstar = { version = "0.11.0", features = ["serde"] }
 serde = { version = "1.0.160", features = ["derive"] }
 serde_json = { version = "1.0.95", features = ["preserve_order"] }
 slice-group-by = "0.3.0"
@@ -63,7 +64,7 @@ uuid = { version = "1.3.1", features = ["v4"] }
 filter-parser = { path = "../filter-parser" }
 
 # documents words self-join
-itertools = "0.10.5"
+itertools = "0.11.0"
 
 # profiling
 puffin = "0.16.0"

milli/src/error.rs

@@ -122,22 +122,28 @@ only composed of alphanumeric characters (a-z A-Z 0-9), hyphens (-) and undersco
         .field,
         match .valid_fields.is_empty() {
             true => "This index does not have configured sortable attributes.".to_string(),
-            false => format!("Available sortable attributes are: `{}`.",
-                    valid_fields.iter().map(AsRef::as_ref).collect::<Vec<&str>>().join(", ")
+            false => format!("Available sortable attributes are: `{}{}`.",
+                    valid_fields.iter().map(AsRef::as_ref).collect::<Vec<&str>>().join(", "),
+                    .hidden_fields.then_some(", <..hidden-attributes>").unwrap_or(""),
                 ),
         }
     )]
-    InvalidSortableAttribute { field: String, valid_fields: BTreeSet<String> },
+    InvalidSortableAttribute { field: String, valid_fields: BTreeSet<String>, hidden_fields: bool },
     #[error("Attribute `{}` is not facet-searchable. {}",
         .field,
         match .valid_fields.is_empty() {
             true => "This index does not have configured facet-searchable attributes. To make it facet-searchable add it to the `filterableAttributes` index settings.".to_string(),
-            false => format!("Available facet-searchable attributes are: `{}`. To make it facet-searchable add it to the `filterableAttributes` index settings.",
-                    valid_fields.iter().map(AsRef::as_ref).collect::<Vec<&str>>().join(", ")
+            false => format!("Available facet-searchable attributes are: `{}{}`. To make it facet-searchable add it to the `filterableAttributes` index settings.",
+                    valid_fields.iter().map(AsRef::as_ref).collect::<Vec<&str>>().join(", "),
+                    .hidden_fields.then_some(", <..hidden-attributes>").unwrap_or(""),
                 ),
         }
     )]
-    InvalidFacetSearchFacetName { field: String, valid_fields: BTreeSet<String> },
+    InvalidFacetSearchFacetName {
+        field: String,
+        valid_fields: BTreeSet<String>,
+        hidden_fields: bool,
+    },
     #[error("Attribute `{}` is not searchable. Available searchable attributes are: `{}{}`.",
         .field,
         .valid_fields.iter().map(AsRef::as_ref).collect::<Vec<&str>>().join(", "),
@@ -340,8 +346,11 @@ fn conditionally_lookup_for_error_message() {
     ];
 
     for (list, suffix) in messages {
-        let err =
-            UserError::InvalidSortableAttribute { field: "name".to_string(), valid_fields: list };
+        let err = UserError::InvalidSortableAttribute {
+            field: "name".to_string(),
+            valid_fields: list,
+            hidden_fields: false,
+        };
 
         assert_eq!(err.to_string(), format!("{} {}", prefix, suffix));
     }

milli/src/index.rs

@@ -655,6 +655,26 @@ impl Index {
         }
     }
 
+    /* remove hidden fields */
+    pub fn remove_hidden_fields(
+        &self,
+        rtxn: &RoTxn,
+        fields: impl IntoIterator<Item = impl AsRef<str>>,
+    ) -> Result<(BTreeSet<String>, bool)> {
+        let mut valid_fields =
+            fields.into_iter().map(|f| f.as_ref().to_string()).collect::<BTreeSet<String>>();
+
+        let fields_len = valid_fields.len();
+
+        if let Some(dn) = self.displayed_fields(rtxn)? {
+            let displayable_names = dn.iter().map(|s| s.to_string()).collect();
+            valid_fields = &valid_fields & &displayable_names;
+        }
+
+        let hidden_fields = fields_len > valid_fields.len();
+        Ok((valid_fields, hidden_fields))
+    }
+
     /* searchable fields */
 
