1248302Sbrooks/* $NetBSD: unvis.c,v 1.41 2012/12/15 04:29:53 matt 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 37248302Sbrooks__RCSID("$NetBSD: unvis.c,v 1.41 2012/12/15 04:29:53 matt Exp $"); 38241236Sbrooks#endif 39241236Sbrooks#endif /* LIBC_SCCS and not lint */ 40244401Sbrooks__FBSDID("$FreeBSD: releng/10.2/contrib/libc-vis/unvis.c 248302 2013-03-14 23:51:47Z brooks $"); 41241236Sbrooks 42241236Sbrooks#include "namespace.h" 43241236Sbrooks#include <sys/types.h> 44241236Sbrooks 45241236Sbrooks#include <assert.h> 46241236Sbrooks#include <ctype.h> 47241236Sbrooks#include <stdint.h> 48241236Sbrooks#include <stdio.h> 49241236Sbrooks#include <errno.h> 50241236Sbrooks#include <vis.h> 51241236Sbrooks 52244401Sbrooks#define _DIAGASSERT(x) assert(x) 53244401Sbrooks 54244401Sbrooks/* 55244401Sbrooks * Return the number of elements in a statically-allocated array, 56244401Sbrooks * __x. 57244401Sbrooks */ 58244401Sbrooks#define __arraycount(__x) (sizeof(__x) / sizeof(__x[0])) 59244401Sbrooks 60241236Sbrooks#ifdef __weak_alias 61241236Sbrooks__weak_alias(strnunvisx,_strnunvisx) 62241236Sbrooks#endif 63241236Sbrooks 64241236Sbrooks#if !HAVE_VIS 65241236Sbrooks/* 66241236Sbrooks * decode driven by state machine 67241236Sbrooks */ 68241236Sbrooks#define S_GROUND 0 /* haven't seen escape char */ 69241236Sbrooks#define S_START 1 /* start decoding special sequence */ 70241236Sbrooks#define S_META 2 /* metachar started (M) */ 71241236Sbrooks#define S_META1 3 /* metachar more, regular char (-) */ 72241236Sbrooks#define S_CTRL 4 /* control char started (^) */ 73241236Sbrooks#define S_OCTAL2 5 /* octal digit 2 */ 74241236Sbrooks#define S_OCTAL3 6 /* octal digit 3 */ 75241236Sbrooks#define S_HEX 7 /* mandatory hex digit */ 76241236Sbrooks#define S_HEX1 8 /* http hex digit */ 77241236Sbrooks#define S_HEX2 9 /* http hex digit 2 */ 78241236Sbrooks#define S_MIME1 10 /* mime hex digit 1 */ 79241236Sbrooks#define S_MIME2 11 /* mime hex digit 2 */ 80241236Sbrooks#define S_EATCRNL 12 /* mime eating CRNL */ 81241236Sbrooks#define S_AMP 13 /* seen & */ 82241236Sbrooks#define S_NUMBER 14 /* collecting number */ 83241236Sbrooks#define S_STRING 15 /* collecting string */ 84241236Sbrooks 85241236Sbrooks#define isoctal(c) (((u_char)(c)) >= '0' && ((u_char)(c)) <= '7') 86241236Sbrooks#define xtod(c) (isdigit(c) ? (c - '0') : ((tolower(c) - 'a') + 10)) 87241236Sbrooks#define XTOD(c) (isdigit(c) ? (c - '0') : ((c - 'A') + 10)) 88241236Sbrooks 89241236Sbrooks/* 90241236Sbrooks * RFC 1866 91241236Sbrooks */ 92241236Sbrooksstatic const struct nv { 93248302Sbrooks char name[7]; 94241236Sbrooks uint8_t value; 95241236Sbrooks} nv[] = { 96241236Sbrooks { "AElig", 198 }, /* capital AE diphthong (ligature) */ 97241236Sbrooks { "Aacute", 193 }, /* capital A, acute accent */ 98241236Sbrooks { "Acirc", 194 }, /* capital A, circumflex accent */ 99241236Sbrooks { "Agrave", 192 }, /* capital A, grave accent */ 100241236Sbrooks { "Aring", 197 }, /* capital A, ring */ 101241236Sbrooks { "Atilde", 195 }, /* capital A, tilde */ 102241236Sbrooks { "Auml", 196 }, /* capital A, dieresis or umlaut mark */ 103241236Sbrooks { "Ccedil", 199 }, /* capital C, cedilla */ 104241236Sbrooks { "ETH", 208 }, /* capital Eth, Icelandic */ 105241236Sbrooks { "Eacute", 201 }, /* capital E, acute accent */ 106241236Sbrooks { "Ecirc", 202 }, /* capital E, circumflex accent */ 107241236Sbrooks { "Egrave", 200 }, /* capital E, grave accent */ 108241236Sbrooks { "Euml", 203 }, /* capital E, dieresis or umlaut mark */ 109241236Sbrooks { "Iacute", 205 }, /* capital I, acute accent */ 110241236Sbrooks { "Icirc", 206 }, /* capital I, circumflex accent */ 111241236Sbrooks { "Igrave", 204 }, /* capital I, grave accent */ 112241236Sbrooks { "Iuml", 207 }, /* capital I, dieresis or umlaut mark */ 113241236Sbrooks { "Ntilde", 209 }, /* capital N, tilde */ 114241236Sbrooks { "Oacute", 211 }, /* capital O, acute accent */ 115241236Sbrooks { "Ocirc", 212 }, /* capital O, circumflex accent */ 116241236Sbrooks { "Ograve", 210 }, /* capital O, grave accent */ 117241236Sbrooks { "Oslash", 216 }, /* capital O, slash */ 118241236Sbrooks { "Otilde", 213 }, /* capital O, tilde */ 119241236Sbrooks { "Ouml", 214 }, /* capital O, dieresis or umlaut mark */ 120241236Sbrooks { "THORN", 222 }, /* capital THORN, Icelandic */ 121241236Sbrooks { "Uacute", 218 }, /* capital U, acute accent */ 122241236Sbrooks { "Ucirc", 219 }, /* capital U, circumflex accent */ 123241236Sbrooks { "Ugrave", 217 }, /* capital U, grave accent */ 124241236Sbrooks { "Uuml", 220 }, /* capital U, dieresis or umlaut mark */ 125241236Sbrooks { "Yacute", 221 }, /* capital Y, acute accent */ 126241236Sbrooks { "aacute", 225 }, /* small a, acute accent */ 127241236Sbrooks { "acirc", 226 }, /* small a, circumflex accent */ 128241236Sbrooks { "acute", 180 }, /* acute accent */ 129241236Sbrooks { "aelig", 230 }, /* small ae diphthong (ligature) */ 130241236Sbrooks { "agrave", 224 }, /* small a, grave accent */ 131241236Sbrooks { "amp", 38 }, /* ampersand */ 132241236Sbrooks { "aring", 229 }, /* small a, ring */ 133241236Sbrooks { "atilde", 227 }, /* small a, tilde */ 134241236Sbrooks { "auml", 228 }, /* small a, dieresis or umlaut mark */ 135241236Sbrooks { "brvbar", 166 }, /* broken (vertical) bar */ 136241236Sbrooks { "ccedil", 231 }, /* small c, cedilla */ 137241236Sbrooks { "cedil", 184 }, /* cedilla */ 138241236Sbrooks { "cent", 162 }, /* cent sign */ 139241236Sbrooks { "copy", 169 }, /* copyright sign */ 140241236Sbrooks { "curren", 164 }, /* general currency sign */ 141241236Sbrooks { "deg", 176 }, /* degree sign */ 142241236Sbrooks { "divide", 247 }, /* divide sign */ 143241236Sbrooks { "eacute", 233 }, /* small e, acute accent */ 144241236Sbrooks { "ecirc", 234 }, /* small e, circumflex accent */ 145241236Sbrooks { "egrave", 232 }, /* small e, grave accent */ 146241236Sbrooks { "eth", 240 }, /* small eth, Icelandic */ 147241236Sbrooks { "euml", 235 }, /* small e, dieresis or umlaut mark */ 148241236Sbrooks { "frac12", 189 }, /* fraction one-half */ 149241236Sbrooks { "frac14", 188 }, /* fraction