unvis.c revision 241236
1241236Sbrooks/* $NetBSD: unvis.c,v 1.39 2012/03/13 21:13:37 christos Exp $ */ 2241236Sbrooks 3241236Sbrooks/*- 4241236Sbrooks * Copyright (c) 1989, 1993 5241236Sbrooks * The Regents of the University of California. All rights reserved. 6241236Sbrooks * 7241236Sbrooks * Redistribution and use in source and binary forms, with or without 8241236Sbrooks * modification, are permitted provided that the following conditions 9241236Sbrooks * are met: 10241236Sbrooks * 1. Redistributions of source code must retain the above copyright 11241236Sbrooks * notice, this list of conditions and the following disclaimer. 12241236Sbrooks * 2. Redistributions in binary form must reproduce the above copyright 13241236Sbrooks * notice, this list of conditions and the following disclaimer in the 14241236Sbrooks * documentation and/or other materials provided with the distribution. 15241236Sbrooks * 3. Neither the name of the University nor the names of its contributors 16241236Sbrooks * may be used to endorse or promote products derived from this software 17241236Sbrooks * without specific prior written permission. 18241236Sbrooks * 19241236Sbrooks * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 20241236Sbrooks * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 21241236Sbrooks * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 22241236Sbrooks * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 23241236Sbrooks * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 24241236Sbrooks * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 25241236Sbrooks * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 26241236Sbrooks * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 27241236Sbrooks * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 28241236Sbrooks * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 29241236Sbrooks * SUCH DAMAGE. 30241236Sbrooks */ 31241236Sbrooks 32241236Sbrooks#include <sys/cdefs.h> 33241236Sbrooks#if defined(LIBC_SCCS) && !defined(lint) 34241236Sbrooks#if 0 35241236Sbrooksstatic char sccsid[] = "@(#)unvis.c 8.1 (Berkeley) 6/4/93"; 36241236Sbrooks#else 37241236Sbrooks__RCSID("$NetBSD: unvis.c,v 1.39 2012/03/13 21:13:37 christos Exp $"); 38241236Sbrooks#endif 39241236Sbrooks#endif /* LIBC_SCCS and not lint */ 40241236Sbrooks 41241236Sbrooks#include "namespace.h" 42241236Sbrooks#include <sys/types.h> 43241236Sbrooks 44241236Sbrooks#include <assert.h> 45241236Sbrooks#include <ctype.h> 46241236Sbrooks#include <stdint.h> 47241236Sbrooks#include <stdio.h> 48241236Sbrooks#include <errno.h> 49241236Sbrooks#include <vis.h> 50241236Sbrooks 51241236Sbrooks#ifdef __weak_alias 52241236Sbrooks__weak_alias(strnunvisx,_strnunvisx) 53241236Sbrooks#endif 54241236Sbrooks 55241236Sbrooks#if !HAVE_VIS 56241236Sbrooks/* 57241236Sbrooks * decode driven by state machine 58241236Sbrooks */ 