Annotation of mandoc/mandoc.c, Revision 1.87
1.87 ! schwarze 1: /* $Id: mandoc.c,v 1.86 2014/08/18 09:11:47 kristaps Exp $ */
1.1 kristaps 2: /*
1.59 schwarze 3: * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
1.78 schwarze 4: * Copyright (c) 2011, 2012, 2013, 2014 Ingo Schwarze <schwarze@openbsd.org>
1.1 kristaps 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: *
1.36 schwarze 10: * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHORS DISCLAIM ALL WARRANTIES
1.1 kristaps 11: * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
1.36 schwarze 12: * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHORS BE LIABLE FOR
1.1 kristaps 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: */
1.9 kristaps 18: #include "config.h"
1.7 kristaps 19:
1.2 kristaps 20: #include <sys/types.h>
21:
1.1 kristaps 22: #include <assert.h>
23: #include <ctype.h>
1.50 kristaps 24: #include <errno.h>
25: #include <limits.h>
1.1 kristaps 26: #include <stdlib.h>
1.4 kristaps 27: #include <stdio.h>
28: #include <string.h>
1.7 kristaps 29: #include <time.h>
1.1 kristaps 30:
1.18 kristaps 31: #include "mandoc.h"
1.76 schwarze 32: #include "mandoc_aux.h"
1.1 kristaps 33: #include "libmandoc.h"
34:
1.37 schwarze 35: #define DATESIZE 32
36:
1.18 kristaps 37: static int a2time(time_t *, const char *, const char *);
1.37 schwarze 38: static char *time2a(time_t);
1.7 kristaps 39:
1.45 kristaps 40:
41: enum mandoc_esc
1.74 schwarze 42: mandoc_escape(const char **end, const char **start, int *sz)
1.1 kristaps 43: {
1.65 schwarze 44: const char *local_start;
45: int local_sz;
46: char term;
1.79 schwarze 47: enum mandoc_esc gly;
1.45 kristaps 48:
1.65 schwarze 49: /*
50: * When the caller doesn't provide return storage,
51: * use local storage.
52: */
53:
54: if (NULL == start)
55: start = &local_start;
56: if (NULL == sz)
57: sz = &local_sz;
58:
59: /*
60: * Beyond the backslash, at least one input character
61: * is part of the escape sequence. With one exception
62: * (see below), that character won't be returned.
63: */
64:
1.45 kristaps 65: gly = ESCAPE_ERROR;
1.65 schwarze 66: *start = ++*end;
67: *sz = 0;
1.64 schwarze 68: term = '\0';
1.18 kristaps 69:
1.65 schwarze 70: switch ((*start)[-1]) {
1.45 kristaps 71: /*
72: * First the glyphs. There are several different forms of
73: * these, but each eventually returns a substring of the glyph
74: * name.
75: */
1.79 schwarze 76: case '(':
1.45 kristaps 77: gly = ESCAPE_SPECIAL;
1.65 schwarze 78: *sz = 2;
1.45 kristaps 79: break;
1.79 schwarze 80: case '[':
1.45 kristaps 81: gly = ESCAPE_SPECIAL;
82: term = ']';
83: break;
1.79 schwarze 84: case 'C':
1.65 schwarze 85: if ('\'' != **start)
1.45 kristaps 86: return(ESCAPE_ERROR);
1.65 schwarze 87: *start = ++*end;
1.87 ! schwarze 88: gly = ESCAPE_SPECIAL;
1.45 kristaps 89: term = '\'';
90: break;
1.72 schwarze 91:
92: /*
93: * Escapes taking no arguments at all.
94: */
1.79 schwarze 95: case 'd':
1.72 schwarze 96: /* FALLTHROUGH */
1.79 schwarze 97: case 'u':
1.72 schwarze 98: return(ESCAPE_IGNORE);
1.63 schwarze 99:
100: /*
101: * The \z escape is supposed to output the following
1.79 schwarze 102: * character without advancing the cursor position.
1.63 schwarze 103: * Since we are mostly dealing with terminal mode,
104: * let us just skip the next character.
105: */
1.79 schwarze 106: case 'z':
1.63 schwarze 107: return(ESCAPE_SKIPCHAR);
1.1 kristaps 108:
1.45 kristaps 109: /*
110: * Handle all triggers matching \X(xy, \Xx, and \X[xxxx], where
111: * 'X' is the trigger. These have opaque sub-strings.
