Regex to extract function from c++ function calls

Question

I have to extract a function name from various c++ function calls. Following are some of the function calls examples and extracted function names highlighted.

• std::basic_fstream<char,std::char_traits<char> >::~basic_fstream<char,std::char_traits<char> > ~basic_fstream
• CSocket::Send send

• CMap<unsigned int,unsigned int &,tagLAUNCHOBJECT,tagLAUNCHOBJECT &>::RemoveAll Cerner::Foundations::String::Rep::~Rep~Rep

• CCMessage::~CCMessage ~CCMessage

• std::_Tree<std::_Tmap_traits<std::basic_string<char,std::char_traits<char>,std::allocator<char> >,u_Tree
• Lib::DispatcherCache::~DispatcherCache~DispatcherCache

• CPrefDataObjectLoader<CPrefManagerKey,CPrefManagerValue,CGetPrefManager,PrefManagerKeyFunctor>::Get Get

  The following Regex works for most of the functions

• /((?:[^:]*))$';/ This regex get the string from the last :
• /+?(?=<)';/ This one removes string that starts with <

But for std::basic_fstream<char,std::char_traits<char> >::~basic_fstream<char,std::char_traits<char> > the output I get is char_traits because this string is after last ':' but the result should be ~basic_fstream. Is there a way I can combine both regex and ignore everything that is within <>?

Answer 1

The grammar of C++ is not only not regular, it's actually highly context-sensitive (especially near templates). Even a proper CFG parser won't help you, let alone a plain old regex… Rather than trying to approximate the impossible using ugly and fragile hacks, why not use an actual tool for the job? If you want to parse C++, then use a C++ parser, such as libclang.