A cleanup for MDEV-27896 Wrong result upon `COLLATE latin1_bin CHARACTER SET...

A cleanup for MDEV-27896 Wrong result upon `COLLATE latin1_bin CHARACTER SET latin1` on the table or the database level

Changing the error messages in a statement like this:

CREATE DATABASE db1
         COLLATE utf8mb4_bin
         CHARACTER SET utf8mb4
         CHARACTER SET latin1;

from
  COLLATION 'utf8mb4_bin' is not valid for CHARACTER SET 'latin1'

to a more expected:

  Conflicting declarations: 'CHARACTER SET utf8mb4' and 'CHARACTER SET latin1'

In order to do this:
- Adding a new type TYPE_CHARACTER_SET_COLLATE_EXACT into
  Lex_exact_charset_extended_collation_attrs_st

- Removing m_had_charset_exact from its descendant class
  Lex_extended_charset_extended_collation_attrs_st

Additional cleanup:
- Changing methods in Lex_exact_charset_extended_collation_attrs_st
  set_charset(), set_charset_collate_default(), set_charset_collate_binary()
  to get Lex_exact_charset instead CHARSET_INFO as a parameter,
  to guarantee that the argument is only CHARACTER SET and does not have
  any COLLATE clauses yet. This change is not directly related to
  the error message change.

mysql-test/main/ctype_collate_context.result

@@ -654,11 +654,11 @@ result	ERROR: Conflicting declarations: 'CHARACTER SET utf8mb4' and 'CHARACTER S
 	
 query	CREATE DATABASE
 attrs	COLLATE DEFAULT CHARACTER SET utf8mb4 CHARACTER SET latin1
-result	ERROR: COLLATION 'utf8mb4_general_ci' is not valid for CHARACTER SET 'latin1'
+result	ERROR: Conflicting declarations: 'CHARACTER SET utf8mb4' and 'CHARACTER SET latin1'
 	
 query	CREATE DATABASE
 attrs	COLLATE utf8mb4_bin CHARACTER SET utf8mb4 CHARACTER SET latin1
-result	ERROR: COLLATION 'utf8mb4_bin' is not valid for CHARACTER SET 'latin1'
+result	ERROR: Conflicting declarations: 'CHARACTER SET utf8mb4' and 'CHARACTER SET latin1'
 	
 query	CREATE DATABASE
 attrs	COLLATE latin1_swedish_ci CHARACTER SET utf8mb4 CHARACTER SET latin1
@@ -674,7 +674,7 @@ result	ERROR: Conflicting declarations: 'CHARACTER SET DEFAULT (utf8mb4)' and 'C
 	
 query	CREATE DATABASE
 attrs	CHARACTER SET utf8mb4 COLLATE DEFAULT CHARACTER SET latin1
-result	ERROR: COLLATION 'utf8mb4_general_ci' is not valid for CHARACTER SET 'latin1'
+result	ERROR: Conflicting declarations: 'CHARACTER SET utf8mb4' and 'CHARACTER SET latin1'
 	
 query	CREATE DATABASE
 attrs	COLLATE utf8mb4_bin COLLATE DEFAULT CHARACTER SET latin1
@@ -694,7 +694,7 @@ result	ERROR: COLLATION 'utf8mb4_bin' is not valid for CHARACTER SET 'latin1'
 	
 query	CREATE DATABASE
 attrs	CHARACTER SET utf8mb4 COLLATE utf8mb4_bin CHARACTER SET latin1
-result	ERROR: COLLATION 'utf8mb4_bin' is not valid for CHARACTER SET 'latin1'
+result	ERROR: Conflicting declarations: 'CHARACTER SET utf8mb4' and 'CHARACTER SET latin1'
 	
 query	CREATE DATABASE
 attrs	COLLATE DEFAULT COLLATE utf8mb4_bin CHARACTER SET latin1
@@ -774,7 +774,7 @@ result	ERROR: Conflicting declarations: 'CHARACTER SET utf8mb4' and 'CHARACTER S
 	
 query	CREATE DATABASE
 attrs	COLLATE DEFAULT CHARACTER SET latin1 CHARACTER SET utf8mb4
-result	ERROR: COLLATION 'latin1_swedish_ci' is not valid for CHARACTER SET 'utf8mb4'
+result	ERROR: Conflicting declarations: 'CHARACTER SET latin1' and 'CHARACTER SET utf8mb4'
 	