112: */
1.79 schwarze 113: case 'F':
1.45 kristaps 114: /* FALLTHROUGH */
1.79 schwarze 115: case 'g':
1.24 kristaps 116: /* FALLTHROUGH */
1.79 schwarze 117: case 'k':
1.24 kristaps 118: /* FALLTHROUGH */
1.79 schwarze 119: case 'M':
1.24 kristaps 120: /* FALLTHROUGH */
1.79 schwarze 121: case 'm':
1.24 kristaps 122: /* FALLTHROUGH */
1.79 schwarze 123: case 'n':
1.24 kristaps 124: /* FALLTHROUGH */
1.79 schwarze 125: case 'V':
1.24 kristaps 126: /* FALLTHROUGH */
1.79 schwarze 127: case 'Y':
1.60 schwarze 128: gly = ESCAPE_IGNORE;
1.24 kristaps 129: /* FALLTHROUGH */
1.79 schwarze 130: case 'f':
1.45 kristaps 131: if (ESCAPE_ERROR == gly)
132: gly = ESCAPE_FONT;
1.65 schwarze 133: switch (**start) {
1.79 schwarze 134: case '(':
1.65 schwarze 135: *start = ++*end;
136: *sz = 2;
1.45 kristaps 137: break;
1.79 schwarze 138: case '[':
1.65 schwarze 139: *start = ++*end;
1.45 kristaps 140: term = ']';
141: break;
142: default:
1.65 schwarze 143: *sz = 1;
1.45 kristaps 144: break;
145: }
146: break;
147:
148: /*
149: * These escapes are of the form \X'Y', where 'X' is the trigger
150: * and 'Y' is any string. These have opaque sub-strings.
1.78 schwarze 151: * The \B and \w escapes are handled in roff.c, roff_res().
1.45 kristaps 152: */
1.79 schwarze 153: case 'A':
1.24 kristaps 154: /* FALLTHROUGH */
1.79 schwarze 155: case 'b':
1.24 kristaps 156: /* FALLTHROUGH */
1.79 schwarze 157: case 'D':
1.24 kristaps 158: /* FALLTHROUGH */
1.79 schwarze 159: case 'o':
1.24 kristaps 160: /* FALLTHROUGH */
1.79 schwarze 161: case 'R':
1.73 schwarze 162: /* FALLTHROUGH */
1.79 schwarze 163: case 'X':
1.24 kristaps 164: /* FALLTHROUGH */
1.79 schwarze 165: case 'Z':
1.77 schwarze 166: if ('\0' == **start)
1.45 kristaps 167: return(ESCAPE_ERROR);
168: gly = ESCAPE_IGNORE;
1.77 schwarze 169: term = **start;
1.65 schwarze 170: *start = ++*end;
1.24 kristaps 171: break;
1.45 kristaps 172:
173: /*
174: * These escapes are of the form \X'N', where 'X' is the trigger
175: * and 'N' resolves to a numerical expression.
176: */
1.79 schwarze 177: case 'h':
1.28 kristaps 178: /* FALLTHROUGH */
1.79 schwarze 179: case 'H':
1.45 kristaps 180: /* FALLTHROUGH */
1.79 schwarze 181: case 'L':
1.45 kristaps 182: /* FALLTHROUGH */
1.79 schwarze 183: case 'l':
1.45 kristaps 184: /* FALLTHROUGH */
1.79 schwarze 185: case 'S':
1.45 kristaps 186: /* FALLTHROUGH */
1.79 schwarze 187: case 'v':
1.28 kristaps 188: /* FALLTHROUGH */
1.79 schwarze 189: case 'x':
1.82 schwarze 190: if (strchr(" %&()*+-./0123456789:<=>", **start)) {
1.86 kristaps 191: if ('\0' != **start)
192: ++*end;
1.65 schwarze 193: return(ESCAPE_ERROR);
1.82 schwarze 194: }
1.73 schwarze 195: gly = ESCAPE_IGNORE;
1.77 schwarze 196: term = **start;
1.65 schwarze 197: *start = ++*end;
1.45 kristaps 198: break;
1.60 schwarze 199:
200: /*
201: * Special handling for the numbered character escape.
202: * XXX Do any other escapes need similar handling?
203: */
1.79 schwarze 204: case 'N':
1.65 schwarze 205: if ('\0' == **start)
1.60 schwarze 206: return(ESCAPE_ERROR);
1.65 schwarze 207: (*end)++;
208: if (isdigit((unsigned char)**start)) {
209: *sz = 1;
1.60 schwarze 210: return(ESCAPE_IGNORE);
1.65 schwarze 211: }
212: (*start)++;
1.60 schwarze 213: while (isdigit((unsigned char)**end))
214: (*end)++;
1.65 schwarze 215: *sz = *end - *start;
1.60 schwarze 216: if ('\0' != **end)
217: (*end)++;
218: return(ESCAPE_NUMBERED);
1.45 kristaps 219:
1.79 schwarze 220: /*
1.45 kristaps 221: * Sizes get a special category of their own.
