rename VEC_Distance to VEC_Distance_Euclidean

and create a parent Item_func_vec_distance_common class

mysql-test/main/vector.result

@@ -106,28 +106,28 @@ id	hex(v)
 9	7B713F3E5258323F80D1113D673B2B3F66E3583F
 10	6CA1D43E9DF91B3FE580DA3E1C247D3F147CF33E
 flush tables;
-select id,vec_distance(v, x'B047263c9f87233fcfd27e3eae493e3f0329f43e') d from t1 order by d limit 3;
+select id,vec_distance_euclidean(v, x'B047263c9f87233fcfd27e3eae493e3f0329f43e') d from t1 order by d limit 3;
 id	d
 9	0.4719976290006591
 10	0.5069011044450041
 3	0.5865673124650332
-select id,vec_distance(x'b047263C9f87233fcfd27e3eae493e3f0329f43e', v) d from t1 order by d limit 3;
+select id,vec_distance_euclidean(x'b047263C9f87233fcfd27e3eae493e3f0329f43e', v) d from t1 order by d limit 3;
 id	d
 9	0.4719976290006591
 10	0.5069011044450041
 3	0.5865673124650332
-select id>0,vec_distance(v, NULL) d from t1 order by d limit 3;
+select id>0,vec_distance_euclidean(v, NULL) d from t1 order by d limit 3;
 id>0	d
 1	NULL
 1	NULL
 1	NULL
-select id>0,vec_distance(v, x'123456') d from t1 order by d limit 3;
+select id>0,vec_distance_euclidean(v, x'123456') d from t1 order by d limit 3;
 id>0	d
 1	NULL
 1	NULL
 1	NULL
-select t1.id as id1, t2.id as id2, vec_distance(t1.v, t2.v) from t1, t1 as t2 order by 3,1,2;
-id1	id2	vec_distance(t1.v, t2.v)
+select t1.id as id1, t2.id as id2, vec_distance_euclidean(t1.v, t2.v) from t1, t1 as t2 order by 3,1,2;
+id1	id2	vec_distance_euclidean(t1.v, t2.v)
 1	1	0
 2	2	0
 3	3	0
@@ -229,18 +229,18 @@ id1	id2	vec_distance(t1.v, t2.v)
 7	8	1.288239696195716
 8	7	1.288239696195716
 delete from t1 where v = x'7b713f3e5258323f80d1113d673b2b3f66e3583f';
-select id,vec_distance(v, x'B047263C9f87233fcfd27e3eae493e3f0329f43e') d from t1 order by d limit 3;
+select id,vec_distance_euclidean(v, x'B047263C9f87233fcfd27e3eae493e3f0329f43e') d from t1 order by d limit 3;
 id	d
 10	0.5069011044450041
 3	0.5865673124650332
 7	0.7344464697214867
 insert t1 (v) values (x'7b713f3e5258323f80d1113d673b2b3f66e3583f');
-select id,vec_distance(v, x'b047263c9F87233fcfd27e3eae493e3f0329f43e') d from t1 order by d limit 3;
+select id,vec_distance_euclidean(v, x'b047263c9F87233fcfd27e3eae493e3f0329f43e') d from t1 order by d limit 3;
 id	d
 11	0.4719976290006591
 10	0.5069011044450041
 3	0.5865673124650332
-select id,vec_distance(v, x'B047263c9F87233fcfd27e3eae493e3f0329f43e') d from t1 order by d limit 5;
+select id,vec_distance_euclidean(v, x'B047263c9F87233fcfd27e3eae493e3f0329f43e') d from t1 order by d limit 5;
 id	d
 11	0.4719976290006591
 10	0.5069011044450041
@@ -248,7 +248,7 @@ id	d
 7	0.7344464697214867
 5	0.7671033529042712
 update t1 set v=x'76EDFC3E4B57243F10F8423FB158713F020BAA3E' where v=x'6CA1D43E9DF91B3FE580DA3E1C247D3F147CF33E';
-select id,vec_distance(v, x'B047263C9F87233fcfd27e3eae493e3f0329f43e') d from t1 order by d limit 5;
+select id,vec_distance_euclidean(v, x'B047263C9F87233fcfd27e3eae493e3f0329f43e') d from t1 order by d limit 5;
 id	d
 11	0.4719976290006591
 3	0.5865673124650332
@@ -266,7 +266,7 @@ insert t1 (v) values (x'e360d63ebe554f3fcdbc523f4522193f5236083d'),
 (x'56926c3fdf098d3e2c8c5e3d1ad4953daa9d0b3e'),
 (x'7b713f3e5258323f80d1113d673b2b3f66e3583f'),
 (x'6ca1d43e9df91b3fe580da3e1c247d3f147cf33e');
-select id,vec_distance(v, x'b047263c9f87233Fcfd27e3eae493e3f0329f43e') d from t1 order by d limit 5;
+select id,vec_distance_euclidean(v, x'b047263c9f87233Fcfd27e3eae493e3f0329f43e') d from t1 order by d limit 5;
 id	d
 20	0.4719976290006591
 21	0.5069011044450041