59241236Sbrooks#define S_GROUND 0 /* haven't seen escape char */ 60241236Sbrooks#define S_START 1 /* start decoding special sequence */ 61241236Sbrooks#define S_META 2 /* metachar started (M) */ 62241236Sbrooks#define S_META1 3 /* metachar more, regular char (-) */ 63241236Sbrooks#define S_CTRL 4 /* control char started (^) */ 64241236Sbrooks#define S_OCTAL2 5 /* octal digit 2 */ 65241236Sbrooks#define S_OCTAL3 6 /* octal digit 3 */ 66241236Sbrooks#define S_HEX 7 /* mandatory hex digit */ 67241236Sbrooks#define S_HEX1 8 /* http hex digit */ 68241236Sbrooks#define S_HEX2 9 /* http hex digit 2 */ 69241236Sbrooks#define S_MIME1 10 /* mime hex digit 1 */ 70241236Sbrooks#define S_MIME2 11 /* mime hex digit 2 */ 71241236Sbrooks#define S_EATCRNL 12 /* mime eating CRNL */ 72241236Sbrooks#define S_AMP 13 /* seen & */ 73241236Sbrooks#define S_NUMBER 14 /* collecting number */ 74241236Sbrooks#define S_STRING 15 /* collecting string */ 75241236Sbrooks 76241236Sbrooks#define isoctal(c) (((u_char)(c)) >= '0' && ((u_char)(c)) <= '7') 77241236Sbrooks#define xtod(c) (isdigit(c) ? (c - '0') : ((tolower(c) - 'a') + 10)) 78241236Sbrooks#define XTOD(c) (isdigit(c) ? (c - '0') : ((c - 'A') + 10)) 79241236Sbrooks 80241236Sbrooks/* 81241236Sbrooks * RFC 1866 82241236Sbrooks */ 83241236Sbrooksstatic const struct nv { 84241236Sbrooks const char *name; 85241236Sbrooks uint8_t value; 86241236Sbrooks} nv[] = { 87241236Sbrooks { "AElig", 198 }, /* capital AE diphthong (ligature) */ 88241236Sbrooks { "Aacute", 193 }, /* capital A, acute accent */ 89241236Sbrooks { "Acirc", 194 }, /* capital A, circumflex accent */ 90241236Sbrooks { "Agrave", 192 }, /* capital A, grave accent */ 91241236Sbrooks { "Aring", 197 }, /* capital A, ring */ 92241236Sbrooks { "Atilde", 195 }, /* capital A, tilde */ 93241236Sbrooks { "Auml", 196 }, /* capital A, dieresis or umlaut mark */ 94241236Sbrooks { "Ccedil", 199 }, /* capital C, cedilla */ 95241236Sbrooks { "ETH", 208 }, /* capital Eth, Icelandic */ 96241236Sbrooks { "Eacute", 201 }, /* capital E, acute accent */ 97241236Sbrooks { "Ecirc", 202 }, /* capital E, circumflex accent */ 98241236Sbrooks { "Egrave", 200 }, /* capital E, grave accent */ 99241236Sbrooks { "Euml", 203 }, /* capital E, dieresis or umlaut mark */ 100241236Sbrooks { "Iacute", 205 }, /* capital I, acute accent */ 101241236Sbrooks { "Icirc", 206 }, /* capital I, circumflex accent */ 102241236Sbrooks { "Igrave", 204 }, /* capital I, grave accent */ 103241236Sbrooks { "Iuml", 207 }, /* capital I, dieresis or umlaut mark */ 104241236Sbrooks { "Ntilde", 209 }, /* capital N, tilde */ 105241236Sbrooks { "Oacute", 211 }, /* capital O, acute accent */ 106241236Sbrooks { "Ocirc", 212 }, /* capital O, circumflex accent */ 107241236Sbrooks { "Ograve", 210 }, /* capital O, grave accent */ 108241236Sbrooks { "Oslash", 216 }, /* capital O, slash */ 109241236Sbrooks { "Otilde", 213 }, /* capital O, tilde */ 110241236Sbrooks { "Ouml", 214 }, /* capital O, dieresis or umlaut mark */ 111241236Sbrooks { "THORN", 222 }, /* capital THORN, Icelandic */ 112241236Sbrooks { "Uacute", 218 }, /* capital U, acute accent */ 113241236Sbrooks { "Ucirc", 219 }, /* capital U, circumflex accent */ 114241236Sbrooks { "Ugrave", 217 }, /* capital U, grave accent */ 115241236Sbrooks { "Uuml", 220 }, /* capital U, dieresis or umlaut mark */ 116241236Sbrooks { "Yacute", 221 }, /* capital Y, acute accent */ 117241236Sbrooks { "aacute", 225 }, /* small a, acute accent */ 118241236Sbrooks { "acirc", 226 }, /* small a, circumflex accent */ 119241236Sbrooks { "acute", 180 }, /* acute accent */ 120241236Sbrooks { "aelig", 230 }, /* small ae diphthong (ligature) */ 121241236Sbrooks { "agrave", 224 }, /* small a, grave accent */ 122241236Sbrooks { "amp", 38 }, /* ampersand */ 123241236Sbrooks { "aring", 229 }, /* small a, ring */ 124241236Sbrooks { "atilde", 227 }, /* small a, tilde */ 125241236Sbrooks { "auml", 228 }, /* small a, dieresis or umlaut mark */ 126241236Sbrooks { "brvbar", 166 }, /* broken (vertical) bar */ 127241236Sbrooks { "ccedil", 231 }, /* small c, cedilla */ 128241236Sbrooks { "cedil", 184 }, /* cedilla */ 129241236Sbrooks { "cent", 162 }, /* cent sign */ 130241236Sbrooks { "copy", 169 }, /* copyright sign */ 131241236Sbrooks { "curren", 164 }, /* general currency sign */ 132241236Sbrooks { "deg", 176 }, /* degree sign */ 133241236Sbrooks { "divide", 247 }, /* divide sign */ 134241236Sbrooks { "eacute", 233 }, /* small e, acute accent */ 135241236Sbrooks { "ecirc", 234 }, /* small e, circumflex accent */ 136241236Sbrooks { "egrave", 232 }, /* small e, grave accent */ 137241236Sbrooks { "eth", 240 }, /* small eth, Icelandic */ 138241236Sbrooks { "euml", 235 }, /* small e, dieresis or umlaut mark */ 139241236Sbrooks { "frac12", 189 }, /* fraction one-half */ 140241236Sbrooks { "frac14", 188 }, /* fraction one-quarter */ 141241236Sbrooks { "frac34", 190 }, /* fraction three-quarters */ 142241236Sbrooks { "gt", 62 }, /* greater than */ 143241236Sbrooks { "iacute", 237 }, /* small i, acute accent */ 144241236Sbrooks { "icirc", 238 }, /* small i, circumflex accent */ 145241236Sbrooks { "iexcl", 161 }, /* inverted exclamation mark */ 146241236Sbrooks { "igrave", 236 }, /* small i, grave accent */ 147241236Sbrooks { "iquest", 191 }, /* inverted question mark */ 148241236Sbrooks { "iuml", 239 }, /* small i, dieresis or umlaut mark */ 149241236Sbrooks { "laquo", 171 }, /* angle quotation mark, left */ 150241236Sbrooks { "lt", 60 }, /* less than */ 151241236Sbrooks { "macr", 175 }, /* macron */ 152241236Sbrooks { "micro", 181 }, /* micro sign */ 153241236Sbrooks { "middot", 183 }, /* middle dot */ 154241236Sbrooks { "nbsp", 160 }, /* no-break space */ 155241236Sbrooks { "not", 172 }, /* not sign */ 