     /// Write the user defined searchable fields and generate the real searchable fields from the specified fields ids map.
@@ -1820,11 +1840,11 @@ pub(crate) mod tests {
             .unwrap();
         index
             .add_documents(documents!([
-                { "id": 0, "_geo": { "lat": 0, "lng": 0 } },
-                { "id": 1, "_geo": { "lat": 0, "lng": -175 } },
-                { "id": 2, "_geo": { "lat": 0, "lng": 175 } },
+                { "id": 0, "_geo": { "lat": "0", "lng": "0" } },
+                { "id": 1, "_geo": { "lat": 0, "lng": "-175" } },
+                { "id": 2, "_geo": { "lat": "0", "lng": 175 } },
                 { "id": 3, "_geo": { "lat": 85, "lng": 0 } },
-                { "id": 4, "_geo": { "lat": -85, "lng": 0 } },
+                { "id": 4, "_geo": { "lat": "-85", "lng": "0" } },
             ]))
             .unwrap();
 

milli/src/lib.rs

@@ -97,7 +97,7 @@ const MAX_LMDB_KEY_LENGTH: usize = 500;
 ///
 /// This number is determined by the keys of the different facet databases
 /// and adding a margin of safety.
-pub const MAX_FACET_VALUE_LENGTH: usize = MAX_LMDB_KEY_LENGTH - 20;
+pub const MAX_FACET_VALUE_LENGTH: usize = MAX_LMDB_KEY_LENGTH - 32;
 
 /// The maximum length a word can be
 pub const MAX_WORD_LENGTH: usize = MAX_LMDB_KEY_LENGTH / 2;
@@ -293,15 +293,15 @@ pub fn normalize_facet(original: &str) -> String {
 #[derive(serde::Serialize, serde::Deserialize, Debug)]
 #[serde(transparent)]
 pub struct VectorOrArrayOfVectors {
-    #[serde(with = "either::serde_untagged")]
-    inner: either::Either<Vec<f32>, Vec<Vec<f32>>>,
+    #[serde(with = "either::serde_untagged_optional")]
+    inner: Option<either::Either<Vec<f32>, Vec<Vec<f32>>>>,
 }
 
 impl VectorOrArrayOfVectors {
-    pub fn into_array_of_vectors(self) -> Vec<Vec<f32>> {
-        match self.inner {
-            either::Either::Left(vector) => vec![vector],
-            either::Either::Right(vectors) => vectors,
+    pub fn into_array_of_vectors(self) -> Option<Vec<Vec<f32>>> {
+        match self.inner? {
+            either::Either::Left(vector) => Some(vec![vector]),
+            either::Either::Right(vectors) => Some(vectors),
         }
     }
 }

milli/src/search/mod.rs

@@ -280,9 +280,13 @@ impl<'a> SearchForFacetValues<'a> {
 
         let filterable_fields = index.filterable_fields(rtxn)?;
         if !filterable_fields.contains(&self.facet) {
+            let (valid_fields, hidden_fields) =
+                index.remove_hidden_fields(rtxn, filterable_fields)?;
+
             return Err(UserError::InvalidFacetSearchFacetName {
                 field: self.facet.clone(),
-                valid_fields: filterable_fields.into_iter().collect(),
+                valid_fields,
+                hidden_fields,
             }
             .into());
         }

milli/src/search/new/bucket_sort.rs

@@ -91,11 +91,12 @@ pub fn bucket_sort<'ctx, Q: RankingRuleQueryTrait>(
     /// Update the universes accordingly and inform the logger.
     macro_rules! back {
         () => {
-            assert!(
-                ranking_rule_universes[cur_ranking_rule_index].is_empty(),
-                "The ranking rule {} did not sort its bucket exhaustively",
-                ranking_rules[cur_ranking_rule_index].id()
-            );
+            // FIXME: temporarily disabled assert: see <https://github.com/meilisearch/meilisearch/pull/4013>
+            // assert!(
+            //     ranking_rule_universes[cur_ranking_rule_index].is_empty(),
+            //     "The ranking rule {} did not sort its bucket exhaustively",
+            //     ranking_rules[cur_ranking_rule_index].id()
+            // );
             logger.end_iteration_ranking_rule(
                 cur_ranking_rule_index,
                 ranking_rules[cur_ranking_rule_index].as_ref(),

