InnoDB: Fixed bugs in the padding and trimming of trailing spaces

that affected the UCS2 character set. (Bug #7350)


innobase/data/data0type.c:
  Added dtype_get_charset_coll_noninline()
innobase/include/data0type.h:
  Added dtype_get_charset_coll_noninline()
innobase/include/row0mysql.h:
  Added charset field to mysql_row_templ_struct.
innobase/include/row0mysql.ic:
  row_mysql_store_col_in_innobase_format():
  When removing trailing spaces, treat the UCS2 character set properly.
innobase/rem/rem0cmp.c:
  cmp_whole_field(): Do not remove trailing 0x20 bytes, as
  innobase_mysql_cmp() implicitly pads the strings with trailing
  spaces as necessary.
innobase/row/row0sel.c:
  row_sel_field_store_in_mysql_format(): Do not pad with 0x20 bytes.
  row_sel_store_mysql_rec(): Pad VARCHARs with trailing spaces
  (0x20, or 0x0020 in UCS2).
sql/ha_innodb.cc:
  build_template(): Initialize templ->charset

1 files changed, 26 insertions, 5 deletions

diff --git a/innobase/include/row0mysql.ic b/innobase/include/row0mysql.ic
index 4ecd66e06ec..fc922b52d0a 100644
--- a/innobase/include/row0mysql.ic
+++ b/innobase/include/row0mysql.ic
@@ -91,12 +91,33 @@ row_mysql_store_col_in_innobase_format(
 		}
 	} else if (type == DATA_VARCHAR || type == DATA_VARMYSQL
 						|| type == DATA_BINARY) {
+		/* Remove trailing spaces. */
+
+		/* Handle UCS2 strings differently.  As no new
+		collations will be introduced in 4.1, we hardcode the
+		charset-collation codes here.  In 5.0, the logic will
+		be based on mbminlen. */
+		ulint	cset	= dtype_get_charset_coll(
+				dtype_get_prtype(dfield_get_type(dfield)));
 		ptr = row_mysql_read_var_ref(&col_len, mysql_data);
-		
-		/* Remove trailing spaces */
-		while (col_len > 0 && ptr[col_len - 1] == ' ') {
-			col_len--;
-		} 
+		if (cset == 35/*ucs2_general_ci*/
+				|| cset == 90/*ucs2_bin*/
+				|| (cset >= 128/*ucs2_unicode_ci*/
+				&& cset <= 144/*ucs2_persian_ci*/)) {
+			/* space=0x0020 */
+			/* Trim "half-chars", just in case. */
+			col_len &= ~1;
+
+			while (col_len >= 2 && ptr[col_len - 2] == 0x00
+					&& ptr[col_len - 1] == 0x20) {
+				col_len -= 2;
+			}
+		} else {
+			/* space=0x20 */
+			while (col_len > 0 && ptr[col_len - 1] == 0x20) {
+				col_len--;
+			}
+		}
 	} else if (type == DATA_BLOB) {
 		ptr = row_mysql_read_blob_ref(&col_len, mysql_data, col_len);
 	}