1 files changed, 286 insertions, 35 deletions

diff --git a/sql/item_sum.cc b/sql/item_sum.cc
index 5b24a1eda90..431d8b56e6a 100644
--- a/sql/item_sum.cc
+++ b/sql/item_sum.cc
@@ -30,7 +30,7 @@ Item_sum::Item_sum(List<Item> &list)
   if ((args=(Item**) sql_alloc(sizeof(Item*)*arg_count)))
   {
     uint i=0;
-    List_iterator<Item> li(list);
+    List_iterator_fast<Item> li(list);
     Item *item;
 
     while ((item=li++))
@@ -52,6 +52,8 @@ void Item_sum::make_field(Send_field *tmp_field)
     tmp_field->flags=0;
     if (!maybe_null)
       tmp_field->flags|= NOT_NULL_FLAG;
+    if (unsigned_flag)
+      tmp_field->flags |= UNSIGNED_FLAG;
     tmp_field->length=max_length;
     tmp_field->decimals=decimals;
     tmp_field->type=(result_type() == INT_RESULT ? FIELD_TYPE_LONG :
@@ -150,7 +152,7 @@ Item_sum_hybrid::fix_fields(THD *thd,TABLE_LIST *tables)
     return 1;
   hybrid_type=item->result_type();
   if (hybrid_type == INT_RESULT)
-    max_length=21;
+    max_length=20;
   else if (hybrid_type == REAL_RESULT)
     max_length=float_length(decimals);
   else
@@ -158,6 +160,7 @@ Item_sum_hybrid::fix_fields(THD *thd,TABLE_LIST *tables)
   decimals=item->decimals;
   maybe_null=item->maybe_null;
   binary=item->binary;
+  unsigned_flag=item->unsigned_flag;
   result_field=0;
   null_value=1;
   fix_length_and_dec();
@@ -323,12 +326,27 @@ double Item_sum_hybrid::val()
 {
   if (null_value)
     return 0.0;
-  if (hybrid_type == STRING_RESULT)
-  {
+  switch (hybrid_type) {
+  case STRING_RESULT:
     String *res;  res=val_str(&str_value);
     return res ? atof(res->c_ptr()) : 0.0;
+  case INT_RESULT:
+    if (unsigned_flag)
+      return ulonglong2double(sum_int);
+    return (double) sum_int;
+  case REAL_RESULT:
+    return sum;
   }
-  return sum;
+  return 0;					// Keep compiler happy
+}
+
+longlong Item_sum_hybrid::val_int()
+{
+  if (null_value)
+    return 0;
+  if (hybrid_type == INT_RESULT)
+    return sum_int;
+  return (longlong) Item_sum_hybrid::val();
 }
 
 
@@ -337,25 +355,26 @@ Item_sum_hybrid::val_str(String *str)
 {
   if (null_value)
     return 0;
-  if (hybrid_type == STRING_RESULT)
+  switch (hybrid_type) {
+  case STRING_RESULT:
     return &value;
-  str->set(sum,decimals);
-  return str;
+  case REAL_RESULT:
+    str->set(sum,decimals);
+    break;
+  case INT_RESULT:
+    if (unsigned_flag)
+      str->set((ulonglong) sum_int);
+    else
+      str->set((longlong) sum_int);
+    break;
+  }
+  return str;					// Keep compiler happy
 }
 
-
 bool Item_sum_min::add()
 {
-  if (hybrid_type != STRING_RESULT)
-  {
-    double nr=args[0]->val();
-    if (!args[0]->null_value && (null_value || nr < sum))
-    {
-      sum=nr;
-      null_value=0;
-    }
-  }
-  else
+  switch (hybrid_type) {
+  case STRING_RESULT:
   {
     String *result=args[0]->val_str(&tmp_value);
     if (!args[0]->null_value &&
@@ -366,22 +385,39 @@ bool Item_sum_min::add()
       null_value=0;
     }
   }
-  return 0;
-}
-
-
-bool Item_sum_max::add()
-{
-  if (hybrid_type != STRING_RESULT)
+  break;
+  case INT_RESULT:
+  {
+    longlong nr=args[0]->val_int();
+    if (!args[0]->null_value && (null_value ||
+				 (unsigned_flag && 
+				  (ulonglong) nr < (ulonglong) sum_int) ||
+				 (!unsigned_flag && nr < sum_int)))
+    {
+      sum_int=nr;
+      null_value=0;
+    }
+  }
+  break;
+  case REAL_RESULT:
   {
     double nr=args[0]->val();
-    if (!args[0]->null_value && (null_value || nr > sum))
+    if (!args[0]->null_value && (null_value || nr < sum))
     {
       sum=nr;
       null_value=0;
     }
   }
-  else
+  break;
+  }
+  return 0;
+}
+
+
+bool Item_sum_max::add()
+{
+  switch (hybrid_type) {
+  case STRING_RESULT:
   {
     String *result=args[0]->val_str(&tmp_value);
     if (!args[0]->null_value &&
@@ -392,6 +428,31 @@ bool Item_sum_max::add()
       null_value=0;
     }
   }
+  break;
+  case INT_RESULT:
+  {
+    longlong nr=args[0]->val_int();
+    if (!args[0]->null_value && (null_value ||
+				 (unsigned_flag && 
+				  (ulonglong) nr > (ulonglong) sum_int) ||
+				 (!unsigned_flag && nr > sum_int)))
+    {
+      sum_int=nr;
+      null_value=0;
+    }
+  }
+  break;
+  case REAL_RESULT:
+  {
+    double nr=args[0]->val();
+    if (!args[0]->null_value && (null_value || nr > sum))
+    {
+      sum=nr;
+      null_value=0;
+    }
+  }
+  break;
+  }
   return 0;
 }
 