milli/src/search/new/matches/mod.rs

@@ -418,19 +418,11 @@ impl<'t> Matcher<'t, '_> {
         } else {
             match &self.matches {
                 Some((tokens, matches)) => {
-                    // If the text has to be cropped,
-                    // compute the best interval to crop around.
-                    let matches = match format_options.crop {
-                        Some(crop_size) if crop_size > 0 => {
-                            self.find_best_match_interval(matches, crop_size)
-                        }
-                        _ => matches,
-                    };
-
                     // If the text has to be cropped,
                     // crop around the best interval.
                     let (byte_start, byte_end) = match format_options.crop {
                         Some(crop_size) if crop_size > 0 => {
+                            let matches = self.find_best_match_interval(matches, crop_size);
                             self.crop_bounds(tokens, matches, crop_size)
                         }
                         _ => (0, self.text.len()),
@@ -450,6 +442,11 @@ impl<'t> Matcher<'t, '_> {
                         for m in matches {
                             let token = &tokens[m.token_position];
 
+                            // skip matches out of the crop window.
+                            if token.byte_start < byte_start || token.byte_end > byte_end {
+                                continue;
+                            }
+
                             if byte_index < token.byte_start {
                                 formatted.push(&self.text[byte_index..token.byte_start]);
                             }
@@ -800,6 +797,37 @@ mod tests {
         );
     }
 
+    #[test]
+    fn format_highlight_crop_phrase_query() {
+        //! testing: https://github.com/meilisearch/meilisearch/issues/3975
+        let temp_index = TempIndex::new();
+        temp_index
+            .add_documents(documents!([
+                { "id": 1, "text": "The groundbreaking invention had the power to split the world between those who embraced progress and those who resisted change!" }
+            ]))
+            .unwrap();
+        let rtxn = temp_index.read_txn().unwrap();
+
+        let format_options = FormatOptions { highlight: true, crop: Some(10) };
+        let text = "The groundbreaking invention had the power to split the world between those who embraced progress and those who resisted change!";
+
+        let builder = MatcherBuilder::new_test(&rtxn, &temp_index, "\"the world\"");
+        let mut matcher = builder.build(text);
+        // should return 10 words with a marker at the start as well the end, and the highlighted matches.
+        insta::assert_snapshot!(
+            matcher.format(format_options),
+            @"…had the power to split <em>the</em> <em>world</em> between those who…"
+        );
+
+        let builder = MatcherBuilder::new_test(&rtxn, &temp_index, "those \"and those\"");
+        let mut matcher = builder.build(text);
+        // should highlight "those" and the phrase "and those".
+        insta::assert_snapshot!(
+            matcher.format(format_options),
+            @"…world between <em>those</em> who embraced progress <em>and</em> <em>those</em> who resisted…"
+        );
+    }
+
     #[test]
     fn smaller_crop_size() {
         //! testing: https://github.com/meilisearch/specifications/pull/120#discussion_r836536295

milli/src/search/new/mod.rs

@@ -20,7 +20,7 @@ mod sort;
 #[cfg(test)]
 mod tests;
 
-use std::collections::{BTreeSet, HashSet};
+use std::collections::HashSet;
 
 use bucket_sort::{bucket_sort, BucketSortOutput};
 use charabia::TokenizerBuilder;
@@ -108,24 +108,11 @@ impl<'ctx> SearchContext<'ctx> {
                 (None, None) => continue,
                 // The field is not searchable => User error
                 (_fid, Some(false)) => {
-                    let mut valid_fields: BTreeSet<_> =
-                        fids_map.names().map(String::from).collect();
+                    let (valid_fields, hidden_fields) = match searchable_names {
+                        Some(sn) => self.index.remove_hidden_fields(self.txn, sn)?,
+                        None => self.index.remove_hidden_fields(self.txn, fids_map.names())?,
+                    };
 
