Merge #4693

4693: Introduce distinct attributes at search time r=irevoire a=Kerollmops

This PR fixes #4611.

### To Do
- [x] Remove the `distinguishableAttributes` settings (not even a commit about that).
- [x] Use the `filterableAttributes` to be able to use the `distinct` parameter at search.
- [x] Work on the errors and make tests.

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

milli/src/error.rs

@@ -134,6 +134,17 @@ only composed of alphanumeric characters (a-z A-Z 0-9), hyphens (-) and undersco
         }
     )]
     InvalidSortableAttribute { field: String, valid_fields: BTreeSet<String>, hidden_fields: bool },
+    #[error("Attribute `{}` is not filterable and thus, cannot be used as distinct attribute. {}",
+        .field,
+        match .valid_fields.is_empty() {
+            true => "This index does not have configured filterable attributes.".to_string(),
+            false => format!("Available filterable attributes are: `{}{}`.",
+                    valid_fields.iter().map(AsRef::as_ref).collect::<Vec<&str>>().join(", "),
+                    .hidden_fields.then_some(", <..hidden-attributes>").unwrap_or(""),
+                ),
+        }
+    )]
+    InvalidDistinctAttribute { field: String, valid_fields: BTreeSet<String>, hidden_fields: bool },
     #[error("Attribute `{}` is not facet-searchable. {}",
         .field,
         match .valid_fields.is_empty() {

milli/src/search/hybrid.rs

@@ -159,6 +159,7 @@ impl<'a> Search<'a> {
             offset: 0,
             limit: self.limit + self.offset,
             sort_criteria: self.sort_criteria.clone(),
+            distinct: self.distinct.clone(),
             searchable_attributes: self.searchable_attributes,
             geo_strategy: self.geo_strategy,
             terms_matching_strategy: self.terms_matching_strategy,

milli/src/search/mod.rs

@@ -11,8 +11,8 @@ use self::new::{execute_vector_search, PartialSearchResult};
 use crate::score_details::{ScoreDetails, ScoringStrategy};
 use crate::vector::Embedder;
 use crate::{
-    execute_search, filtered_universe, AscDesc, DefaultSearchLogger, DocumentId, Index, Result,
-    SearchContext, TimeBudget,
+    execute_search, filtered_universe, AscDesc, DefaultSearchLogger, DocumentId, Error, Index,
+    Result, SearchContext, TimeBudget, UserError,
 };
 
 // Building these factories is not free.
@@ -40,6 +40,7 @@ pub struct Search<'a> {
     offset: usize,
     limit: usize,
     sort_criteria: Option<Vec<AscDesc>>,
+    distinct: Option<String>,
     searchable_attributes: Option<&'a [String]>,
     geo_strategy: new::GeoSortStrategy,
     terms_matching_strategy: TermsMatchingStrategy,
@@ -61,6 +62,7 @@ impl<'a> Search<'a> {
             offset: 0,
             limit: 20,
             sort_criteria: None,
+            distinct: None,
             searchable_attributes: None,
             geo_strategy: new::GeoSortStrategy::default(),
             terms_matching_strategy: TermsMatchingStrategy::default(),
@@ -105,6 +107,11 @@ impl<'a> Search<'a> {
         self
     }
 
+    pub fn distinct(&mut self, distinct: String) -> &mut Search<'a> {
+        self.distinct = Some(distinct);
+        self
+    }
+
     pub fn searchable_attributes(&mut self, searchable: &'a [String]) -> &mut Search<'a> {
         self.searchable_attributes = Some(searchable);
         self
@@ -169,6 +176,19 @@ impl<'a> Search<'a> {
             ctx.attributes_to_search_on(searchable_attributes)?;
         }
 
+        if let Some(distinct) = &self.distinct {
+            let filterable_fields = ctx.index.filterable_fields(ctx.txn)?;
+            if !crate::is_faceted(distinct, &filterable_fields) {
+                let (valid_fields, hidden_fields) =
+                    ctx.index.remove_hidden_fields(ctx.txn, filterable_fields)?;
+                return Err(Error::UserError(UserError::InvalidDistinctAttribute {
+                    field: distinct.clone(),
+                    valid_fields,
+                    hidden_fields,
+                }));
+            }
+        }
+
         let universe = filtered_universe(ctx.index, ctx.txn, &self.filter)?;
         let PartialSearchResult {
             located_query_terms,
@@ -185,6 +205,7 @@ impl<'a> Search<'a> {
                     self.scoring_strategy,
                     universe,
                     &self.sort_criteria,
+                    &self.distinct,
                     self.geo_strategy,
                     self.offset,
                     self.limit,
@@ -202,6 +223,7 @@ impl<'a> Search<'a> {
                 self.exhaustive_number_hits,
                 universe,
                 &self.sort_criteria,
+                &self.distinct,
                 self.geo_strategy,
                 self.offset,
                 self.limit,
@@ -238,6 +260,7 @@ impl fmt::Debug for Search<'_> {
             offset,
             limit,
             sort_criteria,
+            distinct,
             searchable_attributes,
             geo_strategy: _,
             terms_matching_strategy,
@@ -257,6 +280,7 @@ impl fmt::Debug for Search<'_> {
             .field("offset", offset)
             .field("limit", limit)
             .field("sort_criteria", sort_criteria)
+            .field("distinct", distinct)
             .field("searchable_attributes", searchable_attributes)
             .field("terms_matching_strategy", terms_matching_strategy)
             .field("scoring_strategy", scoring_strategy)