156241236Sbrooks { "ntilde", 241 }, /* small n, tilde */ 157241236Sbrooks { "oacute", 243 }, /* small o, acute accent */ 158241236Sbrooks { "ocirc", 244 }, /* small o, circumflex accent */ 159241236Sbrooks { "ograve", 242 }, /* small o, grave accent */ 160241236Sbrooks { "ordf", 170 }, /* ordinal indicator, feminine */ 161241236Sbrooks { "ordm", 186 }, /* ordinal indicator, masculine */ 162241236Sbrooks { "oslash", 248 }, /* small o, slash */ 163241236Sbrooks { "otilde", 245 }, /* small o, tilde */ 164241236Sbrooks { "ouml", 246 }, /* small o, dieresis or umlaut mark */ 165241236Sbrooks { "para", 182 }, /* pilcrow (paragraph sign) */ 166241236Sbrooks { "plusmn", 177 }, /* plus-or-minus sign */ 167241236Sbrooks { "pound", 163 }, /* pound sterling sign */ 168241236Sbrooks { "quot", 34 }, /* double quote */ 169241236Sbrooks { "raquo", 187 }, /* angle quotation mark, right */ 170241236Sbrooks { "reg", 174 }, /* registered sign */ 171241236Sbrooks { "sect", 167 }, /* section sign */ 172241236Sbrooks { "shy", 173 }, /* soft hyphen */ 173241236Sbrooks { "sup1", 185 }, /* superscript one */ 174241236Sbrooks { "sup2", 178 }, /* superscript two */ 175241236Sbrooks { "sup3", 179 }, /* superscript three */ 176241236Sbrooks { "szlig", 223 }, /* small sharp s, German (sz ligature) */ 177241236Sbrooks { "thorn", 254 }, /* small thorn, Icelandic */ 178241236Sbrooks { "times", 215 }, /* multiply sign */ 179241236Sbrooks { "uacute", 250 }, /* small u, acute accent */ 180241236Sbrooks { "ucirc", 251 }, /* small u, circumflex accent */ 181241236Sbrooks { "ugrave", 249 }, /* small u, grave accent */ 182241236Sbrooks { "uml", 168 }, /* umlaut (dieresis) */ 183241236Sbrooks { "uuml", 252 }, /* small u, dieresis or umlaut mark */ 184241236Sbrooks { "yacute", 253 }, /* small y, acute accent */ 185241236Sbrooks { "yen", 165 }, /* yen sign */ 186241236Sbrooks { "yuml", 255 }, /* small y, dieresis or umlaut mark */ 187241236Sbrooks}; 188241236Sbrooks 189241236Sbrooks/* 190241236Sbrooks * unvis - decode characters previously encoded by vis 191241236Sbrooks */ 192241236Sbrooksint 193241236Sbrooksunvis(char *cp, int c, int *astate, int flag) 194241236Sbrooks{ 195241236Sbrooks unsigned char uc = (unsigned char)c; 196241236Sbrooks unsigned char st, ia, is, lc; 197241236Sbrooks 198241236Sbrooks/* 199241236Sbrooks * Bottom 8 bits of astate hold the state machine state. 200241236Sbrooks * Top 8 bits hold the current character in the http 1866 nv string decoding 201241236Sbrooks */ 202241236Sbrooks#define GS(a) ((a) & 0xff) 203241236Sbrooks#define SS(a, b) (((uint32_t)(a) << 24) | (b)) 204241236Sbrooks#define GI(a) ((uint32_t)(a) >> 24) 205241236Sbrooks 206241236Sbrooks _DIAGASSERT(cp != NULL); 207241236Sbrooks _DIAGASSERT(astate != NULL); 208241236Sbrooks st = GS(*astate); 209241236Sbrooks 210241236Sbrooks if (flag & UNVIS_END) { 211241236Sbrooks switch (st) { 212241236Sbrooks case S_OCTAL2: 213241236Sbrooks case S_OCTAL3: 214241236Sbrooks case S_HEX2: 215241236Sbrooks *astate = SS(0, S_GROUND); 216241236Sbrooks return UNVIS_VALID; 217241236Sbrooks case S_GROUND: 218241236Sbrooks return UNVIS_NOCHAR; 219241236Sbrooks default: 220241236Sbrooks return UNVIS_SYNBAD; 221241236Sbrooks } 222241236Sbrooks } 223241236Sbrooks 224241236Sbrooks switch (st) { 225241236Sbrooks 226241236Sbrooks case S_GROUND: 227241236Sbrooks *cp = 0; 228241236Sbrooks if ((flag & VIS_NOESCAPE) == 0 && c == '\\') { 229241236Sbrooks *astate = SS(0, S_START); 230241236Sbrooks return UNVIS_NOCHAR; 231241236Sbrooks } 232241236Sbrooks if ((flag & VIS_HTTP1808) && c == '%') { 233241236Sbrooks *astate = SS(0, S_HEX1); 234241236Sbrooks return UNVIS_NOCHAR; 235241236Sbrooks } 236241236Sbrooks if ((flag & VIS_HTTP1866) && c == '&') { 237241236Sbrooks *astate = SS(0, S_AMP); 238241236Sbrooks return UNVIS_NOCHAR; 239241236Sbrooks } 240241236Sbrooks if ((flag & VIS_MIMESTYLE) && c == '=') { 241241236Sbrooks *astate = SS(0, S_MIME1); 242241236Sbrooks return UNVIS_NOCHAR; 243241236Sbrooks } 244241236Sbrooks *cp = c; 245241236Sbrooks return UNVIS_VALID; 246241236Sbrooks 247241236Sbrooks case S_START: 248241236Sbrooks switch(c) { 249241236Sbrooks case '\\': 250241236Sbrooks *cp = c; 251241236Sbrooks *astate = SS(0, S_GROUND); 252241236Sbrooks return UNVIS_VALID; 253241236Sbrooks case '0': case '1': case '2': case '3': 254241236Sbrooks case '4': case '5': case '6': case '7': 255241236Sbrooks *cp = (c - '0'); 256241236Sbrooks *astate = SS(0, S_OCTAL2); 257241236Sbrooks return UNVIS_NOCHAR; 258241236Sbrooks case 'M': 259241236Sbrooks *cp = (char)0200; 260241236Sbrooks *astate = SS(0, S_META); 261241236Sbrooks return UNVIS_NOCHAR; 262241236Sbrooks case '^': 263241236Sbrooks *astate = SS(0, S_CTRL); 264241236Sbrooks return UNVIS_NOCHAR; 265241236Sbrooks case 'n': 266241236Sbrooks *cp = '\n'; 267241236Sbrooks *astate = SS(0, S_GROUND); 268241236Sbrooks return UNVIS_VALID; 269241236Sbrooks case 'r': 270241236Sbrooks *cp = '\r'; 271241236Sbrooks *astate = SS(0, S_GROUND); 272241236Sbrooks return UNVIS_VALID; 273241236Sbrooks case 'b': 274241236Sbrooks *cp = '\b'; 275241236Sbrooks *astate = SS(0, S_GROUND); 276241236Sbrooks return UNVIS_VALID; 277241236Sbrooks case 'a': 278241236Sbrooks *cp = '\007'; 279241236Sbrooks *astate = SS(0, S_GROUND); 280241236Sbrooks return UNVIS_VALID; 281241236Sbrooks case 'v': 282241236Sbrooks *cp = '\v'; 283241236Sbrooks *astate = SS(0, S_GROUND); 284241236Sbrooks return UNVIS_VALID; 285241236Sbrooks case 't': 286241236Sbrooks *cp = '\t'; 287241236Sbrooks *astate = SS(0, S_GROUND); 288241236Sbrooks return UNVIS_VALID; 289241236Sbrooks case 'f': 290241236Sbrooks *cp = '\f'; 291241236Sbrooks *astate = SS(0, S_GROUND); 292241236Sbrooks return UNVIS_VALID; 293241236Sbrooks case 's': 294241236Sbrooks *cp = ' '; 295241236Sbrooks *astate = SS(0, S_GROUND); 296241236Sbrooks return UNVIS_VALID; 297241236Sbrooks case 'E': 298241236Sbrooks *cp = '\033'; 299241236Sbrooks *astate = SS(0, S_GROUND); 300241236Sbrooks return UNVIS_VALID; 301241236Sbrooks case 'x': 302241236Sbrooks *astate = SS(0, S_HEX); 303241236Sbrooks return UNVIS_NOCHAR; 304241236Sbrooks case '\n': 305241236Sbrooks /* 306241236Sbrooks * hidden newline 307241236Sbrooks */ 308241236Sbrooks *astate = SS(0, S_GROUND); 309241236Sbrooks return UNVIS_NOCHAR; 310241236Sbrooks case '$': 311241236Sbrooks /* 312241236Sbrooks * hidden marker 313241236Sbrooks */ 314241236Sbrooks *astate = SS(0, S_GROUND); 315241236Sbrooks return UNVIS_NOCHAR; 316241236Sbrooks } 317241236Sbrooks goto bad; 318241236Sbrooks 319241236Sbrooks case S_META: 320241236Sbrooks if (c == '-') 321241236Sbrooks *astate = SS(0, S_META1); 322241236Sbrooks else if (c == '^') 323241236Sbrooks *astate = SS(0, S_CTRL); 324241236Sbrooks else 325241236Sbrooks goto bad; 326241236Sbrooks return UNVIS_NOCHAR; 327241236Sbrooks 328241236Sbrooks case S_META1: 329241236Sbrooks *astate = SS(0, S_GROUND); 330241236Sbrooks *cp |= c; 331241236Sbrooks return UNVIS_VALID; 332241236Sbrooks 333241236Sbrooks case S_CTRL: 334241236Sbrooks if (c == '?') 335241236Sbrooks *cp |= 0177; 336241236Sbrooks else 337241236Sbrooks *cp |= c & 037; 338241236Sbrooks *astate = SS(0, S_GROUND); 339241236Sbrooks return UNVIS_VALID; 340241236Sbrooks 341241236Sbrooks case S_OCTAL2: /* second possible octal digit */ 342241236Sbrooks if (isoctal(uc)) { 343241236Sbrooks /* 344241236Sbrooks * yes - and maybe a third 345241236Sbrooks */ 346241236Sbrooks *cp = (*cp << 3) + (c - '0'); 347241236Sbrooks *astate = SS(0, S_OCTAL3); 348241236Sbrooks return UNVIS_NOCHAR; 349241236Sbrooks } 350241236Sbrooks /* 351241236Sbrooks * no - done with current sequence, push back passed char 352241236Sbrooks */ 353241236Sbrooks *astate = SS(0, S_GROUND); 354241236Sbrooks return UNVIS_VALIDPUSH; 355241236Sbrooks 356241236Sbrooks case S_OCTAL3: /* third possible octal digit */ 357241236Sbrooks *astate = SS(0, S_GROUND); 358241236Sbrooks if (isoctal(uc)) { 359241236Sbrooks *cp = (*cp << 3) + (c - '0'); 360241236Sbrooks return UNVIS_VALID; 361241236Sbrooks } 362241236Sbrooks /* 363241236Sbrooks * we were done, push back passed char 364241236Sbrooks */ 365241236Sbrooks return UNVIS_VALIDPUSH; 366241236Sbrooks 367241236Sbrooks case S_HEX: 368241236Sbrooks if (!isxdigit(uc)) 369241236Sbrooks goto bad; 370241236Sbrooks /*FALLTHROUGH*/ 371241236Sbrooks case S_HEX1: 372241236Sbrooks if (isxdigit(uc)) { 373241236Sbrooks *cp = xtod(uc); 374241236Sbrooks *astate = SS(0, S_HEX2); 375241236Sbrooks return UNVIS_NOCHAR; 376241236Sbrooks } 377241236Sbrooks /* 378241236Sbrooks * no - done with current sequence, push back passed char 379241236Sbrooks */ 380241236Sbrooks *astate = SS(0, S_GROUND); 381241236Sbrooks return UNVIS_VALIDPUSH; 382241236Sbrooks 383241236Sbrooks case S_HEX2: 384241236Sbrooks *astate = S_GROUND; 385241236Sbrooks if (isxdigit(uc)) { 386241236Sbrooks *cp = xtod(uc) | (*cp << 4); 387241236Sbrooks return UNVIS_VALID; 388241236Sbrooks } 389241236Sbrooks return UNVIS_VALIDPUSH; 390241236Sbrooks 391241236Sbrooks case S_MIME1: 392241236Sbrooks if (uc == '\n' || uc == '\r') { 393241236Sbrooks *astate = SS(0, S_EATCRNL); 394241236Sbrooks return UNVIS_NOCHAR; 395241236Sbrooks } 396241236Sbrooks if (isxdigit(uc) && (isdigit(uc) || isupper(uc))) { 397241236Sbrooks *cp = XTOD(uc); 398241236Sbrooks *astate = SS(0, S_MIME2); 399241236Sbrooks return UNVIS_NOCHAR; 400241236Sbrooks } 401241236Sbrooks goto bad; 402241236Sbrooks 403241236Sbrooks case S_MIME2: 404241236Sbrooks if (isxdigit(uc) && (isdigit(uc) || isupper(uc))) { 405241236Sbrooks *astate = SS(0, S_GROUND); 406241236Sbrooks *cp = XTOD(uc) | (*cp << 4); 407241236Sbrooks return UNVIS_VALID; 408241236Sbrooks } 409241236Sbrooks goto bad; 410241236Sbrooks 411241236Sbrooks case S_EATCRNL: 412241236Sbrooks switch (uc) { 413241236Sbrooks case '\r': 414241236Sbrooks case '\n': 415241236Sbrooks return UNVIS_NOCHAR; 416241236Sbrooks case '=': 417241236Sbrooks *astate = SS(0, S_MIME1); 418241236Sbrooks return UNVIS_NOCHAR; 419241236Sbrooks default: 420241236Sbrooks *cp = uc; 421241236Sbrooks *astate = SS(0, S_GROUND); 422241236Sbrooks return UNVIS_VALID; 423241236Sbrooks } 424241236Sbrooks 425241236Sbrooks case S_AMP: 426241236Sbrooks *cp = 0; 427241236Sbrooks if (uc == '#') { 428241236Sbrooks *astate = SS(0, S_NUMBER); 429241236Sbrooks return UNVIS_NOCHAR; 430241236Sbrooks } 431241236Sbrooks *astate = SS(0, S_STRING); 432241236Sbrooks /*FALLTHROUGH*/ 433241236Sbrooks 434241236Sbrooks case S_STRING: 435241236Sbrooks ia = *cp; /* index in the array */ 436241236Sbrooks is = GI(*astate); /* index in the string */ 437241236Sbrooks lc = is == 0 ? 0 : nv[ia].name[is - 1]; /* last character */ 438241236Sbrooks 439241236Sbrooks if (uc == ';') 440241236Sbrooks uc = '\0'; 441241236Sbrooks 442241236Sbrooks for (; ia < __arraycount(nv); ia++) { 443241236Sbrooks if (is != 0 && nv[ia].name[is - 1] != lc) 444241236Sbrooks goto bad; 445241236Sbrooks if (nv[ia].name[is] == uc) 446241236Sbrooks break; 447241236Sbrooks } 448241236Sbrooks 449241236Sbrooks if (ia == __arraycount(nv)) 450241236Sbrooks goto bad; 451241236Sbrooks 452241236Sbrooks if (uc != 0) { 453241236Sbrooks *cp = ia; 454241236Sbrooks *astate = SS(is + 1, S_STRING); 455241236Sbrooks return UNVIS_NOCHAR; 456241236Sbrooks } 457241236Sbrooks 458241236Sbrooks *cp = nv[ia].value; 459241236Sbrooks *astate = SS(0, S_GROUND); 460241236Sbrooks return UNVIS_VALID; 461241236Sbrooks 462241236Sbrooks case S_NUMBER: 463241236Sbrooks if (uc == ';') 464241236Sbrooks return UNVIS_VALID; 465241236Sbrooks