From fe0ccb11bd3192c224615fe09efe47d04c29cfe9 Mon Sep 17 00:00:00 2001
From: mdecimus <mauro@stalw.art>
Date: Thu, 22 Aug 2024 12:33:23 +0200
Subject: Do not insert empty keywords in FTS index

---
 crates/store/src/fts/index.rs | 30 ++++++++++++++++++++----------
 crates/store/src/write/key.rs |  2 +-
 2 files changed, 21 insertions(+), 11 deletions(-)

(limited to 'crates')

diff --git a/crates/store/src/fts/index.rs b/crates/store/src/fts/index.rs
index 4b0c6c63..6795a80a 100644
--- a/crates/store/src/fts/index.rs
+++ b/crates/store/src/fts/index.rs
@@ -95,11 +95,14 @@ impl<'x, T: Into<u8> + Display + Clone + std::fmt::Debug> FtsDocument<'x, T> {
     }
 
     pub fn index_keyword(&mut self, field: Field<T>, text: impl Into<Cow<'x, str>>) {
-        self.parts.push(Text {
-            field,
-            text: text.into(),
-            typ: Type::Keyword,
-        });
+        let text = text.into();
+        if !text.is_empty() {
+            self.parts.push(Text {
+                field,
+                text,
+                typ: Type::Keyword,
+            });
+        }
     }
 }
 
@@ -146,11 +149,14 @@ impl Store {
                     position += 10;
                 }
                 Type::Keyword => {
-                    let field = u8::from(text.field);
-                    tokens
-                        .entry(BitmapHash::new(text.text.as_ref()))
-                        .or_default()
-                        .insert_keyword(TokenType::word(field));
+                    let value = text.text.as_ref();
+                    if !value.is_empty() {
+                        let field = u8::from(text.field);
+                        tokens
+                            .entry(BitmapHash::new(value))
+                            .or_default()
+                            .insert_keyword(TokenType::word(field));
+                    }
                 }
             }
         }
@@ -268,6 +274,10 @@ impl Store {
                             hash[..len].copy_from_slice(&key[U32_LEN..U32_LEN + len]);
                             (hash, len as u8)
                         }
+                        0 => {
+                            // Temporary fix for empty keywords
+                            (hash, 0)
+                        }
                         invalid => {
                             return Err(trc::Error::corrupted_key(key, None, trc::location!())
                                 .ctx(trc::Key::Reason, "Invalid bitmap key length")
diff --git a/crates/store/src/write/key.rs b/crates/store/src/write/key.rs
index d045d987..86649255 100644
--- a/crates/store/src/write/key.rs
+++ b/crates/store/src/write/key.rs
@@ -250,7 +250,7 @@ impl<T: ResolveId> ValueClass<T> {
                 let serializer = serializer.write(account_id).write(
                     hash.hash
                         .get(0..std::cmp::min(hash.len as usize, 8))
-                        .unwrap(),
+                        .unwrap_or_default(),
                 );
 
                 if hash.len >= 8 {
-- 
cgit v1.2.3-70-g09d2