From 89d9e1703b06ad6f84933e1399e70a9d055dfbe8 Mon Sep 17 00:00:00 2001
From: metamuffin <metamuffin@disroot.org>
Date: Wed, 11 Mar 2026 05:00:42 +0100
Subject: tokenize lowercase

---
 database/src/kv/index.rs | 28 ++++++++++++++--------------
 1 file changed, 14 insertions(+), 14 deletions(-)

(limited to 'database')

diff --git a/database/src/kv/index.rs b/database/src/kv/index.rs
index 469a6e8..e4f707c 100644
--- a/database/src/kv/index.rs
+++ b/database/src/kv/index.rs
@@ -62,19 +62,19 @@ pub fn update_index(
         SortKey::Text(path) => {
             let mut tokens = HashSet::new();
             for val in path.get_matching_values(ob) {
-                for tok in text_tokenizer(val) {
+                for tok in text_tokenizer(str::from_utf8(val).unwrap()) {
                     tokens.insert(tok);
                 }
             }
-            for &tok in &tokens {
+            for tok in &tokens {
                 for mut k in ks.clone() {
                     k.push(0);
-                    k.extend(tok);
+                    k.extend(tok.as_bytes());
                     index_counter(txn, &k, remove)?;
                 }
                 for mut k in ks.clone() {
                     k.push(1);
-                    k.extend(tok);
+                    k.extend(tok.as_bytes());
                     k.extend(row.to_be_bytes());
                     index_marker(txn, &k, remove)?;
                 }
@@ -181,24 +181,22 @@ pub fn iter_index<'a>(
             )
         }
         Sort::TextSearch(_, text) => {
-            let search_tokens = text_tokenizer(text.as_bytes())
-                .map(|e| e.to_owned())
-                .collect::<Vec<_>>();
-            let mut min_tok = Vec::new();
+            let search_tokens = text_tokenizer(text);
+            let mut min_tok = String::new();
             let mut min_count = u64::MAX;
             for token in &search_tokens {
                 let mut k = prefix.clone();
                 k.push(0);
-                k.extend(token);
+                k.extend(token.as_bytes());
                 let count = read_counter(txn, &k, 0)?;
                 if count < min_count {
                     min_count = count;
-                    min_tok = token.to_owned()
+                    min_tok = token.to_string()
                 }
             }
             let mut min_token_prefix = prefix.clone();
             min_token_prefix.push(1);
-            min_token_prefix.extend(&min_tok);
+            min_token_prefix.extend(min_tok.as_bytes());
             Box::new(
                 PrefixIterator {
                     inner: txn.iter(&min_token_prefix, false)?,
@@ -213,7 +211,7 @@ pub fn iter_index<'a>(
                     for token in &search_tokens {
                         let mut k = prefix.clone();
                         k.push(1);
-                        k.extend(token);
+                        k.extend(token.as_bytes());
                         k.extend(rn.to_be_bytes());
                         let v = match txn.get(&k) {
                             Ok(v) => v,
@@ -230,9 +228,11 @@ pub fn iter_index<'a>(
     })
 }
 
-fn text_tokenizer(text: &[u8]) -> impl Iterator<Item = &[u8]> {
-    text.split(|x| matches!(*x, b' ' | b',' | b':' | b'/' | b'+' | b'&'))
+fn text_tokenizer(text: &str) -> HashSet<String> {
+    text.split(|x| matches!(x, ' ' | ',' | ':' | '/' | '+' | '&'))
         .filter(|x| !x.is_empty())
+        .map(|s| s.to_lowercase())
+        .collect()
 }
 
 fn inc_key(mut k: Vec<u8>) -> Vec<u8> {
-- 
cgit v1.3