222: */
1.79 schwarze 223: case 's':
1.45 kristaps 224: gly = ESCAPE_IGNORE;
1.28 kristaps 225:
1.45 kristaps 226: /* See +/- counts as a sign. */
1.65 schwarze 227: if ('+' == **end || '-' == **end || ASCII_HYPH == **end)
228: (*end)++;
1.8 kristaps 229:
1.65 schwarze 230: switch (**end) {
1.79 schwarze 231: case '(':
1.65 schwarze 232: *start = ++*end;
233: *sz = 2;
1.22 kristaps 234: break;
1.79 schwarze 235: case '[':
1.65 schwarze 236: *start = ++*end;
1.64 schwarze 237: term = ']';
1.22 kristaps 238: break;
1.79 schwarze 239: case '\'':
1.65 schwarze 240: *start = ++*end;
1.64 schwarze 241: term = '\'';
1.22 kristaps 242: break;
243: default:
1.65 schwarze 244: *sz = 1;
1.22 kristaps 245: break;
1.8 kristaps 246: }
247:
1.45 kristaps 248: break;
1.33 kristaps 249:
1.45 kristaps 250: /*
251: * Anything else is assumed to be a glyph.
1.65 schwarze 252: * In this case, pass back the character after the backslash.
1.45 kristaps 253: */
254: default:
255: gly = ESCAPE_SPECIAL;
1.65 schwarze 256: *start = --*end;
257: *sz = 1;
1.22 kristaps 258: break;
1.45 kristaps 259: }
260:
261: assert(ESCAPE_ERROR != gly);
262:
263: /*
1.64 schwarze 264: * Read up to the terminating character,
265: * paying attention to nested escapes.
1.45 kristaps 266: */
267:
268: if ('\0' != term) {
1.64 schwarze 269: while (**end != term) {
270: switch (**end) {
1.79 schwarze 271: case '\0':
1.64 schwarze 272: return(ESCAPE_ERROR);
1.79 schwarze 273: case '\\':
1.64 schwarze 274: (*end)++;
275: if (ESCAPE_ERROR ==
276: mandoc_escape(end, NULL, NULL))
277: return(ESCAPE_ERROR);
278: break;
279: default:
280: (*end)++;
281: break;
282: }
283: }
1.65 schwarze 284: *sz = (*end)++ - *start;
1.64 schwarze 285: } else {
1.65 schwarze 286: assert(*sz > 0);
287: if ((size_t)*sz > strlen(*start))
1.45 kristaps 288: return(ESCAPE_ERROR);
1.65 schwarze 289: *end += *sz;
1.45 kristaps 290: }
291:
292: /* Run post-processors. */
293:
294: switch (gly) {
1.79 schwarze 295: case ESCAPE_FONT:
1.68 schwarze 296: if (2 == *sz) {
297: if ('C' == **start) {
298: /*
299: * Treat constant-width font modes
300: * just like regular font modes.
301: */
302: (*start)++;
303: (*sz)--;
304: } else {
305: if ('B' == (*start)[0] && 'I' == (*start)[1])
306: gly = ESCAPE_FONTBI;
307: break;
308: }
1.65 schwarze 309: } else if (1 != *sz)
1.45 kristaps 310: break;
1.61 kristaps 311:
1.65 schwarze 312: switch (**start) {
1.79 schwarze 313: case '3':
1.45 kristaps 314: /* FALLTHROUGH */
1.79 schwarze 315: case 'B':
1.45 kristaps 316: gly = ESCAPE_FONTBOLD;
317: break;
1.79 schwarze 318: case '2':
1.45 kristaps 319: /* FALLTHROUGH */
1.79 schwarze 320: case 'I':
1.45 kristaps 321: gly = ESCAPE_FONTITALIC;
1.22 kristaps 322: break;
1.79 schwarze 323: case 'P':
1.45 kristaps 324: gly = ESCAPE_FONTPREV;
1.22 kristaps 325: break;
1.79 schwarze 326: case '1':
1.45 kristaps 327: /* FALLTHROUGH */
1.79 schwarze 328: case 'R':
1.45 kristaps 329: gly = ESCAPE_FONTROMAN;
1.1 kristaps 330: break;
331: }
1.46 kristaps 332: break;
1.79 schwarze 333: case ESCAPE_SPECIAL:
1.65 schwarze 334: if (1 == *sz && 'c' == **start)
1.45 kristaps 335: gly = ESCAPE_NOSPACE;
1.87 ! schwarze 336: /*
! 337: * Unicode escapes are defined in groff as \[uXXXX]
! 338: * to \[u10FFFF], where the contained value must be
! 339: * a valid Unicode codepoint. Here, however, only
! 340: * check the length and the validity of all digits.
