📄 utf.c
字号:
#ifdef PLAN9#include <u.h>#include <libc.h>#include <bio.h>#else#include <sys/types.h>#include <stdio.h>#include <stdlib.h>#include <string.h>#include <unistd.h>#include <errno.h>#include "plan9.h"#endif#include "hdr.h"/* the our_* routines are implementations for the corresponding library routines. for a while, i tried to actually name them wctomb etc but stopped that after i found a system which made wchar_t an unsigned char.*/int our_wctomb(char *s, unsigned long wc);int our_mbtowc(wchar_t *p, char *s, unsigned n);enum{ Char1 = Runeself, Rune1 = Runeself, Char21 = 0xA1, Rune21 = 0x0100, Char22 = 0xF6, Rune22 = 0x4016, Char3 = 0xFC, Rune3 = 0x10000, /* really 0x38E2E */ Esc = 0xBE, Bad = Runeerror};#ifdef PLAN9int errno;#endifenum{ T1 = 0x00, Tx = 0x80, T2 = 0xC0, T3 = 0xE0, T4 = 0xF0, T5 = 0xF8, T6 = 0xFC, Bit1 = 7, Bitx = 6, Bit2 = 5, Bit3 = 4, Bit4 = 3, Bit5 = 2, Bit6 = 2, Mask1 = (1<<Bit1)-1, Maskx = (1<<Bitx)-1, Mask2 = (1<<Bit2)-1, Mask3 = (1<<Bit3)-1, Mask4 = (1<<Bit4)-1, Mask5 = (1<<Bit5)-1, Mask6 = (1<<Bit6)-1, Wchar1 = (1UL<<Bit1)-1, Wchar2 = (1UL<<(Bit2+Bitx))-1, Wchar3 = (1UL<<(Bit3+2*Bitx))-1, Wchar4 = (1UL<<(Bit4+3*Bitx))-1, Wchar5 = (1UL<<(Bit5+4*Bitx))-1#ifndef EILSEQ , /* we hate ansi c's comma rules */ EILSEQ = 123#endif /* PLAN9 */};intour_wctomb(char *s, unsigned long wc){ if(s == 0) return 0; /* no shift states */ if(wc & ~Wchar2) { if(wc & ~Wchar4) { if(wc & ~Wchar5) { /* 6 bytes */ s[0] = T6 | ((wc >> 5*Bitx) & Mask6); s[1] = Tx | ((wc >> 4*Bitx) & Maskx); s[2] = Tx | ((wc >> 3*Bitx) & Maskx); s[3] = Tx | ((wc >> 2*Bitx) & Maskx); s[4] = Tx | ((wc >> 1*Bitx) & Maskx); s[5] = Tx | (wc & Maskx); return 6; } /* 5 bytes */ s[0] = T5 | (wc >> 4*Bitx); s[1] = Tx | ((wc >> 3*Bitx) & Maskx); s[2] = Tx | ((wc >> 2*Bitx) & Maskx); s[3] = Tx | ((wc >> 1*Bitx) & Maskx); s[4] = Tx | (wc & Maskx); return 5; } if(wc & ~Wchar3) { /* 4 bytes */ s[0] = T4 | (wc >> 3*Bitx); s[1] = Tx | ((wc >> 2*Bitx) & Maskx); s[2] = Tx | ((wc >> 1*Bitx) & Maskx); s[3] = Tx | (wc & Maskx); return 4; } /* 3 bytes */ s[0] = T3 | (wc >> 2*Bitx); s[1] = Tx | ((wc >> 1*Bitx) & Maskx); s[2] = Tx | (wc & Maskx); return 3; } if(wc & ~Wchar1) { /* 2 bytes */ s[0] = T2 | (wc >> 1*Bitx); s[1] = Tx | (wc & Maskx); return 2; } /* 1 byte */ s[0] = T1 | wc; return 1;}intour_mbtowc(wchar_t *p, char *s, unsigned n){ uchar *us; int c0, c1, c2, c3, c4, c5; unsigned long wc; wchar_t wchar; if(s == 0) return 0; /* no shift states */ if(n < 1) goto badlen; us = (uchar*)s; c0 = us[0]; if(c0 >= T3) { if(n < 3) goto badlen; c1 = us[1] ^ Tx; c2 = us[2] ^ Tx; if((c1|c2) & T2) goto bad; if(c0 >= T5) { if(n < 5) goto badlen; c3 = us[3] ^ Tx; c4 = us[4] ^ Tx; if((c3|c4) & T2) goto bad; if(c0 >= T6) { /* 6 bytes */ if(n < 6) goto badlen; c5 = us[5] ^ Tx; if(c5 & T2) goto bad; wc = ((((((((((c0 & Mask6) << Bitx) | c1) << Bitx) | c2) << Bitx) | c3) << Bitx) | c4) << Bitx) | c5; if(wc <= Wchar5) goto bad; *p = wchar = wc; return 6; } /* 5 bytes */ wc = ((((((((c0 & Mask5) << Bitx) | c1) << Bitx) | c2) << Bitx) | c3) << Bitx) | c4; if(wc <= Wchar4) goto bad; *p = wchar = wc; return 5; } if(c0 >= T4) { /* 4 bytes */ if(n < 4) goto badlen; c3 = us[3] ^ Tx; if(c3 & T2) goto bad; wc = ((((((c0 & Mask4) << Bitx) | c1) << Bitx) | c2) << Bitx) | c3; if(wc <= Wchar3) goto bad; *p = wchar = wc; return 4; } /* 3 bytes */ wc = ((((c0 & Mask3) << Bitx) | c1) << Bitx) | c2; if(wc <= Wchar2) goto bad; *p = wchar = wc; return 3; } if(c0 >= T2) { /* 2 bytes */ if(n < 2) goto badlen; c1 = us[1] ^ Tx; if(c1 & T2) goto bad; wc = ((c0 & Mask2) << Bitx) | c1; if(wc <= Wchar1) goto bad; *p = wchar = wc; return 2; } /* 1 byte */ if(c0 >= Tx) goto bad; *p = wchar = c0; return 1;bad: errno = EILSEQ; return -1;badlen: return -2;}
⌨️ 快捷键说明
复制代码
Ctrl + C
搜索代码
Ctrl + F
全屏模式
F11
切换主题
Ctrl + Shift + D
显示快捷键
?
增大字号
Ctrl + =
减小字号
Ctrl + -