Chris@16: /*=============================================================================
Chris@16:     Boost.Wave: A Standard compliant C++ preprocessor library
Chris@16: 
Chris@16:     Detect the need to insert a whitespace token into the output stream
Chris@16:     
Chris@16:     http://www.boost.org/
Chris@16: 
Chris@16:     Copyright (c) 2001-2012 Hartmut Kaiser. Distributed under the Boost
Chris@16:     Software License, Version 1.0. (See accompanying file
Chris@16:     LICENSE_1_0.txt or copy at http://www.boost.org/LICENSE_1_0.txt)
Chris@16: =============================================================================*/
Chris@16: #if !defined(INSERT_WHITESPACE_DETECTION_HPP_765EF77B_0513_4967_BDD6_6A38148C4C96_INCLUDED)
Chris@16: #define INSERT_WHITESPACE_DETECTION_HPP_765EF77B_0513_4967_BDD6_6A38148C4C96_INCLUDED
Chris@16: 
Chris@16: #include <boost/wave/wave_config.hpp>   
Chris@16: #include <boost/wave/token_ids.hpp>   
Chris@16: 
Chris@16: // this must occur after all of the includes and before any code appears
Chris@16: #ifdef BOOST_HAS_ABI_HEADERS
Chris@16: #include BOOST_ABI_PREFIX
Chris@16: #endif
Chris@16: 
Chris@16: ///////////////////////////////////////////////////////////////////////////////
Chris@16: namespace boost {
Chris@16: namespace wave {
Chris@16: namespace util {
Chris@16: 
Chris@16: namespace impl {
Chris@16: 
Chris@16: // T_IDENTIFIER
Chris@16:     template <typename StringT>
Chris@16:     inline bool
Chris@16:     would_form_universal_char (StringT const &value)
Chris@16:     {
Chris@16:         if ('u' != value[0] && 'U' != value[0])
Chris@16:             return false;
Chris@16:         if ('u' == value[0] && value.size() < 5)
Chris@16:             return false;
Chris@16:         if ('U' == value[0] && value.size() < 9)
Chris@16:             return false;
Chris@16:     
Chris@16:     typename StringT::size_type pos = 
Chris@16:         value.find_first_not_of("0123456789abcdefABCDEF", 1);
Chris@16:         
Chris@16:         if (StringT::npos == pos || 
Chris@16:             ('u' == value[0] && pos > 5) ||
Chris@16:             ('U' == value[0] && pos > 9))
Chris@16:         {
Chris@16:             return true;        // would form an universal char
Chris@16:         }
Chris@16:         return false;
Chris@16:     }
Chris@16:     template <typename StringT>
Chris@16:     inline bool 
Chris@16:     handle_identifier(boost::wave::token_id prev, 
Chris@16:         boost::wave::token_id before, StringT const &value)
Chris@16:     {
Chris@16:         using namespace boost::wave;
Chris@16:         switch (static_cast<unsigned int>(prev)) {
Chris@16:         case T_IDENTIFIER:
Chris@16:         case T_NONREPLACABLE_IDENTIFIER:
Chris@16:         case T_COMPL_ALT:
Chris@16:         case T_OR_ALT:
Chris@16:         case T_AND_ALT:
Chris@16:         case T_NOT_ALT:
Chris@16:         case T_XOR_ALT:
Chris@16:         case T_ANDASSIGN_ALT:
Chris@16:         case T_ORASSIGN_ALT:
Chris@16:         case T_XORASSIGN_ALT:
Chris@16:         case T_NOTEQUAL_ALT:
Chris@16:         case T_FIXEDPOINTLIT:
Chris@16:             return true;
Chris@16: 
Chris@16:         case T_FLOATLIT:
Chris@16:         case T_INTLIT:
Chris@16:         case T_PP_NUMBER:
Chris@16:             return (value.size() > 1 || (value[0] != 'e' && value[0] != 'E'));
Chris@16:             
Chris@16:          // avoid constructing universal characters (\u1234)