one-quarter */ 150241236Sbrooks { "frac34", 190 }, /* fraction three-quarters */ 151241236Sbrooks { "gt", 62 }, /* greater than */ 152241236Sbrooks { "iacute", 237 }, /* small i, acute accent */ 153241236Sbrooks { "icirc", 238 }, /* small i, circumflex accent */ 154241236Sbrooks { "iexcl", 161 }, /* inverted exclamation mark */ 155241236Sbrooks { "igrave", 236 }, /* small i, grave accent */ 156241236Sbrooks { "iquest", 191 }, /* inverted question mark */ 157241236Sbrooks { "iuml", 239 }, /* small i, dieresis or umlaut mark */ 158241236Sbrooks { "laquo", 171 }, /* angle quotation mark, left */ 159241236Sbrooks { "lt", 60 }, /* less than */ 160241236Sbrooks { "macr", 175 }, /* macron */ 161241236Sbrooks { "micro", 181 }, /* micro sign */ 162241236Sbrooks { "middot", 183 }, /* middle dot */ 163241236Sbrooks { "nbsp", 160 }, /* no-break space */ 164241236Sbrooks { "not", 172 }, /* not sign */ 165241236Sbrooks { "ntilde", 241 }, /* small n, tilde */ 166241236Sbrooks { "oacute", 243 }, /* small o, acute accent */ 167241236Sbrooks { "ocirc", 244 }, /* small o, circumflex accent */ 168241236Sbrooks { "ograve", 242 }, /* small o, grave accent */ 169241236Sbrooks { "ordf", 170 }, /* ordinal indicator, feminine */ 170241236Sbrooks { "ordm", 186 }, /* ordinal indicator, masculine */ 171241236Sbrooks { "oslash", 248 }, /* small o, slash */ 172241236Sbrooks { "otilde", 245 }, /* small o, tilde */ 173241236Sbrooks { "ouml", 246 }, /* small o, dieresis or umlaut mark */ 174241236Sbrooks { "para", 182 }, /* pilcrow (paragraph sign) */ 175241236Sbrooks { "plusmn", 177 }, /* plus-or-minus sign */ 176241236Sbrooks { "pound", 163 }, /* pound sterling sign */ 177241236Sbrooks { "quot", 34 }, /* double quote */ 178241236Sbrooks { "raquo", 187 }, /* angle quotation mark, right */ 179241236Sbrooks { "reg", 174 }, /* registered sign */ 180241236Sbrooks { "sect", 167 }, /* section sign */ 181241236Sbrooks { "shy", 173 }, /* soft hyphen */ 182241236Sbrooks { "sup1", 185 }, /* superscript one */ 183241236Sbrooks { "sup2", 178 }, /* superscript two */ 184241236Sbrooks { "sup3", 179 }, /* superscript three */ 185241236Sbrooks { "szlig", 223 }, /* small sharp s, German (sz ligature) */ 186241236Sbrooks { "thorn", 254 }, /* small thorn, Icelandic */ 187241236Sbrooks { "times", 215 }, /* multiply sign */ 188241236Sbrooks { "uacute", 250 }, /* small u, acute accent */ 189241236Sbrooks { "ucirc", 251 }, /* small u, circumflex accent */ 190241236Sbrooks { "ugrave", 249 }, /* small u, grave accent */ 191241236Sbrooks { "uml", 168 }, /* umlaut (dieresis) */ 192241236Sbrooks { "uuml", 252 }, /* small u, dieresis or umlaut mark */ 193241236Sbrooks { "yacute", 253 }, /* small y, acute accent */ 194241236Sbrooks { "yen", 165 }, /* yen sign */ 195241236Sbrooks { "yuml", 255 }, /* small y, dieresis or umlaut mark */ 196241236Sbrooks}; 197241236Sbrooks 198241236Sbrooks/* 199241236Sbrooks * unvis - decode characters previously encoded by vis 200241236Sbrooks */ 201241236Sbrooksint 202241236Sbrooksunvis(char *cp, int c, int *astate, int flag) 203241236Sbrooks{ 204241236Sbrooks unsigned char uc = (unsigned char)c; 205241236Sbrooks unsigned char st, ia, is, lc; 206241236Sbrooks 207241236Sbrooks/* 208241236Sbrooks * Bottom 8 bits of astate hold the state machine state. 