On Sun, Jun 8, 2014 at 6:00 PM, Marc Delisle <marc@infomarc.info> wrote:

Le 2014-06-05 07:34, Chirayu Chiripal a écrit :

> Hi,
> Please see the line 28 of [0] which is the var_dump of
> PMA_SQP_analyze(PMA_SQP_parse($create_query)). The 'table_ref' index is
> an empty array.
>
> I think it should have name of the table to be created or current one is
> correct?

Hi Chirayu,
the parser sets 'table_ref' only for SELECT statements.

>
> Because in the result there is no index which mention table name separately.
>
> Also, at [0]#L34: 'create_table_fields' index should contain only column
> details if I am not wrong. But at [0]#L44, 'latin_general_ci' which is
> actually the collation of the table, is appearing in the middle. Is this
> a desired behaviour or it is a bug?

A bug.

Hi, I found the reason for the collation coming back as the column name, I did the var_dump() with ‘TESTSUITE’ defined as 1 and I noticed that when TESTSUITE is defined then mysql_charsets are not fetched and hence it is not able to detect that it is charset not a normal identifier (See https://github.com/phpmyadmin/phpmyadmin/blob/master/libraries/sqlparser.lib.php#L46) otherwise when TESTSUITE is not defined then column names are detected properly.

>
> Note: The query passed is present in 'unsorted_query' index at [0]#L16.
>
> [0]: http://pastebin.com/YcbuivPB

--

Marc Delisle | phpMyAdmin

Regards,
Chirayu Chiripal
phpMyAdmin Intern - Google Summer of Code 2014
https://chirayuchiripal.wordpress.com/