Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[visualage]

The amuleweb in aMuleCVS-20071226 does not support unicode characters in search/download properly. There are actually two problems:

1. URL decoding does not work properly under certain local charset (e.g. CP936). This is caused by the line

Code: [Select]

               char toReplace[2] = {(char)i, 0};       // decode URL
               m_dec_str[i] = char2unicode(toReplace);
For some charsets (e.g. CP936), if i is greater than 127, it will be considered as a two-byte character. However, {i, 0} is not a valid two-byte character. Therefore, char2unicode() will cause problem.

2. The key/value submitted to amule webserver are parsed as 8-bit charset, not a unicode charset (e.g. UTF-8). This is caused by CUrlDecodeTable::DecodeString(), where is only considers every escaped/unescaped character in the value as an 8-bit character, without further decoding.

If we can assume that all amuleweb skins use UTF-8 as encoding, the following patch fixes the above two problems.

Code: [Select]

--- amule-cvs.org/src/webserver/src/WebServer.cpp       2007-12-08 02:04:25.000000000 +0800
+++ amule-cvs/src/webserver/src/WebServer.cpp   2007-12-27 11:05:46.669544744 +0800
@@ -145,8 +145,9 @@
                snprintf(fromReplace, sizeof(fromReplace), "%%%02X", i);
                m_enc_u_str[i] = char2unicode(fromReplace);

-               char toReplace[2] = {(char)i, 0};       // decode URL
-               m_dec_str[i] = char2unicode(toReplace);
+               //char toReplace[2] = {(char)i, 0};     // decode URL
+               //m_dec_str[i] = char2unicode(toReplace);
+               m_dec_str[i] = wxString::Format(wxT("%c"), i);
        }
 }

@@ -157,6 +158,12 @@
                str.Replace(m_enc_l_str[i], m_dec_str[i]);
                str.Replace(m_enc_u_str[i], m_dec_str[i]);
        }
+       char *buffer = new char[str.Len() + 1];
+       for (size_t i=0;i<str.Len();++i)
+         buffer[i] = str[i];
+       buffer[str.Len()] = 0; // Mark the end of the string
+       str = UTF82unicode(buffer);
+       delete[] buffer;
 }

 CUrlDecodeTable*       CUrlDecodeTable::ms_instance;

It has been tested on my linux box (Fedora Core 3, arch x86_64, CP936 local charset).

[phoenix]

visualage,

Your patch seems fine, I will committ it with some minor modifications and it will be on tomorrow svn tarball.

Next time you submit a patch, please attach it to the message, do not copy/paste it because of the loss of whitespace characters.

Thank you for your support!