Bug #66496 [Fbk->Dup]: conversion of UTF-8 strings containing language specific chars is wrong
Edit report at https://bugs.php.net/bug.php?id=66496&edit=1
ID: 66496
Updated by: ab@php.net
Reported by: care at novadys dot de
Summary: conversion of UTF-8 strings containing language
specific chars is wrong
-Status: Feedback
+Status: Duplicate
Type: Bug
Package: COM related
Operating System: Windows
PHP Version: 5.5.8
Block user comment: N
Private report: N
New Comment:
This is fixed now as it's the same as bug #66431, please check.
Thanks.
Previous Comments:
------------------------------------------------------------------------
[2014-01-17 19:52:14] ab@php.net
We need a code snippet in PHP to fix and write a test.
Thanks.
------------------------------------------------------------------------
[2014-01-17 14:56:54] care at novadys dot de
It works correctly when replacing the line
V_BSTR(v) = SysAllocStringByteLen((char*)olestring, Z_STRLEN_P(z) * sizeof(OLECHAR));
by
V_BSTR(v) = SysAllocStringByteLen((char*)olestring, wcslen(olestring) * sizeof(OLECHAR));
------------------------------------------------------------------------
[2014-01-16 20:28:26] ab@php.net
Were it possible you to post a repro snippet? Thanks.
------------------------------------------------------------------------
[2014-01-16 14:43:19] care at novadys dot de
Description:
------------
Using a UTF-8 String as input for a com-function will generate a wrong string in COM interface. If
the input string is containing n-chars which are encoded with 2 bytes, the length of resulting
string is n byte too long.
example: "I want to Dusseldorf and Koln" is correctly handled
"I want to Düsseldorf and Köln" will call COM function with a string:
"I want to Düsseldorf and Köln\0\4"
reason:
ext/com_dotnet/com_variant.c
...
PHP_COM_DOTNET_API void php_com_variant_from_zval(VARIANT *v, zval *z, int codepage TSRMLS_DC)
...
case IS_STRING:
V_VT(v) = VT_BSTR;
olestring = php_com_string_to_olestring(Z_STRVAL_P(z), Z_STRLEN_P(z),
codepage TSRMLS_CC);
here is the problem:
V_BSTR(v) = SysAllocStringByteLen((char*)olestring, Z_STRLEN_P(z) * sizeof(OLECHAR));
When input string is UTF-8 encoded Z_STRLEN_P(z) has a count of 2 byte for each "special"
char. So length of input string is count of all chars + count of "special" chars. After
conversion to olestring Z_STRLEN_P(z) is the wrong length the olestring is shorter. So
SysAllocStringByteLen is allocating too much memory and while the for the string length of
olestring. In result those strings are always showing a '\0' for the first special char +
n-1 random chars for the remaing special chars.
------------------------------------------------------------------------
--
Edit this bug report at https://bugs.php.net/bug.php?id=66496&edit=1
Thread (5 messages)