 query	CREATE DATABASE
 attrs	COLLATE utf8mb4_bin CHARACTER SET latin1 CHARACTER SET utf8mb4
@@ -782,11 +782,11 @@ result	ERROR: COLLATION 'utf8mb4_bin' is not valid for CHARACTER SET 'latin1'
 	
 query	CREATE DATABASE
 attrs	COLLATE latin1_swedish_ci CHARACTER SET latin1 CHARACTER SET utf8mb4
-result	ERROR: COLLATION 'latin1_swedish_ci' is not valid for CHARACTER SET 'utf8mb4'
+result	ERROR: Conflicting declarations: 'CHARACTER SET latin1' and 'CHARACTER SET utf8mb4'
 	
 query	CREATE DATABASE
 attrs	COLLATE latin1_bin CHARACTER SET latin1 CHARACTER SET utf8mb4
-result	ERROR: COLLATION 'latin1_bin' is not valid for CHARACTER SET 'utf8mb4'
+result	ERROR: Conflicting declarations: 'CHARACTER SET latin1' and 'CHARACTER SET utf8mb4'
 	
 query	CREATE DATABASE
 attrs	CHARACTER SET latin1 CHARACTER SET utf8mb4 CHARACTER SET utf8mb4
@@ -802,7 +802,7 @@ result	ERROR: COLLATION 'latin1_bin' is not valid for CHARACTER SET 'utf8mb4'
 	
 query	CREATE DATABASE
 attrs	CHARACTER SET latin1 COLLATE DEFAULT CHARACTER SET utf8mb4
-result	ERROR: COLLATION 'latin1_swedish_ci' is not valid for CHARACTER SET 'utf8mb4'
+result	ERROR: Conflicting declarations: 'CHARACTER SET latin1' and 'CHARACTER SET utf8mb4'
 	
 query	CREATE DATABASE
 attrs	COLLATE utf8mb4_bin COLLATE DEFAULT CHARACTER SET utf8mb4
@@ -838,7 +838,7 @@ result	ERROR: COLLATION 'latin1_swedish_ci' is not valid for CHARACTER SET 'utf8
 	
 query	CREATE DATABASE
 attrs	CHARACTER SET latin1 COLLATE latin1_swedish_ci CHARACTER SET utf8mb4
-result	ERROR: COLLATION 'latin1_swedish_ci' is not valid for CHARACTER SET 'utf8mb4'
+result	ERROR: Conflicting declarations: 'CHARACTER SET latin1' and 'CHARACTER SET utf8mb4'
 	
 query	CREATE DATABASE
 attrs	CHARACTER SET utf8mb4 COLLATE latin1_swedish_ci CHARACTER SET utf8mb4
@@ -866,7 +866,7 @@ result	ERROR: COLLATION 'latin1_bin' is not valid for CHARACTER SET 'utf8mb4'
 	
 query	CREATE DATABASE
 attrs	CHARACTER SET latin1 COLLATE latin1_bin CHARACTER SET utf8mb4
-result	ERROR: COLLATION 'latin1_bin' is not valid for CHARACTER SET 'utf8mb4'
+result	ERROR: Conflicting declarations: 'CHARACTER SET latin1' and 'CHARACTER SET utf8mb4'
 	
 query	CREATE DATABASE
 attrs	CHARACTER SET utf8mb4 COLLATE latin1_bin CHARACTER SET utf8mb4

