Fix for bug#20648 We introduce a new field method for knowing "real size", and...

Fix for bug#20648 We introduce a new field method for knowing "real size", and we now in archive null unused bits of a row to null before writing. 


sql/field.cc:
  data_length for field needs to be calculated directly.
sql/field.h:
  The new method data_length() returns the "real" length of the field.
sql/ha_archive.cc:
  Before a write_row() archive nulls space beyond the size of the row in varchars to make sure that compression only sees NULL.

sql/field.cc

@@ -6412,6 +6412,11 @@ void Field_varstring::sql_type(String &res) const
 }
 
 
+uint Field_varstring::data_length(const char *from)
+{
+  return length_bytes == 1 ? (uint) (uchar) *ptr : uint2korr(ptr);
+}
+
 /*
   Functions to create a packed row.
   Here the number of length bytes are depending on the given max_length

sql/field.h

@@ -144,6 +144,11 @@ public:
     table, which is located on disk).
   */
   virtual uint32 pack_length_in_rec() const { return pack_length(); }
+
+  /*
+    data_length() return the "real size" of the data in memory.
+  */
+  virtual uint32 data_length(const char *from) { return pack_length(); }
   virtual uint32 sort_length() const { return pack_length(); }
   virtual void reset(void) { bzero(ptr,pack_length()); }
   virtual void reset_fields() {}
@@ -1102,6 +1107,7 @@ public:
   int key_cmp(const byte *str, uint length);
   uint packed_col_length(const char *to, uint length);
   uint max_packed_col_length(uint max_length);
+  uint data_length(const char *from);
   uint size_of() const { return sizeof(*this); }
   enum_field_types real_type() const { return MYSQL_TYPE_VARCHAR; }
   bool has_charset(void) const

sql/ha_archive.cc

@@ -710,6 +710,28 @@ int ha_archive::write_row(byte *buf)
     if (init_archive_writer())
       DBUG_RETURN(HA_ERR_CRASHED_ON_USAGE);
 
+  /*
+    Varchar structures are constant in size but are not cleaned up request
+    to request. The following sets all unused space to null to improve
+    compression.
+  */
+  for (Field **field=table->field ; *field ; field++)
+  {
+    DBUG_PRINT("archive",("Pack is %d\n", (*field)->pack_length()));
+    DBUG_PRINT("archive",("MyPack is %d\n", (*field)->data_length((char*) buf + (*field)->offset())));
+    if ((*field)->real_type() == MYSQL_TYPE_VARCHAR) 
+    {
+      uint actual_length= (*field)->data_length((char*) buf + (*field)->offset());
+      uint offset= (*field)->offset() + actual_length + 
+        (actual_length > 255 ? 2 : 1);
+      DBUG_PRINT("archive",("Offset is %d -> %d\n", actual_length, offset));
+      /*
+      if ((*field)->pack_length() + (*field)->offset() != offset)
+        bzero(buf + offset, (size_t)((*field)->pack_length() + (actual_length > 255 ? 2 : 1) - (*field)->data_length));
+    */
+    }
+  }
+
   share->rows_recorded++;
   rc= real_write_row(buf, share->archive_write);
   pthread_mutex_unlock(&share->mutex);