utf16.odin 1.6 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081
  1. package utf16
  2. REPLACEMENT_CHAR :: '\ufffd';
  3. MAX_RUNE :: '\U0010ffff';
  4. _surr1 :: 0xd800;
  5. _surr2 :: 0xdc00;
  6. _surr3 :: 0xe000;
  7. _surr_self :: 0x10000;
  8. is_surrogate :: proc(r: rune) -> bool {
  9. return _surr1 <= r && r < _surr3;
  10. }
  11. decode_surrogate_pair :: proc(r1, r2: rune) -> rune {
  12. if _surr1 <= r1 && r1 < _surr2 && _surr2 <= r2 && r2 < _surr3 {
  13. return (r1-_surr1)<<10 | (r2 - _surr2) + _surr_self;
  14. }
  15. return REPLACEMENT_CHAR;
  16. }
  17. encode_surrogate_pair :: proc(r: rune) -> (r1, r2: rune) {
  18. if r < _surr_self || r > MAX_RUNE {
  19. return REPLACEMENT_CHAR, REPLACEMENT_CHAR;
  20. }
  21. r -= _surr_self;
  22. return _surr1 + (r>>10)&0x3ff, _surr2 + r&0x3ff;
  23. }
  24. encode :: proc(d: []u16, s: []rune) -> int {
  25. n, m := 0, len(d);
  26. loop: for r in s {
  27. switch r {
  28. case 0.._surr1-1, _surr3 .. _surr_self-1:
  29. if m+1 < n do break loop;
  30. d[n] = u16(r);
  31. n += 1;
  32. case _surr_self .. MAX_RUNE:
  33. if m+2 < n do break loop;
  34. r1, r2 := encode_surrogate_pair(r);
  35. d[n] = u16(r1);
  36. d[n+1] = u16(r2);
  37. n += 2;
  38. case:
  39. if m+1 < n do break loop;
  40. d[n] = u16(REPLACEMENT_CHAR);
  41. n += 1;
  42. }
  43. }
  44. return n;
  45. }
  46. encode_string :: proc(d: []u16, s: string) -> int {
  47. n, m := 0, len(d);
  48. loop: for r in s {
  49. switch r {
  50. case 0.._surr1-1, _surr3 .. _surr_self-1:
  51. if m+1 < n do break loop;
  52. d[n] = u16(r);
  53. n += 1;
  54. case _surr_self .. MAX_RUNE:
  55. if m+2 < n do break loop;
  56. r1, r2 := encode_surrogate_pair(r);
  57. d[n] = u16(r1);
  58. d[n+1] = u16(r2);
  59. n += 2;
  60. case:
  61. if m+1 < n do break loop;
  62. d[n] = u16(REPLACEMENT_CHAR);
  63. n += 1;
  64. }
  65. }
  66. return n;
  67. }