-                    // Filter by the searchable names
-                    if let Some(sn) = searchable_names {
-                        let searchable_names = sn.iter().map(|s| s.to_string()).collect();
-                        valid_fields = &valid_fields & &searchable_names;
-                    }
-
-                    let searchable_count = valid_fields.len();
-
-                    // Remove hidden fields
-                    if let Some(dn) = self.index.displayed_fields(self.txn)? {
-                        let displayable_names = dn.iter().map(|s| s.to_string()).collect();
-                        valid_fields = &valid_fields & &displayable_names;
-                    }
-
-                    let hidden_fields = searchable_count > valid_fields.len();
                     let field = field_name.to_string();
                     return Err(UserError::InvalidSearchableAttribute {
                         field,
@@ -604,16 +591,24 @@ fn check_sort_criteria(ctx: &SearchContext, sort_criteria: Option<&Vec<AscDesc>>
     for asc_desc in sort_criteria {
         match asc_desc.member() {
             Member::Field(ref field) if !crate::is_faceted(field, &sortable_fields) => {
+                let (valid_fields, hidden_fields) =
+                    ctx.index.remove_hidden_fields(ctx.txn, sortable_fields)?;
+
                 return Err(UserError::InvalidSortableAttribute {
                     field: field.to_string(),
-                    valid_fields: sortable_fields.into_iter().collect(),
-                })?
+                    valid_fields,
+                    hidden_fields,
+                })?;
             }
             Member::Geo(_) if !sortable_fields.contains("_geo") => {
+                let (valid_fields, hidden_fields) =
+                    ctx.index.remove_hidden_fields(ctx.txn, sortable_fields)?;
+
                 return Err(UserError::InvalidSortableAttribute {
                     field: "_geo".to_string(),
-                    valid_fields: sortable_fields.into_iter().collect(),
-                })?
+                    valid_fields,
+                    hidden_fields,
+                })?;
             }
             _ => (),
         }

milli/src/update/facet/mod.rs

@@ -94,7 +94,7 @@ use crate::heed_codec::facet::{FacetGroupKey, FacetGroupKeyCodec, FacetGroupValu
 use crate::heed_codec::ByteSliceRefCodec;
 use crate::update::index_documents::create_sorter;
 use crate::update::merge_btreeset_string;
-use crate::{BEU16StrCodec, Index, Result, BEU16};
+use crate::{BEU16StrCodec, Index, Result, BEU16, MAX_FACET_VALUE_LENGTH};
 
 pub mod bulk;
 pub mod delete;
@@ -191,7 +191,16 @@ impl<'i> FacetsUpdate<'i> {
         for result in database.iter(wtxn)? {
             let (facet_group_key, ()) = result?;
             if let FacetGroupKey { field_id, level: 0, left_bound } = facet_group_key {
-                let normalized_facet = left_bound.normalize(&options);
+                let mut normalized_facet = left_bound.normalize(&options);
+                let normalized_truncated_facet: String;
+                if normalized_facet.len() > MAX_FACET_VALUE_LENGTH {
+                    normalized_truncated_facet = normalized_facet
+                        .char_indices()
+                        .take_while(|(idx, _)| *idx < MAX_FACET_VALUE_LENGTH)
+                        .map(|(_, c)| c)
+                        .collect();
+                    normalized_facet = normalized_truncated_facet.into();
+                }
                 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)?;

milli/src/update/index_documents/extract/extract_docid_word_positions.rs

@@ -28,8 +28,8 @@ pub fn extract_docid_word_positions<R: io::Read + io::Seek>(
     indexer: GrenadParameters,
     searchable_fields: &Option<HashSet<FieldId>>,
     stop_words: Option<&fst::Set<&[u8]>>,
-    allowed_separators: Option<&Vec<&str>>,
-    dictionary: Option<&Vec<&str>>,
+    allowed_separators: Option<&[&str]>,
+    dictionary: Option<&[&str]>,
     max_positions_per_attributes: Option<u32>,
 ) -> Result<(RoaringBitmap, grenad::Reader<File>, ScriptLanguageDocidsMap)> {
     puffin::profile_function!();
@@ -55,12 +55,10 @@ pub fn extract_docid_word_positions<R: io::Read + io::Seek>(
         tokenizer_builder.stop_words(stop_words);
     }
     if let Some(dictionary) = dictionary {
-        // let dictionary: Vec<_> = dictionary.iter().map(String::as_str).collect();
-        tokenizer_builder.words_dict(dictionary.as_slice());
+        tokenizer_builder.words_dict(dictionary);
     }
     if let Some(separators) = allowed_separators {
-        // let separators: Vec<_> = separators.iter().map(String::as_str).collect();
-        tokenizer_builder.separators(separators.as_slice());
+        tokenizer_builder.separators(separators);
     }
     let tokenizer = tokenizer_builder.build();
 
