C++ regex 연습! syntax highlighting 을 구현해보자~

[codesafer]

//-----------------------------------------------------------------------------
#define CSTR

#include <array>
#include <string>
#include <regex>
//-----------------------------------------------------------------------------

enum    class	CPP_SYNTAX : int
{
    etc,
    block_comment,
    line_comment,
    preprocessor,

    string_literal,
    character_literal,
    real_literal,
    integer_literal,
    bool_literal,

    types,
    flow_controls,
    keywords,
    operators,
    end,
};

const   std::array< CPP_SYNTAX, 12 > ordered_syntaxes =
{
    CPP_SYNTAX::string_literal,
    CPP_SYNTAX::block_comment,
    CPP_SYNTAX::line_comment,
    CPP_SYNTAX::preprocessor,

    CPP_SYNTAX::character_literal,
    CPP_SYNTAX::real_literal,
    CPP_SYNTAX::integer_literal,
    CPP_SYNTAX::bool_literal,

    CPP_SYNTAX::types,
    CPP_SYNTAX::flow_controls,
    CPP_SYNTAX::keywords,
    CPP_SYNTAX::operators,
};
//-----------------------------------------------------------------------------

const   std::array< std::regex, 13 > syntax_rules
{
    std::regex( "" ),
    std::regex( "\\/\\*(?:[^\\*][^\\/]|[\\s])*\\*\\/" ),
    std::regex( "\\/\\/.*" ),
    std::regex( "^[\\t ]*#\\s*[a-zA-Z]*" ),

    std::regex( "(?:\\bL)?\"[^\"]*\"" ),
    std::regex( "(?:\\bL)?'[^']*'" ),
    std::regex( "[\\+\\-]?(?:(?:\\d+\\.\\d*|\\.\\d+)"
                "(?:[eE][\\+\\-]?\\d+)?)[fFlL]?(?!\\.)" ),
    std::regex( "[\\+\\-]?\\b(?:0[0-7]*|[1-9]\\d*|"
                "0x[\\dA-F]+)[uUlL]{0,3}\\b(?!\\.)" ),
    std::regex( "\\b(?:true|false)\\b" ),
    
    std::regex( "\\b(?:bool|char|double|float|int|long|short|"
                "signed|unsigned|void|wchar_t)\\b" ),
    std::regex( "\\b(?:break|case|catch|continue|default|do|else|"
                "for|goto|if|return|switch|throw|try|while)\\b" ),
    std::regex( "\\b(?:asm|auto|class|const_cast|const|delete|"
                "dynamic_cast|enum|explicit|export|extern|friend|"
                "inline|main|mutable|namespace|new|operator|private|"
                "protected|public|register|reinterpret_cast|sizeof|"
                "static_cast|static|struct|template|this|typedef|"
                "typeid|typename|union|using|virtual|volatile)\\b" ),
    std::regex( "[\\{\\}\\[\\]\\(\\)\\.\\?\\*\\+\\-\\^\\|\\\\\\/&<>%:;~!=,]+" ),
};
//-----------------------------------------------------------------------------

const   std::string coloring_begins[] =
{
    "<E>", "<M>", "<L>", "<D>",
    "<G>", "<C>", "<R>", "<Z>", "<N>",
    "<T>", "<F>", "<K>", "<O>",
};

const   std::string coloring_ends[] =
{
    "</E>", "</M>", "</L>", "</D>",
    "</G>", "</C>", "</R>", "</Z>", "</N>",
    "</T>", "</F>", "</K>", "</O>",
};

auto    html_heading( std::string& html )
{
    return	html = "<style type=\"text/css\">\n"
    "E { color: #555; font-family: 'Courier New'; font-size: 10pt; }\n"
    "M { color: #AAA; font-family: 'Courier New'; font-size: 10pt; }\n"
    "L { color: #AAA; font-family: 'Courier New'; font-size: 10pt; }\n"
    "D { color: #238; font-family: 'Courier New'; font-size: 10pt; }\n"
    "G { color: #F66; font-family: 'Courier New'; font-size: 10pt; }\n"
    "C { color: #D55; font-family: 'Courier New'; font-size: 10pt; }\n"
    "R { color: #08F; font-family: 'Courier New'; font-size: 10pt; }\n"
    "Z { color: #66F; font-family: 'Courier New'; font-size: 10pt; }\n"
    "N { color: #45F; font-family: 'Courier New'; font-size: 10pt; }\n"
    "T { color: #45F; font-family: 'Courier New'; font-size: 10pt; }\n"
    "F { color: #45F; font-family: 'Courier New'; font-size: 10pt; }\n"
    "K { color: #45F; font-family: 'Courier New'; font-size: 10pt; }\n"
    "O { color: #8AF; font-family: 'Courier New'; font-size: 10pt; }\n"
    "</style>\n<pre>\n";
}

