From b31158528020d204e2e604b949fcb257374cb1eb Mon Sep 17 00:00:00 2001
From: ManyTheFish <many@meilisearch.com>
Date: Wed, 13 Nov 2024 13:50:10 +0100
Subject: [PATCH] Fix facet strings

---
 crates/milli/src/update/new/channel.rs        |  4 ++--
 .../new/extract/faceted/extract_facets.rs     | 23 +++++++++----------
 2 files changed, 13 insertions(+), 14 deletions(-)

diff --git a/crates/milli/src/update/new/channel.rs b/crates/milli/src/update/new/channel.rs
index 3287a1f7f..fbf102f18 100644
--- a/crates/milli/src/update/new/channel.rs
+++ b/crates/milli/src/update/new/channel.rs
@@ -364,9 +364,9 @@ impl DocidsSender for FacetDocidsSender<'_> {
 pub struct FieldIdDocidFacetSender<'a>(&'a ExtractorSender);
 
 impl FieldIdDocidFacetSender<'_> {
-    pub fn write_facet_string(&self, key: &[u8]) -> StdResult<(), SendError<()>> {
+    pub fn write_facet_string(&self, key: &[u8], value: &[u8]) -> StdResult<(), SendError<()>> {
         debug_assert!(FieldDocIdFacetStringCodec::bytes_decode(key).is_ok());
-        let entry = EntryOperation::Write(KeyValueEntry::from_small_key_value(&key, &[]));
+        let entry = EntryOperation::Write(KeyValueEntry::from_small_key_value(&key, &value));
         self.0
             .send_db_operation(DbOperation { database: Database::FieldIdDocidFacetStrings, entry })
     }
diff --git a/crates/milli/src/update/new/extract/faceted/extract_facets.rs b/crates/milli/src/update/new/extract/faceted/extract_facets.rs
index 0e7dcc4b9..89223bc55 100644
--- a/crates/milli/src/update/new/extract/faceted/extract_facets.rs
+++ b/crates/milli/src/update/new/extract/faceted/extract_facets.rs
@@ -290,22 +290,22 @@ impl<'doc> DelAddFacetValue<'doc> {
         sender: &FieldIdDocidFacetSender,
         doc_alloc: &Bump,
     ) -> std::result::Result<(), crossbeam_channel::SendError<()>> {
-        println!("sending FieldIdDocidFacet data");
-        let mut count = 0;
         let mut buffer = bumpalo::collections::Vec::new_in(doc_alloc);
         for ((fid, value), deladd) in self.strings {
-            buffer.clear();
-            buffer.extend_from_slice(&fid.to_be_bytes());
-            buffer.extend_from_slice(&docid.to_be_bytes());
-            buffer.extend_from_slice(&value);
-            match deladd {
-                DelAdd::Deletion => sender.delete_facet_string(&buffer)?,
-                DelAdd::Addition => sender.write_facet_string(&buffer)?,
+            if let Ok(s) = std::str::from_utf8(&value) {
+                buffer.clear();
+                buffer.extend_from_slice(&fid.to_be_bytes());
+                buffer.extend_from_slice(&docid.to_be_bytes());
+                let normalized = crate::normalize_facet(s);
+                let truncated = truncate_str(&normalized);
+                buffer.extend_from_slice(truncated.as_bytes());
+                match deladd {
+                    DelAdd::Deletion => sender.delete_facet_string(&buffer)?,
+                    DelAdd::Addition => sender.write_facet_string(&buffer, &value)?,
+                }
             }
-            count += 1;
         }
 
-        count = 0;
         for ((fid, value), deladd) in self.f64s {
             buffer.clear();
             buffer.extend_from_slice(&fid.to_be_bytes());
@@ -315,7 +315,6 @@ impl<'doc> DelAddFacetValue<'doc> {
                 DelAdd::Deletion => sender.delete_facet_f64(&buffer)?,
                 DelAdd::Addition => sender.write_facet_f64(&buffer)?,
             }
-            count += 1;
         }
 
         Ok(())