@@ -676,9 +737,17 @@ Item_sum_hybrid::min_max_update_int_field(int offset)
   nr=args[0]->val_int();
   if (!args[0]->null_value)
   {
-    if (result_field->is_null(offset) ||
-	(cmp_sign > 0 ? old_nr > nr : old_nr < nr))
+    if (result_field->is_null(offset))
       old_nr=nr;
+    else
+    {
+      bool res=(unsigned_flag ?
+		(ulonglong) old_nr > (ulonglong) nr :
+		old_nr > nr);
+      /* (cmp_sign > 0 && res) || (!(cmp_sign > 0) && !res) */
+      if ((cmp_sign > 0) ^ (!res))
+	old_nr=nr;
+    }
     result_field->set_notnull();
   }
   else if (result_field->is_null(offset))
@@ -788,11 +857,74 @@ String *Item_std_field::val_str(String *str)
 
 #include "sql_select.h"
 
+static int simple_raw_key_cmp(void* arg, byte* key1, byte* key2)
+{
+  return memcmp(key1, key2, *(uint*) arg);
+}
+
+static int simple_str_key_cmp(void* arg, byte* key1, byte* key2)
+{
+  return my_sortcmp((char*) key1, (char*) key2, *(uint*) arg);
+}
+
+/*
+  Did not make this one static - at least gcc gets confused when
+  I try to declare a static function as a friend. If you can figure
+  out the syntax to make a static function a friend, make this one
+  static
+*/
+
+int composite_key_cmp(void* arg, byte* key1, byte* key2)
+{
+  Item_sum_count_distinct* item = (Item_sum_count_distinct*)arg;
+  Field **field    = item->table->field;
+  Field **field_end= field + item->table->fields;
+  uint32 *lengths=item->field_lengths;
+  for (; field < field_end; ++field)
+  {
+    Field* f = *field;
+    int len = *lengths++;
+    int res = f->key_cmp(key1, key2);
+    if (res)
+      return res;
+    key1 += len;
+    key2 += len;
+  }
+  return 0;
+}
+
+/*
+  helper function for walking the tree when we dump it to MyISAM -
+  tree_walk will call it for each leaf
+*/
+
+int dump_leaf(byte* key, uint32 count __attribute__((unused)),
+		     Item_sum_count_distinct* item)
+{
+  byte* buf = item->table->record[0];
+  int error;
+  /*
+    The first item->rec_offset bytes are taken care of with
+    restore_record(table,2) in setup()
+  */
+  memcpy(buf + item->rec_offset, key, item->tree.size_of_element);
+  if ((error = item->table->file->write_row(buf)))
+  {
+    if (error != HA_ERR_FOUND_DUPP_KEY &&
+	error != HA_ERR_FOUND_DUPP_UNIQUE)
+      return 1;
+  }
+  return 0;
+}
+
+
 Item_sum_count_distinct::~Item_sum_count_distinct()
 {
   if (table)
     free_tmp_table(current_thd, table);
   delete tmp_table_param;
+  if (use_tree)
+    delete_tree(&tree);
 }
 
