[MDEV-31414] Implement optional lengths for string types

mysql-test/main/varchar_no_length.cnf

@@ -0,0 +1,2 @@
+# Use default conf to set collation and charset
+!include include/default_my.cnf

mysql-test/main/varchar_no_length.result

@@ -0,0 +1,101 @@
+#
+# SQL Standard T081: VARCHAR without explicit length specification
+#
+#
+# Basic test: CREATE TABLE with VARCHAR without length or charset (scenario 1a)
+#
+CREATE TABLE t1 (a VARCHAR);
+SHOW CREATE TABLE t1;
+Table	Create Table
+t1	CREATE TABLE `t1` (
+  `a` varchar(16383) DEFAULT NULL
+) ENGINE=MyISAM DEFAULT CHARSET=utf8mb4 COLLATE=utf8mb4_uca1400_ai_ci
+DROP TABLE t1;
+
+#
+# Basic test: CREATE TABLE with charset (scenario 1b)
+#
+CREATE TABLE t1 (a VARCHAR CHARSET latin1);
+SHOW CREATE TABLE t1;
+Table	Create Table
+t1	CREATE TABLE `t1` (
+  `a` varchar(65532) CHARACTER SET latin1 COLLATE latin1_swedish_ci DEFAULT NULL
+) ENGINE=MyISAM DEFAULT CHARSET=utf8mb4 COLLATE=utf8mb4_uca1400_ai_ci
+DROP TABLE t1;
+
+#
+# Test VARCHAR without length with specific column charsets or collations (scenario 2a)
+#
+CREATE TABLE t1 (a VARCHAR CHARSET utf8mb4);
+SHOW CREATE TABLE t1;
+Table	Create Table
+t1	CREATE TABLE `t1` (
+  `a` varchar(16383) DEFAULT NULL
+) ENGINE=MyISAM DEFAULT CHARSET=utf8mb4 COLLATE=utf8mb4_uca1400_ai_ci
+DROP TABLE t1;
+
+CREATE TABLE t1 (a VARCHAR COLLATE utf8mb4_general_ci);
+SHOW CREATE TABLE t1;
+Table	Create Table
+t1	CREATE TABLE `t1` (
+  `a` varchar(16383) CHARACTER SET utf8mb4 COLLATE utf8mb4_general_ci DEFAULT NULL
+) ENGINE=MyISAM DEFAULT CHARSET=utf8mb4 COLLATE=utf8mb4_uca1400_ai_ci
+DROP TABLE t1;
+
+CREATE TABLE t1 (a VARCHAR CHARSET latin1);
+SHOW CREATE TABLE t1;
+Table	Create Table
+t1	CREATE TABLE `t1` (
+  `a` varchar(65532) CHARACTER SET latin1 COLLATE latin1_swedish_ci DEFAULT NULL
+) ENGINE=MyISAM DEFAULT CHARSET=utf8mb4 COLLATE=utf8mb4_uca1400_ai_ci
+DROP TABLE t1;
+
+CREATE TABLE t1 (a VARCHAR COLLATE latin1_general_ci);
+SHOW CREATE TABLE t1;
+Table	Create Table
+t1	CREATE TABLE `t1` (
+  `a` varchar(65532) CHARACTER SET latin1 COLLATE latin1_general_ci DEFAULT NULL
+) ENGINE=MyISAM DEFAULT CHARSET=utf8mb4 COLLATE=utf8mb4_uca1400_ai_ci
+DROP TABLE t1;
+
+#
+# Test VARCHAR without length with specific column charsets and collations (scenario 2b)
+#
+CREATE TABLE t1 (a VARCHAR CHARSET utf8mb4 COLLATE utf8mb4_general_ci);
+SHOW CREATE TABLE t1;
+Table	Create Table
+t1	CREATE TABLE `t1` (
+  `a` varchar(16383) CHARACTER SET utf8mb4 COLLATE utf8mb4_general_ci DEFAULT NULL
+) ENGINE=MyISAM DEFAULT CHARSET=utf8mb4 COLLATE=utf8mb4_uca1400_ai_ci
+DROP TABLE t1;
+
+CREATE TABLE t1 (a VARCHAR CHARSET latin1 COLLATE latin1_general_ci);
+SHOW CREATE TABLE t1;
+Table	Create Table
+t1	CREATE TABLE `t1` (
+  `a` varchar(65532) CHARACTER SET latin1 COLLATE latin1_general_ci DEFAULT NULL
+) ENGINE=MyISAM DEFAULT CHARSET=utf8mb4 COLLATE=utf8mb4_uca1400_ai_ci
+DROP TABLE t1;
+
+#
+# Basic test: CREATE TABLE with charset (Scenario 3)
+#
+CREATE TABLE t1 (a VARCHAR) CHARSET latin1;
+SHOW CREATE TABLE t1;
+Table	Create Table
+t1	CREATE TABLE `t1` (
+  `a` varchar(65532) DEFAULT NULL
+) ENGINE=MyISAM DEFAULT CHARSET=latin1 COLLATE=latin1_swedish_ci
+DROP TABLE t1;
+
+#
+# Test ALTER TABLE MODIFY COLUMN with VARCHAR without length
+#
+CREATE TABLE t1 (id VARCHAR(1));
+ALTER TABLE t1 MODIFY id VARCHAR;
+SHOW CREATE TABLE t1;
+Table	Create Table
+t1	CREATE TABLE `t1` (
+  `id` varchar(16383) DEFAULT NULL
+) ENGINE=MyISAM DEFAULT CHARSET=utf8mb4 COLLATE=utf8mb4_uca1400_ai_ci
+DROP TABLE t1;

