Cannot parse an empty C++ struct with Boost Spirit X3

Question

I'm trying to parse a C++ struct defined in a header file. I'm starting to define the grammar but I've a problem.

This is my code:

#include <boost/spirit/home/x3.hpp>

int main() {
  namespace x3 = boost::spirit::x3;

  // Parse "#if !defined XXX_X_" or "'#ifndef X_X"
  auto Ifndef = x3::skip(x3::space)[(x3::lit('#') >> (x3::lit("ifndef") | (x3::lit("if") >> x3::lit("!defined"))))];
  auto HeaderGuardFirstRow = Ifndef >> +(x3::alnum | '_');

  // Parse "#define XXX_X" or "#  define XXX_X"
  auto Define = x3::skip(x3::space)[(x3::lit('#') >> x3::lit("define"))];
  auto HeaderGuardSecondRow = Define >> +(x3::alnum | '_');

  // Parse
  // "
  //  #if !defined XXX_X_
  //  #define XXX_X
  // "
  auto HeaderGuardBegin = HeaderGuardFirstRow >> HeaderGuardSecondRow;

  // Parse "#endif" or "#  endif"
  auto HeaderGuardEnd = x3::skip(x3::space)[x3::lit('#') >> (x3::lit("endif"))];

  // Parse variable name like "xxx" or "my_var"
  auto VariableName = x3::lexeme[x3::char_("a-zA-Z_") >> *(x3::alnum | x3::lit("_"))];

  // Skipper for C++ comments (nested /* */ are not handled for now)
  auto SingleLineComment = "//" >> *(x3::char_ - x3::eol) >> (x3::eol | x3::eoi);
  auto BlockComment = "/*" >> *(x3::char_ - "*/") >> "*/";
  auto Skipper = SingleLineComment | BlockComment | x3::ascii::space;

  // Parse
  // "
  // typedef struct {
  // } MyStruct;
  // "
  // ERROR: This parse does not work
  auto StructType = -x3::lit("typedef") >> x3::skip(Skipper)[x3::lit("struct") >> x3::lit('{')] >>
  x3::skip(Skipper)[x3::lit('}') >> VariableName >> x3::lit(";")];

  // Header grammar. Should parse
  // "
  // #if !defined XXX_H
  //  #define XXX_H
  //  typedef struct {
  //  } MyStruct;
  //  #endif
  // "
  auto grammar = HeaderGuardBegin >> *(StructType) >> HeaderGuardEnd;

  std::string data01(R"xx(
    #if !defined XXX_H
    #define XXX_H
    #endif
  )xx");

  bool r = phrase_parse(
    data01.begin(),
    data01.end(),
    grammar,
    Skipper
    );

    std::string data02(R"xx(
    #if !defined XXX_H
    // Single line comment
    #define XXX_H
    #endif // !XXX_H
  )xx");

  r = phrase_parse(
    data02.begin(),
    data02.end(),
    grammar,
    Skipper
    );

  std::string data03(R"xx(
    #if !defined XXX_H
    #define XXX_H
    typedef struct {
    } MyStruct;
    #endif
  )xx");

  // r = false: This parsing does not work.
  r = phrase_parse(
    data03.begin(),
    data03.end(),
    grammar,
    Skipper
    );
  return 0;
}

In the code there are three strings to parse: one with only header guards, the second like the first but with some C++ comments, and the third with an empty struct.

It's the last one that's failing to parse, and I don't understand why. In the grammar that I use for the struct StructType I first check for an optional typedef, then the keyword struct with the { character that can be attached or not, then I search for the } character followed by a variable name followed by ;.

I don't understand where the error is. What I'm doing wrong in parsing the empty structure?

Answer 1

Few things:

• skippers are inherited by the surrounding context
• you didn't have lexeme[] around the "tokens" in the header guards, so it would match including the typedefstruct because space also includes line-ends.

You can simplify things:

Live On Coliru

#include <iostream>
#include <boost/spirit/home/x3.hpp>
#include <iomanip>

int main() {
    namespace x3 = boost::spirit::x3;

    // Parse "#if !defined XXX_X_" or "'#ifndef X_X"
    auto const Ifndef               = (x3::lit('#') >> (x3::lit("ifndef") | (x3::lit("if") >> x3::lit("!defined"))));
    auto const HeaderGuardFirstRow  = Ifndef >> x3::lexeme[+(x3::alnum | '_')];

    // Parse "#define XXX_X" or "#  define XXX_X"
    auto const Define               = (x3::lit('#') >> x3::lit("define"));
    auto const HeaderGuardSecondRow = Define >> x3::lexeme[+(x3::alnum | '_')];

    // Parse
    // "
    //  #if !defined XXX_X_
    //  #define XXX_X
    // "
    auto const HeaderGuardBegin     = HeaderGuardFirstRow >> HeaderGuardSecondRow;

    // Parse "#endif" or "#  endif"
    auto const HeaderGuardEnd       = x3::lit('#') >> (x3::lit("endif"));

    // Parse variable name like "xxx" or "my_var"
    auto const VariableName         = x3::lexeme[x3::char_("a-zA-Z_") >> *(x3::alnum | x3::lit("_"))];

    // Skipper for C++ comments (nested /* */ are not handled for now)
    auto const SingleLineComment    = "//" >> *(x3::char_ - x3::eol) >> (x3::eol | x3::eoi);
    auto const BlockComment         = "/*" >> *(x3::char_ - "*/") >> "*/";
    auto const Skipper              = SingleLineComment | BlockComment | x3::ascii::space;

    auto const StructType = 
        -x3::lit("typedef") 
        >> "struct" >> '{' >> '}' >> VariableName
        >> ";"
        ;

    // Header grammar.
    auto grammar = HeaderGuardBegin >> *StructType >> HeaderGuardEnd;

    for (std::string const data : {
            R"xx(
    #if !defined XXX_H
    #define XXX_H
    #endif
  )xx",
            R"xx(
    #if !defined XXX_H
    // Single line comment
    #define XXX_H
    #endif // !XXX_H
  )xx",
            R"xx(
    #if !defined XXX_H
    #define XXX_H
    typedef struct {
        // aloha
    } MyStruct;

    typedef struct { /* caramba */ } MyOtherStruct
;
    #endif
  )xx" }) {
        auto f = data.begin(), l = data.end();
        std::cout << "Parsing " << std::quoted(data) << "\n";
        if (phrase_parse(f, l, grammar, Skipper)) {
            std::cout << "Parsed\n";
        } else {
            std::cout << "Failed to parse\n";
        }

        if (f!=l) {
            std::cout << "Remaining unparsed: " << std::quoted(std::string(f,l)) << "\n";
        }
    }
}

Prints

Parsing "
    #if !defined XXX_H
    #define XXX_H
    #endif
  "
Parsed
Parsing "
    #if !defined XXX_H
    // Single line comment
    #define XXX_H
    #endif // !XXX_H
  "
Parsed
Parsing "
    #if !defined XXX_H
    #define XXX_H
    typedef struct {
        // aloha
    } MyStruct;

    typedef struct { /* caramba */ } MyOtherStruct
;
    #endif
  "
Parsed