无法使用 Boost Spirit X3 解析空的 C++ 结构

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

本文关键字:C++ 结构 X3 Boost Spirit      更新时间:2023-10-16

>我正在尝试解析头文件中定义的C++结构。我开始定义语法,但我有一个问题。

这是我的代码:

#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;
}

在代码中有三个字符串要解析:一个只有标头保护,第二个与第一个类似,但有一些C++注释,第三个字符串具有空结构。

这是最后一个无法解析的,我不明白为什么。在我用于结构StructType的语法中,我首先检查可选的typedef,然后关键字struct可以附加或不附加的{字符,然后我搜索}字符,后跟变量名称,后跟;

我不明白错误在哪里。我在解析空结构时做错了什么?

几件事:

  • 船长由周围环境继承
  • 您在标头保护中的"令牌"周围没有lexeme[],因此它将匹配包括typedefstruct space因为它还包括行端。

您可以简化以下操作:

住在科里鲁

#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 << "Parsedn";
        } else {
            std::cout << "Failed to parsen";
        }
        if (f!=l) {
            std::cout << "Remaining unparsed: " << std::quoted(std::string(f,l)) << "n";
        }
    }
}

指纹

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

sehe 在保留代码方面做得非常出色。但是由于我在我想我会发布之后对你是什么有自己的想法。我确实从中了解到船长可能非常复杂。另一方面,我通常跳过omit.

由于您似乎希望扔掉所有起跑线和评论#因此我将其归结为。在某些时候,我确定您将解析为一个属性。所以我从结构名称开始,因此dest字符串。

只是另一种看待它的方式。

#include <iostream>
#define BOOST_SPIRIT_X3_DEBUG
#include <boost/spirit/home/x3.hpp>
using namespace boost::spirit::x3;
// Parse all/skip"#..."
auto const def = lit("#") >> omit[lexeme[*char_("a-zA-Z_! ")]];
// Skipper for C++ comments (nested /* */ are not handled for now)
auto const comment = ("//" >> omit[*(char_ - eol)]) | ("/*" >> omit[*(char_ - "*/")] >> "*/");
auto const skipper = *(def | comment | space);
// Parse variable name like "xxx" or "my_var"
auto const name = rule<struct name, std::string>("name") = *char_("a-zA-Z_");
auto const struct_rule = lit("typedef") >> "struct" >> '{' >> omit[*(char_-'}')] >> '}' >> name >> ';';
auto const final = skipper >> -struct_rule >> skipper;
void parse(char* in)
{
    std::string str(in);
    auto it = str.begin();
    std::string dest;
    bool r = phrase_parse(it, str.end(), final, space, dest);// , dest);
    std::cout << std::boolalpha << "r: " << r << std::endl
        << std::string(it, str.end()) << std::endl
        << "DEST: " << dest << std::endl;
}
int main()
{
    parse(R"xx(
        #if !defined XXX_H
        #define XXX_H
        #endif
      )xx");
    parse(R"xx(
        #if !defined XXX_H
        // Single line comment
        #define XXX_H
        #endif // !XXX_H
      )xx");
    parse(R"xx(
        #if !defined XXX_H
        #define XXX_H
        typedef struct {
        } MyStruct;
        #endif
      )xx");
    return 0;
}

指纹:

r: true
DEST:
r: true
DEST:
<name>
  <try> MyStruct;ntt#endifn</try>
  <success>;ntt#endifnt  </success>
  <attributes>[M, y, S, t, r, u, c, t]</attributes>
</name>
r: true
DEST:   MyStruct