[6] | 1 | module m_charset |
---|
| 2 | ! |
---|
| 3 | ! One-byte only, sorry |
---|
| 4 | ! |
---|
| 5 | private |
---|
| 6 | |
---|
| 7 | integer, parameter, private :: small_int = selected_int_kind(1) |
---|
| 8 | |
---|
| 9 | !-------------------------------------------------------------------------- |
---|
| 10 | type, public :: charset_t |
---|
| 11 | ! private |
---|
| 12 | integer(kind=small_int), dimension(0:255) :: mask |
---|
| 13 | end type charset_t |
---|
| 14 | |
---|
| 15 | |
---|
| 16 | public :: operator(.in.), operator(+) |
---|
| 17 | public :: assignment(=) |
---|
| 18 | public :: print_charset, reset_charset |
---|
| 19 | |
---|
| 20 | interface operator(.in.) |
---|
| 21 | module procedure belongs |
---|
| 22 | end interface |
---|
| 23 | private :: belongs |
---|
| 24 | |
---|
| 25 | interface assignment(=) |
---|
| 26 | module procedure set_string_to_charset, set_codes_to_charset |
---|
| 27 | end interface |
---|
| 28 | private :: set_string_to_charset, set_codes_to_charset |
---|
| 29 | |
---|
| 30 | interface operator(+) |
---|
| 31 | module procedure add_string_to_charset, & |
---|
| 32 | add_code_to_charset, add_codes_to_charset |
---|
| 33 | end interface |
---|
| 34 | private :: add_string_to_charset, add_code_to_charset, add_codes_to_charset |
---|
| 35 | |
---|
| 36 | !-------------------------------------------------------------------------- |
---|
| 37 | |
---|
| 38 | character(len=*), parameter, private :: & |
---|
| 39 | lowercase = "abcdefghijklmnopqrstuvwxyz", & |
---|
| 40 | uppercase = "ABCDEFGHIJKLMNOPQRSTUVWXYZ", & |
---|
| 41 | digits = "0123456789" |
---|
| 42 | |
---|
| 43 | integer, parameter, public :: SPACE = 32 |
---|
| 44 | integer, parameter, public :: NEWLINE = 10 |
---|
| 45 | integer, parameter, public :: CARRIAGE_RETURN = 13 |
---|
| 46 | integer, parameter, public :: TAB = 9 |
---|
| 47 | |
---|
| 48 | type(charset_t), public :: initial_name_chars |
---|
| 49 | type(charset_t), public :: name_chars |
---|
| 50 | type(charset_t), public :: whitespace |
---|
| 51 | type(charset_t), public :: valid_chars |
---|
| 52 | type(charset_t), public :: uppercase_chars |
---|
| 53 | |
---|
| 54 | public :: setup_xml_charsets |
---|
| 55 | |
---|
| 56 | |
---|
| 57 | CONTAINS !========================================================== |
---|
| 58 | |
---|
| 59 | !-------------------------------------------------------------- |
---|
| 60 | function belongs(c,charset) result(res) |
---|
| 61 | character(len=1), intent(in) :: c |
---|
| 62 | type(charset_t), intent(in) :: charset |
---|
| 63 | logical :: res |
---|
| 64 | |
---|
| 65 | integer :: code |
---|
| 66 | |
---|
| 67 | code = ichar(c) |
---|
| 68 | res = (charset%mask(code) == 1) |
---|
| 69 | |
---|
| 70 | end function belongs |
---|
| 71 | |
---|
| 72 | !-------------------------------------------------------------- |
---|
| 73 | |
---|
| 74 | function add_string_to_charset(charset,str) result (sum) |
---|
| 75 | type(charset_t), intent(in) :: charset |
---|
| 76 | character(len=*), intent(in) :: str |
---|
| 77 | type(charset_t) :: sum |
---|
| 78 | |
---|
| 79 | integer :: length, code, i |
---|
| 80 | |
---|
| 81 | sum%mask = charset%mask |
---|
| 82 | |
---|
| 83 | length = len_trim(str) |
---|
| 84 | do i = 1, length |
---|
| 85 | code = ichar(str(i:i)) |
---|
| 86 | sum%mask(code) = 1 |
---|
| 87 | enddo |
---|
| 88 | end function add_string_to_charset |
---|
| 89 | |
---|
| 90 | !-------------------------------------------------------------- |
---|
| 91 | |
---|
| 92 | function add_code_to_charset(charset,code) result(sum) |
---|
| 93 | type(charset_t), intent(in) :: charset |
---|
| 94 | integer, intent(in) :: code |
---|
| 95 | type(charset_t) :: sum |
---|
| 96 | |
---|