mysql-test/main/vector.test

@@ -41,27 +41,27 @@ insert t1 (v) values (x'e360d63ebe554f3fcdbc523f4522193f5236083d'),
 select id, hex(v) from t1;
 flush tables;
 # test with a valid query vector
-select id,vec_distance(v, x'B047263c9f87233fcfd27e3eae493e3f0329f43e') d from t1 order by d limit 3;
+select id,vec_distance_euclidean(v, x'B047263c9f87233fcfd27e3eae493e3f0329f43e') d from t1 order by d limit 3;
 # swapped arguments
-select id,vec_distance(x'b047263C9f87233fcfd27e3eae493e3f0329f43e', v) d from t1 order by d limit 3;
+select id,vec_distance_euclidean(x'b047263C9f87233fcfd27e3eae493e3f0329f43e', v) d from t1 order by d limit 3;
 # test with NULL (id is unpredictable)
-select id>0,vec_distance(v, NULL) d from t1 order by d limit 3;
+select id>0,vec_distance_euclidean(v, NULL) d from t1 order by d limit 3;
 # test with invalid query vector (id is unpredictable)
-select id>0,vec_distance(v, x'123456') d from t1 order by d limit 3;
-select t1.id as id1, t2.id as id2, vec_distance(t1.v, t2.v) from t1, t1 as t2 order by 3,1,2;
+select id>0,vec_distance_euclidean(v, x'123456') d from t1 order by d limit 3;
+select t1.id as id1, t2.id as id2, vec_distance_euclidean(t1.v, t2.v) from t1, t1 as t2 order by 3,1,2;
 
 # test delete
 delete from t1 where v = x'7b713f3e5258323f80d1113d673b2b3f66e3583f';
-select id,vec_distance(v, x'B047263C9f87233fcfd27e3eae493e3f0329f43e') d from t1 order by d limit 3;
+select id,vec_distance_euclidean(v, x'B047263C9f87233fcfd27e3eae493e3f0329f43e') d from t1 order by d limit 3;
 
 # test insert deleted vec
 insert t1 (v) values (x'7b713f3e5258323f80d1113d673b2b3f66e3583f');
-select id,vec_distance(v, x'b047263c9F87233fcfd27e3eae493e3f0329f43e') d from t1 order by d limit 3;
+select id,vec_distance_euclidean(v, x'b047263c9F87233fcfd27e3eae493e3f0329f43e') d from t1 order by d limit 3;
 
 # test update
-select id,vec_distance(v, x'B047263c9F87233fcfd27e3eae493e3f0329f43e') d from t1 order by d limit 5;
+select id,vec_distance_euclidean(v, x'B047263c9F87233fcfd27e3eae493e3f0329f43e') d from t1 order by d limit 5;
 update t1 set v=x'76EDFC3E4B57243F10F8423FB158713F020BAA3E' where v=x'6CA1D43E9DF91B3FE580DA3E1C247D3F147CF33E';
-select id,vec_distance(v, x'B047263C9F87233fcfd27e3eae493e3f0329f43e') d from t1 order by d limit 5;
+select id,vec_distance_euclidean(v, x'B047263C9F87233fcfd27e3eae493e3f0329f43e') d from t1 order by d limit 5;
 