mysql-test/main/varchar_no_length.test

@@ -0,0 +1,90 @@
+#
+# MDEV-31414
+# Test for SQL Standard T081: "Optional string types maximum length"
+# This tests the feature that allows VARCHAR columns to be created without
+# specifying a length, defaulting to the maximum VARCHAR length (65535)
+#
+# We have 3 scenarios for resolving charset and collation
+# 1. Empty
+# 2. Precisely Typed
+# 3. Contextually typed
+#
+# Check the comment section in "lex_charset.cc" at 
+# CHARSET_INFO *Lex_exact_charset_extended_collation_attrs_st::
+#                resolved_to_character_set
+
+--echo #
+--echo # SQL Standard T081: VARCHAR without explicit length specification
+--echo #
+
+--echo #
+--echo # Basic test: CREATE TABLE with VARCHAR without length or charset (scenario 1a)
+--echo #
+
+CREATE TABLE t1 (a VARCHAR);
+SHOW CREATE TABLE t1;
+DROP TABLE t1;
+
+--echo 
+--echo #
+--echo # Basic test: CREATE TABLE with charset (scenario 1b)
+--echo #
+
+CREATE TABLE t1 (a VARCHAR CHARSET latin1);
+SHOW CREATE TABLE t1;
+DROP TABLE t1;
+
+--echo
+--echo #
+--echo # Test VARCHAR without length with specific column charsets or collations (scenario 2a)
+--echo #
+
+CREATE TABLE t1 (a VARCHAR CHARSET utf8mb4);
+SHOW CREATE TABLE t1;
+DROP TABLE t1;
+
+--echo 
+CREATE TABLE t1 (a VARCHAR COLLATE utf8mb4_general_ci);
+SHOW CREATE TABLE t1;
+DROP TABLE t1;
+
+--echo 
+CREATE TABLE t1 (a VARCHAR CHARSET latin1);
+SHOW CREATE TABLE t1;
+DROP TABLE t1;
+
+--echo 
+CREATE TABLE t1 (a VARCHAR COLLATE latin1_general_ci);
+SHOW CREATE TABLE t1;
+DROP TABLE t1;
+
+--echo 
+--echo #
+--echo # Test VARCHAR without length with specific column charsets and collations (scenario 2b)
+--echo #
+
+CREATE TABLE t1 (a VARCHAR CHARSET utf8mb4 COLLATE utf8mb4_general_ci);
+SHOW CREATE TABLE t1;
+DROP TABLE t1;
+
+--echo 
+CREATE TABLE t1 (a VARCHAR CHARSET latin1 COLLATE latin1_general_ci);
+SHOW CREATE TABLE t1;
+DROP TABLE t1;
+
+--echo 
+--echo #
+--echo # Basic test: CREATE TABLE with charset (Scenario 3)
+--echo #
+CREATE TABLE t1 (a VARCHAR) CHARSET latin1;
+SHOW CREATE TABLE t1;
+DROP TABLE t1;
+
+--echo
+--echo #
+--echo # Test ALTER TABLE MODIFY COLUMN with VARCHAR without length
+--echo #
+CREATE TABLE t1 (id VARCHAR(1));
+ALTER TABLE t1 MODIFY id VARCHAR;
+SHOW CREATE TABLE t1;
+DROP TABLE t1;