milli/src/search/new/bucket_sort.rs

@@ -22,6 +22,7 @@ pub fn bucket_sort<'ctx, Q: RankingRuleQueryTrait>(
     ctx: &mut SearchContext<'ctx>,
     mut ranking_rules: Vec<BoxRankingRule<'ctx, Q>>,
     query: &Q,
+    distinct: Option<&str>,
     universe: &RoaringBitmap,
     from: usize,
     length: usize,
@@ -34,7 +35,12 @@ pub fn bucket_sort<'ctx, Q: RankingRuleQueryTrait>(
     logger.ranking_rules(&ranking_rules);
     logger.initial_universe(universe);
 
-    let distinct_fid = if let Some(field) = ctx.index.distinct_field(ctx.txn)? {
+    let distinct_field = match distinct {
+        Some(distinct) => Some(distinct),
+        None => ctx.index.distinct_field(ctx.txn)?,
+    };
+
+    let distinct_fid = if let Some(field) = distinct_field {
         ctx.index.fields_ids_map(ctx.txn)?.id(field)
     } else {
         None

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

@@ -516,6 +516,7 @@ mod tests {
                 false,
                 universe,
                 &None,
+                &None,
                 crate::search::new::GeoSortStrategy::default(),
                 0,
                 100,

milli/src/search/new/mod.rs

@@ -568,6 +568,7 @@ pub fn execute_vector_search(
     scoring_strategy: ScoringStrategy,
     universe: RoaringBitmap,
     sort_criteria: &Option<Vec<AscDesc>>,
+    distinct: &Option<String>,
     geo_strategy: geo_sort::Strategy,
     from: usize,
     length: usize,
@@ -598,6 +599,7 @@ pub fn execute_vector_search(
         ctx,
         ranking_rules,
         &PlaceholderQuery,
+        distinct.as_deref(),
         &universe,
         from,
         length,
@@ -627,6 +629,7 @@ pub fn execute_search(
     exhaustive_number_hits: bool,
     mut universe: RoaringBitmap,
     sort_criteria: &Option<Vec<AscDesc>>,
+    distinct: &Option<String>,
     geo_strategy: geo_sort::Strategy,
     from: usize,
     length: usize,
@@ -717,6 +720,7 @@ pub fn execute_search(
             ctx,
             ranking_rules,
             &graph,
+            distinct.as_deref(),
             &universe,
             from,
             length,
@@ -732,6 +736,7 @@ pub fn execute_search(
             ctx,
             ranking_rules,
             &PlaceholderQuery,
+            distinct.as_deref(),
             &universe,
             from,
             length,
@@ -748,7 +753,12 @@ pub fn execute_search(
     // The candidates is the universe unless the exhaustive number of hits
     // is requested and a distinct attribute is set.
     if exhaustive_number_hits {
-        if let Some(f) = ctx.index.distinct_field(ctx.txn)? {
+        let distinct_field = match distinct.as_deref() {
+            Some(distinct) => Some(distinct),
+            None => ctx.index.distinct_field(ctx.txn)?,
+        };
+
+        if let Some(f) = distinct_field {
             if let Some(distinct_fid) = fields_ids_map.id(f) {
                 all_candidates = apply_distinct_rule(ctx, distinct_fid, &all_candidates)?.remaining;
             }

milli/src/search/new/tests/distinct.rs

@@ -205,8 +205,18 @@ fn create_index() -> TempIndex {
     index
 }
 
-fn verify_distinct(index: &Index, txn: &RoTxn, docids: &[u32]) -> Vec<String> {
-    let vs = collect_field_values(index, txn, index.distinct_field(txn).unwrap().unwrap(), docids);
+fn verify_distinct(
+    index: &Index,
+    txn: &RoTxn,
+    distinct: Option<&str>,
+    docids: &[u32],
+) -> Vec<String> {
+    let vs = collect_field_values(
+        index,
+        txn,
+        distinct.or_else(|| index.distinct_field(txn).unwrap()).unwrap(),
+        docids,
+    );
 
     let mut unique = HashSet::new();
     for v in vs.iter() {
@@ -223,12 +233,49 @@ fn verify_distinct(index: &Index, txn: &RoTxn, docids: &[u32]) -> Vec<String> {
 fn test_distinct_placeholder_no_ranking_rules() {
     let index = create_index();
 