Chris@16:         case TOKEN_FROM_ID('\\', UnknownTokenType):
Chris@16:             return would_form_universal_char(value);
Chris@16:         }
Chris@16:         return false;
Chris@16:     }
Chris@16: // T_INTLIT
Chris@16:     inline bool 
Chris@16:     handle_intlit(boost::wave::token_id prev, boost::wave::token_id /*before*/)
Chris@16:     {
Chris@16:         using namespace boost::wave;
Chris@16:         switch (static_cast<unsigned int>(prev)) {
Chris@16:         case T_IDENTIFIER:
Chris@16:         case T_NONREPLACABLE_IDENTIFIER:
Chris@16:         case T_INTLIT:
Chris@16:         case T_FLOATLIT:
Chris@16:         case T_FIXEDPOINTLIT:
Chris@16:         case T_PP_NUMBER:
Chris@16:             return true;
Chris@16:         }
Chris@16:         return false;
Chris@16:     }
Chris@16: // T_FLOATLIT
Chris@16:     inline bool 
Chris@16:     handle_floatlit(boost::wave::token_id prev, 
Chris@16:         boost::wave::token_id /*before*/)
Chris@16:     {
Chris@16:         using namespace boost::wave;
Chris@16:         switch (static_cast<unsigned int>(prev)) {
Chris@16:         case T_IDENTIFIER:
Chris@16:         case T_NONREPLACABLE_IDENTIFIER:
Chris@16:         case T_INTLIT:
Chris@16:         case T_FLOATLIT:
Chris@16:         case T_FIXEDPOINTLIT:
Chris@16:         case T_PP_NUMBER:
Chris@16:             return true;
Chris@16:         }
Chris@16:         return false;
Chris@16:     }
Chris@16: // <% T_LEFTBRACE
Chris@16:     inline bool 
Chris@16:     handle_alt_leftbrace(boost::wave::token_id prev, 
Chris@16:         boost::wave::token_id /*before*/)
Chris@16:     {
Chris@16:         using namespace boost::wave;
Chris@16:         switch (static_cast<unsigned int>(prev)) {
Chris@16:         case T_LESS:        // <<%
Chris@16:         case T_SHIFTLEFT:   // <<<%
Chris@16:             return true;
Chris@16:         }
Chris@16:         return false;
Chris@16:     }
Chris@16: // <: T_LEFTBRACKET
Chris@16:     inline bool 
Chris@16:     handle_alt_leftbracket(boost::wave::token_id prev, 
Chris@16:         boost::wave::token_id /*before*/)
Chris@16:     {
Chris@16:         using namespace boost::wave;
Chris@16:         switch (static_cast<unsigned int>(prev)) {
Chris@16:         case T_LESS:        // <<:
Chris@16:         case T_SHIFTLEFT:   // <<<:
Chris@16:             return true;
Chris@16:         }
Chris@16:         return false;
Chris@16:     }
Chris@16: // T_FIXEDPOINTLIT
Chris@16:     inline bool 
Chris@16:     handle_fixedpointlit(boost::wave::token_id prev, 
Chris@16:         boost::wave::token_id /*before*/)
Chris@16:     {
Chris@16:         using namespace boost::wave;
Chris@16:         switch (static_cast<unsigned int>(prev)) {
Chris@16:         case T_IDENTIFIER:
Chris@16:         case T_NONREPLACABLE_IDENTIFIER:
Chris@16:         case T_INTLIT:
Chris@16:         case T_FLOATLIT:
Chris@16:         case T_FIXEDPOINTLIT:
Chris@16:         case T_PP_NUMBER:
Chris@16:             return true;
Chris@16:         }
Chris@16:         return false;
Chris@16:     }
Chris@16: // T_DOT
Chris@16:     inline bool 
Chris@16:     handle_dot(boost::wave::token_id prev, boost::wave::token_id before)
Chris@16:     {
Chris@16:         using namespace boost::wave;
Chris@16:         switch (static_cast<unsigned int>(prev)) {
Chris@16:         case T_DOT:
Chris@16:             if (T_DOT == before)
Chris@16:                 return true;    // ...
