1   
//
1   
//
2   
// Copyright (c) 2021 Vinnie Falco (vinnie dot falco at gmail dot com)
2   
// Copyright (c) 2021 Vinnie Falco (vinnie dot falco at gmail dot com)
3   
//
3   
//
4   
// Distributed under the Boost Software License, Version 1.0. (See accompanying
4   
// Distributed under the Boost Software License, Version 1.0. (See accompanying
5   
// file LICENSE_1_0.txt or copy at http://www.boost.org/LICENSE_1_0.txt)
5   
// file LICENSE_1_0.txt or copy at http://www.boost.org/LICENSE_1_0.txt)
6   
//
6   
//
7   
// Official repository: https://github.com/boostorg/url
7   
// Official repository: https://github.com/boostorg/url
8   
//
8   
//
9  
9  
10   
#ifndef BOOST_URL_GRAMMAR_HEXDIG_CHARS_HPP
10   
#ifndef BOOST_URL_GRAMMAR_HEXDIG_CHARS_HPP
11   
#define BOOST_URL_GRAMMAR_HEXDIG_CHARS_HPP
11   
#define BOOST_URL_GRAMMAR_HEXDIG_CHARS_HPP
12  
12  
13   
#include <boost/url/detail/config.hpp>
13   
#include <boost/url/detail/config.hpp>
14   
#include <boost/url/grammar/detail/charset.hpp>
14   
#include <boost/url/grammar/detail/charset.hpp>
15  
15  
16   
namespace boost {
16   
namespace boost {
17   
namespace urls {
17   
namespace urls {
18   
namespace grammar {
18   
namespace grammar {
19   
namespace implementation_defined {
19   
namespace implementation_defined {
20   
struct hexdig_chars_t
20   
struct hexdig_chars_t
21   
{
21   
{
22   
    /** Determine if a character is a hexadecimal digit
22   
    /** Determine if a character is a hexadecimal digit
23  
23  
24   
        @param c The character to test
24   
        @param c The character to test
25   
        @return `true` if `c` is a hexadecimal digit.
25   
        @return `true` if `c` is a hexadecimal digit.
26   
    */
26   
    */
27   
    constexpr
27   
    constexpr
28   
    bool
28   
    bool
29   
    operator()(char c) const noexcept
29   
    operator()(char c) const noexcept
30   
    {
30   
    {
31   
        return
31   
        return
32   
            (c >= '0' && c <= '9') ||
32   
            (c >= '0' && c <= '9') ||
33   
            (c >= 'A' && c <= 'F') ||
33   
            (c >= 'A' && c <= 'F') ||
34   
            (c >= 'a' && c <= 'f');
34   
            (c >= 'a' && c <= 'f');
35   
    }
35   
    }
36  
36  
37   
#ifdef BOOST_URL_USE_SSE2
37   
#ifdef BOOST_URL_USE_SSE2
38   
    char const*
38   
    char const*
39   
    find_if(
39   
    find_if(
40   
        char const* first,
40   
        char const* first,
41   
        char const* last) const noexcept
41   
        char const* last) const noexcept
42   
    {
42   
    {
43   
        return detail::find_if_pred(
43   
        return detail::find_if_pred(
44   
            *this, first, last);
44   
            *this, first, last);
45   
    }
45   
    }
46  
46  
47   
    char const*
47   
    char const*
48   
    find_if_not(
48   
    find_if_not(
49   
        char const* first,
49   
        char const* first,
50   
        char const* last) const noexcept
50   
        char const* last) const noexcept
51   
    {
51   
    {
52   
        return detail::find_if_not_pred(
52   
        return detail::find_if_not_pred(
53   
            *this, first, last);
53   
            *this, first, last);
54   
    }
54   
    }
55   
#endif
55   
#endif
56   
};
56   
};
57   
}
57   
}
58  
58  
59   
/** The set of hexadecimal digits
59   
/** The set of hexadecimal digits
60  
60  
61   
    @par Example
61   
    @par Example
62   
    Character sets are used with rules and the
62   
    Character sets are used with rules and the
63   
    functions @ref find_if and @ref find_if_not.
63   
    functions @ref find_if and @ref find_if_not.
64   
    @code
64   
    @code
65   
    system::result< core::string_view > rv = parse( "8086FC19", token_rule( hexdig_chars ) );
65   
    system::result< core::string_view > rv = parse( "8086FC19", token_rule( hexdig_chars ) );
66   
    @endcode
66   
    @endcode
67  
67  
68   
    @par BNF
68   
    @par BNF
69   
    @code
69   
    @code
70   
    HEXDIG      = DIGIT
70   
    HEXDIG      = DIGIT
71   
                / "A" / "B" / "C" / "D" / "E" / "F"
71   
                / "A" / "B" / "C" / "D" / "E" / "F"
72   
                / "a" / "b" / "c" / "d" / "e" / "f"
72   
                / "a" / "b" / "c" / "d" / "e" / "f"
73   
    @endcode
73   
    @endcode
74  
74  
75   
    @note The RFCs are inconsistent on the case
75   
    @note The RFCs are inconsistent on the case
76   
    sensitivity of hexadecimal digits. Existing
76   
    sensitivity of hexadecimal digits. Existing
77   
    uses suggest case-insensitivity is a de-facto
77   
    uses suggest case-insensitivity is a de-facto
78   
    standard.
78   
    standard.
