MDEV-38180: Add xxh32() and xxh3() SQL functions

mysql-test/main/func_xxh.result

@@ -0,0 +1,44 @@
+SELECT XXH32('abc') = '80712ed5' AS xxh32_expected;
+xxh32_expected
+1
+SELECT XXH32(11223344) = '1852349e' AS xxh32_num;
+xxh32_num
+1
+SELECT XXH32('abc') = XXH32('abc') AS xxh32_eq;
+xxh32_eq
+1
+SELECT XXH32(NULL) IS NULL AS xxh32_null;
+xxh32_null
+1
+SELECT XXH32('') = '00000000' AS xxh32_empty;
+xxh32_empty
+1
+SELECT XXH3('abc') = '244da40f405c870e' AS xxh3_expected;
+xxh3_expected
+1
+SELECT XXH3(11223344) = '8178f6e2d84ca479' AS xxh3_num;
+xxh3_num
+1
+SELECT XXH3('abc') = XXH32('abc') AS xxh3_eq;
+xxh3_eq
+0
+SELECT XXH3(NULL) IS NULL AS xxh3_null;
+xxh3_null
+1
+SELECT XXH3('') = '0000000000000000' AS xxh3_empty;
+xxh3_empty
+1
+SELECT XXH32(' ') = XXH32('    ') AS xxh32_space_eq;
+xxh32_space_eq
+1
+SELECT XXH3(' ') = XXH3('    ') AS xxh3_space_eq;
+xxh3_space_eq
+1
+SELECT XXH32(_koi8u 0x20 COLLATE koi8u_general_ci) =
+XXH32(_koi8u 0x60 COLLATE koi8u_general_ci) AS xxh32_koi8u_eq;
+xxh32_koi8u_eq
+1
+SELECT XXH3(_koi8u 0x20 COLLATE koi8u_general_ci) =
+XXH3(_koi8u 0x60 COLLATE koi8u_general_ci) AS xxh3_koi8u_eq;
+xxh3_koi8u_eq
+1

mysql-test/main/func_xxh.test

@@ -0,0 +1,19 @@
+SELECT XXH32('abc') = '80712ed5' AS xxh32_expected;
+SELECT XXH32(11223344) = '1852349e' AS xxh32_num;
+SELECT XXH32('abc') = XXH32('abc') AS xxh32_eq;
+SELECT XXH32(NULL) IS NULL AS xxh32_null;
+SELECT XXH32('') = '00000000' AS xxh32_empty;
+
+SELECT XXH3('abc') = '244da40f405c870e' AS xxh3_expected;
+SELECT XXH3(11223344) = '8178f6e2d84ca479' AS xxh3_num;
+SELECT XXH3('abc') = XXH32('abc') AS xxh3_eq;
+SELECT XXH3(NULL) IS NULL AS xxh3_null;
+SELECT XXH3('') = '0000000000000000' AS xxh3_empty;
+
+SELECT XXH32(' ') = XXH32('    ') AS xxh32_space_eq;
+SELECT XXH3(' ') = XXH3('    ') AS xxh3_space_eq;
+
+SELECT XXH32(_koi8u 0x20 COLLATE koi8u_general_ci) =
+XXH32(_koi8u 0x60 COLLATE koi8u_general_ci) AS xxh32_koi8u_eq;
+SELECT XXH3(_koi8u 0x20 COLLATE koi8u_general_ci) =
+XXH3(_koi8u 0x60 COLLATE koi8u_general_ci) AS xxh3_koi8u_eq;

sql/item_create.cc

@@ -633,6 +633,25 @@ class Create_func_crc32c : public Create_native_func
   virtual ~Create_func_crc32c() = default;
 };
 
