| ... | ... | @@ -242,6 +242,13 @@ fn Validity_Criteria( |
| 242 | 242 | var it = std.unicode.Utf8View.initUnchecked(label).iterator(); |
| 243 | 243 | var prev_ccc: u8 = 0; |
| 244 | 244 | var prev_cp: u21 = 0; |
| 245 | |
| 246 | // 8. If CheckJoiners, the label must satisify the ContextJ rules from RFC 5892 Appendix A. |
| 247 | var point_2before: ucd.unicode_data.Codepoint = ucd.unicode_data.data[0]; |
| 248 | var point_before: ucd.unicode_data.Codepoint = ucd.unicode_data.data[0]; |
| 249 | var zwnj_state: enum { start, dual_joining, transparent1, zwnj, transparent2, right_joining } = .start; |
| 250 | var check_zwnj_state: ?bool = null; |
| 251 | |
| 245 | 252 | while (it.nextCodepointSlice()) |sl| { |
| 246 | 253 | const cp = std.unicode.utf8Decode(sl) catch unreachable; |
| 247 | 254 | const status, _, _ = mappingRow(cp); |
| ... | ... | @@ -292,98 +299,89 @@ fn Validity_Criteria( |
| 292 | 299 | if (sl[0] == '-') continue; |
| 293 | 300 | return error.IDNAFailure; |
| 294 | 301 | } |
| 295 | | } |
| 296 | 302 | |
| 297 | | // 8. If CheckJoiners, the label must satisify the ContextJ rules from RFC 5892 Appendix A. |
| 298 | | if (CheckJoiners) { |
| 299 | | it.i = 0; |
| 300 | | var point_2before: ucd.unicode_data.Codepoint = ucd.unicode_data.data[0]; |
| 301 | | var point_before: ucd.unicode_data.Codepoint = ucd.unicode_data.data[0]; |
| 302 | | var zwnj_state: enum { start, dual_joining, transparent1, zwnj, transparent2, right_joining } = .start; |
| 303 | | var check_zwnj_state: ?bool = null; |
| 303 | const p = ucd.unicode_data.find(cp); |
| 304 | 304 | |
| 305 | | while (it.nextCodepointSlice()) |s| { |
| 306 | | const c = std.unicode.utf8Decode(s) catch unreachable; |
| 307 | | const p = ucd.unicode_data.find(c); |
| 308 | | { |
| 309 | | // Appendix A.1. ZERO WIDTH NON-JOINER |
| 310 | | if (c == 0x200C) { |
| 311 | | if (check_zwnj_state == null) check_zwnj_state = true; |
| 312 | | if (point_before[3] == 9) check_zwnj_state = false; |
| 313 | | } |
| 314 | | switch (zwnj_state) { |
| 315 | | .start => { |
| 316 | | if (Joining_Type(c)) |t| { |
| 317 | | if (t == .L) zwnj_state = .dual_joining; |
| 318 | | if (t == .D) zwnj_state = .dual_joining; |
| 319 | | } |
| 320 | | }, |
| 321 | | .dual_joining => { |
| 322 | | if (Joining_Type(c)) |t| { |
| 323 | | switch (t) { |
| 324 | | .T => zwnj_state = .transparent1, |
| 325 | | .L, .D => {}, |
| 326 | | else => zwnj_state = .transparent1, |
| 327 | | } |
| 328 | | } else { |
| 329 | | zwnj_state = .transparent1; |
| 330 | | } |
| 331 | | if (c == 0x200C) zwnj_state = .zwnj; |
| 332 | | }, |
| 333 | | .transparent1 => { |
| 334 | | if (c == 0x200C) { |
| 335 | | zwnj_state = .zwnj; |
| 336 | | } else { |
| 337 | | zwnj_state = .start; |
| 305 | if (CheckJoiners) { |
| 306 | // Appendix A.1. ZERO WIDTH NON-JOINER |
| 307 | if (cp == 0x200C) { |
| 308 | if (check_zwnj_state == null) check_zwnj_state = true; |
| 309 | if (point_before[3] == 9) check_zwnj_state = false; |
| 310 | } |
| 311 | switch (zwnj_state) { |
| 312 | .start => { |
| 313 | if (Joining_Type(cp)) |t| { |
| 314 | if (t == .L) zwnj_state = .dual_joining; |
| 315 | if (t == .D) zwnj_state = .dual_joining; |
| 316 | } |
| 317 | }, |
| 318 | .dual_joining => { |
| 319 | if (Joining_Type(cp)) |t| { |
| 320 | switch (t) { |
| 321 | .T => zwnj_state = .transparent1, |
| 322 | .L, .D => {}, |
| 323 | else => zwnj_state = .transparent1, |
| 338 | 324 | } |
| 339 | | }, |
| 340 | | .zwnj => { |
| 341 | | if (Joining_Type(c)) |t| { |
| 342 | | switch (t) { |
| 343 | | .T => zwnj_state = .transparent2, |
| 344 | | .R, .D => zwnj_state = .right_joining, |
| 345 | | .L => zwnj_state = .dual_joining, |
| 346 | | else => zwnj_state = .transparent2, |
| 347 | | } |
| 348 | | } else { |
| 349 | | zwnj_state = .transparent2; |
| 325 | } else { |
| 326 | zwnj_state = .transparent1; |
| 327 | } |