 # test delete all and reinsert
 delete from t1;
@@ -75,7 +75,7 @@ insert t1 (v) values (x'e360d63ebe554f3fcdbc523f4522193f5236083d'),
                      (x'56926c3fdf098d3e2c8c5e3d1ad4953daa9d0b3e'),
                      (x'7b713f3e5258323f80d1113d673b2b3f66e3583f'),
                      (x'6ca1d43e9df91b3fe580da3e1c247d3f147cf33e');
-select id,vec_distance(v, x'b047263c9f87233Fcfd27e3eae493e3f0329f43e') d from t1 order by d limit 5;
+select id,vec_distance_euclidean(v, x'b047263c9f87233Fcfd27e3eae493e3f0329f43e') d from t1 order by d limit 5;
 
 
 --error ER_TRUNCATED_WRONG_VALUE_FOR_FIELD

mysql-test/main/vector_funcs.test

@@ -36,7 +36,7 @@ select hex(vec_fromtext('[]'));
 select vec_totext(x'');
 
 select id, vec_totext(t1.v) as a, vec_totext(vec_fromtext(vec_totext(t1.v))) as b,
-       vec_distance(t1.v, vec_fromtext(vec_totext(t1.v))) < 0.000001
+       vec_distance_euclidean(t1.v, vec_fromtext(vec_totext(t1.v))) < 0.000001 as c
 from t1;
 
 drop table t1;

mysql-test/main/vector_innodb.result

@@ -38,14 +38,14 @@ insert t1 values
 (41, x'71046a3e85329b3e05240e3f45c9283f1847363f98d47d3f4224b73d487b613f');
 delete from t1 where id=7;
 rollback;
-select id,vec_distance(v, x'c923e33dc0da313fe7c7983e526b3d3fde63963e6eaf3a3f27fa133fe27a583f') d from t1 order by d limit 5;
+select id,vec_distance_euclidean(v, x'c923e33dc0da313fe7c7983e526b3d3fde63963e6eaf3a3f27fa133fe27a583f') d from t1 order by d limit 5;
 id	d
 2	0.8781474260354732
 10	0.8856208347761952
 30	1.0162643974895857
 7	1.026397313888122
 5	1.0308161006949719
-select id,vec_distance(v, x'754b5f3ea2312b3fc169f43e4604883e1d20173e8dd7443f421b703fb11e0d3e') d from t1 order by d limit 5;
+select id,vec_distance_euclidean(v, x'754b5f3ea2312b3fc169f43e4604883e1d20173e8dd7443f421b703fb11e0d3e') d from t1 order by d limit 5;
 id	d
 2	0.9426904171992334
 33	0.9477554826856
@@ -71,14 +71,14 @@ insert t2 values
 (20, x'58dc7d3fc9feaa3e19e26b3f31820c3f93070b3fc4e36e3f'),
 (21, x'35e05d3f18e8513fb81a3d3f8acf7d3e794a1d3c72f9613f');
 commit;
-select id,vec_distance(v, x'1f4d053f7056493f937da03dd8c97a3f220cbb3c926c1c3facca213ec0618a3e') d from t1 order by d limit 5;
+select id,vec_distance_euclidean(v, x'1f4d053f7056493f937da03dd8c97a3f220cbb3c926c1c3facca213ec0618a3e') d from t1 order by d limit 5;
 id	d
 6	0.9309383181777582
 5	0.9706304662574956
 30	0.98144492002831
 50	1.079862635421575
 2	1.0907138991979892
-select id,vec_distance(v, x'f618663f256be73e62cd453f8bcdbf3e16ae503c3858313f') d from t2 order by d limit 5;
+select id,vec_distance_euclidean(v, x'f618663f256be73e62cd453f8bcdbf3e16ae503c3858313f') d from t2 order by d limit 5;
 id	d
 21	0.43559180321379337
 20	0.6435053022072372

mysql-test/main/vector_innodb.test

@@ -37,8 +37,8 @@ insert t1 values
 delete from t1 where id=7;
 rollback;
 
-select id,vec_distance(v, x'c923e33dc0da313fe7c7983e526b3d3fde63963e6eaf3a3f27fa133fe27a583f') d from t1 order by d limit 5;
-select id,vec_distance(v, x'754b5f3ea2312b3fc169f43e4604883e1d20173e8dd7443f421b703fb11e0d3e') d from t1 order by d limit 5;
+select id,vec_distance_euclidean(v, x'c923e33dc0da313fe7c7983e526b3d3fde63963e6eaf3a3f27fa133fe27a583f') d from t1 order by d limit 5;
+select id,vec_distance_euclidean(v, x'754b5f3ea2312b3fc169f43e4604883e1d20173e8dd7443f421b703fb11e0d3e') d from t1 order by d limit 5;
 
