[ psqlodbc-Bugs-1010313 ] ucs2_to_utf8 endianness problem

From: <noreply(at)pgfoundry(dot)org>
To: noreply(at)pgfoundry(dot)org
Subject: [ psqlodbc-Bugs-1010313 ] ucs2_to_utf8 endianness problem
Date: 2008-03-12 16:43:37
Message-ID: 20080312164337.4F11E17AE666@pgfoundry.org
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-odbc

Bugs item #1010313, was opened at 2008-03-12 16:43
You can respond by visiting:
http://pgfoundry.org/tracker/?func=detail&atid=538&aid=1010313&group_id=1000125

Category: None
Group: None
Status: Open
Resolution: None
Priority: 3
Submitted By: Ken Robbins (kpr)
Assigned to: Nobody (None)
Summary: ucs2_to_utf8 endianness problem

Initial Comment:
psqlodbc 08.02.0200
PostgreSQL 8.2.3

I'm using SQLBindParameter() (using SQL_WVARCHAR and SQL_C_WCHAR as the types) along with SQLExecute() to execute a prepared statement.

In the byte buffer, I'm using UCS-2 encoding. E.g., I'm doing something like this:

wchar_t c = 0x03C0; // lower case greek letter pi
char buf[255];
memcpy(buf, (char*) c, sizeof(wchar_t));

This eventually gets passed into ucs2_to_utf8().

char *ucs2_to_utf8(const SQLWCHAR *ucs2str, SQLLEN ilen, SQLLEN *olen, BOOL lower_identifier)

Whenever, a UTF-8 byte sequence is more than one byte, a memcpy is used on either the UInt2 or Int4 types. E.g.,

memcpy(utf8str + len, (char *) &byte2code, sizeof(byte2code));

However, the ordering of the bytes of byte2code is different depending on whether the platform is little endian or big endian. When I run my code on an Intel environment (Linux), the code runs fine. However, when I run my code on a PowerPC environment (also Linux), the UTF-8 byte sequence is wrong.

I added mylog() calls to the ucs2_to_utf8() code to see what bytes were at the memcpy step and also the final byte sequence. The bytes are correct; however, the ordering is flipped (understandably). For the 2 byte sequences, the ordering is just flipped. For the 3 byte sequences, in addition to the ordering being flipped, the wrong 3 bytes are being used.

Whenever I use SQL_VARCHAR and SQL_C_CHAR and put the UTF-8 byte sequence in the byte buffer myself, it works fine on both platforms.

I believe that ucs2_to_utf8() needs to account for the endianness of the platform, so the right bytes are put in the final returned UTF-8 sequence. However, if I am not doing something right, please advise me on that also.

----------------------------------------------------------------------

You can respond by visiting:
http://pgfoundry.org/tracker/?func=detail&atid=538&aid=1010313&group_id=1000125

Browse pgsql-odbc by date

  From Date Subject
Next Message Andreas 2008-03-20 17:20:39 Access complains : Update or CancelUpdate without AddNew or Edit
Previous Message Glyn Astill 2008-03-11 09:40:52 Re: Cannot get psqlODBC to work