PHP 4.0 Bug #8630 Updated: impossible to write and read thai language in an oracle database

ID: 8630 Updated by: sniper Reported By: stephane@asiatdc.com Old-Status: Open Status: Feedback Bug Type: OCI8 related Assigned To: Comments: Did you setup all the relevant environment variables before starting Apache? Check the manual: http://www.php.net/manual/ref.oci8.php --Jani Previous Comments: --------------------------------------------------------------------------- [2001-01-10 07:05:56] stephane@asiatdc.com I actually front to a major problem using PHP (4.0.2) to insert Thai characters into an Oracle database. The mail is a little bit long but try to explain many things. I am using Oracle under Linux RH 6.2 Kernel 2.2.17 recompiled for oracle and Apache 1.3.12. A lot of people seems to have the same problem (when they want to insert arabic or japanese characters) using the same configuration (cf. PHP sites mailing lists / forum). The problem doesn't seems to be a configuration problem. Oracle is correctly configured (I mean Thai character set -as oracle told me- and env. variables...) and insertion rules are conform to Oracle doc. However, all tests have been also made with the UTF-8 character set. Regarding the tests I have done and information I have gotten from various programmers, Oracle can definitely store Thai. Php can display web pages with Thai language (the basic test is to input Thai from a html form and generate a new web page with what has been input before). BUT, there is one thing that I don't completly understand: *) When I sniff the incoming and outgoing information going through the linux oracle+web server, I can see that the incoming information (posted from a browser) is: 'S Q L = & T H A I = % E 6 % E 4 % D 3 % B E % D 0 % D 1 % D 5 & S u b = S u b m i t' So this is Hexa. representing ascii code greather than 128 (equiv to 230,228,211... in decimal) which is Thai characters. *)If I ask Php to generate a new page with the Thai sentence I have input before, php send back the ASCII DECIMAL code: ' b r > . s q l : I N S E R T I N T O S C O T T . T H A I 2 ( C O L ) V A L U E S ( N ' . . . . . . . ' ) < b r > . s t a t e m e n t e x e c u t e d < b r >' NOTE: The dot '.' represent ascii decimal code greather than 127. So a conversion has been done within php but it works. *) What is stored in the database is the received ASCII code (230,228,211...) LESS 128 (102,100,83..). So in oracle, I have a non thai sentence. The main question is: Which entity modify the ascii code? Php? Oracle? Apache? all of them will receive this mail. Of course, a solution is to add 128 when you retrieve characters from the database but first, it's dirty and second, thai sentence can contain numbers ;-)... I really hope someone can help. I thank you all in advance, --------------------------------------------------------------------------- Full Bug description available at: http://bugs.php.net/?id=8630