hnsw indexing ops

cozo-core/src/data/relation.rs

@@ -95,7 +95,7 @@ pub(crate) enum ColType {
     Validity,
 }
 
-#[derive(Debug, Clone, Eq, PartialEq, serde_derive::Deserialize, serde_derive::Serialize)]
+#[derive(Debug, Clone, Eq, PartialEq, Hash, serde_derive::Deserialize, serde_derive::Serialize)]
 pub(crate) enum VecElementType {
     F32,
     F64,

cozo-core/src/parse/sys.rs

@@ -14,6 +14,7 @@ use miette::{ensure, miette, Diagnostic, Result};
 use thiserror::Error;
 
 use crate::data::program::InputProgram;
+use crate::data::relation::VecElementType;
 use crate::data::symb::Symbol;
 use crate::data::value::{DataValue, ValidityTs};
 use crate::parse::expr::build_expr;
@@ -36,10 +37,38 @@ pub(crate) enum SysOp {
     SetTriggers(Symbol, Vec<String>, Vec<String>, Vec<String>),
     SetAccessLevel(Vec<Symbol>, AccessLevel),
     CreateIndex(Symbol, Symbol, Vec<Symbol>),
-    CreateVectorIndex(Symbol, Symbol, Vec<Symbol>),
+    CreateVectorIndex(HnswIndexConfig),
     RemoveIndex(Symbol, Symbol),
 }
 
+#[derive(Debug, Clone, PartialEq, Eq, Hash)]
+pub(crate) struct HnswIndexConfig {
+    pub(crate) base_relation: Symbol,
+    pub(crate) index_name: Symbol,
+    pub(crate) vec_dim: usize,
+    pub(crate) dtype: VecElementType,
+    pub(crate) vec_fields: Vec<Symbol>,
+    pub(crate) tag_fields: Vec<Symbol>,
+    pub(crate) distance: HnswDistance,
+    pub(crate) ef_construction: usize,
+    pub(crate) max_elements: usize,
+}
+
+#[derive(Debug, Clone, Copy, PartialEq, Eq, Hash)]
+pub(crate) enum HnswDistance {
+    L2,
+    InnerProduct,
+    Cosine,
+}
+
+pub(crate) struct HnswKnnQueryOptions {
+    k: usize,
+    ef: usize,
+    auto_diversity: bool,
+    min_separation: f64,
+    max_distance: f64,
+}
+
 #[derive(Debug, Diagnostic, Error)]
 #[error("Cannot interpret {0} as process ID")]
 #[diagnostic(code(parser::not_proc_id))]
@@ -186,6 +215,9 @@ pub(crate) fn parse_sys(
             }
         }
         Rule::list_fixed_rules => SysOp::ListFixedRules,
+        Rule::vec_idx_op => {
+            todo!("vec_idx_op")
+        }
         rule => unreachable!("{:?}", rule),
     })
 }

cozo-core/src/runtime/db.rs

@@ -1157,7 +1157,7 @@ impl<'s, S: Storage<'s>> Db<S> {
                     vec![vec![DataValue::from(OK_STR)]],
                 ))
             }
-            SysOp::CreateVectorIndex(..) => {
+            SysOp::CreateVectorIndex(config) => {
                 todo!()
             }
             SysOp::RemoveIndex(rel_name, idx_name) => {

cozo-core/src/runtime/tests.rs

@@ -757,3 +757,12 @@ fn test_vec_types() {
         res.into_json()["rows"][0][1]
     );
 }
+
+#[test]
+fn test_vec_index() {
+    let db = DbInstance::new("mem", "", "").unwrap();
+    db.run_script(":create a {k: String => tags: [String], v: <F32; 8>}", Default::default())
+        .unwrap();
+    // db.run_script("::hnsw create a:vec {dim: 128, dtype: f32, fields: [v], tags: tags, distance: Cosine, ef_construction: 20, max_elements: 50}", Default::default())
+    //     .unwrap();
+}