From 0de57c3a9f26fddf86fe62867e34cff20d7c3c26 Mon Sep 17 00:00:00 2001
From: Ziyang Hu <hu.ziyang@cantab.net>
Date: Mon, 26 Jun 2023 12:35:35 +0800
Subject: [PATCH] fix LSH index not properly cleaned after tuple deletion

---
 cozo-core/src/query/stored.rs        |  3 ++-
 cozo-core/src/runtime/minhash_lsh.rs |  2 +-
 cozo-core/src/runtime/tests.rs       | 26 ++++++++++++++++++++++++++
 3 files changed, 29 insertions(+), 2 deletions(-)

diff --git a/cozo-core/src/query/stored.rs b/cozo-core/src/query/stored.rs
index c4c74209..78a4a2b0 100644
--- a/cozo-core/src/query/stored.rs
+++ b/cozo-core/src/query/stored.rs
@@ -952,6 +952,7 @@ impl<'a> SessionTx<'a> {
         let has_indices = !relation_store.indices.is_empty();
         let has_hnsw_indices = !relation_store.hnsw_indices.is_empty();
         let has_fts_indices = !relation_store.fts_indices.is_empty();
+        let has_lsh_indices = !relation_store.lsh_indices.is_empty();
         let fts_processors = self.make_fts_lsh_processors(relation_store)?;
         let mut new_tuples: Vec<DataValue> = vec![];
         let mut old_tuples: Vec<DataValue> = vec![];
@@ -977,7 +978,7 @@ impl<'a> SessionTx<'a> {
                     });
                 }
             }
-            if need_to_collect || has_indices || has_hnsw_indices || has_fts_indices {
+            if need_to_collect || has_indices || has_hnsw_indices || has_fts_indices || has_lsh_indices {
                 if let Some(existing) = self.store_tx.get(&key, false)? {
                     let mut tup = extracted.clone();
                     extend_tuple_from_v(&mut tup, &existing);
diff --git a/cozo-core/src/runtime/minhash_lsh.rs b/cozo-core/src/runtime/minhash_lsh.rs
index 15ec2708..cfb0039f 100644
--- a/cozo-core/src/runtime/minhash_lsh.rs
+++ b/cozo-core/src/runtime/minhash_lsh.rs
@@ -35,7 +35,7 @@ impl<'a> SessionTx<'a> {
     ) -> Result<()> {
         let bytes = match bytes {
             None => {
-                if let Some(mut found) = inv_idx_handle.get_val_only(self, tuple)? {
+                if let Some(mut found) = inv_idx_handle.get_val_only(self, &tuple[..inv_idx_handle.metadata.keys.len()])? {
                     let inv_key = inv_idx_handle.encode_key_for_store(tuple, Default::default())?;
                     self.store_tx.del(&inv_key)?;
                     match found.pop() {
diff --git a/cozo-core/src/runtime/tests.rs b/cozo-core/src/runtime/tests.rs
index 6528b88b..b4898299 100644
--- a/cozo-core/src/runtime/tests.rs
+++ b/cozo-core/src/runtime/tests.rs
@@ -905,6 +905,32 @@ fn test_lsh_indexing3() {
     }
 }
 
+
+#[test]
+fn test_lsh_indexing4() {
+    for i in 1..10 {
+        let f = i as f64 / 10.;
+        let db = DbInstance::new("mem", "", "").unwrap();
+        db.run_default(r":create a {k: String => v: String}")
+            .unwrap();
+        db.run_script(
+            r"::lsh create a:lsh {extractor: v, tokenizer: NGram, n_gram: 3, target_threshold: $t }",
+            BTreeMap::from([("t".into(), f.into())]),
+            ScriptMutability::Mutable
+        )
+            .unwrap();
+        db.run_default("?[k, v] <- [['a', 'ewiygfspeoighjsfcfxzdfncalsdf']] :put a {k => v}")
+            .unwrap();
+        db.run_default("?[k] <- [['a']] :rm a {k}")
+            .unwrap();
+        let res = db
+            .run_default("?[k] := ~a:lsh{k | query: 'ewiygfspeoighjsfcfxzdfncalsdf', k: 1}")
+            .unwrap();
+        assert!(res.rows.len() == 0);
+    }
+}
+
+
 #[test]
 fn test_lsh_indexing() {
     let db = DbInstance::new("mem", "", "").unwrap();