mandoc.c

   1 /*      $Id: mandoc.c,v 1.100 2017/06/02 19:21:23 schwarze Exp $ */
   2 /*
   3  * Copyright (c) 2008-2011, 2014 Kristaps Dzonsons <kristaps@bsd.lv>
   4  * Copyright (c) 2011-2015, 2017 Ingo Schwarze <schwarze@openbsd.org>
   5  *
   6  * Permission to use, copy, modify, and distribute this software for any
   7  * purpose with or without fee is hereby granted, provided that the above
   8  * copyright notice and this permission notice appear in all copies.
   9  *
  10  * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHORS DISCLAIM ALL WARRANTIES
  11  * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
  12  * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHORS BE LIABLE FOR
  13  * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
  14  * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
  15  * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
  16  * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
  17  */
  18 #include "config.h"
  19
  20 #include <sys/types.h>
  21
  22 #include <assert.h>
  23 #include <ctype.h>
  24 #include <errno.h>
  25 #include <limits.h>
  26 #include <stdlib.h>
  27 #include <stdio.h>
  28 #include <string.h>
  29 #include <time.h>
  30
  31 #include "mandoc.h"
  32 #include "mandoc_aux.h"
  33 #include "libmandoc.h"
  34
  35 static  int      a2time(time_t *, const char *, const char *);
  36 static  char    *time2a(time_t);
  37
  38
  39 enum mandoc_esc
  40 mandoc_escape(const char **end, const char **start, int *sz)
  41 {
  42         const char      *local_start;
  43         int              local_sz;
  44         char             term;
  45         enum mandoc_esc  gly;
  46
  47         /*
  48          * When the caller doesn't provide return storage,
  49          * use local storage.
  50          */
  51
  52         if (NULL == start)
  53                 start = &local_start;
  54         if (NULL == sz)
  55                 sz = &local_sz;
  56
  57         /*
  58          * Beyond the backslash, at least one input character
  59          * is part of the escape sequence.  With one exception
  60          * (see below), that character won't be returned.
  61          */
  62
  63         gly = ESCAPE_ERROR;
  64         *start = ++*end;
  65         *sz = 0;
  66         term = '\0';
  67
  68         switch ((*start)[-1]) {
  69         /*
  70          * First the glyphs.  There are several different forms of
  71          * these, but each eventually returns a substring of the glyph
  72          * name.
  73          */
  74         case '(':
  75                 gly = ESCAPE_SPECIAL;
  76                 *sz = 2;
  77                 break;
  78         case '[':
  79                 gly = ESCAPE_SPECIAL;
  80                 term = ']';
  81                 break;
  82         case 'C':
  83                 if ('\'' != **start)
  84                         return ESCAPE_ERROR;
  85                 *start = ++*end;
  86                 gly = ESCAPE_SPECIAL;
  87                 term = '\'';
  88                 break;
  89
  90         /*
  91          * Escapes taking no arguments at all.
  92          */
  93         case 'd':
  94         case 'u':
  95         case ',':
  96         case '/':
  97                 return ESCAPE_IGNORE;
  98
  99         /*
 100          * The \z escape is supposed to output the following
 101          * character without advancing the cursor position.
 102          * Since we are mostly dealing with terminal mode,
 103          * let us just skip the next character.
 104          */
 105         case 'z':
 106                 return ESCAPE_SKIPCHAR;
 107
 108         /*
 109          * Handle all triggers matching \X(xy, \Xx, and \X[xxxx], where
 110          * 'X' is the trigger.  These have opaque sub-strings.