Chris@16:             break;
Chris@16:         }
Chris@16:         return false;
Chris@16:     }
Chris@16: // T_QUESTION_MARK
Chris@16:     inline bool 
Chris@16:     handle_questionmark(boost::wave::token_id prev, 
Chris@16:         boost::wave::token_id /*before*/)
Chris@16:     {
Chris@16:         using namespace boost::wave;
Chris@16:         switch(static_cast<unsigned int>(prev)) {
Chris@16:         case TOKEN_FROM_ID('\\', UnknownTokenType):     // \?
Chris@16:         case T_QUESTION_MARK:   // ??
Chris@16:             return true;
Chris@16:         }
Chris@16:         return false;
Chris@16:     }
Chris@16: // T_NEWLINE
Chris@16:     inline bool
Chris@16:     handle_newline(boost::wave::token_id prev, 
Chris@16:         boost::wave::token_id before)
Chris@16:     {
Chris@16:         using namespace boost::wave;
Chris@16:         switch(static_cast<unsigned int>(prev)) {
Chris@16:         case TOKEN_FROM_ID('\\', UnknownTokenType): // \ \n
Chris@16:         case T_DIVIDE:
Chris@16:             if (T_QUESTION_MARK == before)
Chris@16:                 return true;    // ?/\n     // may be \\n
Chris@16:             break;
Chris@16:         }
Chris@16:         return false;
Chris@16:     }
Chris@16: 
Chris@16:     inline bool 
Chris@16:     handle_parens(boost::wave::token_id prev)
Chris@16:     {
Chris@16:         switch (static_cast<unsigned int>(prev)) {
Chris@16:         case T_LEFTPAREN:
Chris@16:         case T_RIGHTPAREN:
Chris@16:         case T_LEFTBRACKET:
Chris@16:         case T_RIGHTBRACKET:
Chris@16:         case T_LEFTBRACE:
Chris@16:         case T_RIGHTBRACE:
Chris@16:         case T_SEMICOLON:
Chris@16:         case T_COMMA:
Chris@16:         case T_COLON:
Chris@16:             // no insertion between parens/brackets/braces and operators
Chris@16:             return false;   
Chris@16: 
Chris@16:         default:
Chris@16:             break;
Chris@16:         }
Chris@16:         return true;
Chris@16:     }
Chris@16:         
Chris@16: }   // namespace impl
Chris@16: 
Chris@16: class insert_whitespace_detection 
Chris@16: {
Chris@16: public:
Chris@16:     insert_whitespace_detection(bool insert_whitespace_ = true) 
Chris@16:     :   insert_whitespace(insert_whitespace_),
Chris@16:         prev(boost::wave::T_EOF), beforeprev(boost::wave::T_EOF) 
Chris@16:     {}
Chris@16:     
Chris@16:     template <typename StringT>
Chris@16:     bool must_insert(boost::wave::token_id current, StringT const &value)
Chris@16:     {
Chris@16:         if (!insert_whitespace)
Chris@16:             return false;       // skip whitespace insertion alltogether
Chris@16:             
Chris@16:         using namespace boost::wave;
Chris@16:         switch (static_cast<unsigned int>(current)) {
Chris@16:         case T_NONREPLACABLE_IDENTIFIER:
Chris@16:         case T_IDENTIFIER: 
Chris@16:             return impl::handle_identifier(prev, beforeprev, value); 
Chris@16:         case T_PP_NUMBER:
Chris@16:         case T_INTLIT:
Chris@16:             return impl::handle_intlit(prev, beforeprev); 
Chris@16:         case T_FLOATLIT:
Chris@16:             return impl::handle_floatlit(prev, beforeprev); 
Chris@16:         case T_STRINGLIT:
Chris@16:             if (TOKEN_FROM_ID('L', IdentifierTokenType) == prev)       // 'L'
Chris@16:                 return true;
Chris@16:             break;
Chris@16:         case T_LEFTBRACE_ALT:
Chris@16:             return impl::handle_alt_leftbrace(prev, beforeprev); 
Chris@16:         case T_LEFTBRACKET_ALT:
Chris@16:             return impl::handle_alt_leftbracket(prev, beforeprev); 
Chris@16:         case T_FIXEDPOINTLIT:
Chris@16:             return impl::handle_fixedpointlit(prev, beforeprev); 
Chris@16:         case T_DOT:
Chris@16:             return impl::handle_dot(prev, beforeprev); 
Chris@16:         case T_QUESTION_MARK:
Chris@16:             return impl::handle_questionmark(prev, beforeprev); 
Chris@16:         case T_NEWLINE:
Chris@16:             return impl::handle_newline(prev, beforeprev); 
Chris@16: 
Chris@16:         case T_LEFTPAREN:
Chris@16:         case T_RIGHTPAREN:
Chris@16:         case T_LEFTBRACKET:
Chris@16:         case T_RIGHTBRACKET:
Chris@16:         case T_SEMICOLON:
Chris@16:         case T_COMMA:
Chris@16:         case T_COLON:
Chris@16:             switch (static_cast<unsigned int>(prev)) {
Chris@16:             case T_LEFTPAREN:
Chris@16:             case T_RIGHTPAREN:
Chris@16:             case T_LEFTBRACKET:
Chris@16:             case T_RIGHTBRACKET:
Chris@16:             case T_LEFTBRACE:
Chris@16:             case T_RIGHTBRACE:
Chris@16:                 return false;   // no insertion between parens/brackets/braces
Chris@16: 
Chris@16:             default:
Chris@16:                 if (IS_CATEGORY(prev, OperatorTokenType))
Chris@16:                     return false;
Chris@16:                 break;
Chris@16:             }        
Chris@16:             break;
Chris@16:             
Chris@16:         case T_LEFTBRACE:
Chris@16:         case T_RIGHTBRACE:
Chris@16:             switch (static_cast<unsigned int>(prev)) {
Chris@16:             case T_LEFTPAREN:
Chris@16:             case T_RIGHTPAREN:
Chris@16:             case T_LEFTBRACKET:
Chris@16:             case T_RIGHTBRACKET:
Chris@16:             case T_LEFTBRACE:
Chris@16:             case T_RIGHTBRACE:
Chris@16:             case T_SEMICOLON:
Chris@16:             case T_COMMA:
Chris@16:             case T_COLON:
Chris@16:                 return false;   // no insertion between parens/brackets/braces
Chris@16: 
Chris@16:             case T_QUESTION_MARK:
Chris@16:                 if (T_QUESTION_MARK == beforeprev)
Chris@16:                     return true;
Chris@16:                 if (IS_CATEGORY(prev, OperatorTokenType))
Chris@16:                     return false;
Chris@16:                 break;
Chris@16:                 
Chris@16:             default:
Chris@16:                 break;
Chris@16:             }
Chris@16:             break;
Chris@16:                             
Chris@16:         case T_MINUS:
Chris@16:         case T_MINUSMINUS:
Chris@16:         case T_MINUSASSIGN:
Chris@16:             if (T_MINUS == prev || T_MINUSMINUS == prev)
Chris@16:                 return true;
Chris@16:             if (!impl::handle_parens(prev))
Chris@16:                 return false;
Chris@16:             if (T_QUESTION_MARK == prev && T_QUESTION_MARK == beforeprev)
Chris@16:                 return true;
Chris@16:             break;
Chris@16:             
Chris@16:         case T_PLUS:
Chris@16:         case T_PLUSPLUS:
Chris@16:         case T_PLUSASSIGN:
Chris@16:             if (T_PLUS == prev || T_PLUSPLUS == prev)
Chris@16:                 return true;
Chris@16:             if (!impl::handle_parens(prev))
Chris@16:                 return false;
Chris@16:             if (T_QUESTION_MARK == prev && T_QUESTION_MARK == beforeprev)
Chris@16:                 return true;
Chris@16:             break;
Chris@16:             
Chris@16:         case T_DIVIDE:
Chris@16:         case T_DIVIDEASSIGN:
Chris@16:             if (T_DIVIDE == prev)
Chris@16:                 return true;
Chris@16:             if (!impl::handle_parens(prev))
Chris@16:                 return false;
Chris@16:             if (T_QUESTION_MARK == prev && T_QUESTION_MARK == beforeprev)
Chris@16:                 return true;
Chris@16:             break;
Chris@16:         
Chris@16:         case T_EQUAL:
Chris@16:         case T_ASSIGN:
Chris@16:             switch (static_cast<unsigned int>(prev)) {
Chris@16:             case T_PLUSASSIGN:
Chris@16:             case T_MINUSASSIGN:
Chris@16:             case T_DIVIDEASSIGN:
Chris@16:             case T_STARASSIGN:
Chris@16:             case T_SHIFTRIGHTASSIGN:
Chris@16:             case T_SHIFTLEFTASSIGN:
Chris@16:             case T_EQUAL:
Chris@16:             case T_NOTEQUAL:
Chris@16:             case T_LESSEQUAL:
Chris@16:             case T_GREATEREQUAL:
Chris@16:             case T_LESS:
Chris@16:             case T_GREATER:
Chris@16:             case T_PLUS:
Chris@16:             case T_MINUS:
Chris@16:             case T_STAR:
Chris@16:             case T_DIVIDE:
Chris@16:             case T_ORASSIGN:
Chris@16:             case T_ANDASSIGN:
Chris@16:             case T_XORASSIGN:
Chris@16:             case T_OR:
Chris@16:             case T_AND:
Chris@16:             case T_XOR:
Chris@16:             case T_OROR:
Chris@16:             case T_ANDAND:
Chris@16:                 return true;
Chris@16: 
Chris@16:             case T_QUESTION_MARK:
Chris@16:                 if (T_QUESTION_MARK == beforeprev)
Chris@16:                     return true;
Chris@16:                 break;
Chris@16:                 
Chris@16:             default:
Chris@16:                 if (!impl::handle_parens(prev))
Chris@16:                     return false;
Chris@16:                 break;
Chris@16:             }
Chris@16:             break;
Chris@16: 
Chris@16:         case T_GREATER:
Chris@16:             if (T_MINUS == prev || T_GREATER == prev)
Chris@16:                 return true;    // prevent -> or >>
Chris@16:             if (!impl::handle_parens(prev))
Chris@16:                 return false;
Chris@16:             if (T_QUESTION_MARK == prev && T_QUESTION_MARK == beforeprev)
Chris@16:                 return true;
Chris@16:             break;
Chris@16: 
Chris@16:         case T_LESS:
Chris@16:             if (T_LESS == prev)
Chris@16:                 return true;    // prevent <<
Chris@16:             // fall through
Chris@16:         case T_CHARLIT:
Chris@16:         case T_NOT:
Chris@16:         case T_NOTEQUAL:
Chris@16:             if (!impl::handle_parens(prev))
Chris@16:                 return false;
Chris@16:             if (T_QUESTION_MARK == prev && T_QUESTION_MARK == beforeprev)
Chris@16:                 return true;
Chris@16:             break;
Chris@16: 
Chris@16:         case T_AND:
Chris@16:         case T_ANDAND:
Chris@16:             if (!impl::handle_parens(prev))
Chris@16:                 return false;
Chris@16:             if (T_AND == prev || T_ANDAND == prev)
Chris@16:                 return true;
Chris@16:             break;
Chris@16:             
Chris@16:         case T_OR:
Chris@16:             if (!impl::handle_parens(prev))
Chris@16:                 return false;
Chris@16:             if (T_OR == prev)
Chris@16:                 return true;
Chris@16:             break;
Chris@16:             
Chris@16:         case T_XOR:
Chris@16:             if (!impl::handle_parens(prev))
Chris@16:                 return false;
