demo_utf8towc.c 864 B

12345678910111213141516171819202122232425262728293031323334353637383940
  1. /* demo_utf8towc() against libc mbrtowc()
  2. *
  3. * Copyright 2017 Rob Landley <rob@landley.net>
  4. USE_DEMO_UTF8TOWC(NEWTOY(demo_utf8towc, 0, TOYFLAG_USR|TOYFLAG_BIN))
  5. config DEMO_UTF8TOWC
  6. bool "demo_utf8towc"
  7. default n
  8. help
  9. usage: demo_utf8towc
  10. Print differences between toybox's utf8 conversion routines vs libc du jour.
  11. */
  12. #include "toys.h"
  13. void demo_utf8towc_main(void)
  14. {
  15. mbstate_t mb;
  16. int len1, len2;
  17. unsigned u, h, wc2;
  18. wchar_t wc1;
  19. memset(&mb, 0, sizeof(mb));
  20. for (u = 1; u<=0x10ffff; u++) {
  21. char *str = (void *)&h;
  22. wc1 = wc2 = 0;
  23. len2 = 4;
  24. h = htonl(u);
  25. while (!*str) str++, len2--;
  26. len1 = mbrtowc(&wc1, str, len2, &mb);
  27. if (len1<0) memset(&mb, 0, sizeof(mb));
  28. len2 = utf8towc(&wc2, str, len2);
  29. if (len1 != len2 || wc1 != wc2)
  30. printf("%x %d %x %d %x\n", u, len1, wc1, len2, wc2);
  31. }
  32. }