209241236Sbrooks * Top 8 bits hold the current character in the http 1866 nv string decoding 210241236Sbrooks */ 211241236Sbrooks#define GS(a) ((a) & 0xff) 212241236Sbrooks#define SS(a, b) (((uint32_t)(a) << 24) | (b)) 213241236Sbrooks#define GI(a) ((uint32_t)(a) >> 24) 214241236Sbrooks 215241236Sbrooks _DIAGASSERT(cp != NULL); 216241236Sbrooks _DIAGASSERT(astate != NULL); 217241236Sbrooks st = GS(*astate); 218241236Sbrooks 219241236Sbrooks if (flag & UNVIS_END) { 220241236Sbrooks switch (st) { 221241236Sbrooks case S_OCTAL2: 222241236Sbrooks case S_OCTAL3: 223241236Sbrooks case S_HEX2: 224241236Sbrooks *astate = SS(0, S_GROUND); 225241236Sbrooks return UNVIS_VALID; 226241236Sbrooks case S_GROUND: 227241236Sbrooks return UNVIS_NOCHAR; 228241236Sbrooks default: 229241236Sbrooks return UNVIS_SYNBAD; 230241236Sbrooks } 231241236Sbrooks } 232241236Sbrooks 233241236Sbrooks switch (st) { 234241236Sbrooks 235241236Sbrooks case S_GROUND: 236241236Sbrooks *cp = 0; 237241236Sbrooks if ((flag & VIS_NOESCAPE) == 0 && c == '\\') { 238241236Sbrooks *astate = SS(0, S_START); 239241236Sbrooks return UNVIS_NOCHAR; 240241236Sbrooks } 241241236Sbrooks if ((flag & VIS_HTTP1808) && c == '%') { 242241236Sbrooks *astate = SS(0, S_HEX1); 243241236Sbrooks return UNVIS_NOCHAR; 244241236Sbrooks } 245241236Sbrooks if ((flag & VIS_HTTP1866) && c == '&') { 246241236Sbrooks *astate = SS(0, S_AMP); 247241236Sbrooks return UNVIS_NOCHAR; 248241236Sbrooks } 249241236Sbrooks if ((flag & VIS_MIMESTYLE) && c == '=') { 250241236Sbrooks *astate = SS(0, S_MIME1); 251241236Sbrooks return UNVIS_NOCHAR; 252241236Sbrooks } 253241236Sbrooks *cp = c; 254241236Sbrooks return UNVIS_VALID; 255241236Sbrooks 256241236Sbrooks case S_START: 257241236Sbrooks switch(c) { 258241236Sbrooks case '\\': 259241236Sbrooks *cp = c; 260241236Sbrooks *astate = SS(0, S_GROUND); 261241236Sbrooks return UNVIS_VALID; 262241236Sbrooks case '0': case '1': case '2': case '3': 263241236Sbrooks case '4': case '5': case '6': case '7': 264241236Sbrooks *cp = (c - '0'); 265241236Sbrooks *astate = SS(0, S_OCTAL2); 266241236Sbrooks return UNVIS_NOCHAR; 267241236Sbrooks case 'M': 268241236Sbrooks *cp = (char)0200; 269241236Sbrooks *astate = SS(0, S_META); 270241236Sbrooks return UNVIS_NOCHAR; 271241236Sbrooks case '^': 272241236Sbrooks *astate = SS(0, S_CTRL); 273241236Sbrooks return UNVIS_NOCHAR; 274241236Sbrooks case 'n': 275241236Sbrooks *cp = '\n'; 276241236Sbrooks *astate = SS(0, S_GROUND); 277241236Sbrooks return UNVIS_VALID; 278241236Sbrooks case 'r': 279241236Sbrooks *cp = '\r'; 280241236Sbrooks *astate = SS(0, S_GROUND); 281241236Sbrooks return UNVIS_VALID; 282241236Sbrooks case 'b': 283241236Sbrooks *cp = '\b'; 284241236Sbrooks *astate = SS(0, S_GROUND); 285241236Sbrooks return UNVIS_VALID; 286241236Sbrooks case 'a': 287241236Sbrooks *cp = '\007'; 288241236Sbrooks *astate = SS(0, S_GROUND); 289241236Sbrooks return UNVIS_VALID; 290241236Sbrooks case 'v': 291241236Sbrooks *cp = '\v'; 292241236Sbrooks *astate = SS(0, S_GROUND); 293241236Sbrooks return UNVIS_VALID; 294241236Sbrooks case 't': 295241236Sbrooks *cp = '\t'; 296241236Sbrooks *astate = SS(0, S_GROUND); 297241236Sbrooks return UNVIS_VALID; 298241236Sbrooks case 'f': 299241236Sbrooks *cp = '\f'; 300241236Sbrooks *astate = SS(0, S_GROUND); 301241236Sbrooks return UNVIS_VALID; 302241236Sbrooks case 's': 303241236Sbrooks *cp = ' '; 304241236Sbrooks *astate = SS(0, S_GROUND); 305241236Sbrooks return UNVIS_VALID; 306241236Sbrooks case 'E': 307241236Sbrooks *cp = '\033'; 308241236Sbrooks *astate = SS(0, S_GROUND); 309241236Sbrooks return UNVIS_VALID; 310241236Sbrooks case 'x': 311241236Sbrooks *astate = SS(0, S_HEX); 312241236Sbrooks return UNVIS_NOCHAR; 313241236Sbrooks case '\n': 314241236Sbrooks /* 315241236Sbrooks * hidden newline 316241236Sbrooks */ 317241236Sbrooks *astate = SS(0, S_GROUND); 318241236Sbrooks return UNVIS_NOCHAR; 319241236Sbrooks case '$': 320241236Sbrooks /* 321241236Sbrooks * hidden marker 322241236Sbrooks */ 323241236Sbrooks *astate = SS(0, S_GROUND); 324241236Sbrooks return UNVIS_NOCHAR; 325241236Sbrooks } 326241236Sbrooks goto bad; 327241236Sbrooks 328241236Sbrooks case S_META: 329241236Sbrooks if (c == '-') 330241236Sbrooks *astate = SS(0, S_META1); 331241236Sbrooks else if (c == '^') 332241236Sbrooks *astate = SS(0, S_CTRL); 333241236Sbrooks else 334241236Sbrooks goto bad; 335241236Sbrooks return UNVIS_NOCHAR; 336241236Sbrooks 337241236Sbrooks case S_META1: 338241236Sbrooks *astate = SS(0, S_GROUND); 339241236Sbrooks *cp |= c; 340241236Sbrooks return UNVIS_VALID; 341241236Sbrooks 342241236Sbrooks case S_CTRL: 343241236Sbrooks if (c == '?') 344241236Sbrooks *cp |= 0177; 345241236Sbrooks else 346241236Sbrooks *cp |= c & 037; 347241236Sbrooks *astate = SS(0, S_GROUND); 348241236Sbrooks return UNVIS_VALID; 349241236Sbrooks 350241236Sbrooks case S_OCTAL2: /* second possible octal digit */ 351241236Sbrooks if (isoctal(uc)) { 352241236Sbrooks /* 353241236Sbrooks * yes - and maybe a third 354241236Sbrooks */ 355241236Sbrooks *cp = (*cp << 3) + (c - '0'); 356241236Sbrooks *astate = SS(0, S_OCTAL3); 357241236Sbrooks return UNVIS_NOCHAR; 358241236Sbrooks } 359241236Sbrooks /* 360241236Sbrooks * no - done with current sequence, push back passed char 361241236Sbrooks */ 362241236Sbrooks *astate = SS(0, S_GROUND); 363241236Sbrooks return UNVIS_VALIDPUSH; 364241236Sbrooks 365241236Sbrooks case S_OCTAL3: /* third possible octal digit */ 366241236Sbrooks *astate = SS(0, S_GROUND); 367241236Sbrooks if (isoctal(uc)) { 368241236Sbrooks *cp = (*cp << 3) + (c - '0'); 369241236Sbrooks return UNVIS_VALID; 370241236Sbrooks } 371241236Sbrooks /* 372241236Sbrooks * we were done, push back passed char 373241236Sbrooks */ 374241236Sbrooks return UNVIS_VALIDPUSH; 375241236Sbrooks 376241236Sbrooks case S_HEX: 377241236Sbrooks if (!isxdigit(uc)) 378241236Sbrooks goto bad; 379241236Sbrooks /*FALLTHROUGH*/ 380241236Sbrooks case S_HEX1: 381241236Sbrooks if (isxdigit(uc)) { 382241236Sbrooks *cp = xtod(uc); 383241236Sbrooks *astate = SS(0, S_HEX2); 384241236Sbrooks return UNVIS_NOCHAR; 385241236Sbrooks } 386241236Sbrooks /* 387241236Sbrooks * no - done with current sequence, push back passed char 388241236Sbrooks */ 389241236Sbrooks *astate = SS(0, S_GROUND); 390241236Sbrooks return UNVIS_VALIDPUSH; 391241236Sbrooks 392241236Sbrooks case S_HEX2: 393241236Sbrooks *astate = S_GROUND; 394241236Sbrooks if (isxdigit(uc)) { 395241236Sbrooks *cp = xtod(uc) | (*cp << 4); 396241236Sbrooks return UNVIS_VALID; 397241236Sbrooks } 398241236Sbrooks return UNVIS_VALIDPUSH; 399241236Sbrooks 400241236Sbrooks case S_MIME1: 401241236Sbrooks if (uc == '\n' || uc == '\r') { 402241236Sbrooks *astate = SS(0, S_EATCRNL); 403241236Sbrooks return UNVIS_NOCHAR; 404241236Sbrooks } 405241236Sbrooks if (isxdigit(uc) && (isdigit(uc) || isupper(uc))) { 406241236Sbrooks *cp = XTOD(uc); 407241236Sbrooks *astate = SS(0, S_MIME2); 408241236Sbrooks return UNVIS_NOCHAR; 409241236Sbrooks } 410241236Sbrooks goto bad; 411241236Sbrooks 412241236Sbrooks case S_MIME2: 413241236Sbrooks if (isxdigit(uc) && (isdigit(uc) || isupper(uc))) { 414241236Sbrooks *astate = SS(0, S_GROUND); 415241236Sbrooks *cp = XTOD(uc) | (*cp << 4); 416241236Sbrooks return UNVIS_VALID; 417241236Sbrooks } 418241236Sbrooks goto bad; 419241236Sbrooks 420241236Sbrooks case S_EATCRNL: 421241236Sbrooks switch (uc) { 422241236Sbrooks case '\r': 423241236Sbrooks case '\n': 424241236Sbrooks return UNVIS_NOCHAR; 425241236Sbrooks case '=': 426241236Sbrooks *astate = SS(0, S_MIME1); 427241236Sbrooks return UNVIS_NOCHAR; 428241236Sbrooks default: 429241236Sbrooks *cp = uc; 430241236Sbrooks *astate = SS(0, S_GROUND); 431241236Sbrooks return UNVIS_VALID; 432241236Sbrooks } 433241236Sbrooks 434241236Sbrooks case S_AMP: 435241236Sbrooks *cp = 0; 436241236Sbrooks if (uc == '#') { 437241236Sbrooks *astate = SS(0, S_NUMBER); 438241236Sbrooks return UNVIS_NOCHAR; 439241236Sbrooks } 440241236Sbrooks *astate = SS(0, S_STRING); 441241236Sbrooks /*FALLTHROUGH*/ 442241236Sbrooks 443241236Sbrooks case S_STRING: 444241236Sbrooks ia = *cp; /* index in the array */ 445241236Sbrooks is = GI(*astate); /* index in the string */ 446241236Sbrooks lc = is == 0 ? 0 : nv[ia].name[is - 1]; /* last character */ 447241236Sbrooks 448241236Sbrooks if (uc == ';') 449241236Sbrooks uc = '\0'; 450241236Sbrooks 451241236Sbrooks for (; ia < __arraycount(nv); ia++) { 452241236Sbrooks if (is != 0 && nv[ia].name[is - 1] != lc) 453241236Sbrooks goto bad; 454241236Sbrooks if (nv[ia].name[is] == uc) 455241236Sbrooks break; 456241236Sbrooks } 457241236Sbrooks 458241236Sbrooks if (ia == __arraycount(nv)) 459241236Sbrooks goto bad; 460241236Sbrooks 461241236Sbrooks if (uc != 0) { 462241236Sbrooks *cp = ia; 463241236Sbrooks *astate = SS(is + 1, S_STRING); 464241236Sbrooks return UNVIS_NOCHAR; 465241236Sbrooks } 466241236Sbrooks 467241236Sbrooks *cp = nv[ia].value; 468241236Sbrooks *astate = SS(0, S_GROUND); 