[LOGCXX-369] Load Properties File Fails When There Are multibyte Characters in the Path

• Type: Bug
• Status: Resolved
• Priority: Major
• Resolution: Fixed
• Affects Version/s: 0.10.0
• Fix Version/s: 0.11.0
• Component/s: Configurator
• Environment: Windows XP SP2; VC8.0SP1
• Assignee: Thorsten Schöning
• Reporter: Starwalker
• Created: 17/Sep/10 11:14
• Updated: 21/Mar/18 18:39
• Resolved: 21/Mar/18 18:39

Description

Load properties file fails when calling PropertyConfigurator::configure using a MBCS string which type is char * or std::string and contains multibyte characters, such as Chinese characters.

Analyse
When calling PropertyConfigurator::configure using a MBCS string which type is char * or std::string, the constructor of File will call Transcoder::decode to convert MBCS string to DBCS string. When meets the byte which is greater than or equal to 0x80, the function will call MbstowcsCharsetDecoder::decode to process the following characters. But there is an error in MbstowcsCharsetDecoder::decode when calculating the offset after calling mbsrtowcs.

size_t converted = mbsrtowcs(buf, // This function returns the number of wide characters, maybe not equal to the number of the bytes of multibyte characters
&src,
requested,
&mbstate);
if (converted == (size_t) -1) {
stat = APR_BADARG;
in.position(src - in.data());
break;
} else {
stat = append(out, buf);
in.position(in.position() + converted); // Here, the number of the bytes of multibyte characters should be added, not the number of wide characters
}

The path "D:\Temp\中文文件夹\STC\Config\Log.properties" becomes "D:\Temp\中文文件夹\STC\Config\Log.propertiesrties" after convertion.
There are 10 bytes for the 5 Chinese characters, but only 5 wide characters. Because the error above, last 5 characters repeated.

Issue Links

• is duplicated by: LOGCXX-399 Non-ascii character output wrong.
• links to: GitHub Pull Request #4