getEncoding.js 2.0 KB

12345678910111213141516171819202122232425262728293031323334353637383940414243444546474849505152535455565758596061626364656667686970
  1. function getEncoding(str) {
  2. const lowerCase = 3;
  3. const upperCase = 1;
  4. const codePage = {
  5. 'k': 'koi8-r',
  6. 'w': 'Windows-1251',
  7. 'd': 'cp866',
  8. 'i': 'ISO-8859-5',
  9. 'm': 'maccyrillic',
  10. };
  11. let charsets = {
  12. 'k': 0,
  13. 'w': 0,
  14. 'd': 0,
  15. 'i': 0,
  16. 'm': 0
  17. };
  18. const len = str.len;
  19. const blockSize = (len > 5*3000 ? 3000 : len);
  20. let counter = 0;
  21. let i = 0;
  22. while (i < len) {
  23. const char = str.charCodeAt(i);
  24. //non-russian characters
  25. if (char < 128 || char > 256)
  26. continue;
  27. //CP866
  28. if ((char > 159 && char < 176) || (char > 223 && char < 242)) charsets['d'] += lowerCase;
  29. if ((char > 127 && char < 160)) charsets['d'] += upperCase;
  30. //KOI8-R
  31. if ((char > 191 && char < 223)) charsets['k'] += lowerCase;
  32. if ((char > 222 && char < 256)) charsets['k'] += upperCase;
  33. //WIN-1251
  34. if (char > 223 && char < 256) charsets['w'] += lowerCase;
  35. if (char > 191 && char < 224) charsets['w'] += upperCase;
  36. //MAC
  37. if (char > 221 && char < 255) charsets['m'] += lowerCase;
  38. if (char > 127 && char < 160) charsets['m'] += upperCase;
  39. //ISO-8859-5
  40. if (char > 207 && char < 240) charsets['i'] += lowerCase;
  41. if (char > 175 && char < 208) charsets['i'] += upperCase;
  42. counter++;
  43. if (counter > blockSize) {
  44. counter = 0;
  45. i += Math.round(len/2 - 2*blockSize);
  46. }
  47. i++;
  48. }
  49. let sorted = Object.keys(charsets).map(function(key) {
  50. return { codePage: codePage[key], c: charsets[key] };
  51. });
  52. sorted.sort((a, b) => a.c - b.c);
  53. return sorted[0].codePage;
  54. }
  55. module.exports = getEncoding;