 111          */
 112         case 'F':
 113         case 'g':
 114         case 'k':
 115         case 'M':
 116         case 'm':
 117         case 'n':
 118         case 'V':
 119         case 'Y':
 120                 gly = ESCAPE_IGNORE;
 121                 /* FALLTHROUGH */
 122         case 'f':
 123                 if (ESCAPE_ERROR == gly)
 124                         gly = ESCAPE_FONT;
 125                 switch (**start) {
 126                 case '(':
 127                         *start = ++*end;
 128                         *sz = 2;
 129                         break;
 130                 case '[':
 131                         *start = ++*end;
 132                         term = ']';
 133                         break;
 134                 default:
 135                         *sz = 1;
 136                         break;
 137                 }
 138                 break;
 139
 140         /*
 141          * These escapes are of the form \X'Y', where 'X' is the trigger
 142          * and 'Y' is any string.  These have opaque sub-strings.
 143          * The \B and \w escapes are handled in roff.c, roff_res().
 144          */
 145         case 'A':
 146         case 'b':
 147         case 'D':
 148         case 'R':
 149         case 'X':
 150         case 'Z':
 151                 gly = ESCAPE_IGNORE;
 152                 /* FALLTHROUGH */
 153         case 'o':
 154                 if (**start == '\0')
 155                         return ESCAPE_ERROR;
 156                 if (gly == ESCAPE_ERROR)
 157                         gly = ESCAPE_OVERSTRIKE;
 158                 term = **start;
 159                 *start = ++*end;
 160                 break;
 161
 162         /*
 163          * These escapes are of the form \X'N', where 'X' is the trigger
 164          * and 'N' resolves to a numerical expression.
 165          */
 166         case 'h':
 167         case 'H':
 168         case 'L':
 169         case 'l':
 170         case 'S':
 171         case 'v':
 172         case 'x':
 173                 if (strchr(" %&()*+-./0123456789:<=>", **start)) {
 174                         if ('\0' != **start)
 175                                 ++*end;
 176                         return ESCAPE_ERROR;
 177                 }
 178                 switch ((*start)[-1]) {
 179                 case 'h':
 180                         gly = ESCAPE_HORIZ;
 181                         break;
 182                 case 'l':
 183                         gly = ESCAPE_HLINE;
 184                         break;
 185                 default:
 186                         gly = ESCAPE_IGNORE;
 187                         break;
 188                 }
 189                 term = **start;
 190                 *start = ++*end;
 191                 break;
 192
 193         /*
 194          * Special handling for the numbered character escape.
 195          * XXX Do any other escapes need similar handling?
 196          */
 197         case 'N':
 198                 if ('\0' == **start)
 199                         return ESCAPE_ERROR;
 200                 (*end)++;
 201                 if (isdigit((unsigned char)**start)) {
 202                         *sz = 1;
 203                         return ESCAPE_IGNORE;
 204                 }
 205                 (*start)++;
 206                 while (isdigit((unsigned char)**end))
 207                         (*end)++;
 208                 *sz = *end - *start;
 209                 if ('\0' != **end)
 210                         (*end)++;
 211                 return ESCAPE_NUMBERED;
 212
 213         /*
 214          * Sizes get a special category of their own.
 215          */
 216         case 's':
 217                 gly = ESCAPE_IGNORE;
 218
 219                 /* See +/- counts as a sign. */
 220                 if ('+' == **end || '-' == **end || ASCII_HYPH == **end)
 221                         *start = ++*end;
 222
 223                 switch (**end) {
 224                 case '(':
 225                         *start = ++*end;
 226                         *sz = 2;
 227                         break;
 228                 case '[':
 229                         *start = ++*end;
 230                         term = ']';
 231                         break;
 232                 case '\'':
 233                         *start = ++*end;
 234                         term = '\'';
 235                         break;
 236                 case '3':
 237                 case '2':
 238                 case '1':
 239                         *sz = (*end)[-1] == 's' &&
 240                             isdigit((unsigned char)(*end)[1]) ? 2 : 1;
 241                         break;
 242                 default:
 243                         *sz = 1;
 244                         break;
 245                 }
 246
 247                 break;
 248
 249         /*
 250          * Anything else is assumed to be a glyph.
 251          * In this case, pass back the character after the backslash.