@@ -228,9 +226,9 @@ fn process_tokens<'a>(
 ) -> impl Iterator<Item = (usize, Token<'a>)> {
     tokens
         .skip_while(|token| token.is_separator())
-        .scan((0, None), |(offset, prev_kind), token| {
+        .scan((0, None), |(offset, prev_kind), mut token| {
             match token.kind {
-                TokenKind::Word | TokenKind::StopWord | TokenKind::Unknown => {
+                TokenKind::Word | TokenKind::StopWord if !token.lemma().is_empty() => {
                     *offset += match *prev_kind {
                         Some(TokenKind::Separator(SeparatorKind::Hard)) => 8,
                         Some(_) => 1,
@@ -246,7 +244,7 @@ fn process_tokens<'a>(
                 {
                     *prev_kind = Some(token.kind);
                 }
-                _ => (),
+                _ => token.kind = TokenKind::Unknown,
             }
             Some((*offset, token))
         })

milli/src/update/index_documents/extract/extract_facet_string_docids.rs

@@ -46,7 +46,7 @@ pub fn extract_facet_string_docids<R: io::Read + io::Seek>(
         if normalised_value.len() > MAX_FACET_VALUE_LENGTH {
             normalised_truncated_value = normalised_value
                 .char_indices()
-                .take_while(|(idx, _)| idx + 4 < MAX_FACET_VALUE_LENGTH)
+                .take_while(|(idx, _)| *idx < MAX_FACET_VALUE_LENGTH)
                 .map(|(_, c)| c)
                 .collect();
             normalised_value = normalised_truncated_value.as_str();

milli/src/update/index_documents/extract/extract_fid_docid_facet_values.rs

@@ -28,11 +28,13 @@ pub struct ExtractedFacetValues {
 ///
 /// Returns the generated grenad reader containing the docid the fid and the orginal value as key
 /// and the normalized value as value extracted from the given chunk of documents.
+/// We need the fid of the geofields to correctly parse them as numbers if they were sent as strings initially.
 #[logging_timer::time]
 pub fn extract_fid_docid_facet_values<R: io::Read + io::Seek>(
     obkv_documents: grenad::Reader<R>,
     indexer: GrenadParameters,
     faceted_fields: &HashSet<FieldId>,
+    geo_fields_ids: Option<(FieldId, FieldId)>,
 ) -> Result<ExtractedFacetValues> {
     puffin::profile_function!();
 
@@ -84,7 +86,10 @@ pub fn extract_fid_docid_facet_values<R: io::Read + io::Seek>(
 
                 let value = from_slice(field_bytes).map_err(InternalError::SerdeJson)?;
 
-                match extract_facet_values(&value) {
+                match extract_facet_values(
+                    &value,
+                    geo_fields_ids.map_or(false, |(lat, lng)| field_id == lat || field_id == lng),
+                ) {
                     FilterableValues::Null => {
                         facet_is_null_docids.entry(field_id).or_default().insert(document);
                     }
@@ -177,12 +182,13 @@ enum FilterableValues {
     Values { numbers: Vec<f64>, strings: Vec<(String, String)> },
 }
 
-fn extract_facet_values(value: &Value) -> FilterableValues {
+fn extract_facet_values(value: &Value, geo_field: bool) -> FilterableValues {
     fn inner_extract_facet_values(
         value: &Value,
         can_recurse: bool,
         output_numbers: &mut Vec<f64>,
         output_strings: &mut Vec<(String, String)>,
+        geo_field: bool,
     ) {
         match value {
             Value::Null => (),
@@ -193,13 +199,30 @@ fn extract_facet_values(value: &Value) -> FilterableValues {
                 }
             }
             Value::String(original) => {
+                // if we're working on a geofield it MUST be something we can parse or else there was an internal error
+                // in the enrich pipeline. But since the enrich pipeline worked, we want to avoid crashing at all costs.
+                if geo_field {
+                    if let Ok(float) = original.parse() {
+                        output_numbers.push(float);
+                    } else {
+                        log::warn!(
+                            "Internal error, could not parse a geofield that has been validated. Please open an issue."
+                        )
+                    }
+                }
                 let normalized = crate::normalize_facet(original);
                 output_strings.push((normalized, original.clone()));
             }
             Value::Array(values) => {
                 if can_recurse {
                     for value in values {
-                        inner_extract_facet_values(value, false, output_numbers, output_strings);
+                        inner_extract_facet_values(
+                            value,
+                            false,
+                            output_numbers,
+                            output_strings,
+                            geo_field,
+                        );
                     }
                 }
             }
@@ -215,7 +238,7 @@ fn extract_facet_values(value: &Value) -> FilterableValues {
         otherwise => {
             let mut numbers = Vec::new();
             let mut strings = Vec::new();
-            inner_extract_facet_values(otherwise, true, &mut numbers, &mut strings);
+            inner_extract_facet_values(otherwise, true, &mut numbers, &mut strings, geo_field);
             FilterableValues::Values { numbers, strings }
         }
     }