sql/lex_charset.cc

@@ -72,6 +72,20 @@ bool Lex_exact_charset::
 }
 
 
+bool Lex_exact_charset_opt_extended_collate::
+  raise_if_charsets_differ(const Lex_exact_charset &cs) const
+{
+  if (!my_charset_same(m_ci, cs.charset_info()))
+  {
+    my_error(ER_CONFLICTING_DECLARATIONS, MYF(0),
+             "CHARACTER SET ", m_ci->cs_name.str,
+             "CHARACTER SET ", cs.charset_info()->cs_name.str);
+    return true;
+  }
+  return false;
+}
+
+
 bool Lex_exact_charset_opt_extended_collate::
   raise_if_not_applicable(const Lex_exact_collation &cl) const
 {
@@ -386,8 +400,7 @@ CHARSET_INFO *Lex_exact_charset_extended_collation_attrs_st::
   case TYPE_EMPTY:
     return def;
   case TYPE_CHARACTER_SET:
-    DBUG_ASSERT(m_ci);
-    return m_ci;
+  case TYPE_CHARACTER_SET_COLLATE_EXACT:
   case TYPE_COLLATE_EXACT:
     DBUG_ASSERT(m_ci);
     return m_ci;
@@ -418,11 +431,13 @@ bool Lex_exact_charset_extended_collation_attrs_st::
   case TYPE_CHARACTER_SET:
     {
       // CHARACTER SET latin1 .. COLLATE latin1_swedish_ci
-      if (Lex_exact_charset(m_ci).raise_if_not_applicable(cl))
+      Lex_exact_charset_opt_extended_collate tmp(m_ci, false);
+      if (tmp.merge_exact_collation(cl))
         return true;
-      *this= Lex_exact_charset_extended_collation_attrs(cl);
+      *this= Lex_exact_charset_extended_collation_attrs(tmp);
       return false;
     }
+  case TYPE_CHARACTER_SET_COLLATE_EXACT:
   case TYPE_COLLATE_EXACT:
     {
       // [CHARACTER SET latin1] COLLATE latin1_bin .. COLLATE latin1_bin
@@ -460,9 +475,10 @@ bool Lex_exact_charset_extended_collation_attrs_st::
       Lex_exact_charset_opt_extended_collate tmp(m_ci, false);
       if (tmp.merge_context_collation(cl))
         return true;
-      *this= Lex_exact_charset_extended_collation_attrs(tmp.collation());
+      *this= Lex_exact_charset_extended_collation_attrs(tmp);
       return false;
     }
+  case TYPE_CHARACTER_SET_COLLATE_EXACT:
   case TYPE_COLLATE_EXACT:
     // [CHARACTER SET latin1] COLLATE latin1_swedish_ci .. COLLATE DEFAULT
     return Lex_exact_collation(m_ci).
@@ -538,7 +554,15 @@ bool Lex_exact_charset_extended_collation_attrs_st::
 
   case TYPE_COLLATE_EXACT:
     // COLLATE latin1_bin .. CHARACTER SET cs
-    return cs.raise_if_not_applicable(Lex_exact_collation(m_ci));
+    if (cs.raise_if_not_applicable(Lex_exact_collation(m_ci)))
+      return true;
+    m_type= TYPE_CHARACTER_SET_COLLATE_EXACT;
+    return false;
+
+  case TYPE_CHARACTER_SET_COLLATE_EXACT:
+    // CHARACTER SET cs1 COLLATE cl .. CHARACTER SET cs2
+    return Lex_exact_charset_opt_extended_collate(m_ci, true).
+             raise_if_charsets_differ(cs);
 
   case TYPE_COLLATE_CONTEXTUALLY_TYPED:
     // COLLATE DEFAULT .. CHARACTER SET cs
@@ -546,7 +570,7 @@ bool Lex_exact_charset_extended_collation_attrs_st::
       Lex_exact_charset_opt_extended_collate tmp(cs);
       if (tmp.merge_context_collation(Lex_context_collation(m_ci)))
         return true;
-      *this= Lex_exact_charset_extended_collation_attrs(tmp.collation());
+      *this= Lex_exact_charset_extended_collation_attrs(tmp);
       return false;
     }
   }
@@ -567,7 +591,6 @@ bool Lex_extended_charset_extended_collation_attrs_st::merge_charset_default()
 bool Lex_extended_charset_extended_collation_attrs_st::
        merge_exact_charset(const Lex_exact_charset &cs)
 {
-  m_had_charset_exact= true;
   if (m_charset_order == CHARSET_TYPE_EMPTY)
     m_charset_order= CHARSET_TYPE_EXACT;
   return Lex_exact_charset_extended_collation_attrs_st::merge_exact_charset(cs);
@@ -621,6 +644,7 @@ Lex_extended_charset_extended_collation_attrs_st::
     }
     return m_ci;
 