 ### two indexes in one transaction:
 create table t2 (id int auto_increment primary key, v blob not null, vector index (v)) engine=innodb;
@@ -63,8 +63,8 @@ insert t2 values
      (21, x'35e05d3f18e8513fb81a3d3f8acf7d3e794a1d3c72f9613f');
 commit;
 
-select id,vec_distance(v, x'1f4d053f7056493f937da03dd8c97a3f220cbb3c926c1c3facca213ec0618a3e') d from t1 order by d limit 5;
-select id,vec_distance(v, x'f618663f256be73e62cd453f8bcdbf3e16ae503c3858313f') d from t2 order by d limit 5;
+select id,vec_distance_euclidean(v, x'1f4d053f7056493f937da03dd8c97a3f220cbb3c926c1c3facca213ec0618a3e') d from t1 order by d limit 5;
+select id,vec_distance_euclidean(v, x'f618663f256be73e62cd453f8bcdbf3e16ae503c3858313f') d from t2 order by d limit 5;
 
 drop table t1, t2;
 exit;

sql/item_create.cc

@@ -6235,21 +6235,21 @@ Create_func_year_week::create_native(THD *thd, const LEX_CSTRING *name,
 }
 
 
-class Create_func_vec_distance: public Create_func_arg2
+class Create_func_vec_distance_euclidean: public Create_func_arg2
 {
 public:
   Item *create_2_arg(THD *thd, Item *arg1, Item *arg2) override
-  { return new (thd->mem_root) Item_func_vec_distance(thd, arg1, arg2); }
+  { return new (thd->mem_root) Item_func_vec_distance_euclidean(thd, arg1, arg2); }
 
-  static Create_func_vec_distance s_singleton;
+  static Create_func_vec_distance_euclidean s_singleton;
 
 protected:
-  Create_func_vec_distance() = default;
-  virtual ~Create_func_vec_distance() = default;
+  Create_func_vec_distance_euclidean() = default;
+  virtual ~Create_func_vec_distance_euclidean() = default;
 };
 
+Create_func_vec_distance_euclidean Create_func_vec_distance_euclidean::s_singleton;
 
-Create_func_vec_distance Create_func_vec_distance::s_singleton;
 
 class Create_func_vec_totext: public Create_func_arg1
 {
@@ -6510,7 +6510,7 @@ const Native_func_registry func_array[] =
   { { STRING_WITH_LEN("UPDATEXML") }, BUILDER(Create_func_xml_update)},
   { { STRING_WITH_LEN("UPPER") }, BUILDER(Create_func_ucase)},
   { { STRING_WITH_LEN("UUID_SHORT") }, BUILDER(Create_func_uuid_short)},
-  { { STRING_WITH_LEN("VEC_DISTANCE") }, BUILDER(Create_func_vec_distance)},
+  { { STRING_WITH_LEN("VEC_DISTANCE_EUCLIDEAN") }, BUILDER(Create_func_vec_distance_euclidean)},
   { { STRING_WITH_LEN("VEC_FROMTEXT") }, BUILDER(Create_func_vec_fromtext)},
   { { STRING_WITH_LEN("VEC_TOTEXT") }, BUILDER(Create_func_vec_totext)},
   { { STRING_WITH_LEN("VERSION") }, BUILDER(Create_func_version)},

sql/item_vectorfunc.cc

@@ -20,17 +20,9 @@
   This file defines all vector functions
 */
 
-#include <cmath>
-#include <my_global.h>
-
-#include "item.h"
 #include "item_vectorfunc.h"
-#include "json_lib.h"
-#include "m_ctype.h"
-#include "sql_const.h"
-#include "sql_error.h"
 
-key_map Item_func_vec_distance::part_of_sortkey() const
+key_map Item_func_vec_distance_common::part_of_sortkey() const
 {
   key_map map(0);
   if (Item_field *item= get_field_arg())
@@ -44,7 +36,7 @@ key_map Item_func_vec_distance::part_of_sortkey() const
   return map;
 }
 