milli/src/update/index_documents/extract/extract_vector_points.rs

@@ -35,7 +35,7 @@ pub fn extract_vector_points<R: io::Read + io::Seek>(
         // lazily get it when needed
         let document_id = || -> Value {
             let document_id = obkv.get(primary_key_id).unwrap();
-            serde_json::from_slice(document_id).unwrap()
+            from_slice(document_id).unwrap()
         };
 
         // first we retrieve the _vectors field
@@ -52,12 +52,14 @@ pub fn extract_vector_points<R: io::Read + io::Seek>(
                 }
             };
 
-            for (i, vector) in vectors.into_iter().enumerate().take(u16::MAX as usize) {
-                let index = u16::try_from(i).unwrap();
-                let mut key = docid_bytes.to_vec();
-                key.extend_from_slice(&index.to_be_bytes());
-                let bytes = cast_slice(&vector);
-                writer.insert(key, bytes)?;
+            if let Some(vectors) = vectors {
+                for (i, vector) in vectors.into_iter().enumerate().take(u16::MAX as usize) {
+                    let index = u16::try_from(i).unwrap();
+                    let mut key = docid_bytes.to_vec();
+                    key.extend_from_slice(&index.to_be_bytes());
+                    let bytes = cast_slice(&vector);
+                    writer.insert(key, bytes)?;
+                }
             }
         }
         // else => the `_vectors` object was `null`, there is nothing to do

milli/src/update/index_documents/extract/mod.rs

@@ -49,8 +49,8 @@ pub(crate) fn data_from_obkv_documents(
     geo_fields_ids: Option<(FieldId, FieldId)>,
     vectors_field_id: Option<FieldId>,
     stop_words: Option<fst::Set<&[u8]>>,
-    allowed_separators: Option<Vec<&str>>,
-    dictionary: Option<Vec<&str>>,
+    allowed_separators: Option<&[&str]>,
+    dictionary: Option<&[&str]>,
     max_positions_per_attributes: Option<u32>,
     exact_attributes: HashSet<FieldId>,
 ) -> Result<()> {
@@ -59,7 +59,13 @@ pub(crate) fn data_from_obkv_documents(
     original_obkv_chunks
         .par_bridge()
         .map(|original_documents_chunk| {
-            send_original_documents_data(original_documents_chunk, lmdb_writer_sx.clone())
+            send_original_documents_data(
+                original_documents_chunk,
+                indexer,
+                lmdb_writer_sx.clone(),
+                vectors_field_id,
+                primary_key_id,
+            )
         })
         .collect::<Result<()>>()?;
 
@@ -76,7 +82,6 @@ pub(crate) fn data_from_obkv_documents(
                     &faceted_fields,
                     primary_key_id,
                     geo_fields_ids,
-                    vectors_field_id,
                     &stop_words,
                     &allowed_separators,
                     &dictionary,
@@ -265,11 +270,33 @@ fn spawn_extraction_task<FE, FS, M>(
 /// - documents
 fn send_original_documents_data(
     original_documents_chunk: Result<grenad::Reader<File>>,
+    indexer: GrenadParameters,
     lmdb_writer_sx: Sender<Result<TypedChunk>>,
+    vectors_field_id: Option<FieldId>,
+    primary_key_id: FieldId,
 ) -> Result<()> {
     let original_documents_chunk =
         original_documents_chunk.and_then(|c| unsafe { as_cloneable_grenad(&c) })?;
 
