| 1 | module wcwidth_mod |
| 2 | implicit none |
| 3 | private |
| 4 | |
| 5 | public :: codepoint_width |
| 6 | |
| 7 | contains |
| 8 | |
| 9 | ! Determine display width of a Unicode codepoint |
| 10 | ! Returns 0 for zero-width, 1 for normal, 2 for wide characters |
| 11 | function codepoint_width(cp) result(w) |
| 12 | integer, intent(in) :: cp |
| 13 | integer :: w |
| 14 | |
| 15 | w = 1 ! Default width |
| 16 | |
| 17 | ! Zero-width characters |
| 18 | if (cp == 0) then |
| 19 | w = 0 |
| 20 | return |
| 21 | end if |
| 22 | |
| 23 | ! C0/C1 control characters are zero-width |
| 24 | if (cp < 32 .or. (cp >= 127 .and. cp < 160)) then |
| 25 | w = 0 |
| 26 | return |
| 27 | end if |
| 28 | |
| 29 | ! Combining characters (selected ranges) |
| 30 | ! Combining Diacritical Marks: U+0300-U+036F |
| 31 | if (cp >= int(z'0300') .and. cp <= int(z'036F')) then |
| 32 | w = 0 |
| 33 | return |
| 34 | end if |
| 35 | |
| 36 | ! Combining Diacritical Marks Extended: U+1AB0-U+1AFF |
| 37 | if (cp >= int(z'1AB0') .and. cp <= int(z'1AFF')) then |
| 38 | w = 0 |
| 39 | return |
| 40 | end if |
| 41 | |
| 42 | ! Combining Diacritical Marks Supplement: U+1DC0-U+1DFF |
| 43 | if (cp >= int(z'1DC0') .and. cp <= int(z'1DFF')) then |
| 44 | w = 0 |
| 45 | return |
| 46 | end if |
| 47 | |
| 48 | ! Combining Diacritical Marks for Symbols: U+20D0-U+20FF |
| 49 | if (cp >= int(z'20D0') .and. cp <= int(z'20FF')) then |
| 50 | w = 0 |
| 51 | return |
| 52 | end if |
| 53 | |
| 54 | ! Variation Selectors: U+FE00-U+FE0F |
| 55 | if (cp >= int(z'FE00') .and. cp <= int(z'FE0F')) then |
| 56 | w = 0 |
| 57 | return |
| 58 | end if |
| 59 | |
| 60 | ! Zero Width Joiner/Non-Joiner: U+200B-U+200D |
| 61 | if (cp >= int(z'200B') .and. cp <= int(z'200D')) then |
| 62 | w = 0 |
| 63 | return |
| 64 | end if |
| 65 | |
| 66 | ! Soft Hyphen: U+00AD |
| 67 | if (cp == int(z'00AD')) then |
| 68 | w = 0 |
| 69 | return |
| 70 | end if |
| 71 | |
| 72 | ! Wide character ranges (East Asian Width: Wide or Fullwidth) |
| 73 | |
| 74 | ! CJK Unified Ideographs: U+4E00-U+9FFF |
| 75 | if (cp >= int(z'4E00') .and. cp <= int(z'9FFF')) then |
| 76 | w = 2 |
| 77 | return |
| 78 | end if |
| 79 | |
| 80 | ! CJK Unified Ideographs Extension A: U+3400-U+4DBF |
| 81 | if (cp >= int(z'3400') .and. cp <= int(z'4DBF')) then |
| 82 | w = 2 |
| 83 | return |
| 84 | end if |
| 85 | |
| 86 | ! CJK Unified Ideographs Extension B: U+20000-U+2A6DF |
| 87 | if (cp >= int(z'20000') .and. cp <= int(z'2A6DF')) then |
| 88 | w = 2 |
| 89 | return |
| 90 | end if |
| 91 | |
| 92 | ! CJK Compatibility Ideographs: U+F900-U+FAFF |
