sip,uri,test: Escape SIP URIs

[maximilianfridrich]

According to RFC 3261, SIP URIs need to escape UTF-8 characters.

[maximilianfridrich]

Unfortunately, the MSVC compiled retest seems to hang when it runs the UTF-8 related test test_uri_user on Windows. I've been trying to figure out what's going on, but I can't even get BareSIP to compile on my Windows VM. Any help is greatly appreciated.

[alfredh]

try to avoid Unicode characters in the source code.

you can use a binary string instead, i.e. "\x01\x02\x03\x04" etc.

[maximilianfridrich]

Thank you! So simple, and I was convinced it would solve the Windows problem. I am not sure what the MSVC compiler is doing here...

Do you have any more insight into what could be going wrong on the Windows machines? Or could you provide the script to setup Windows the way it is setup in the GitHub Actions?

[alfredh]

perhaps some C-string is not null terminated ?

add some debugging, see if the program is stuck in a while loop

[sreimers]

Yes looks like a out of bounds, this is the error window message, why windows test hangs:

Debug Assertion Failed!

Program: retest.exe File: minkernel\crts\ucrt\src\appcrt\convert\isctype.cpp Line: 36

Expression: c >= -1 && c <= 255

https://stackoverflow.com/questions/39538846/program-crashes

This is the Call stack:

retest.exe!is_unreserved(char c) Line 43	C
 	retest.exe!is_user(char c) Line 85	C
 	retest.exe!comp_escape(re_printf * pf, const pl * pl, bool(*)(char) eh) Line 173	C
 	retest.exe!uri_user_escape(re_printf * pf, const pl * pl) Line 235	C
 	retest.exe!re_vhprintf(const char * fmt, char * ap, int(*)(const char *, unsigned __int64, void *) vph, void * arg) Line 311	
 	retest.exe!mbuf_printf(mbuf * mb, const char * fmt, ...) Line 540	C
 	retest.exe!test_uri_user() Line 213	C
 	retest.exe!test_unit(const char * name, bool verbose) Line 506	C
 	retest.exe!test_reg(const char * name, bool verbose) Line 740	C
 	retest.exe!main(int argc, char * * argv) Line 235	C
``

[sreimers]

char c is -61

[sreimers]

Oh I think isalnum() is the problem. It behaves differently on windows.

Maybe you can add a range check c >= -1 && c <= 255 before calling isalnum().

[sreimers]

The behavior is undefined if the value of ch is not representable as unsigned char and is not equal to [EOF]

https://en.cppreference.com/w/c/string/byte/isalnum

[maximilianfridrich]

I was about to comment the same thing, yes, isalnum seems to be the problem. Apparently, Unix-like systems accept negative values without complaining, while Windows doesn't. The man page of isalnum states:

These functions check whether c, which must have the value of an unsigned char or EOF, falls into a certain character class according to the specified locale.

So this should certainly be an unsigned char.

[maximilianfridrich]

I casted it because "ä" is represented by 0xC3 0xA4 in UTF-8 which as signed char gives us -61 and -92. So they are valid values, but need to be casted to be correctly interpreted by isalnum.

Edit: Actually, the Notes section of the man page states that precisely this should be done.

[maximilianfridrich]

Thank you so much for the help, the Windows tests are finally passing!