 252          */
 253         default:
 254                 gly = ESCAPE_SPECIAL;
 255                 *start = --*end;
 256                 *sz = 1;
 257                 break;
 258         }
 259
 260         assert(ESCAPE_ERROR != gly);
 261
 262         /*
 263          * Read up to the terminating character,
 264          * paying attention to nested escapes.
 265          */
 266
 267         if ('\0' != term) {
 268                 while (**end != term) {
 269                         switch (**end) {
 270                         case '\0':
 271                                 return ESCAPE_ERROR;
 272                         case '\\':
 273                                 (*end)++;
 274                                 if (ESCAPE_ERROR ==
 275                                     mandoc_escape(end, NULL, NULL))
 276                                         return ESCAPE_ERROR;
 277                                 break;
 278                         default:
 279                                 (*end)++;
 280                                 break;
 281                         }
 282                 }
 283                 *sz = (*end)++ - *start;
 284         } else {
 285                 assert(*sz > 0);
 286                 if ((size_t)*sz > strlen(*start))
 287                         return ESCAPE_ERROR;
 288                 *end += *sz;
 289         }
 290
 291         /* Run post-processors. */
 292
 293         switch (gly) {
 294         case ESCAPE_FONT:
 295                 if (2 == *sz) {
 296                         if ('C' == **start) {
 297                                 /*
 298                                  * Treat constant-width font modes
 299                                  * just like regular font modes.
 300                                  */
 301                                 (*start)++;
 302                                 (*sz)--;
 303                         } else {
 304                                 if ('B' == (*start)[0] && 'I' == (*start)[1])
 305                                         gly = ESCAPE_FONTBI;
 306                                 break;
 307                         }
 308                 } else if (1 != *sz)
 309                         break;
 310
 311                 switch (**start) {
 312                 case '3':
 313                 case 'B':
 314                         gly = ESCAPE_FONTBOLD;
 315                         break;
 316                 case '2':
 317                 case 'I':
 318                         gly = ESCAPE_FONTITALIC;
 319                         break;
 320                 case 'P':
 321                         gly = ESCAPE_FONTPREV;
 322                         break;
 323                 case '1':
 324                 case 'R':
 325                         gly = ESCAPE_FONTROMAN;
 326                         break;
 327                 }
 328                 break;
 329         case ESCAPE_SPECIAL:
 330                 if (1 == *sz && 'c' == **start)
 331                         gly = ESCAPE_NOSPACE;
 332                 /*
 333                  * Unicode escapes are defined in groff as \[u0000]
 334                  * to \[u10FFFF], where the contained value must be
 335                  * a valid Unicode codepoint.  Here, however, only
 336                  * check the length and range.
 337                  */
 338                 if (**start != 'u' || *sz < 5 || *sz > 7)
 339                         break;
 340                 if (*sz == 7 && ((*start)[1] != '1' || (*start)[2] != '0'))
 341                         break;
 342                 if (*sz == 6 && (*start)[1] == '0')
 343                         break;
 344                 if (*sz == 5 && (*start)[1] == 'D' &&
 345                     strchr("89ABCDEF", (*start)[2]) != NULL)
 346                         break;
 347                 if ((int)strspn(*start + 1, "0123456789ABCDEFabcdef")
 348                     + 1 == *sz)
 349                         gly = ESCAPE_UNICODE;
 350                 break;
 351         default:
 352                 break;
 353         }
 354
 355         return gly;
 356 }
 357
 358 /*
 359  * Parse a quoted or unquoted roff-style request or macro argument.
 360  * Return a pointer to the parsed argument, which is either the original
 361  * pointer or advanced by one byte in case the argument is quoted.
 362  * NUL-terminate the argument in place.
 363  * Collapse pairs of quotes inside quoted arguments.
 364  * Advance the argument pointer to the next argument,
 365  * or to the NUL byte terminating the argument line.