+class Create_func_xxh32 : public Create_func_arg1
+{
+public:
+  Item *create_1_arg(THD *thd, Item *arg1) override;
+  static Create_func_xxh32 s_singleton;
+protected:
+  Create_func_xxh32() = default;
+  ~Create_func_xxh32() override = default;
+};
+
+class Create_func_xxh3 : public Create_func_arg1
+{
+public:
+  Item *create_1_arg(THD *thd, Item *arg1) override;
+  static Create_func_xxh3 s_singleton;
+protected:
+  Create_func_xxh3() = default;
+  ~Create_func_xxh3() override = default;
+};
 
 class Create_func_datediff : public Create_func_arg2
 {
@@ -3682,6 +3701,19 @@ Create_func_crc32c::create_native(THD *thd, const LEX_CSTRING *name,
     : new (thd->mem_root) Item_func_crc32(thd, true, arg1);
 }
 
+Create_func_xxh32 Create_func_xxh32::s_singleton;
+
+Item *Create_func_xxh32::create_1_arg(THD *thd, Item *arg1)
+{
+  return new (thd->mem_root) Item_func_xxh32(thd, arg1);
+}
+
+Create_func_xxh3 Create_func_xxh3::s_singleton;
+
+Item *Create_func_xxh3::create_1_arg(THD *thd, Item *arg1)
+{
+  return new (thd->mem_root) Item_func_xxh3(thd, arg1);
+}
 
 Create_func_datediff Create_func_datediff::s_singleton;
 
@@ -6338,6 +6370,8 @@ const Native_func_registry func_array[] =
   { { STRING_WITH_LEN("COT") }, BUILDER(Create_func_cot)},
   { { STRING_WITH_LEN("CRC32") }, BUILDER(Create_func_crc32)},
   { { STRING_WITH_LEN("CRC32C") }, BUILDER(Create_func_crc32c)},
+  { { STRING_WITH_LEN("XXH32") }, BUILDER(Create_func_xxh32) },
+  { { STRING_WITH_LEN("XXH3") }, BUILDER(Create_func_xxh3) },
   { { STRING_WITH_LEN("DATABASE") }, BUILDER(Create_func_database)},
   { { STRING_WITH_LEN("DATEDIFF") }, BUILDER(Create_func_datediff)},
   { { STRING_WITH_LEN("DATE_FORMAT") }, BUILDER(Create_func_date_format)},

sql/item_strfunc.cc

@@ -43,6 +43,7 @@
                                 // my_make_scrambled_password_323
 #include <m_ctype.h>
 #include <my_md5.h>
+#include "../mysys/xxhash.h"
 C_MODE_START
 #include "../mysys/my_static.h"			// For soundex_map
 C_MODE_END
@@ -4570,6 +4571,76 @@ longlong Item_func_crc32::val_int()
     (ulonglong{crc_func(uint32_t(crc), res->ptr(), res->length())});
 }
 
+namespace
+{
+void bytes_to_hex_lower(const unsigned char *digest, size_t length, String *to)
+{
+  static const char hex[]= "0123456789abcdef";
+  char buffer[32 * 2];
+
+  DBUG_ASSERT(length * 2 <= sizeof(buffer));
+
+  for (size_t i= 0; i < length; ++i)
+  {
+    buffer[i * 2]= hex[digest[i] >> 4];
+    buffer[i * 2 + 1]= hex[digest[i] & 0x0f];
+  }
+
+  to->copy(buffer, length * 2, &my_charset_latin1);
+}
+} // namespace
+
+String *Item_func_xxh32::val_str_ascii(String *to)
+{
+  DBUG_ASSERT(fixed());
+  DBUG_ASSERT(arg_count == 1);
+
+  String *input= args[0]->val_str(&value);
+  if (!input)
+  { 
+    null_value= true;
+    return nullptr; 
+  }
+  null_value= false;
+  my_hasher_st hasher= my_hasher_xxh32();
+  input->charset()->hash_sort(
+    &hasher,
+    reinterpret_cast<const uchar *>(input->ptr()),
+    input->length());
+  const uint64_t hash= hasher.m_finalize(&hasher);
+
+  XXH32_canonical_t canonical;
+  XXH32_canonicalFromHash(&canonical, (XXH32_hash_t) hash);
+  bytes_to_hex_lower(canonical.digest, sizeof(canonical.digest), to);
+  return to;
+}
+
+String *Item_func_xxh3::val_str_ascii(String *to)
+{
+  DBUG_ASSERT(fixed());
+  DBUG_ASSERT(arg_count == 1);
+
+  String *input= args[0]->val_str(&value);
+  if (!input)
+  { 
+    null_value= true;
+    return nullptr; 
+  }
+  null_value= false;
+  my_hasher_st hasher= my_hasher_xxh3();
+  input->charset()->hash_sort(
+    &hasher,
+    reinterpret_cast<const uchar *>(input->ptr()),
+    input->length());
+
+  const uint64_t hash= hasher.m_finalize(&hasher);
+
+  XXH64_canonical_t canonical;
+  XXH64_canonicalFromHash(&canonical, (XXH64_hash_t) hash);
+  bytes_to_hex_lower(canonical.digest, sizeof(canonical.digest), to);
+  return to;
+}
+
 #ifdef HAVE_COMPRESS
 #include "zlib.h"
 