| 328 | if (cp == 0x200C) zwnj_state = .zwnj; |
| 329 | }, |
| 330 | .transparent1 => { |
| 331 | if (cp == 0x200C) { |
| 332 | zwnj_state = .zwnj; |
| 333 | } else { |
| 334 | zwnj_state = .start; |
| 335 | } |
| 336 | }, |
| 337 | .zwnj => { |
| 338 | if (Joining_Type(cp)) |t| { |
| 339 | switch (t) { |
| 340 | .T => zwnj_state = .transparent2, |
| 341 | .R, .D => zwnj_state = .right_joining, |
| 342 | .L => zwnj_state = .dual_joining, |
| 343 | else => zwnj_state = .transparent2, |
| 350 | 344 | } |
| 351 | | }, |
| 352 | | .transparent2 => { |
| 353 | | if (Joining_Type(c)) |t| { |
| 354 | | switch (t) { |
| 355 | | .R, .D => zwnj_state = .right_joining, |
| 356 | | .L => zwnj_state = .dual_joining, |
| 357 | | else => zwnj_state = .start, |
| 358 | | } |
| 359 | | } else { |
| 360 | | zwnj_state = .start; |
| 345 | } else { |
| 346 | zwnj_state = .transparent2; |
| 347 | } |
| 348 | }, |
| 349 | .transparent2 => { |
| 350 | if (Joining_Type(cp)) |t| { |
| 351 | switch (t) { |
| 352 | .R, .D => zwnj_state = .right_joining, |
| 353 | .L => zwnj_state = .dual_joining, |
| 354 | else => zwnj_state = .start, |
| 361 | 355 | } |
| 362 | | }, |
| 363 | | .right_joining => { |
| 364 | | //rule passed |
| 365 | | }, |
| 366 | | } |
| 367 | | // Appendix A.2. ZERO WIDTH JOINER |
| 368 | | if (c == 0x200D and point_before[3] != 9) return error.IDNAFailure; |
| 369 | | // Appendix A.3. MIDDLE DOT |
| 370 | | if (point_before[0] == 0x00B7 and !(point_2before[0] == 0x006C and c == 0x006C)) return error.IDNAFailure; |
| 371 | | // Appendix A.4. GREEK LOWER NUMERAL SIGN (KERAIA) |
| 372 | | if (point_before[0] == 0x0375 and !(Script(c) == .Greek)) return error.IDNAFailure; |
| 373 | | // Appendix A.5. HEBREW PUNCTUATION GERESH |
| 374 | | if (c == 0x05F3 and !(Script(point_before[0]) == .Hebrew)) return error.IDNAFailure; |
| 375 | | // Appendix A.6. HEBREW PUNCTUATION GERSHAYIM |
| 376 | | if (c == 0x05F4 and !(Script(point_before[0]) == .Hebrew)) return error.IDNAFailure; |
| 377 | | // Appendix A.7. KATAKANA MIDDLE DOT |
| 378 | | if (false) return error.IDNAFailure; |
| 379 | | // Appendix A.8. ARABIC-INDIC DIGITS |
| 380 | | if (false) return error.IDNAFailure; |
| 381 | | // Appendix A.9. EXTENDED ARABIC-INDIC DIGITS |
| 382 | | if (false) return error.IDNAFailure; |
| 356 | } else { |
| 357 | zwnj_state = .start; |
| 358 | } |
| 359 | }, |
| 360 | .right_joining => { |
| 361 | //rule passed |
| 362 | }, |
| 383 | 363 | } |
| 384 | | point_2before = point_before; |
| 385 | | point_before = p; |
| 364 | // Appendix A.2. ZERO WIDTH JOINER |
| 365 | if (cp == 0x200D and point_before[3] != 9) return error.IDNAFailure; |
| 366 | // Appendix A.3. MIDDLE DOT |
| 367 | if (point_before[0] == 0x00B7 and !(point_2before[0] == 0x006C and cp == 0x006C)) return error.IDNAFailure; |
| 368 | // Appendix A.4. GREEK LOWER NUMERAL SIGN (KERAIA) |
| 369 | if (point_before[0] == 0x0375 and !(Script(cp) == .Greek)) return error.IDNAFailure; |
| 370 | // Appendix A.5. HEBREW PUNCTUATION GERESH |
| 371 | if (cp == 0x05F3 and !(Script(point_before[0]) == .Hebrew)) return error.IDNAFailure; |
| 372 | // Appendix A.6. HEBREW PUNCTUATION GERSHAYIM |
| 373 | if (cp == 0x05F4 and !(Script(point_before[0]) == .Hebrew)) return error.IDNAFailure; |
| 374 | // Appendix A.7. KATAKANA MIDDLE DOT |
| 375 | if (false) return error.IDNAFailure; |
| 376 | // Appendix A.8. ARABIC-INDIC DIGITS |
| 377 | if (false) return error.IDNAFailure; |
| 378 | // Appendix A.9. EXTENDED ARABIC-INDIC DIGITS |
| 379 | if (false) return error.IDNAFailure; |
| 386 | 380 | } |
| 381 | point_2before = point_before; |
| 382 | point_before = p; |
| 383 | } |
| 384 | if (CheckJoiners) { |
| 387 | 385 | if (check_zwnj_state == true and zwnj_state != .right_joining) return error.IDNAFailure; |
| 388 | 386 | } |
| 389 | 387 | |
Meghan Denny <hello@nektro.net> 2026-01-20 23:34:41 -08:00