Chris@16:             if (T_XOR == prev)
Chris@16:                 return true;
Chris@16:             break;
Chris@16:             
Chris@16:         case T_COMPL_ALT:
Chris@16:         case T_OR_ALT:
Chris@16:         case T_AND_ALT:
Chris@16:         case T_NOT_ALT:
Chris@16:         case T_XOR_ALT:
Chris@16:         case T_ANDASSIGN_ALT:
Chris@16:         case T_ORASSIGN_ALT:
Chris@16:         case T_XORASSIGN_ALT:
Chris@16:         case T_NOTEQUAL_ALT:
Chris@16:             switch (static_cast<unsigned int>(prev)) {
Chris@16:             case T_LEFTPAREN:
Chris@16:             case T_RIGHTPAREN:
Chris@16:             case T_LEFTBRACKET:
Chris@16:             case T_RIGHTBRACKET:
Chris@16:             case T_LEFTBRACE:
Chris@16:             case T_RIGHTBRACE:
Chris@16:             case T_SEMICOLON:
Chris@16:             case T_COMMA:
Chris@16:             case T_COLON:
Chris@16:                 // no insertion between parens/brackets/braces and operators
Chris@16:                 return false;   
Chris@16: 
Chris@16:             case T_IDENTIFIER:
Chris@16:                 if (T_NONREPLACABLE_IDENTIFIER == prev ||
Chris@16:                     IS_CATEGORY(prev, KeywordTokenType))
Chris@16:                 {
Chris@16:                     return true;
Chris@16:                 }
Chris@16:                 break;
Chris@16:                 
Chris@16:             default:
Chris@16:                 break;
Chris@16:             }
Chris@16:             break;
Chris@16:             
Chris@16:         case T_STAR:
Chris@16:             if (T_STAR == prev)
Chris@16:                 return false;     // '*****' do not need to be separated
Chris@16:             if (T_GREATER== prev && 
Chris@16:                 (T_MINUS == beforeprev || T_MINUSMINUS == beforeprev)
Chris@16:                )
Chris@16:             {
Chris@16:                 return true;    // prevent ->*
Chris@16:             }
Chris@16:             break;
Chris@16: 
Chris@16:         case T_POUND:
Chris@16:             if (T_POUND == prev)
Chris@16:                 return true;
Chris@16:             break;
Chris@16:         }
Chris@16: 
Chris@16:     // FIXME: else, handle operators separately (will catch to many cases)
Chris@16: //         if (IS_CATEGORY(current, OperatorTokenType) && 
Chris@16: //             IS_CATEGORY(prev, OperatorTokenType))
Chris@16: //         {
Chris@16: //             return true;    // operators must be delimited always
Chris@16: //         }
Chris@16:         return false;
Chris@16:     }
Chris@16:     void shift_tokens (boost::wave::token_id next_id)
Chris@16:     {
Chris@16:         if (insert_whitespace) {
Chris@16:             beforeprev = prev;
Chris@16:             prev = next_id;
Chris@16:         }
Chris@16:     }
Chris@16:     
Chris@16: private:
Chris@16:     bool insert_whitespace;            // enable this component
Chris@16:     boost::wave::token_id prev;        // the previous analyzed token
Chris@16:     boost::wave::token_id beforeprev;  // the token before the previous
Chris@16: };
Chris@16: 
Chris@16: ///////////////////////////////////////////////////////////////////////////////
Chris@16: }   //  namespace util
Chris@16: }   //  namespace wave 
Chris@16: }   //  namespace boost
Chris@16: 
Chris@16: // the suffix header occurs after all of the code
Chris@16: #ifdef BOOST_HAS_ABI_HEADERS
Chris@16: #include BOOST_ABI_SUFFIX
Chris@16: #endif
Chris@16: 
Chris@16: #endif // !defined(INSERT_WHITESPACE_DETECTION_HPP_765EF77B_0513_4967_BDD6_6A38148C4C96_INCLUDED)