469241236Sbrooks return UNVIS_VALID; 470241236Sbrooks 471241236Sbrooks case S_NUMBER: 472241236Sbrooks if (uc == ';') 473241236Sbrooks return UNVIS_VALID; 474241236Sbrooks if (!isdigit(uc)) 475241236Sbrooks goto bad; 476241236Sbrooks *cp += (*cp * 10) + uc - '0'; 477241236Sbrooks return UNVIS_NOCHAR; 478241236Sbrooks 479241236Sbrooks default: 480241236Sbrooks bad: 481241236Sbrooks /* 482241236Sbrooks * decoder in unknown state - (probably uninitialized) 483241236Sbrooks */ 484241236Sbrooks *astate = SS(0, S_GROUND); 485241236Sbrooks return UNVIS_SYNBAD; 486241236Sbrooks } 487241236Sbrooks} 488241236Sbrooks 489241236Sbrooks/* 490241236Sbrooks * strnunvisx - decode src into dst 491241236Sbrooks * 492241236Sbrooks * Number of chars decoded into dst is returned, -1 on error. 493241236Sbrooks * Dst is null terminated. 494241236Sbrooks */ 495241236Sbrooks 496241236Sbrooksint 497241236Sbrooksstrnunvisx(char *dst, size_t dlen, const char *src, int flag) 498241236Sbrooks{ 499241236Sbrooks char c; 500241236Sbrooks char t = '\0', *start = dst; 501241236Sbrooks int state = 0; 502241236Sbrooks 503241236Sbrooks _DIAGASSERT(src != NULL); 504241236Sbrooks _DIAGASSERT(dst != NULL); 505241236Sbrooks#define CHECKSPACE() \ 506241236Sbrooks do { \ 507241236Sbrooks if (dlen-- == 0) { \ 508241236Sbrooks errno = ENOSPC; \ 509241236Sbrooks return -1; \ 510241236Sbrooks } \ 511241236Sbrooks } while (/*CONSTCOND*/0) 512241236Sbrooks 513241236Sbrooks while ((c = *src++) != '\0') { 514241236Sbrooks again: 515241236Sbrooks switch (unvis(&t, c, &state, flag)) { 516241236Sbrooks case UNVIS_VALID: 517241236Sbrooks CHECKSPACE(); 518241236Sbrooks *dst++ = t; 519241236Sbrooks break; 520241236Sbrooks case UNVIS_VALIDPUSH: 521241236Sbrooks CHECKSPACE(); 522241236Sbrooks *dst++ = t; 523241236Sbrooks goto again; 524241236Sbrooks case 0: 525241236Sbrooks case UNVIS_NOCHAR: 526241236Sbrooks break; 527241236Sbrooks case UNVIS_SYNBAD: 528241236Sbrooks errno = EINVAL; 529241236Sbrooks return -1; 530241236Sbrooks default: 531241236Sbrooks _DIAGASSERT(/*CONSTCOND*/0); 532241236Sbrooks errno = EINVAL; 533241236Sbrooks return -1; 534241236Sbrooks } 535241236Sbrooks } 536241236Sbrooks if (unvis(&t, c, &state, UNVIS_END) == UNVIS_VALID) { 537241236Sbrooks CHECKSPACE(); 538241236Sbrooks *dst++ = t; 539241236Sbrooks } 540241236Sbrooks CHECKSPACE(); 541241236Sbrooks *dst = '\0'; 542241236Sbrooks return (int)(dst - start); 543241236Sbrooks} 544241236Sbrooks 545241236Sbrooksint 546241236Sbrooksstrunvisx(char *dst, const char *src, int flag) 547241236Sbrooks{ 548241236Sbrooks return strnunvisx(dst, (size_t)~0, src, flag); 549241236Sbrooks} 550241236Sbrooks 551241236Sbrooksint 552241236Sbrooksstrunvis(char *dst, const char *src) 553241236Sbrooks{ 554241236Sbrooks return strnunvisx(dst, (size_t)~0, src, 0); 555241236Sbrooks} 556241236Sbrooks 557241236Sbrooksint 558241236Sbrooksstrnunvis(char *dst, size_t dlen, const char *src) 559241236Sbrooks{ 560241236Sbrooks return strnunvisx(dst, dlen, src, 0); 561241236Sbrooks} 562241236Sbrooks#endif 563