flex和bison：字符串文字

Question

I need your advice on how to get string literal. String literals is everything what starts with b" , has zero or more symbols then and ends with " . Token is called BYTES_OBJECT . bytes_type is a name of type which stores such string literals. Here is what I've tried:

{%
#include "parser.hpp"
#include <iostream>
#include <string>
string BUFFER;
%}

%option noyywrap
%x COMMENT
%x BYTESMODE

"b"[\"] {
  BUFFER = string();
  BEGIN BYTESMODE;
} // BYTESMODE

  <BYTESMODE>{BYTES_HEXCODE} {
    // convert to hexcode
  } // BYTES_HEXCODE

  <BYTESMODE>{BYTES_ESCAPE} {
    yytext++; // Move to the next char
    switch (*yytext) {
      case '\'': BUFFER += '\''; break;
      case '\"': BUFFER += '\"'; break;
      case '\\': BUFFER += '\\'; break;
      case '\?': BUFFER += '?';  break;
      case 'a':  BUFFER += '\a'; break;
      case 'b':  BUFFER += '\b'; break;
      case 'f':  BUFFER += '\f'; break;
      case 'n':  BUFFER += '\n'; break;
      case 'r':  BUFFER += '\r'; break;
      case 't':  BUFFER += '\t'; break;
      case 'v':  BUFFER += '\v'; break;
    } // end of switch
  } // BYTES_ESCAPE

  <BYTESMODE>[^\"] {
    BUFFER += yytext;
  } // BYTES_SYMBOL

  <BYTESMODE>\" {
    bytes_type object = BUFFER;
    yylval.bytes_buffer = &object;
    BEGIN INITIAL;
    return BYTES_OBJECT;
  } // closing quote

However, when I try to create string, it cause segment fault. If return BYTES_OBJECT is before BEGIN INITIAL , it processes literals incorrect. How can I solve it? I guess that even concept can be done more easy than I've written.

Thanks in advance!

Answer 1

Your input file has some errors. You should not be doing such detailed handling just to recognize strings. A simpler example:

%option noyywrap
%{
#include <iostream>
#include <string>
%}

%%
b\"(\\.|[^\\"])*\" { 
  std::cout << "string:" << yytext << std::endl;
  return(1); 
}

%%

int main() {
  yylex();
  return 0;
}