79  
79  
80   
    @par Specification
80   
    @par Specification
81   
    @li <a href="https://datatracker.ietf.org/doc/html/rfc5234#appendix-B.1"
81   
    @li <a href="https://datatracker.ietf.org/doc/html/rfc5234#appendix-B.1"
82   
        >B.1. Core Rules (rfc5234)</a>
82   
        >B.1. Core Rules (rfc5234)</a>
83   
    @li <a href="https://datatracker.ietf.org/doc/html/rfc7230#section-1.2"
83   
    @li <a href="https://datatracker.ietf.org/doc/html/rfc7230#section-1.2"
84   
        >1.2. Syntax Notation (rfc7230)</a>
84   
        >1.2. Syntax Notation (rfc7230)</a>
85   
    @li <a href="https://datatracker.ietf.org/doc/html/rfc5952#section-2.3"
85   
    @li <a href="https://datatracker.ietf.org/doc/html/rfc5952#section-2.3"
86   
        >2.3. Uppercase or Lowercase (rfc5952)</a>
86   
        >2.3. Uppercase or Lowercase (rfc5952)</a>
87   
    @li <a href="https://datatracker.ietf.org/doc/html/rfc5952#section-4.3"
87   
    @li <a href="https://datatracker.ietf.org/doc/html/rfc5952#section-4.3"
88   
        >4.3. Lowercase (rfc5952)</a>
88   
        >4.3. Lowercase (rfc5952)</a>
89  
89  
90   
    @see
90   
    @see
91   
        @ref find_if,
91   
        @ref find_if,
92   
        @ref find_if_not,
92   
        @ref find_if_not,
93   
        @ref hexdig_value,
93   
        @ref hexdig_value,
94   
        @ref parse,
94   
        @ref parse,
95   
        @ref token_rule.
95   
        @ref token_rule.
96   
*/
96   
*/
97   
constexpr implementation_defined::hexdig_chars_t hexdig_chars{};
97   
constexpr implementation_defined::hexdig_chars_t hexdig_chars{};
98  
98  
99   
/** Return the decimal value of a hex character
99   
/** Return the decimal value of a hex character
100  
100  
101   
    This function returns the decimal
101   
    This function returns the decimal
102   
    value of a hexadecimal character,
102   
    value of a hexadecimal character,
103   
    or -1 if the argument is not a
103   
    or -1 if the argument is not a
104   
    valid hexadecimal digit.
104   
    valid hexadecimal digit.
105  
105  
106   
    @par BNF
106   
    @par BNF
107   
    @code
107   
    @code
108   
    HEXDIG      = DIGIT
108   
    HEXDIG      = DIGIT
109   
                / "A" / "B" / "C" / "D" / "E" / "F"
109   
                / "A" / "B" / "C" / "D" / "E" / "F"
110   
                / "a" / "b" / "c" / "d" / "e" / "f"
110   
                / "a" / "b" / "c" / "d" / "e" / "f"
111   
    @endcode
111   
    @endcode
112  
112  
113   
    @param ch The character to check
113   
    @param ch The character to check
114  
114  
115   
    @return The decimal value or -1
115   
    @return The decimal value or -1
116   
*/
116   
*/
117   
inline
117   
inline
118   
signed char
118   
signed char
119   
hexdig_value(char ch) noexcept
119   
hexdig_value(char ch) noexcept
120   
{
120   
{
121   
    // Idea for a switch statement to
121   
    // Idea for a switch statement to
122   
    // minimize emitted assembly from
122   
    // minimize emitted assembly from
123   
    // Glen Fernandes
123   
    // Glen Fernandes
124   
    signed char res;
124   
    signed char res;
125   
    switch(ch)
125   
    switch(ch)
126   
    {
126   
    {
127   
    default:            res = -1; break;
127   
    default:            res = -1; break;
128   
    case '0':           res =  0; break;
128   
    case '0':           res =  0; break;
129   
    case '1':           res =  1; break;
129   
    case '1':           res =  1; break;
130   
    case '2':           res =  2; break;
130   
    case '2':           res =  2; break;
131   
    case '3':           res =  3; break;
131   
    case '3':           res =  3; break;
132   
    case '4':           res =  4; break;
132   
    case '4':           res =  4; break;
133   
    case '5':           res =  5; break;
133   
    case '5':           res =  5; break;
134   
    case '6':           res =  6; break;
134   
    case '6':           res =  6; break;
135   
    case '7':           res =  7; break;
135   
    case '7':           res =  7; break;
136   
    case '8':           res =  8; break;
136   
    case '8':           res =  8; break;
137   
    case '9':           res =  9; break;
137   
    case '9':           res =  9; break;
138   
    case 'a': case 'A': res = 10; break;
138   
    case 'a': case 'A': res = 10; break;
139   
    case 'b': case 'B': res = 11; break;
139   
    case 'b': case 'B': res = 11; break;
140   
    case 'c': case 'C': res = 12; break;
140   
    case 'c': case 'C': res = 12; break;
141   
    case 'd': case 'D': res = 13; break;
141   
    case 'd': case 'D': res = 13; break;
142   
    case 'e': case 'E': res = 14; break;
142   
    case 'e': case 'E': res = 14; break;
143   
    case 'f': case 'F': res = 15; break;
143   
    case 'f': case 'F': res = 15; break;
144   
    }
144   
    }
145   
    return res;
145   
    return res;
146   
}
146   
}
147  
147  
148   
} // grammar
148   
} // grammar
149   
} // urls
149   
} // urls
150   
} // boost
150   
} // boost
151  
151  
152   
#endif
152   
#endif