-double Item_func_vec_distance::val_real()
+double Item_func_vec_distance_common::val_real()
 {
   String *r1= args[0]->val_str();
   String *r2= args[1]->val_str();
@@ -54,7 +46,7 @@ double Item_func_vec_distance::val_real()
     return 0;
   float *v1= (float *) r1->ptr();
   float *v2= (float *) r2->ptr();
-  return euclidean_vec_distance(v1, v2, (r1->length()) / sizeof(float));
+  return calc_distance(v1, v2, (r1->length()) / sizeof(float));
 }
 
 bool Item_func_vec_totext::fix_length_and_dec(THD *thd)
@@ -114,19 +106,6 @@ String *Item_func_vec_totext::val_str_ascii(String *str)
   return str;
 }
 
-double euclidean_vec_distance(float *v1, float *v2, size_t v_len)
-{
-  float *p1= v1;
-  float *p2= v2;
-  double d= 0;
-  for (size_t i= 0; i < v_len; p1++, p2++, i++)
-  {
-    float dist= *p1 - *p2;
-    d+= dist * dist;
-  }
-  return sqrt(d);
-}
-
 Item_func_vec_totext::Item_func_vec_totext(THD *thd, Item *a)
     : Item_str_ascii_checksum_func(thd, a)
 {

sql/item_vectorfunc.h

@@ -22,7 +22,7 @@
 #include "lex_string.h"
 #include "item_func.h"
 
-class Item_func_vec_distance: public Item_real_func
+class Item_func_vec_distance_common: public Item_real_func
 {
   Item_field *get_field_arg() const
   {
@@ -36,21 +36,17 @@ class Item_func_vec_distance: public Item_real_func
   {
     return check_argument_types_or_binary(NULL, 0, arg_count);
   }
+  virtual double calc_distance(float *v1, float *v2, size_t v_len) = 0;
 
 public:
-  Item_func_vec_distance(THD *thd, Item *a, Item *b)
+  Item_func_vec_distance_common(THD *thd, Item *a, Item *b)
    :Item_real_func(thd, a, b) {}
   bool fix_length_and_dec(THD *thd) override
   {
-    set_maybe_null();
+    set_maybe_null(); // if wrong dimensions
     return Item_real_func::fix_length_and_dec(thd);
   }
   double val_real() override;
-  LEX_CSTRING func_name_cstring() const override
-  {
-    static LEX_CSTRING name= { STRING_WITH_LEN("VEC_Distance") };
-    return name;
-  }
   Item *get_const_arg() const
   {
     if (args[0]->type() == Item::FIELD_ITEM && args[1]->const_item())
@@ -60,8 +56,32 @@ class Item_func_vec_distance: public Item_real_func
     return NULL;
   }
   key_map part_of_sortkey() const override;
+};
+
+
+class Item_func_vec_distance_euclidean: public Item_func_vec_distance_common
+{
+  double calc_distance(float *v1, float *v2, size_t v_len) override
+  {
+    double d= 0;
+    for (size_t i= 0; i < v_len; i++, v1++, v2++)
+    {
+      float dist= *v1 - *v2;
+      d+= dist * dist;
+    }
+    return sqrt(d);
+  }
+
+public:
+  Item_func_vec_distance_euclidean(THD *thd, Item *a, Item *b)
+   :Item_func_vec_distance_common(thd, a, b) {}
+  LEX_CSTRING func_name_cstring() const override
+  {
+    static LEX_CSTRING name= { STRING_WITH_LEN("VEC_DISTANCE_EUCLIDEAN") };
+    return name;
+  }
   Item *do_get_copy(THD *thd) const override
-  { return get_item_copy<Item_func_vec_distance>(thd, this); }
+  { return get_item_copy<Item_func_vec_distance_euclidean>(thd, this); }
 };
 
 
@@ -101,7 +121,4 @@ class Item_func_vec_fromtext: public Item_str_func
   Item *do_get_copy(THD *thd) const override
   { return get_item_copy<Item_func_vec_fromtext>(thd, this); }
 };
-
-
-double euclidean_vec_distance(float *v1, float *v2, size_t v_len);
 #endif

sql/vector_mhnsw.cc

@@ -1244,7 +1244,7 @@ int mhnsw_first(TABLE *table, KEY *keyinfo, Item *dist, ulonglong limit)
 {
   THD *thd= table->in_use;
   TABLE *graph= table->hlindex;
-  Item_func_vec_distance *fun= (Item_func_vec_distance *)dist;
+  auto *fun= (Item_func_vec_distance_euclidean *)dist;
   String buf, *res= fun->get_const_arg()->val_str(&buf);
   MHNSW_Context *ctx;