Decodes an NCName to a token. @param name name @param lax lax decoding @return token
(final byte[] name, final boolean lax)
| 348 | * @return token |
| 349 | */ |
| 350 | public static byte[] decode(final byte[] name, final boolean lax) { |
| 351 | final int nl = name.length; |
| 352 | if(nl == 0) return null; |
| 353 | if(lax) return name; |
| 354 | |
| 355 | // mode: 0=normal, 1=unicode, 2=underscore, 3=building unicode |
| 356 | final TokenBuilder tb = new TokenBuilder(); |
| 357 | int mode = 0, uc = 0; |
| 358 | for(int n = 0; n < nl;) { |
| 359 | final int cp = cp(name, n); |
| 360 | if(mode >= 3) { |
| 361 | uc <<= 4; |
| 362 | if(cp >= '0' && cp <= '9') { |
| 363 | uc += cp - '0'; |
| 364 | } else if(cp >= 'A' && cp <= 'F') { |
| 365 | uc += cp - 0x37; |
| 366 | } else if(cp >= 'a' && cp <= 'f') { |
| 367 | uc += cp - 0x57; |
| 368 | } else { |
| 369 | return null; |
| 370 | } |
| 371 | if(++mode == 7) { |
| 372 | tb.add(uc); |
| 373 | mode = 0; |
| 374 | uc = 0; |
| 375 | } |
| 376 | } else if(cp == '_') { |
| 377 | // limit underscore counter |
| 378 | if(++mode == 3) { |
| 379 | tb.add('_'); |
| 380 | mode = 0; |
| 381 | continue; |
| 382 | } |
| 383 | } else if(mode == 1) { |
| 384 | // unicode |
| 385 | mode = 3; |
| 386 | continue; |
| 387 | } else if(mode == 2) { |
| 388 | // underscore |
| 389 | tb.add('_'); |
| 390 | mode = 0; |
| 391 | continue; |
| 392 | } else { |
| 393 | // normal character |
| 394 | if(!(tb.isEmpty() ? isNCStartChar(cp) : isNCChar(cp))) return null; |
| 395 | tb.add(cp); |
| 396 | mode = 0; |
| 397 | } |
| 398 | n += cl(name, n); |
| 399 | } |
| 400 | |
| 401 | if(mode == 2) { |
| 402 | tb.add('_'); |
| 403 | } else if(mode == 1 && !tb.isEmpty() || mode >= 3) { |
| 404 | return null; |
| 405 | } |
| 406 | return tb.finish(); |
| 407 | } |