 366  */
 367 char *
 368 mandoc_getarg(struct mparse *parse, char **cpp, int ln, int *pos)
 369 {
 370         char     *start, *cp;
 371         int       quoted, pairs, white;
 372
 373         /* Quoting can only start with a new word. */
 374         start = *cpp;
 375         quoted = 0;
 376         if ('"' == *start) {
 377                 quoted = 1;
 378                 start++;
 379         }
 380
 381         pairs = 0;
 382         white = 0;
 383         for (cp = start; '\0' != *cp; cp++) {
 384
 385                 /*
 386                  * Move the following text left
 387                  * after quoted quotes and after "\\" and "\t".
 388                  */
 389                 if (pairs)
 390                         cp[-pairs] = cp[0];
 391
 392                 if ('\\' == cp[0]) {
 393                         /*
 394                          * In copy mode, translate double to single
 395                          * backslashes and backslash-t to literal tabs.
 396                          */
 397                         switch (cp[1]) {
 398                         case 't':
 399                                 cp[0] = '\t';
 400                                 /* FALLTHROUGH */
 401                         case '\\':
 402                                 pairs++;
 403                                 cp++;
 404                                 break;
 405                         case ' ':
 406                                 /* Skip escaped blanks. */
 407                                 if (0 == quoted)
 408                                         cp++;
 409                                 break;
 410                         default:
 411                                 break;
 412                         }
 413                 } else if (0 == quoted) {
 414                         if (' ' == cp[0]) {
 415                                 /* Unescaped blanks end unquoted args. */
 416                                 white = 1;
 417                                 break;
 418                         }
 419                 } else if ('"' == cp[0]) {
 420                         if ('"' == cp[1]) {
 421                                 /* Quoted quotes collapse. */
 422                                 pairs++;
 423                                 cp++;
 424                         } else {
 425                                 /* Unquoted quotes end quoted args. */
 426                                 quoted = 2;
 427                                 break;
 428                         }
 429                 }
 430         }
 431
 432         /* Quoted argument without a closing quote. */
 433         if (1 == quoted)
 434                 mandoc_msg(MANDOCERR_ARG_QUOTE, parse, ln, *pos, NULL);
 435
 436         /* NUL-terminate this argument and move to the next one. */
 437         if (pairs)
 438                 cp[-pairs] = '\0';
 439         if ('\0' != *cp) {
 440                 *cp++ = '\0';
 441                 while (' ' == *cp)
 442                         cp++;
 443         }
 444         *pos += (int)(cp - start) + (quoted ? 1 : 0);
 445         *cpp = cp;
 446
 447         if ('\0' == *cp && (white || ' ' == cp[-1]))
 448                 mandoc_msg(MANDOCERR_SPACE_EOL, parse, ln, *pos, NULL);
 449
 450         return start;
 451 }
 452
 453 static int
 454 a2time(time_t *t, const char *fmt, const char *p)
 455 {
 456         struct tm        tm;
 457         char            *pp;
 458
 459         memset(&tm, 0, sizeof(struct tm));
 460
 461         pp = NULL;
 462 #if HAVE_STRPTIME
 463         pp = strptime(p, fmt, &tm);
 464 #endif
 465         if (NULL != pp && '\0' == *pp) {
 466                 *t = mktime(&tm);
 467                 return 1;
 468         }
 469
 470         return 0;
 471 }
 472
 473 static char *
 474 time2a(time_t t)
 475 {
 476         struct tm       *tm;
 477         char            *buf, *p;
 478         size_t           ssz;
 479         int              isz;
 480
 481         tm = localtime(&t);
 482         if (tm == NULL)
 483                 return NULL;
 484
 485         /*
 486          * Reserve space:
 487          * up to 9 characters for the month (September) + blank
 488          * up to 2 characters for the day + comma + blank
 489          * 4 characters for the year and a terminating '\0'
 490          */
 491
 492         p = buf = mandoc_malloc(10 + 4 + 4 + 1);
 493
 494         if ((ssz = strftime(p, 10 + 1, "%B ", tm)) == 0)
 495                 goto fail;
 496         p += (int)ssz;
 497
 498         /*
 499          * The output format is just "%d" here, not "%2d" or "%02d".