+    // Set the letter as filterable and unset the distinct attribute.
+    index
+        .update_settings(|s| {
+            s.set_filterable_fields(hashset! { S("letter") });
+            s.reset_distinct_field();
+        })
+        .unwrap();
+
+    let txn = index.read_txn().unwrap();
+
+    let mut s = Search::new(&txn, &index);
+    s.distinct(S("letter"));
+    let SearchResult { documents_ids, .. } = s.execute().unwrap();
+    insta::assert_snapshot!(format!("{documents_ids:?}"), @"[0, 2, 5, 8, 9, 15, 18, 20, 21, 24, 25, 26]");
+    let distinct_values = verify_distinct(&index, &txn, Some("letter"), &documents_ids);
+    insta::assert_debug_snapshot!(distinct_values, @r###"
+    [
+        "\"A\"",
+        "\"B\"",
+        "\"C\"",
+        "\"D\"",
+        "\"E\"",
+        "\"F\"",
+        "\"G\"",
+        "\"H\"",
+        "\"I\"",
+        "__does_not_exist__",
+        "__does_not_exist__",
+        "__does_not_exist__",
+    ]
+    "###);
+}
+
+#[test]
+fn test_distinct_at_search_placeholder_no_ranking_rules() {
+    let index = create_index();
+
     let txn = index.read_txn().unwrap();
 
     let s = Search::new(&txn, &index);
     let SearchResult { documents_ids, .. } = s.execute().unwrap();
     insta::assert_snapshot!(format!("{documents_ids:?}"), @"[0, 2, 5, 8, 9, 15, 18, 20, 21, 24, 25, 26]");
-    let distinct_values = verify_distinct(&index, &txn, &documents_ids);
+    let distinct_values = verify_distinct(&index, &txn, None, &documents_ids);
     insta::assert_debug_snapshot!(distinct_values, @r###"
     [
         "\"A\"",
@@ -263,7 +310,7 @@ fn test_distinct_placeholder_sort() {
 
     let SearchResult { documents_ids, .. } = s.execute().unwrap();
     insta::assert_snapshot!(format!("{documents_ids:?}"), @"[14, 26, 4, 7, 17, 23, 1, 19, 25, 8, 20, 24]");
-    let distinct_values = verify_distinct(&index, &txn, &documents_ids);
+    let distinct_values = verify_distinct(&index, &txn, None, &documents_ids);
     insta::assert_debug_snapshot!(distinct_values, @r###"
     [
         "\"E\"",
@@ -303,7 +350,7 @@ fn test_distinct_placeholder_sort() {
 
     let SearchResult { documents_ids, .. } = s.execute().unwrap();
     insta::assert_snapshot!(format!("{documents_ids:?}"), @"[21, 20, 18, 15, 9, 8, 5, 2, 0, 24, 25, 26]");
-    let distinct_values = verify_distinct(&index, &txn, &documents_ids);
+    let distinct_values = verify_distinct(&index, &txn, None, &documents_ids);
     insta::assert_debug_snapshot!(distinct_values, @r###"
     [
         "\"I\"",
@@ -346,7 +393,7 @@ fn test_distinct_placeholder_sort() {
 
     let SearchResult { documents_ids, .. } = s.execute().unwrap();
     insta::assert_snapshot!(format!("{documents_ids:?}"), @"[23, 20, 19, 17, 14, 8, 7, 4, 1, 26, 25, 24]");
-    let distinct_values = verify_distinct(&index, &txn, &documents_ids);
+    let distinct_values = verify_distinct(&index, &txn, None, &documents_ids);
     insta::assert_debug_snapshot!(distinct_values, @r###"
     [
         "\"I\"",
@@ -399,7 +446,7 @@ fn test_distinct_words() {
 
     let SearchResult { documents_ids, .. } = s.execute().unwrap();
     insta::assert_snapshot!(format!("{documents_ids:?}"), @"[0, 2, 26, 5, 8, 9, 15, 18, 20, 21, 25, 24]");
-    let distinct_values = verify_distinct(&index, &txn, &documents_ids);
+    let distinct_values = verify_distinct(&index, &txn, None, &documents_ids);
     insta::assert_debug_snapshot!(distinct_values, @r###"
     [
         "\"A\"",
@@ -453,7 +500,7 @@ fn test_distinct_sort_words() {
 
     let SearchResult { documents_ids, .. } = s.execute().unwrap();
     insta::assert_snapshot!(format!("{documents_ids:?}"), @"[22, 20, 19, 16, 9, 8, 7, 3, 1, 26, 25, 24]");
-    let distinct_values = verify_distinct(&index, &txn, &documents_ids);
+    let distinct_values = verify_distinct(&index, &txn, None, &documents_ids);
     insta::assert_debug_snapshot!(distinct_values, @r###"
     [
         "\"I\"",
@@ -549,7 +596,7 @@ fn test_distinct_typo() {
     let SearchResult { documents_ids, .. } = s.execute().unwrap();
     insta::assert_snapshot!(format!("{documents_ids:?}"), @"[3, 26, 0, 7, 8, 9, 15, 22, 18, 20, 25, 24]");
 
-    let distinct_values = verify_distinct(&index, &txn, &documents_ids);
+    let distinct_values = verify_distinct(&index, &txn, None, &documents_ids);
     insta::assert_debug_snapshot!(distinct_values, @r###"
     [
         "\"B\"",