| 97 | if ((code > 255) .or. (code < 0)) return |
---|
| 98 | sum%mask = charset%mask |
---|
| 99 | sum%mask(code) = 1 |
---|
| 100 | |
---|
| 101 | end function add_code_to_charset |
---|
| 102 | |
---|
| 103 | !-------------------------------------------------------------- |
---|
| 104 | function add_codes_to_charset(charset,codes) result(sum) |
---|
| 105 | type(charset_t), intent(in) :: charset |
---|
| 106 | integer, dimension(:), intent(in) :: codes |
---|
| 107 | type(charset_t) :: sum |
---|
| 108 | |
---|
| 109 | integer :: i |
---|
| 110 | |
---|
| 111 | sum%mask = charset%mask |
---|
| 112 | do i = 1, size(codes) |
---|
| 113 | if ((codes(i) > 255) .or. (codes(i) < 0)) cycle |
---|
| 114 | sum%mask(codes(i)) = 1 |
---|
| 115 | enddo |
---|
| 116 | end function add_codes_to_charset |
---|
| 117 | |
---|
| 118 | !-------------------------------------------------------------- |
---|
| 119 | |
---|
| 120 | subroutine set_string_to_charset(charset,str) |
---|
| 121 | type(charset_t), intent(out) :: charset |
---|
| 122 | character(len=*), intent(in) :: str |
---|
| 123 | |
---|
| 124 | |
---|
| 125 | integer :: length, code, i |
---|
| 126 | |
---|
| 127 | charset%mask = 0 |
---|
| 128 | |
---|
| 129 | length = len_trim(str) |
---|
| 130 | do i = 1, length |
---|
| 131 | code = ichar(str(i:i)) |
---|
| 132 | charset%mask(code) = 1 |
---|
| 133 | enddo |
---|
| 134 | |
---|
| 135 | end subroutine set_string_to_charset |
---|
| 136 | |
---|
| 137 | !-------------------------------------------------------------- |
---|
| 138 | |
---|
| 139 | subroutine set_codes_to_charset(charset,codes) |
---|
| 140 | type(charset_t), intent(out) :: charset |
---|
| 141 | integer, dimension(:), intent(in) :: codes |
---|
| 142 | |
---|
| 143 | integer :: i |
---|
| 144 | |
---|
| 145 | charset%mask = 0 |
---|
| 146 | |
---|
| 147 | do i = 1, size(codes) |
---|
| 148 | charset%mask(codes(i)) = 1 |
---|
| 149 | enddo |
---|
| 150 | |
---|
| 151 | end subroutine set_codes_to_charset |
---|
| 152 | |
---|
| 153 | |
---|
| 154 | !-------------------------------------------------------------- |
---|
| 155 | subroutine print_charset(charset) |
---|
| 156 | type(charset_t), intent(in) :: charset |
---|
| 157 | |
---|
| 158 | integer :: i |
---|
| 159 | |
---|
| 160 | do i = 0, 255 |
---|
| 161 | if (charset%mask(i) == 1) print *, "Code: ", i |
---|
| 162 | enddo |
---|
| 163 | end subroutine print_charset |
---|
| 164 | |
---|
| 165 | !-------------------------------------------------------------- |
---|
| 166 | |
---|
| 167 | subroutine reset_charset(charset) |
---|
| 168 | type(charset_t), intent(inout) :: charset |
---|
| 169 | |
---|
| 170 | integer :: i |
---|
| 171 | |
---|
| 172 | do i = 0, 255 |
---|
| 173 | charset%mask(i) = 0 |
---|
| 174 | enddo |
---|
| 175 | end subroutine reset_charset |
---|
| 176 | |
---|
| 177 | !-------------------------------------------------------------- |
---|
| 178 | |
---|
| 179 | !-------------------------------------------------------- |
---|
| 180 | subroutine setup_xml_charsets() |
---|
| 181 | |
---|
| 182 | integer :: i |
---|
| 183 | |
---|
| 184 | uppercase_chars = uppercase |
---|
| 185 | initial_name_chars = (lowercase // uppercase // "_:" ) |
---|
| 186 | name_chars = initial_name_chars + ( digits // ".-") |
---|
| 187 | whitespace = (/ SPACE, NEWLINE, TAB, CARRIAGE_RETURN /) |
---|
| 188 | |
---|
| 189 | valid_chars = whitespace + (/ (i, i=33,255) /) |
---|
| 190 | |
---|
| 191 | end subroutine setup_xml_charsets |
---|
| 192 | !-------------------------------------------------------- |
---|
| 193 | |
---|
| 194 | end module m_charset |
---|
| 195 | |
---|
| 196 | |
---|
| 197 | |
---|
| 198 | |
---|
| 199 | |
---|
| 200 | |
---|
| 201 | |
---|
| 202 | |
---|
| 203 | |
---|