auto    html_footing( std::string& html )
{
    return  html += "\n</pre>\n";
}
//-----------------------------------------------------------------------------

struct          RANGE
{
    size_t      position, length;
    CSTR        RANGE( const size_t position, const size_t length )
    :   position( position )
    ,   length( length )
    {
    }
};

struct          COLORING
{
    RANGE       range;
    CPP_SYNTAX  syntax;
    CSTR        COLORING( const RANGE& range, const CPP_SYNTAX syntax )
    :   range( range )
    ,   syntax( syntax )
    {
    }
    bool        operator<( const COLORING& rhs ) const
    {
        return  range.position < rhs.range.position;
    }
};
///----------------------------------------------------------------------------

auto    regex_find_ranges( const std::string& str, const std::regex& rule )
{
    std::vector< RANGE > ranges;
    using   sregex_it = std::sregex_iterator;
    const   sregex_it end;
    for( sregex_it next( str.begin(), str.end(), rule ); next != end; ++next )
        ranges.emplace_back( next->position(), next->length() );
    return  ranges;
}
///----------------------------------------------------------------------------

void    translate_code( std::string& dst, const std::string& src,
	size_t& it, const size_t it_end )
{
    for( ; it < it_end; ++it )
        if( src[ it ] == '&' ) dst += "&amp;";
        else if( src[ it ] == '<' ) dst += "&lt;";
        else dst += src[ it ];
}
///----------------------------------------------------------------------------

#include <algorithm>

auto    build_html( const std::string& code )
{
    std::string result;
    html_heading( result );

    std::vector< COLORING > color_marks;
    std::string temp_code = 
        std::regex_replace( code, std::regex( "\\\\[\\n\\s\\S]" ), "__" );

    for( const auto& syntax : ordered_syntaxes )
    {
        const int index = (int)syntax;
        const auto ranges = regex_find_ranges( temp_code, syntax_rules[ index ] );
        for( const auto& range : ranges )
        {
            color_marks.emplace_back( range, syntax );
            const auto range_end = range.position + range.length;
            for( auto i = range.position; i < range_end; ++i )
                temp_code[ i ] = '_';
        }
    }
    sort( color_marks.begin(), color_marks.end() );

    size_t  position = 0;
    for( const auto& mark : color_marks )
    {
        if( position < mark.range.position )
        {
            result += coloring_begins[ (int)CPP_SYNTAX::etc ];
            translate_code( result, code, position,
                mark.range.position );
            result += coloring_ends[ (int)CPP_SYNTAX::etc ];
        }
        if( position == mark.range.position )
        {
            result += coloring_begins[ (int)mark.syntax ];
            translate_code( result, code, position,
                mark.range.position + mark.range.length );
            result += coloring_ends[ (int)mark.syntax ];
        }
    }
    return  html_footing( result );
}
//-----------------------------------------------------------------------------

#include <iostream>
using   namespace               std;

const   string  source_code =
"#include <stdio.h>\n"
"\n"
"int main()\n"
"{\n"
"    printf( \"Hello world\\n\" );\n"
"    return 0;\n"
"}\n";

int main()
{
    cout << build_html( source_code );
    return 0;
}
//-----------------------------------------------------------------------------

[codesafer]

짧은 깨달음. stl 의 regex 는 장애자들이 만든것임에 틀림없다.

[codesafer]

땅곰 "그냥 <M>foo</M> 처럼 적어놓고 M { color: red; } 하는 식도 있죠 " <-- 요걸 써봐야할듯. 일단 구현해놓고.
땅곰이 고마워~

[codesafer]

Naming rules
You need to have at least one '-' inside the name of a custom element. Any tag names without '-' will result in an error.

[codesafer]

Defining a custom element is simple. Just call document.registerElement() with its tag name as the first argument.var XComponent = document.registerElement('x-component');Now you can use <x-component> wherever you want in the document.<x-component></x-component>

[codesafer]

http://w3c.github.io/webcomponents/spec/custom/

[codesafer]

옹 그냥도 되나 보네 http://www.w3schools.com/html/html_css.asp

[codesafer]

테스트 해보니 잘 되어서 코드 수정~

[codesafer]

HTML 16 색 component 표현 가르쳐준 루나님 감사~

[codesafer]

line comment 와 preprocessor directives 부분의 regex 결과가 vc++ 과 gcc 가 다르군유~ vc++ 은 perl 과 같은 결과가 나오는데 gcc 는 뭐지 =_= 기본 옵션이 다른가...

[codesafer]

일단 완료~ 귀찮음.

[11copy]

[codesafer]

주석보다 따옴표 처리를 먼저오게 했습니다. 몇 가지 정규식을 수정했습니다. 정규식끼리 호환성이;;;

[codesafer]

푸하하 댓글 글씨체도 다 바뀐다. * 은 못쓰겠넹.