+    if let Some(vectors_field_id) = vectors_field_id {
+        let documents_chunk_cloned = original_documents_chunk.clone();
+        let lmdb_writer_sx_cloned = lmdb_writer_sx.clone();
+        rayon::spawn(move || {
+            let result = extract_vector_points(
+                documents_chunk_cloned,
+                indexer,
+                primary_key_id,
+                vectors_field_id,
+            );
+            let _ = match result {
+                Ok(vector_points) => {
+                    lmdb_writer_sx_cloned.send(Ok(TypedChunk::VectorPoints(vector_points)))
+                }
+                Err(error) => lmdb_writer_sx_cloned.send(Err(error)),
+            };
+        });
+    }
+
     // TODO: create a custom internal error
     lmdb_writer_sx.send(Ok(TypedChunk::Documents(original_documents_chunk))).unwrap();
     Ok(())
@@ -291,10 +318,9 @@ fn send_and_extract_flattened_documents_data(
     faceted_fields: &HashSet<FieldId>,
     primary_key_id: FieldId,
     geo_fields_ids: Option<(FieldId, FieldId)>,
-    vectors_field_id: Option<FieldId>,
     stop_words: &Option<fst::Set<&[u8]>>,
-    allowed_separators: &Option<Vec<&str>>,
-    dictionary: &Option<Vec<&str>>,
+    allowed_separators: &Option<&[&str]>,
+    dictionary: &Option<&[&str]>,
     max_positions_per_attributes: Option<u32>,
 ) -> Result<(
     grenad::Reader<CursorClonableMmap>,
@@ -322,25 +348,6 @@ fn send_and_extract_flattened_documents_data(
         });
     }
 
-    if let Some(vectors_field_id) = vectors_field_id {
-        let documents_chunk_cloned = flattened_documents_chunk.clone();
-        let lmdb_writer_sx_cloned = lmdb_writer_sx.clone();
-        rayon::spawn(move || {
-            let result = extract_vector_points(
-                documents_chunk_cloned,
-                indexer,
-                primary_key_id,
-                vectors_field_id,
-            );
-            let _ = match result {
-                Ok(vector_points) => {
-                    lmdb_writer_sx_cloned.send(Ok(TypedChunk::VectorPoints(vector_points)))
-                }
-                Err(error) => lmdb_writer_sx_cloned.send(Err(error)),
-            };
-        });
-    }
-
     let (docid_word_positions_chunk, docid_fid_facet_values_chunks): (Result<_>, Result<_>) =
         rayon::join(
             || {
@@ -350,8 +357,8 @@ fn send_and_extract_flattened_documents_data(
                         indexer,
                         searchable_fields,
                         stop_words.as_ref(),
-                        allowed_separators.as_ref(),
-                        dictionary.as_ref(),
+                        *allowed_separators,
+                        *dictionary,
                         max_positions_per_attributes,
                     )?;
 
@@ -378,6 +385,7 @@ fn send_and_extract_flattened_documents_data(
                     flattened_documents_chunk.clone(),
                     indexer,
                     faceted_fields,
+                    geo_fields_ids,
                 )?;
 
                 // send docid_fid_facet_numbers_chunk to DB writer

milli/src/update/index_documents/mod.rs

@@ -359,8 +359,8 @@ where
                     geo_fields_ids,
                     vectors_field_id,
                     stop_words,
-                    separators,
-                    dictionary,
+                    separators.as_deref(),
+                    dictionary.as_deref(),
                     max_positions_per_attributes,
                     exact_attributes,
                 )
@@ -2550,6 +2550,25 @@ mod tests {
         db_snap!(index, word_position_docids, 3, @"74f556b91d161d997a89468b4da1cb8f");
     }
 
+    /// Index multiple different number of vectors in documents.
+    /// Vectors must be of the same length.
+    #[test]
+    fn test_multiple_vectors() {
+        let index = TempIndex::new();
+
+        index.add_documents(documents!([{"id": 0, "_vectors": [[0, 1, 2], [3, 4, 5]] }])).unwrap();
+        index.add_documents(documents!([{"id": 1, "_vectors": [6, 7, 8] }])).unwrap();
+        index
+            .add_documents(
+                documents!([{"id": 2, "_vectors": [[9, 10, 11], [12, 13, 14], [15, 16, 17]] }]),
+            )
+            .unwrap();
+
+        let rtxn = index.read_txn().unwrap();
+        let res = index.search(&rtxn).vector([0.0, 1.0, 2.0]).execute().unwrap();
+        assert_eq!(res.documents_ids.len(), 3);
+    }
+
     #[test]
     fn reproduce_the_bug() {
         /*

milli/src/update/settings.rs

@@ -573,7 +573,7 @@ impl<'a, 't, 'u, 'i> Settings<'a, 't, 'u, 'i> {
                     tokenizer
                         .tokenize(text)
                         .filter_map(|token| {
-                            if token.is_word() {
+                            if token.is_word() && !token.lemma().is_empty() {
                                 Some(token.lemma().to_string())
                             } else {
                                 None
@@ -608,13 +608,18 @@ impl<'a, 't, 'u, 'i> Settings<'a, 't, 'u, 'i> {
                 for (word, synonyms) in user_synonyms {
                     // Normalize both the word and associated synonyms.
                     let normalized_word = normalize(&tokenizer, word);
-                    let normalized_synonyms =
-                        synonyms.iter().map(|synonym| normalize(&tokenizer, synonym));
+                    let normalized_synonyms: Vec<_> = synonyms
+                        .iter()
+                        .map(|synonym| normalize(&tokenizer, synonym))
+                        .filter(|synonym| !synonym.is_empty())
+                        .collect();
 
