Blob
1 #ifdef PLAN92 #include <u.h>3 #include <libc.h>4 #include <bio.h>5 #else6 #include <sys/types.h>7 #include <stdio.h>8 #include <stdlib.h>9 #include <string.h>10 #include <unistd.h>11 #include <errno.h>12 #include "plan9.h"13 #endif14 #include "hdr.h"16 /*17 the our_* routines are implementations for the corresponding library18 routines. for a while, i tried to actually name them wctomb etc19 but stopped that after i found a system which made wchar_t an20 unsigned char.21 */23 int our_wctomb(char *s, unsigned long wc);24 int our_mbtowc(unsigned long *p, char *s, unsigned n);25 int runetoisoutf(char *str, Rune *rune);26 int fullisorune(char *str, int n);27 int isochartorune(Rune *rune, char *str);29 void30 utf_in(int fd, long *notused, struct convert *out)31 {32 char buf[N];33 int i, j, c, n, tot;34 ulong l;36 USED(notused);37 tot = 0;38 while((n = read(fd, buf+tot, N-tot)) >= 0){39 tot += n;40 for(i=j=0; i<tot; ){41 c = our_mbtowc(&l, buf+i, tot-i);42 if(c == -1)43 break;44 if(c == -2){45 if(squawk)46 EPR "%s: bad UTF sequence near byte %ld in input\n", argv0, ninput+i);47 if(clean)48 continue;49 nerrors++;50 l = Runeerror;51 }52 runes[j++] = l;53 i += c;54 }55 OUT(out, runes, j);56 tot -= i;57 ninput += i;58 if(tot)59 memmove(buf, buf+i, tot);60 if(n == 0)61 break;62 }63 }65 void66 utf_out(Rune *base, int n, long *notused)67 {68 char *p;69 Rune *r;71 USED(notused);72 nrunes += n;73 for(r = base, p = obuf; n-- > 0; r++){74 p += our_wctomb(p, *r);75 }76 noutput += p-obuf;77 write(1, obuf, p-obuf);78 }80 void81 isoutf_in(int fd, long *notused, struct convert *out)82 {83 char buf[N];84 int i, j, c, n, tot;86 USED(notused);87 tot = 0;88 while((n = read(fd, buf+tot, N-tot)) >= 0){89 tot += n;90 for(i=j=0; i<tot; ){91 if(!fullisorune(buf+i, tot-i))92 break;93 c = isochartorune(&runes[j], buf+i);94 if(runes[j] == Runeerror){95 if(squawk)96 EPR "%s: bad UTF sequence near byte %ld in input\n", argv0, ninput+i);97 if(clean)98 continue;99 nerrors++;100 }101 j++;102 i += c;103 }104 OUT(out, runes, j);105 tot -= i;106 ninput += i;107 if(tot)108 memmove(buf, buf+i, tot);109 if(n == 0)110 break;111 }112 }114 void115 isoutf_out(Rune *base, int n, long *notused)116 {117 char *p;118 Rune *r;120 USED(notused);121 nrunes += n;122 for(r = base, p = obuf; n-- > 0; r++)123 p += runetoisoutf(p, r);124 noutput += p-obuf;125 write(1, obuf, p-obuf);126 }129 enum130 {131 Char1 = Runeself, Rune1 = Runeself,132 Char21 = 0xA1, Rune21 = 0x0100,133 Char22 = 0xF6, Rune22 = 0x4016,134 Char3 = 0xFC, Rune3 = 0x10000, /* really 0x38E2E */135 Esc = 0xBE, Bad = Runeerror136 };138 static uchar U[256];139 static uchar T[256];141 static142 void143 mktable(void)144 {145 int i, u;147 for(i=0; i<256; i++) {148 u = i + (0x5E - 0xA0);149 if(i < 0xA0)150 u = i + (0xDF - 0x7F);151 if(i < 0x7F)152 u = i + (0x00 - 0x21);153 if(i < 0x21)154 u = i + (0xBE - 0x00);155 U[i] = u;156 T[u] = i;157 }158 }160 int161 isochartorune(Rune *rune, char *str)162 {163 int c, c1, c2;164 long l;166 if(U[0] == 0)167 mktable();169 /*170 * one character sequence171 * 00000-0009F => 00-9F172 */173 c = *(uchar*)str;174 if(c < Char1) {175 *rune = c;176 return 1;177 }179 /*180 * two character sequence181 * 000A0-000FF => A0; A0-FF182 */183 c1 = *(uchar*)(str+1);184 if(c < Char21) {185 if(c1 >= Rune1 && c1 < Rune21) {186 *rune = c1;187 return 2;188 }189 goto bad;190 }192 /*193 * two character sequence194 * 00100-04015 => A1-F5; 21-7E/A0-FF195 */196 c1 = U[c1];197 if(c1 >= Esc)198 goto bad;199 if(c < Char22) {200 *rune = (c-Char21)*Esc + c1 + Rune21;201 return 2;202 }204 /*205 * three character sequence206 * 04016-38E2D => A6-FB; 21-7E/A0-FF207 */208 c2 = U[*(uchar*)(str+2)];209 if(c2 >= Esc)210 goto bad;211 if(c < Char3) {212 l = (c-Char22)*Esc*Esc + c1*Esc + c2 + Rune22;213 if(l >= Rune3)214 goto bad;215 *rune = l;216 return 3;217 }219 /*220 * bad decoding221 */222 bad:223 *rune = Bad;224 return 1;225 }227 int228 runetoisoutf(char *str, Rune *rune)229 {230 long c;232 if(T[0] == 0)233 mktable();235 /*236 * one character sequence237 * 00000-0009F => 00-9F238 */239 c = *rune;240 if(c < Rune1) {241 str[0] = c;242 return 1;243 }245 /*246 * two character sequence247 * 000A0-000FF => A0; A0-FF248 */249 if(c < Rune21) {250 str[0] = (char)Char1;251 str[1] = c;252 return 2;253 }255 /*256 * two character sequence257 * 00100-04015 => A1-F5; 21-7E/A0-FF258 */259 if(c < Rune22) {260 c -= Rune21;261 str[0] = c/Esc + Char21;262 str[1] = T[c%Esc];263 return 2;264 }266 /*267 * three character sequence268 * 04016-38E2D => A6-FB; 21-7E/A0-FF269 */270 c -= Rune22;271 str[0] = c/(Esc*Esc) + Char22;272 str[1] = T[c/Esc%Esc];273 str[2] = T[c%Esc];274 return 3;275 }277 int278 fullisorune(char *str, int n)279 {280 int c;282 if(n > 0) {283 c = *(uchar*)str;284 if(c < Char1)285 return 1;286 if(n > 1)287 if(c < Char22 || n > 2)288 return 1;289 }290 return 0;291 }293 #ifdef PLAN9294 int errno;295 #endif297 enum298 {299 T1 = 0x00,300 Tx = 0x80,301 T2 = 0xC0,302 T3 = 0xE0,303 T4 = 0xF0,304 T5 = 0xF8,305 T6 = 0xFC,307 Bit1 = 7,308 Bitx = 6,309 Bit2 = 5,310 Bit3 = 4,311 Bit4 = 3,312 Bit5 = 2,313 Bit6 = 2,315 Mask1 = (1<<Bit1)-1,316 Maskx = (1<<Bitx)-1,317 Mask2 = (1<<Bit2)-1,318 Mask3 = (1<<Bit3)-1,319 Mask4 = (1<<Bit4)-1,320 Mask5 = (1<<Bit5)-1,321 Mask6 = (1<<Bit6)-1,323 Wchar1 = (1UL<<Bit1)-1,324 Wchar2 = (1UL<<(Bit2+Bitx))-1,325 Wchar3 = (1UL<<(Bit3+2*Bitx))-1,326 Wchar4 = (1UL<<(Bit4+3*Bitx))-1,327 Wchar5 = (1UL<<(Bit5+4*Bitx))-1329 #ifndef EILSEQ330 , /* we hate ansi c's comma rules */331 EILSEQ = 123332 #endif /* PLAN9 */333 };335 int336 our_wctomb(char *s, unsigned long wc)337 {338 if(s == 0)339 return 0; /* no shift states */340 if(wc & ~Wchar2) {341 if(wc & ~Wchar4) {342 if(wc & ~Wchar5) {343 /* 6 bytes */344 s[0] = T6 | ((wc >> 5*Bitx) & Mask6);345 s[1] = Tx | ((wc >> 4*Bitx) & Maskx);346 s[2] = Tx | ((wc >> 3*Bitx) & Maskx);347 s[3] = Tx | ((wc >> 2*Bitx) & Maskx);348 s[4] = Tx | ((wc >> 1*Bitx) & Maskx);349 s[5] = Tx | (wc & Maskx);350 return 6;351 }352 /* 5 bytes */353 s[0] = T5 | (wc >> 4*Bitx);354 s[1] = Tx | ((wc >> 3*Bitx) & Maskx);355 s[2] = Tx | ((wc >> 2*Bitx) & Maskx);356 s[3] = Tx | ((wc >> 1*Bitx) & Maskx);357 s[4] = Tx | (wc & Maskx);358 return 5;359 }360 if(wc & ~Wchar3) {361 /* 4 bytes */362 s[0] = T4 | (wc >> 3*Bitx);363 s[1] = Tx | ((wc >> 2*Bitx) & Maskx);364 s[2] = Tx | ((wc >> 1*Bitx) & Maskx);365 s[3] = Tx | (wc & Maskx);366 return 4;367 }368 /* 3 bytes */369 s[0] = T3 | (wc >> 2*Bitx);370 s[1] = Tx | ((wc >> 1*Bitx) & Maskx);371 s[2] = Tx | (wc & Maskx);372 return 3;373 }374 if(wc & ~Wchar1) {375 /* 2 bytes */376 s[0] = T2 | (wc >> 1*Bitx);377 s[1] = Tx | (wc & Maskx);378 return 2;379 }380 /* 1 byte */381 s[0] = T1 | wc;382 return 1;383 }385 int386 our_mbtowc(unsigned long *p, char *s, unsigned n)387 {388 uchar *us;389 int c0, c1, c2, c3, c4, c5;390 unsigned long wc;392 if(s == 0)393 return 0; /* no shift states */395 if(n < 1)396 goto badlen;397 us = (uchar*)s;398 c0 = us[0];399 if(c0 >= T3) {400 if(n < 3)401 goto badlen;402 c1 = us[1] ^ Tx;403 c2 = us[2] ^ Tx;404 if((c1|c2) & T2)405 goto bad;406 if(c0 >= T5) {407 if(n < 5)408 goto badlen;409 c3 = us[3] ^ Tx;410 c4 = us[4] ^ Tx;411 if((c3|c4) & T2)412 goto bad;413 if(c0 >= T6) {414 /* 6 bytes */415 if(n < 6)416 goto badlen;417 c5 = us[5] ^ Tx;418 if(c5 & T2)419 goto bad;420 wc = ((((((((((c0 & Mask6) << Bitx) |421 c1) << Bitx) | c2) << Bitx) |422 c3) << Bitx) | c4) << Bitx) | c5;423 if(wc <= Wchar5)424 goto bad;425 *p = wc;426 return 6;427 }428 /* 5 bytes */429 wc = ((((((((c0 & Mask5) << Bitx) |430 c1) << Bitx) | c2) << Bitx) |431 c3) << Bitx) | c4;432 if(wc <= Wchar4)433 goto bad;434 *p = wc;435 return 5;436 }437 if(c0 >= T4) {438 /* 4 bytes */439 if(n < 4)440 goto badlen;441 c3 = us[3] ^ Tx;442 if(c3 & T2)443 goto bad;444 wc = ((((((c0 & Mask4) << Bitx) |445 c1) << Bitx) | c2) << Bitx) |446 c3;447 if(wc <= Wchar3)448 goto bad;449 *p = wc;450 return 4;451 }452 /* 3 bytes */453 wc = ((((c0 & Mask3) << Bitx) |454 c1) << Bitx) | c2;455 if(wc <= Wchar2)456 goto bad;457 *p = wc;458 return 3;459 }460 if(c0 >= T2) {461 /* 2 bytes */462 if(n < 2)463 goto badlen;464 c1 = us[1] ^ Tx;465 if(c1 & T2)466 goto bad;467 wc = ((c0 & Mask2) << Bitx) |468 c1;469 if(wc <= Wchar1)470 goto bad;471 *p = wc;472 return 2;473 }474 /* 1 byte */475 if(c0 >= Tx)476 goto bad;477 *p = c0;478 return 1;480 bad:481 errno = EILSEQ;482 return -1;483 badlen:484 return -2;485 }