| 93 | if (cp >= int(z'F900') .and. cp <= int(z'FAFF')) then |
| 94 | w = 2 |
| 95 | return |
| 96 | end if |
| 97 | |
| 98 | ! CJK Compatibility Ideographs Supplement: U+2F800-U+2FA1F |
| 99 | if (cp >= int(z'2F800') .and. cp <= int(z'2FA1F')) then |
| 100 | w = 2 |
| 101 | return |
| 102 | end if |
| 103 | |
| 104 | ! Hangul Syllables: U+AC00-U+D7AF |
| 105 | if (cp >= int(z'AC00') .and. cp <= int(z'D7AF')) then |
| 106 | w = 2 |
| 107 | return |
| 108 | end if |
| 109 | |
| 110 | ! Hangul Jamo: U+1100-U+11FF |
| 111 | if (cp >= int(z'1100') .and. cp <= int(z'11FF')) then |
| 112 | w = 2 |
| 113 | return |
| 114 | end if |
| 115 | |
| 116 | ! Hangul Compatibility Jamo: U+3130-U+318F |
| 117 | if (cp >= int(z'3130') .and. cp <= int(z'318F')) then |
| 118 | w = 2 |
| 119 | return |
| 120 | end if |
| 121 | |
| 122 | ! Hangul Jamo Extended-A: U+A960-U+A97F |
| 123 | if (cp >= int(z'A960') .and. cp <= int(z'A97F')) then |
| 124 | w = 2 |
| 125 | return |
| 126 | end if |
| 127 | |
| 128 | ! Hangul Jamo Extended-B: U+D7B0-U+D7FF |
| 129 | if (cp >= int(z'D7B0') .and. cp <= int(z'D7FF')) then |
| 130 | w = 2 |
| 131 | return |
| 132 | end if |
| 133 | |
| 134 | ! CJK Radicals Supplement: U+2E80-U+2EFF |
| 135 | if (cp >= int(z'2E80') .and. cp <= int(z'2EFF')) then |
| 136 | w = 2 |
| 137 | return |
| 138 | end if |
| 139 | |
| 140 | ! Kangxi Radicals: U+2F00-U+2FDF |
| 141 | if (cp >= int(z'2F00') .and. cp <= int(z'2FDF')) then |
| 142 | w = 2 |
| 143 | return |
| 144 | end if |
| 145 | |
| 146 | ! CJK Symbols and Punctuation: U+3000-U+303F |
| 147 | if (cp >= int(z'3000') .and. cp <= int(z'303F')) then |
| 148 | w = 2 |
| 149 | return |
| 150 | end if |
| 151 | |
| 152 | ! Hiragana: U+3040-U+309F |
| 153 | if (cp >= int(z'3040') .and. cp <= int(z'309F')) then |
| 154 | w = 2 |
| 155 | return |
| 156 | end if |
| 157 | |
| 158 | ! Katakana: U+30A0-U+30FF |
| 159 | if (cp >= int(z'30A0') .and. cp <= int(z'30FF')) then |
| 160 | w = 2 |
| 161 | return |
| 162 | end if |
| 163 | |
| 164 | ! Katakana Phonetic Extensions: U+31F0-U+31FF |
| 165 | if (cp >= int(z'31F0') .and. cp <= int(z'31FF')) then |
| 166 | w = 2 |
| 167 | return |
| 168 | end if |
| 169 | |
| 170 | ! Bopomofo: U+3100-U+312F |
| 171 | if (cp >= int(z'3100') .and. cp <= int(z'312F')) then |
| 172 | w = 2 |
| 173 | return |
| 174 | end if |
| 175 | |
| 176 | ! Bopomofo Extended: U+31A0-U+31BF |
| 177 | if (cp >= int(z'31A0') .and. cp <= int(z'31BF')) then |
| 178 | w = 2 |
| 179 | return |
| 180 | end if |
| 181 | |
| 182 | ! CJK Strokes: U+31C0-U+31EF |