+  case TYPE_CHARACTER_SET_COLLATE_EXACT:
   case TYPE_COLLATE_EXACT:
   {
     /*
@@ -629,7 +653,7 @@ Lex_extended_charset_extended_collation_attrs_st::
         [ CHARACTER SET cs_exact ]
         CHARACTER SET DEFAULT;
     */
-    if (m_had_charset_exact &&
+    if (m_type == TYPE_CHARACTER_SET_COLLATE_EXACT &&
         raise_if_charset_conflicts_with_default(ctx.charset_default()))
     {
       /*

sql/lex_charset.h

@@ -203,6 +203,7 @@ class Lex_exact_charset_opt_extended_collate
   bool with_collate() const { return m_with_collate; }
   CHARSET_INFO *find_bin_collation() const;
   CHARSET_INFO *find_default_collation() const;
+  bool raise_if_charsets_differ(const Lex_exact_charset &cs) const;
   bool raise_if_not_applicable(const Lex_exact_collation &cl) const;
   /*
     Add another COLLATE clause (exact or context).
@@ -281,14 +282,17 @@ struct Lex_exact_charset_extended_collation_attrs_st
     TYPE_EMPTY= 0,
     TYPE_CHARACTER_SET= 1,
     TYPE_COLLATE_EXACT= 2,
-    TYPE_COLLATE_CONTEXTUALLY_TYPED= 3
+    TYPE_CHARACTER_SET_COLLATE_EXACT= 3,
+    TYPE_COLLATE_CONTEXTUALLY_TYPED= 4
   };
 
 // Number of bits required to store enum Type values
 
-#define LEX_CHARSET_COLLATION_TYPE_BITS 2
-  static_assert(((1<<LEX_CHARSET_COLLATION_TYPE_BITS)-1) >=
-                 TYPE_COLLATE_CONTEXTUALLY_TYPED,
+#define LEX_CHARSET_COLLATION_TYPE_BITS 3
+#define LEX_CHARSET_COLLATION_TYPE_MASK ((1<<LEX_CHARSET_COLLATION_TYPE_BITS)-1)
+
+  static_assert(LEX_CHARSET_COLLATION_TYPE_MASK >=
+                TYPE_COLLATE_CONTEXTUALLY_TYPED,
                  "Lex_exact_charset_extended_collation_attrs_st::Type bits");
 
 protected:
@@ -330,37 +334,37 @@ struct Lex_exact_charset_extended_collation_attrs_st
   }
   void init(const Lex_exact_charset_opt_extended_collate &cscl)
   {
-    cscl.with_collate() ? init(cscl.collation()) :
-                          init(cscl.charset());
+    if (cscl.with_collate())
+      init(cscl.collation().charset_info(), TYPE_CHARACTER_SET_COLLATE_EXACT);
+    else
+      init(cscl.charset());
   }
   bool is_empty() const
   {
     return m_type == TYPE_EMPTY;
   }
-  void set_charset(CHARSET_INFO *cs)
+  void set_charset(const Lex_exact_charset &cs)
   {
-    DBUG_ASSERT(cs);
-    m_ci= cs;
+    m_ci= cs.charset_info();
     m_type= TYPE_CHARACTER_SET;
   }
-  bool set_charset_collate_default(CHARSET_INFO *cs)
+  bool set_charset_collate_default(const Lex_exact_charset &cs)
   {
-    DBUG_ASSERT(cs);
-    if (!(cs= Lex_exact_charset_opt_extended_collate(cs, true).
+    CHARSET_INFO *ci;
+    if (!(ci= Lex_exact_charset_opt_extended_collate(cs).
                 find_default_collation()))
       return true;
-    m_ci= cs;
-    m_type= TYPE_COLLATE_EXACT;
+    m_ci= ci;
+    m_type= TYPE_CHARACTER_SET_COLLATE_EXACT;
     return false;
   }
-  bool set_charset_collate_binary(CHARSET_INFO *cs)
+  bool set_charset_collate_binary(const Lex_exact_charset &cs)
   {
-    DBUG_ASSERT(cs);
-    if (!(cs= Lex_exact_charset_opt_extended_collate(cs, true).
-                find_bin_collation()))
+    CHARSET_INFO *ci;
+    if (!(ci= Lex_exact_charset_opt_extended_collate(cs).find_bin_collation()))
       return true;
-    m_ci= cs;
-    m_type= TYPE_COLLATE_EXACT;
+    m_ci= ci;
+    m_type= TYPE_CHARACTER_SET_COLLATE_EXACT;
     return false;
   }
   void set_collate_default()
@@ -407,6 +411,7 @@ struct Lex_exact_charset_extended_collation_attrs_st
       return merge_exact_collation(Lex_exact_collation(cl.charset_info()));
     case TYPE_COLLATE_CONTEXTUALLY_TYPED:
     case TYPE_CHARACTER_SET:
+    case TYPE_CHARACTER_SET_COLLATE_EXACT:
       break;
     }
     DBUG_ASSERT(0);
@@ -430,6 +435,7 @@ struct Lex_exact_charset_extended_collation_attrs_st
       return merge_exact_collation(Lex_exact_collation(cl.charset_info()));
     case TYPE_COLLATE_CONTEXTUALLY_TYPED:
     case TYPE_CHARACTER_SET:
+    case TYPE_CHARACTER_SET_COLLATE_EXACT:
       break;
     }
     DBUG_ASSERT(0);
@@ -527,33 +533,18 @@ class Lex_extended_charset_extended_collation_attrs_st:
     order of CHARACTER SET clauses in case of conflicts.
   */
   charset_type_t m_charset_order;
-  /*
-    The parent class Lex_exact_charset_extended_collation_attrs_st
-    does not let know if a "COLLATE cl_exact" was used in combination with
-    "CHARACTER SET cs_exact" or just alone.
-    Here we need to distinguish:
-    - CHARACTER SET cs_exact COLLATE cl_exact, or
-    - COLLATE cl_exact CHARACTER SET cs_exact
-    versus just:
-    - COLLATE cl_exact
-    to produce better error messages in case of conflicts.
-    So let's add a flag member:
-  */
-  bool m_had_charset_exact;
 public:
   void init()
   {
     Lex_opt_context_charset_st::init();
     Lex_exact_charset_extended_collation_attrs_st::init();
     m_charset_order= CHARSET_TYPE_EMPTY;
-    m_had_charset_exact= false;
   }
   void init(const Lex_exact_charset_opt_extended_collate &c)
   {
     Lex_opt_context_charset_st::init();
     Lex_exact_charset_extended_collation_attrs_st::init(c);
     m_charset_order= CHARSET_TYPE_EXACT;
-    m_had_charset_exact= true;
   }
   bool is_empty() const
   {

