From ade54493aba062de21bebc79d6607e026775d655 Mon Sep 17 00:00:00 2001
From: ManyTheFish <many@meilisearch.com>
Date: Wed, 14 Aug 2024 11:33:17 +0200
Subject: [PATCH] Only detect language for a facet if several locales have been
 specified by the user in the settings

---
 meilisearch/src/search/mod.rs                      | 14 ++++++++++----
 milli/src/search/facet/search.rs                   | 10 +++++++++-
 .../extract/extract_facet_string_docids.rs         |  9 +++++++--
 3 files changed, 26 insertions(+), 7 deletions(-)

diff --git a/meilisearch/src/search/mod.rs b/meilisearch/src/search/mod.rs
index dada9159b..915505be0 100644
--- a/meilisearch/src/search/mod.rs
+++ b/meilisearch/src/search/mod.rs
@@ -1369,12 +1369,18 @@ pub fn perform_facet_search(
         None => TimeBudget::default(),
     };
 
+    // In the faceted search context, we want to use the intersection between the locales provided by the user
+    // and the locales of the facet string.
+    // If the facet string is not localized, we **ignore** the locales provided by the user because the facet data has no locale.
+    // If the user does not provide locales, we use the locales of the facet string.
     let localized_attributes = index.localized_attributes_rules(&rtxn)?.unwrap_or_default();
-    let locales = locales.or_else(|| {
-        localized_attributes
+    let localized_attributes_locales =
+        localized_attributes.into_iter().find(|attr| attr.match_str(&facet_name));
+    let locales = localized_attributes_locales.map(|attr| {
+        attr.locales
             .into_iter()
-            .find(|attr| attr.match_str(&facet_name))
-            .map(|attr| attr.locales)
+            .filter(|locale| locales.as_ref().map_or(true, |locales| locales.contains(locale)))
+            .collect()
     });
 
     let (search, _, _, _) =
diff --git a/milli/src/search/facet/search.rs b/milli/src/search/facet/search.rs
index 39fb7374a..cdba7ee16 100644
--- a/milli/src/search/facet/search.rs
+++ b/milli/src/search/facet/search.rs
@@ -339,10 +339,18 @@ impl ValuesCollection {
 fn normalize_facet_string(facet_string: &str, locales: Option<&[Language]>) -> String {
     let options = NormalizerOption { lossy: true, ..Default::default() };
     let mut detection = StrDetection::new(facet_string, locales);
+
+    // Detect the language of the facet string only if several locales are explicitly provided.
+    let language = match locales {
+        Some(&[language]) => Some(language),
+        Some(multiple_locales) if multiple_locales.len() > 1 => detection.language(),
+        _ => None,
+    };
+
     let token = Token {
         lemma: std::borrow::Cow::Borrowed(facet_string),
         script: detection.script(),
-        language: detection.language(),
+        language,
         ..Default::default()
     };
 
diff --git a/milli/src/update/index_documents/extract/extract_facet_string_docids.rs b/milli/src/update/index_documents/extract/extract_facet_string_docids.rs
index 45a7696ac..69663bb08 100644
--- a/milli/src/update/index_documents/extract/extract_facet_string_docids.rs
+++ b/milli/src/update/index_documents/extract/extract_facet_string_docids.rs
@@ -271,7 +271,7 @@ fn extract_facet_string_docids_settings<R: io::Read + io::Seek>(
 /// Normalizes the facet string and truncates it to the max length.
 #[tracing::instrument(level = "trace", skip_all, target = "indexing::extract")]
 fn normalize_facet_string(facet_string: &str, locales: Option<&[Language]>) -> String {
-    let options = NormalizerOption { lossy: true, ..Default::default() };
+    let options: NormalizerOption = NormalizerOption { lossy: true, ..Default::default() };
     let mut detection = StrDetection::new(facet_string, locales);
 
     let script = {
@@ -285,7 +285,12 @@ fn normalize_facet_string(facet_string: &str, locales: Option<&[Language]>) -> S
         let span = tracing::trace_span!(target: "indexing::extract::extract_facet_string_docids", "detect_language");
         let _entered = span.enter();
 
-        detection.language()
+        // Detect the language of the facet string only if several locales are explicitly provided.
+        match locales {
+            Some(&[language]) => Some(language),
+            Some(multiple_locales) if multiple_locales.len() > 1 => detection.language(),
+            _ => None,
+        }
     };
 
     let token = Token {