| 183 | if (cp >= int(z'31C0') .and. cp <= int(z'31EF')) then |
| 184 | w = 2 |
| 185 | return |
| 186 | end if |
| 187 | |
| 188 | ! Enclosed CJK Letters and Months: U+3200-U+32FF |
| 189 | if (cp >= int(z'3200') .and. cp <= int(z'32FF')) then |
| 190 | w = 2 |
| 191 | return |
| 192 | end if |
| 193 | |
| 194 | ! CJK Compatibility: U+3300-U+33FF |
| 195 | if (cp >= int(z'3300') .and. cp <= int(z'33FF')) then |
| 196 | w = 2 |
| 197 | return |
| 198 | end if |
| 199 | |
| 200 | ! Fullwidth ASCII: U+FF01-U+FF60 |
| 201 | if (cp >= int(z'FF01') .and. cp <= int(z'FF60')) then |
| 202 | w = 2 |
| 203 | return |
| 204 | end if |
| 205 | |
| 206 | ! Fullwidth Punctuation: U+FFE0-U+FFE6 |
| 207 | if (cp >= int(z'FFE0') .and. cp <= int(z'FFE6')) then |
| 208 | w = 2 |
| 209 | return |
| 210 | end if |
| 211 | |
| 212 | ! Halfwidth Katakana: U+FF65-U+FFDC (these are actually width 1) |
| 213 | if (cp >= int(z'FF65') .and. cp <= int(z'FFDC')) then |
| 214 | w = 1 |
| 215 | return |
| 216 | end if |
| 217 | |
| 218 | ! Yi Syllables: U+A000-U+A48F |
| 219 | if (cp >= int(z'A000') .and. cp <= int(z'A48F')) then |
| 220 | w = 2 |
| 221 | return |
| 222 | end if |
| 223 | |
| 224 | ! Yi Radicals: U+A490-U+A4CF |
| 225 | if (cp >= int(z'A490') .and. cp <= int(z'A4CF')) then |
| 226 | w = 2 |
| 227 | return |
| 228 | end if |
| 229 | |
| 230 | ! Emoji (most are wide) |
| 231 | ! Miscellaneous Symbols and Pictographs: U+1F300-U+1F5FF |
| 232 | if (cp >= int(z'1F300') .and. cp <= int(z'1F5FF')) then |
| 233 | w = 2 |
| 234 | return |
| 235 | end if |
| 236 | |
| 237 | ! Emoticons: U+1F600-U+1F64F |
| 238 | if (cp >= int(z'1F600') .and. cp <= int(z'1F64F')) then |
| 239 | w = 2 |
| 240 | return |
| 241 | end if |
| 242 | |
| 243 | ! Transport and Map Symbols: U+1F680-U+1F6FF |
| 244 | if (cp >= int(z'1F680') .and. cp <= int(z'1F6FF')) then |
| 245 | w = 2 |
| 246 | return |
| 247 | end if |
| 248 | |
| 249 | ! Supplemental Symbols and Pictographs: U+1F900-U+1F9FF |
| 250 | if (cp >= int(z'1F900') .and. cp <= int(z'1F9FF')) then |
| 251 | w = 2 |
| 252 | return |
| 253 | end if |
| 254 | |
| 255 | ! Symbols and Pictographs Extended-A: U+1FA00-U+1FA6F |
| 256 | if (cp >= int(z'1FA00') .and. cp <= int(z'1FA6F')) then |
| 257 | w = 2 |
| 258 | return |
| 259 | end if |
| 260 | |
| 261 | ! Symbols and Pictographs Extended-B: U+1FA70-U+1FAFF |
| 262 | if (cp >= int(z'1FA70') .and. cp <= int(z'1FAFF')) then |
| 263 | w = 2 |
| 264 | return |
| 265 | end if |
| 266 | |
| 267 | ! Default: width 1 |
| 268 | w = 1 |
| 269 | end function codepoint_width |
| 270 | |
| 271 | end module wcwidth_mod |
| 272 |