sql/item_strfunc.h

@@ -2317,6 +2317,64 @@ class Item_func_crc32 :public Item_long_func
   { return get_item_copy<Item_func_crc32>(thd, this); }
 };
 
+class Item_func_xxh32 : public Item_str_ascii_checksum_func
+{
+  String value;
+
+public:
+  Item_func_xxh32(THD *thd, Item *arg) : Item_str_ascii_checksum_func(thd, arg)
+  {
+  }
+
+  String *val_str_ascii(String *to) override;
+
+  bool fix_length_and_dec(THD *) override
+  {
+    fix_length_and_charset(8, default_charset());
+    return false;
+  }
+
+  LEX_CSTRING func_name_cstring() const override
+  {
+    static LEX_CSTRING name= {STRING_WITH_LEN("XXH32")};
+    return name;
+  }
+
+  Item *shallow_copy(THD *thd) const override
+  {
+    return get_item_copy<Item_func_xxh32>(thd, this);
+  }
+};
+
+class Item_func_xxh3 : public Item_str_ascii_checksum_func
+{
+  String value;
+
+public:
+  Item_func_xxh3(THD *thd, Item *arg) : Item_str_ascii_checksum_func(thd, arg)
+  {
+  }
+
+  String *val_str_ascii(String *to) override;
+
+  bool fix_length_and_dec(THD *) override
+  {
+    fix_length_and_charset(16, default_charset());
+    return false;
+  }
+
+  LEX_CSTRING func_name_cstring() const override
+  {
+    static LEX_CSTRING name= {STRING_WITH_LEN("XXH3")};
+    return name;
+  }
+
+  Item *shallow_copy(THD *thd) const override
+  {
+    return get_item_copy<Item_func_xxh3>(thd, this);
+  }
+};
+
 class Item_func_uncompressed_length : public Item_long_func_length
 {
   String value;

strings/hasher-xxh.c

@@ -48,6 +48,7 @@ my_hasher_st my_hasher_xxh32(void)
     { {.m_nr = 0}, FALSE, my_hasher_xxh32_hash_str, my_hasher_xxh32_hash_byte,
       my_hasher_hash_num, my_hasher_xxh32_finalize,
       (void *) XXH32_createState() };
+  XXH32_reset((XXH32_state_t *) tmp.m_specific, 0);
   return tmp;
 }
 
@@ -91,5 +92,6 @@ my_hasher_st my_hasher_xxh3(void)
     { {.m_nr = 0}, FALSE, my_hasher_xxh3_hash_str, my_hasher_xxh3_hash_byte,
       my_hasher_hash_num, my_hasher_xxh3_finalize,
       (void *) XXH3_createState() };
+  XXH3_64bits_reset((XXH3_state_t *) tmp.m_specific);
   return tmp;
 }