if (!isdigit(uc)) 466241236Sbrooks goto bad; 467241236Sbrooks *cp += (*cp * 10) + uc - '0'; 468241236Sbrooks return UNVIS_NOCHAR; 469241236Sbrooks 470241236Sbrooks default: 471241236Sbrooks bad: 472241236Sbrooks /* 473241236Sbrooks * decoder in unknown state - (probably uninitialized) 474241236Sbrooks */ 475241236Sbrooks *astate = SS(0, S_GROUND); 476241236Sbrooks return UNVIS_SYNBAD; 477241236Sbrooks } 478241236Sbrooks} 479241236Sbrooks 480241236Sbrooks/* 481241236Sbrooks * strnunvisx - decode src into dst 482241236Sbrooks * 483241236Sbrooks * Number of chars decoded into dst is returned, -1 on error. 484241236Sbrooks * Dst is null terminated. 485241236Sbrooks */ 486241236Sbrooks 487241236Sbrooksint 488241236Sbrooksstrnunvisx(char *dst, size_t dlen, const char *src, int flag) 489241236Sbrooks{ 490241236Sbrooks char c; 491241236Sbrooks char t = '\0', *start = dst; 492241236Sbrooks int state = 0; 493241236Sbrooks 494241236Sbrooks _DIAGASSERT(src != NULL); 495241236Sbrooks _DIAGASSERT(dst != NULL); 496241236Sbrooks#define CHECKSPACE() \ 497241236Sbrooks do { \ 498241236Sbrooks if (dlen-- == 0) { \ 499241236Sbrooks errno = ENOSPC; \ 500241236Sbrooks return -1; \ 501241236Sbrooks } \ 502241236Sbrooks } while (/*CONSTCOND*/0) 503241236Sbrooks 504241236Sbrooks while ((c = *src++) != '\0') { 505241236Sbrooks again: 506241236Sbrooks switch (unvis(&t, c, &state, flag)) { 507241236Sbrooks case UNVIS_VALID: 508241236Sbrooks CHECKSPACE(); 509241236Sbrooks *dst++ = t; 510241236Sbrooks break; 511241236Sbrooks case UNVIS_VALIDPUSH: 512241236Sbrooks CHECKSPACE(); 513241236Sbrooks *dst++ = t; 514241236Sbrooks goto again; 515241236Sbrooks case 0: 516241236Sbrooks case UNVIS_NOCHAR: 517241236Sbrooks break; 518241236Sbrooks case UNVIS_SYNBAD: 519241236Sbrooks errno = EINVAL; 520241236Sbrooks return -1; 521241236Sbrooks default: 522241236Sbrooks _DIAGASSERT(/*CONSTCOND*/0); 523241236Sbrooks errno = EINVAL; 524241236Sbrooks return -1; 525241236Sbrooks } 526241236Sbrooks } 527241236Sbrooks if (unvis(&t, c, &state, UNVIS_END) == UNVIS_VALID) { 528241236Sbrooks CHECKSPACE(); 529241236Sbrooks *dst++ = t; 530241236Sbrooks } 531241236Sbrooks CHECKSPACE(); 532241236Sbrooks *dst = '\0'; 533241236Sbrooks return (int)(dst - start); 534241236Sbrooks} 535241236Sbrooks 536241236Sbrooksint 537241236Sbrooksstrunvisx(char *dst, const char *src, int flag) 538241236Sbrooks{ 539241236Sbrooks return strnunvisx(dst, (size_t)~0, src, flag); 540241236Sbrooks} 541241236Sbrooks 542241236Sbrooksint 543241236Sbrooksstrunvis(char *dst, const char *src) 544241236Sbrooks{ 545241236Sbrooks return strnunvisx(dst, (size_t)~0, src, 0); 546241236Sbrooks} 547241236Sbrooks 548241236Sbrooksint 549241236Sbrooksstrnunvis(char *dst, size_t dlen, const char *src) 550241236Sbrooks{ 551241236Sbrooks return strnunvisx(dst, dlen, src, 0); 552241236Sbrooks} 553241236Sbrooks#endif 554