fix: support null insert

Signed-off-by: cutecutecat <junyuchen@tensorchord.ai>

src/index/am.rs

@@ -168,7 +168,9 @@ pub unsafe extern "C" fn aminsert(
     let oid = (*index_relation).rd_locator.relNumber;
     let id = Handle::from_sys(oid);
     let vector = from_datum(*values.add(0));
-    am_update::update_insert(id, vector, *heap_tid);
+    if let Some(v) = vector {
+        am_update::update_insert(id, v, *heap_tid);
+    }
     true
 }
 

src/index/am_build.rs

@@ -57,13 +57,21 @@ unsafe extern "C" fn callback(
     _tuple_is_alive: bool,
     state: *mut std::os::raw::c_void,
 ) {
+    let state = &mut *(state as *mut Builder);
+    if *_is_null.add(0) {
+        (*state.result).heap_tuples += 1.0;
+        return;
+    }
     #[cfg(any(feature = "pg14", feature = "pg15"))]
     let oid = (*index_relation).rd_node.relNode;
     #[cfg(feature = "pg16")]
     let oid = (*index_relation).rd_locator.relNumber;
     let id = Handle::from_sys(oid);
-    let state = &mut *(state as *mut Builder);
     let vector = from_datum(*values.add(0));
+    let vector = match vector {
+        Some(v) => v,
+        None => unreachable!(),
+    };
     let pointer = Pointer::from_sys(*ctid);
     match state.rpc.insert(id, vector, pointer) {
         Ok(()) => (),

src/index/am_scan.rs

@@ -68,7 +68,7 @@ pub unsafe fn start_scan(scan: pgrx::pg_sys::IndexScanDesc, orderbys: pgrx::pg_s
         scanner,
         Scanner::Initial {
             node: scanner.node(),
-            vector: Some(vector),
+            vector,
         },
     );
 

src/index/utils.rs

@@ -12,21 +12,24 @@ struct Header {
     kind: u16,
 }
 
-pub unsafe fn from_datum(datum: pgrx::pg_sys::Datum) -> OwnedVector {
+pub unsafe fn from_datum(datum: pgrx::pg_sys::Datum) -> Option<OwnedVector> {
     let p = datum.cast_mut_ptr::<pgrx::pg_sys::varlena>();
+    if p.is_null() {
+        return None;
+    }
     let q = pgrx::pg_sys::pg_detoast_datum(p);
     let vector = match (*q.cast::<Header>()).kind {
         0 => {
             let v = &*q.cast::<Vecf32Header>();
-            OwnedVector::Vecf32(v.for_borrow().for_own())
+            Some(OwnedVector::Vecf32(v.for_borrow().for_own()))
         }
         1 => {
             let v = &*q.cast::<Vecf16Header>();
-            OwnedVector::Vecf16(v.for_borrow().for_own())
+            Some(OwnedVector::Vecf16(v.for_borrow().for_own()))
         }
         2 => {
             let v = &*q.cast::<SVecf32Header>();
-            OwnedVector::SVecF32(v.for_borrow().for_own())
+            Some(OwnedVector::SVecF32(v.for_borrow().for_own()))
         }
         _ => unreachable!(),
     };

tests/sqllogictest/null.slt

@@ -0,0 +1,37 @@
+statement ok
+SET search_path TO pg_temp, vectors;
+
+statement ok
+CREATE TABLE t (val vector(3));
+
+statement ok
+INSERT INTO t (val) SELECT ARRAY[random(), random(), random()]::real[] FROM generate_series(1, 100);
+
+statement ok
+INSERT INTO t (val) SELECT ('[NaN, Infinity, -Infinity]') FROM generate_series(1, 100);
+
+statement ok
+INSERT INTO t (val) SELECT (NULL) FROM generate_series(1, 100);
+
+query I
+SELECT COUNT(1) FROM (SELECT 1 FROM t ORDER BY val <-> '[0.5,0.5,0.5]' limit 10) t2;
+----
+10
+
+statement ok
+CREATE INDEX hnsw_index ON t USING vectors (val vector_l2_ops)
+WITH (options = "[indexing.hnsw]");
+
+query I
+SELECT COUNT(1) FROM (SELECT 1 FROM t ORDER BY val <-> '[0.5,0.5,0.5]' limit 10) t2;
+----
+10
+
+statement ok
+REINDEX INDEX hnsw_index;
+
+query I
+SELECT COUNT(1) FROM (SELECT 1 FROM t ORDER BY val <-> '[0.5,0.5,0.5]' limit 10) t2;
+----
+10
+