sql/field.cc

@@ -10764,9 +10764,16 @@ void
 Column_definition_attributes::set_length_and_dec(const Lex_length_and_dec_st
                                                  &type)
 {
-  if (type.has_explicit_length())
+  if (type.has_explicit_length()) 
+  {
     length= type.length_overflowed() ? (ulonglong) UINT_MAX32 + 1 :
                                        (ulonglong) type.length();
+    has_explicit_length= true;
+  } 
+  else {
+    length= 0;
+    has_explicit_length= false;
+  }
 
   if (type.has_explicit_dec())
     decimals= type.dec();
@@ -11128,7 +11135,8 @@ Column_definition_attributes::Column_definition_attributes(const Field *field)
   charset(field->charset()), // May be NULL ptr
   pack_flag(0),
   decimals(field->decimals()),
-  unireg_check(field->unireg_check)
+  unireg_check(field->unireg_check),
+  has_explicit_length(true)
 {}
 
 
@@ -11139,7 +11147,8 @@ Column_definition_attributes::
   charset(attr.collation.collation),
   pack_flag(attr.unsigned_flag ? 0 : FIELDFLAG_DECIMAL),
   decimals(attr.decimals),
-  unireg_check(Field::NONE)
+  unireg_check(Field::NONE),
+  has_explicit_length(true)
 {}
 
 
@@ -11163,6 +11172,7 @@ Column_definition::Column_definition(THD *thd, Field *old_field,
   invisible= old_field->invisible;
   interval_list.empty(); // prepare_interval_field() needs this
   char_length= (uint) length;
+  has_explicit_length= true;
 
   if (orig_field)
   {
@@ -11255,6 +11265,7 @@ Column_definition::redefine_stage1_common(const Column_definition *dup_field,
   comment=      dup_field->comment;
   option_list=  dup_field->option_list;
   versioning=   dup_field->versioning;
+  has_explicit_length= dup_field->has_explicit_length;
 }
 
 

sql/field.h

@@ -5339,12 +5339,14 @@ class Column_definition_attributes: public Type_extra_attributes
   uint32 pack_flag;
   decimal_digits_t decimals;
   Field::utype unireg_check;
+  bool has_explicit_length;
   Column_definition_attributes()
    :length(0),
     charset(&my_charset_bin),
     pack_flag(0),
     decimals(0),
-    unireg_check(Field::NONE)
+    unireg_check(Field::NONE),
+    has_explicit_length(true)
   { }
   Column_definition_attributes(const Field *field);
   Column_definition_attributes(const Type_all_attributes &attr);

sql/sql_table.cc

@@ -2581,6 +2581,12 @@ bool Column_definition::prepare_stage1_typelib(THD *thd,
 bool Column_definition::prepare_stage1_string(THD *thd,
                                               MEM_ROOT *mem_root)
 {
+  if (real_field_type() == MYSQL_TYPE_VARCHAR && !this->has_explicit_length)
+  {
+    char_length= (MAX_FIELD_VARCHARLENGTH) / (charset->mbmaxlen);
+    length= char_length;
+  }
+
   if (real_field_type() == FIELD_TYPE_STRING &&
       length*charset->mbmaxlen > 1024)
   {
@@ -3220,7 +3226,7 @@ static bool mysql_prepare_create_table_stage1(THD *thd,
                                   COLUMN_DEFINITION_TABLE_FIELD,
                                   &dattr))
       DBUG_RETURN(true);
-
+    
     DBUG_ASSERT(sql_field->charset);
 
     if (check_column_name(sql_field->field_name))

sql/sql_type.cc

@@ -2862,8 +2862,14 @@ Type_handler_varchar::Column_definition_set_attributes(
     }
     break;
   case COLUMN_DEFINITION_ROUTINE_LOCAL:
-  case COLUMN_DEFINITION_TABLE_FIELD:
     break;
+  case COLUMN_DEFINITION_TABLE_FIELD:
+    /*
+      Support SQL Standard T081: "Optional string types maximum length"
+      Allows users to specify VARCHAR fields without a length
+      In this case, has_explicit_length is false.
+    */
+    return false;
   }
   thd->parse_error();
   return true;