 
@@ -829,22 +961,125 @@ bool Item_sum_count_distinct::setup(THD *thd)
     tmp_table_param->cleanup();
   }
   if (!(table=create_tmp_table(thd, tmp_table_param, list, (ORDER*) 0, 1,
-			       0, 0, current_lex->options | thd->options)))
+			       0, 0,
+			       current_lex->select->options | thd->options)))
     return 1;
   table->file->extra(HA_EXTRA_NO_ROWS);		// Don't update rows
+  table->no_rows=1;
+
+  
+  // no blobs, otherwise it would be MyISAM
+  if (table->db_type == DB_TYPE_HEAP)
+  {
+    qsort_cmp2 compare_key;
+    void* cmp_arg;
+
+    // to make things easier for dump_leaf if we ever have to dump to MyISAM
+    restore_record(table,2);
+      
+    if (table->fields == 1)
+    {
+      /*
+	If we have only one field, which is the most common use of
+	count(distinct), it is much faster to use a simpler key
+	compare method that can take advantage of not having to worry
+	about other fields
+      */
+      Field* field = table->field[0];
+      switch(field->type())
+      {
+	/*
+	  If we have a string, we must take care of charsets and case
+	  sensitivity
+	*/
+      case FIELD_TYPE_STRING:
+      case FIELD_TYPE_VAR_STRING:
+	compare_key = (qsort_cmp2)(field->binary() ? simple_raw_key_cmp:
+				   simple_str_key_cmp);
+	break;
+      default:
+	/*
+	  Since at this point we cannot have blobs anything else can
+	  be compared with memcmp
+	*/
+	compare_key = (qsort_cmp2)simple_raw_key_cmp;
+	break;
+      }
+      key_length = field->pack_length();
+      cmp_arg = (void*) &key_length;
+      rec_offset = 1;
+    }
+    else // too bad, cannot cheat - there is more than one field
+    {
+      bool all_binary = 1;
+      Field** field, **field_end;
+      field_end = (field = table->field) + table->fields;
+      uint32 *lengths;
+      if (!(field_lengths= 
+	    (uint32*) thd->alloc(sizeof(uint32) * table->fields)))
+	return 1;
+
+      for (key_length = 0, lengths=field_lengths; field < field_end; ++field)
+      {
+	uint32 length= (*field)->pack_length();
+	key_length += length;
+	*lengths++ = length;
+	if (!(*field)->binary())
+	  all_binary = 0;			// Can't break loop here
+      }
+      rec_offset = table->reclength - key_length;
+      if (all_binary)
+      {
+	compare_key = (qsort_cmp2)simple_raw_key_cmp;
+	cmp_arg = (void*) &key_length;
+      }
+      else
+      {
+	compare_key = (qsort_cmp2) composite_key_cmp ;
+	cmp_arg = (void*) this;
+      }
+    }
+
+    init_tree(&tree, min(max_heap_table_size, sortbuff_size/16), 0,
+	      key_length, compare_key, 0, NULL, cmp_arg);
+    use_tree = 1;
+
+    /*
+      The only time key_length could be 0 is if someone does
+      count(distinct) on a char(0) field - stupid thing to do,
+      but this has to be handled - otherwise someone can crash
+      the server with a DoS attack
+    */
+    max_elements_in_tree = ((key_length) ? max_heap_table_size/key_length :
+			    1);
+  }
   return 0;
 }
 
 
+int Item_sum_count_distinct::tree_to_myisam()
+{
+  if (create_myisam_from_heap(table, tmp_table_param,
+			      HA_ERR_RECORD_FILE_FULL, 1) ||
+      tree_walk(&tree, (tree_walk_action)&dump_leaf, (void*)this,
+		left_root_right))
+    return 1;
+  delete_tree(&tree);
+  use_tree = 0;
+  return 0;
+}
+
 void Item_sum_count_distinct::reset()
 {
-  if (table)
+  if (use_tree)
+    reset_tree(&tree);
+  else if (table)
   {
     table->file->extra(HA_EXTRA_NO_CACHE);
     table->file->delete_all_rows();
     table->file->extra(HA_EXTRA_WRITE_CACHE);
-    (void) add();
   }
+  (void) add();
 }
 
 bool Item_sum_count_distinct::add()
@@ -859,7 +1094,21 @@ bool Item_sum_count_distinct::add()
     if ((*field)->is_real_null(0))
       return 0;					// Don't count NULL
 
-  if ((error=table->file->write_row(table->record[0])))
+  if (use_tree)
+  {
+    /*
+      If the tree got too big, convert to MyISAM, otherwise insert into the
+      tree.
+    */
+    if (tree.elements_in_tree > max_elements_in_tree)
+    {
+      if(tree_to_myisam())
+	return 1;
+    }
+    else if (!tree_insert(&tree, table->record[0] + rec_offset, 0))
+      return 1;
+  }
+  else if ((error=table->file->write_row(table->record[0])))
   {
     if (error != HA_ERR_FOUND_DUPP_KEY &&
 	error != HA_ERR_FOUND_DUPP_UNIQUE)
@@ -875,6 +1124,8 @@ longlong Item_sum_count_distinct::val_int()
 {
   if (!table)					// Empty query
     return LL(0);
+  if (use_tree)
+    return tree.elements_in_tree;
   table->file->info(HA_STATUS_VARIABLE | HA_STATUS_NO_LOCK);
   return table->file->records;
 }
@@ -897,7 +1148,7 @@ void Item_udf_sum::reset()
 
 bool Item_udf_sum::add()
 {
-  DBUG_ENTER("Item_udf_sum::reset");
+  DBUG_ENTER("Item_udf_sum::add");
   udf.add(&null_value);
   DBUG_RETURN(0);
 }