InnoDB: Make CREATE TABLE return error when the minimum row length

exceeds the maximum record size.  (Bug #5682)


innobase/data/data0type.c:
  Remove function dtype_str_needs_mysql_cmp().
  Document dtype_get_at_most_n_mbchars().
  dtype_get_at_most_n_mbchars(): Use mbminlen and mbmaxlen.
innobase/dict/dict0crea.c:
  dict_build_table_def_step(): Reject if minimum row size is too big.
innobase/include/data0type.h:
  Remove dtype_str_needs_mysql_cmp().
  Document dtype_get_at_most_n_mbchars().
  Add dtype_get_min_size().
innobase/include/data0type.ic:
  Add dtype_get_min_size().
innobase/include/row0mysql.h:
  row_mysql_store_col_in_innobase_format(): Add parameter comp,
  as we will only truncate UTF-8 CHAR(n) columns in row_format=compact.
innobase/include/row0mysql.ic:
  row_mysql_store_col_in_innobase_format(): Add parameter comp,
  as we will only truncate UTF-8 CHAR(n) columns in row_format=compact.
innobase/row/row0mysql.c:
  Pass parameter comp to row_mysql_store_col_in_innobase_format().
innobase/row/row0sel.c:
  Pass parameter comp to row_mysql_store_col_in_innobase_format().
  row_sel_field_store_in_mysql_format(): Undo the stripping of
  UTF-8 CHAR(n) columns by padding with spaces.

1 files changed, 32 insertions, 0 deletions

diff --git a/innobase/include/row0mysql.ic b/innobase/include/row0mysql.ic
index 1f5d0b0c990..910546e298c 100644
--- a/innobase/include/row0mysql.ic
+++ b/innobase/include/row0mysql.ic
@@ -67,6 +67,7 @@ row_mysql_store_col_in_innobase_format(
 					as dfield is used! */
 	ulint		col_len,	/* in: MySQL column length */
 	ulint		type,		/* in: data type */
+	bool		comp,		/* in: TRUE=compact format */
 	ulint		is_unsigned)	/* in: != 0 if unsigned integer type */
 {
 	byte*	ptr 	= mysql_data;
@@ -113,6 +114,37 @@ row_mysql_store_col_in_innobase_format(
 				col_len--;
 			}
 		}
+	} else if (comp && type == DATA_MYSQL
+			&& dtype_get_mbminlen(dfield_get_type(dfield)) == 1
+			&& dtype_get_mbmaxlen(dfield_get_type(dfield)) > 1) {
+		/* We assume that this CHAR field is encoded in a
+		variable-length character set where spaces have
+		1:1 correspondence to 0x20 bytes, such as UTF-8.
+
+		Consider a CHAR(n) field, a field of n characters.
+		It will contain between n*mbminlen and n*mbmaxlen bytes.
+		We will try to truncate it to n bytes by stripping
+		space padding.  If the field contains single-byte
+		characters only, it will be truncated to n characters.
+		Consider a CHAR(5) field containing the string ".a   "
+		where "." denotes a 3-byte character represented by
+		the bytes "$%&".  After our stripping, the string will
+		be stored as "$%&a " (5 bytes).  The string ".abc "
+		will be stored as "$%&abc" (6 bytes).
+
+		The space padding will be restored in row0sel.c, function
+		row_sel_field_store_in_mysql_format(). */
+
+		ulint		n_chars;
+		dtype_t*	dtype = dfield_get_type(dfield);
+
+		ut_a(!(dtype_get_len(dtype) % dtype_get_mbmaxlen(dtype)));
+		n_chars = dtype_get_len(dtype) / dtype_get_mbmaxlen(dtype);
+
+		/* Strip space padding. */
+		while (col_len > n_chars && ptr[col_len - 1] == 0x20) {
+			col_len--;
+		}
 	} else if (type == DATA_BLOB) {
 		ptr = row_mysql_read_blob_ref(&col_len, mysql_data, col_len);
 	}