 500          * That's also the reason why we can't just format the
 501          * date as a whole with "%B %e, %Y" or "%B %d, %Y".
 502          * Besides, the present approach is less prone to buffer
 503          * overflows, in case anybody should ever introduce the bug
 504          * of looking at LC_TIME.
 505          */
 506
 507         if ((isz = snprintf(p, 4 + 1, "%d, ", tm->tm_mday)) == -1)
 508                 goto fail;
 509         p += isz;
 510
 511         if (strftime(p, 4 + 1, "%Y", tm) == 0)
 512                 goto fail;
 513         return buf;
 514
 515 fail:
 516         free(buf);
 517         return NULL;
 518 }
 519
 520 char *
 521 mandoc_normdate(struct mparse *parse, char *in, int ln, int pos)
 522 {
 523         time_t           t;
 524
 525         /* No date specified: use today's date. */
 526
 527         if (in == NULL || *in == '\0' || strcmp(in, "$" "Mdocdate$") == 0) {
 528                 mandoc_msg(MANDOCERR_DATE_MISSING, parse, ln, pos, NULL);
 529                 return time2a(time(NULL));
 530         }
 531
 532         /* Valid mdoc(7) date format. */
 533
 534         if (a2time(&t, "$" "Mdocdate: %b %d %Y $", in) ||
 535             a2time(&t, "%b %d, %Y", in))
 536                 return time2a(t);
 537
 538         /* Do not warn about the legacy man(7) format. */
 539
 540         if ( ! a2time(&t, "%Y-%m-%d", in))
 541                 mandoc_msg(MANDOCERR_DATE_BAD, parse, ln, pos, in);
 542
 543         /* Use any non-mdoc(7) date verbatim. */
 544
 545         return mandoc_strdup(in);
 546 }
 547
 548 int
 549 mandoc_eos(const char *p, size_t sz)
 550 {
 551         const char      *q;
 552         int              enclosed, found;
 553
 554         if (0 == sz)
 555                 return 0;
 556
 557         /*
 558          * End-of-sentence recognition must include situations where
 559          * some symbols, such as `)', allow prior EOS punctuation to
 560          * propagate outward.
 561          */
 562
 563         enclosed = found = 0;
 564         for (q = p + (int)sz - 1; q >= p; q--) {
 565                 switch (*q) {
 566                 case '\"':
 567                 case '\'':
 568                 case ']':
 569                 case ')':
 570                         if (0 == found)
 571                                 enclosed = 1;
 572                         break;
 573                 case '.':
 574                 case '!':
 575                 case '?':
 576                         found = 1;
 577                         break;
 578                 default:
 579                         return found &&
 580                             (!enclosed || isalnum((unsigned char)*q));
 581                 }
 582         }
 583
 584         return found && !enclosed;
 585 }
 586
 587 /*
 588  * Convert a string to a long that may not be <0.
 589  * If the string is invalid, or is less than 0, return -1.
 590  */
 591 int
 592 mandoc_strntoi(const char *p, size_t sz, int base)
 593 {
 594         char             buf[32];
 595         char            *ep;
 596         long             v;
 597
 598         if (sz > 31)
 599                 return -1;
 600
 601         memcpy(buf, p, sz);
 602         buf[(int)sz] = '\0';
 603
 604         errno = 0;
 605         v = strtol(buf, &ep, base);
 606
 607         if (buf[0] == '\0' || *ep != '\0')
 608                 return -1;
 609
 610         if (v > INT_MAX)
 611                 v = INT_MAX;
 612         if (v < INT_MIN)
 613                 v = INT_MIN;
 614
 615         return (int)v;
 616 }