sql/sql_yacc.yy

@@ -6506,22 +6506,28 @@ opt_binary:
         ;
 
 binary:
-          BYTE_SYM                     { $$.set_charset(&my_charset_bin); }
-        | charset_or_alias             { $$.set_charset($1); }
+          BYTE_SYM
+          {
+            $$.set_charset(Lex_exact_charset(&my_charset_bin));
+          }
+        | charset_or_alias
+          {
+            $$.set_charset(Lex_exact_charset($1));
+          }
         | charset_or_alias BINARY
           {
-            if ($$.set_charset_collate_binary($1))
+            if ($$.set_charset_collate_binary(Lex_exact_charset($1)))
               MYSQL_YYABORT;
           }
         | BINARY { $$.set_contextually_typed_binary_style(); }
         | BINARY charset_or_alias
           {
-            if ($$.set_charset_collate_binary($2))
+            if ($$.set_charset_collate_binary(Lex_exact_charset($2)))
               MYSQL_YYABORT;
           }
         | charset_or_alias COLLATE_SYM DEFAULT
           {
-            $$.set_charset_collate_default($1);
+            $$.set_charset_collate_default(Lex_exact_charset($1));
           }
         | charset_or_alias COLLATE_SYM collation_name
           {

sql/structs.h

@@ -704,7 +704,9 @@ struct Lex_field_type_st: public Lex_length_and_dec_st
     m_handler= handler;
     m_ci= coll.charset_info();
     Lex_length_and_dec_st::operator=(length_and_dec);
-    m_collation_type= ((uint8) coll.type()) & 0x3;
+    // Using bit-and to avoid the warning:
+    // conversion from ‘uint8’ to ‘unsigned char:3’ may change value
+    m_collation_type= ((uint8) coll.type()) & LEX_CHARSET_COLLATION_TYPE_MASK;
   }
   void set(const Type_handler *handler,
            const Lex_column_charset_collation_attrs_st &coll)
@@ -712,7 +714,9 @@ struct Lex_field_type_st: public Lex_length_and_dec_st
     m_handler= handler;
     m_ci= coll.charset_info();
     Lex_length_and_dec_st::reset();
-    m_collation_type= ((uint8) coll.type()) & 0x3;
+    // Using bit-and to avoid the warning:
+    // conversion from ‘uint8’ to ‘unsigned char:3’ may change value
+    m_collation_type= ((uint8) coll.type()) & LEX_CHARSET_COLLATION_TYPE_MASK;
   }
   void set(const Type_handler *handler, CHARSET_INFO *cs= NULL)
   {