                     // Store the normalized synonyms under the normalized word,
                     // merging the possible duplicate words.
-                    let entry = new_synonyms.entry(normalized_word).or_insert_with(Vec::new);
-                    entry.extend(normalized_synonyms);
+                    if !normalized_word.is_empty() && !normalized_synonyms.is_empty() {
+                        let entry = new_synonyms.entry(normalized_word).or_insert_with(Vec::new);
+                        entry.extend(normalized_synonyms.into_iter());
+                    }
                 }
 
                 // Make sure that we don't have duplicate synonyms.
@@ -1422,6 +1427,43 @@ mod tests {
         assert!(result.documents_ids.is_empty());
     }
 
+    #[test]
+    fn thai_synonyms() {
+        let mut index = TempIndex::new();
+        index.index_documents_config.autogenerate_docids = true;
+
+        let mut wtxn = index.write_txn().unwrap();
+        // Send 3 documents with ids from 1 to 3.
+        index
+            .add_documents_using_wtxn(
+                &mut wtxn,
+                documents!([
+                    { "name": "ยี่ปุ่น" },
+                    { "name": "ญี่ปุ่น" },
+                ]),
+            )
+            .unwrap();
+
+        // In the same transaction provide some synonyms
+        index
+            .update_settings_using_wtxn(&mut wtxn, |settings| {
+                settings.set_synonyms(btreemap! {
+                    "japanese".to_string() => vec![S("ญี่ปุ่น"), S("ยี่ปุ่น")],
+                });
+            })
+            .unwrap();
+        wtxn.commit().unwrap();
+
+        // Ensure synonyms are effectively stored
+        let rtxn = index.read_txn().unwrap();
+        let synonyms = index.synonyms(&rtxn).unwrap();
+        assert!(!synonyms.is_empty()); // at this point the index should return something
+
+        // Check that we can use synonyms
+        let result = index.search(&rtxn).query("japanese").execute().unwrap();
+        assert_eq!(result.documents_ids.len(), 2);
+    }
+
     #[test]
     fn setting_searchable_recomputes_other_settings() {
         let index = TempIndex::new();

permissive-json-pointer/src/lib.rs

@@ -186,12 +186,16 @@ fn create_value(value: &Document, mut selectors: HashSet<&str>) -> Document {
                     let array = create_array(array, &sub_selectors);
                     if !array.is_empty() {
                         new_value.insert(key.to_string(), array.into());
+                    } else {
+                        new_value.insert(key.to_string(), Value::Array(vec![]));
                     }
                 }
                 Value::Object(object) => {
                     let object = create_value(object, sub_selectors);
                     if !object.is_empty() {
                         new_value.insert(key.to_string(), object.into());
+                    } else {
+                        new_value.insert(key.to_string(), Value::Object(Map::new()));
                     }
                 }
                 _ => (),
@@ -211,6 +215,8 @@ fn create_array(array: &[Value], selectors: &HashSet<&str>) -> Vec<Value> {
                 let array = create_array(array, selectors);
                 if !array.is_empty() {
                     res.push(array.into());
+                } else {
+                    res.push(Value::Array(vec![]));
                 }
             }
             Value::Object(object) => {
@@ -637,6 +643,24 @@ mod tests {
         );
     }
 
+    #[test]
+    fn empty_array_object_return_empty() {
+        let value: Value = json!({
+            "array": [],
+            "object": {},
+        });
+        let value: &Document = value.as_object().unwrap();
+
+        let res: Value = select_values(value, vec!["array.name", "object.name"]).into();
+        assert_eq!(
+            res,
+            json!({
+                "array": [],
+                "object": {},
+            })
+        );
+    }
+
     #[test]
     fn all_conflict_variation() {
         let value: Value = json!({