! 341: */
! 342: else if (*sz > 4 && *sz < 8 && **start == 'u' &&
! 343: (int)strspn(*start + 1, "0123456789ABCDEFabcdef")
! 344: + 1 == *sz)
! 345: gly = ESCAPE_UNICODE;
1.22 kristaps 346: break;
1.1 kristaps 347: default:
1.22 kristaps 348: break;
1.1 kristaps 349: }
350:
1.45 kristaps 351: return(gly);
1.36 schwarze 352: }
353:
354: /*
355: * Parse a quoted or unquoted roff-style request or macro argument.
356: * Return a pointer to the parsed argument, which is either the original
357: * pointer or advanced by one byte in case the argument is quoted.
1.71 schwarze 358: * NUL-terminate the argument in place.
1.36 schwarze 359: * Collapse pairs of quotes inside quoted arguments.
360: * Advance the argument pointer to the next argument,
1.71 schwarze 361: * or to the NUL byte terminating the argument line.
1.36 schwarze 362: */
363: char *
1.48 kristaps 364: mandoc_getarg(struct mparse *parse, char **cpp, int ln, int *pos)
1.36 schwarze 365: {
366: char *start, *cp;
367: int quoted, pairs, white;
368:
369: /* Quoting can only start with a new word. */
370: start = *cpp;
1.47 kristaps 371: quoted = 0;
1.36 schwarze 372: if ('"' == *start) {
373: quoted = 1;
374: start++;
1.79 schwarze 375: }
1.36 schwarze 376:
377: pairs = 0;
378: white = 0;
379: for (cp = start; '\0' != *cp; cp++) {
1.67 schwarze 380:
381: /*
382: * Move the following text left
383: * after quoted quotes and after "\\" and "\t".
384: */
1.36 schwarze 385: if (pairs)
386: cp[-pairs] = cp[0];
1.67 schwarze 387:
1.36 schwarze 388: if ('\\' == cp[0]) {
1.67 schwarze 389: /*
390: * In copy mode, translate double to single
391: * backslashes and backslash-t to literal tabs.
392: */
393: switch (cp[1]) {
1.79 schwarze 394: case 't':
1.67 schwarze 395: cp[0] = '\t';
396: /* FALLTHROUGH */
1.79 schwarze 397: case '\\':
1.36 schwarze 398: pairs++;
399: cp++;
1.67 schwarze 400: break;
1.79 schwarze 401: case ' ':
1.36 schwarze 402: /* Skip escaped blanks. */
1.67 schwarze 403: if (0 == quoted)
404: cp++;
405: break;
406: default:
407: break;
408: }
1.36 schwarze 409: } else if (0 == quoted) {
410: if (' ' == cp[0]) {
411: /* Unescaped blanks end unquoted args. */
412: white = 1;
413: break;
414: }
415: } else if ('"' == cp[0]) {
416: if ('"' == cp[1]) {
417: /* Quoted quotes collapse. */
418: pairs++;
419: cp++;
420: } else {
421: /* Unquoted quotes end quoted args. */
422: quoted = 2;
423: break;
424: }
425: }
426: }
427:
428: /* Quoted argument without a closing quote. */
1.48 kristaps 429: if (1 == quoted)
1.83 schwarze 430: mandoc_msg(MANDOCERR_ARG_QUOTE, parse, ln, *pos, NULL);
1.36 schwarze 431:
1.71 schwarze 432: /* NUL-terminate this argument and move to the next one. */
1.36 schwarze 433: if (pairs)
434: cp[-pairs] = '\0';
435: if ('\0' != *cp) {
436: *cp++ = '\0';
437: while (' ' == *cp)
438: cp++;
439: }
1.39 kristaps 440: *pos += (int)(cp - start) + (quoted ? 1 : 0);
1.36 schwarze 441: *cpp = cp;
442:
1.48 kristaps 443: if ('\0' == *cp && (white || ' ' == cp[-1]))
1.83 schwarze 444: mandoc_msg(MANDOCERR_SPACE_EOL, parse, ln, *pos, NULL);
1.36 schwarze 445:
446: return(start);
1.4 kristaps 447: }
1.7 kristaps 448:
449: static int
450: a2time(time_t *t, const char *fmt, const char *p)
451: {
452: struct tm tm;
453: char *pp;
454:
455: memset(&tm, 0, sizeof(struct tm));
456:
1.56 kristaps 457: pp = NULL;
1.85 schwarze 458: #if HAVE_STRPTIME
1.7 kristaps 459: pp = strptime(p, fmt, &tm);
1.56 kristaps 460: #endif
1.7 kristaps 461: if (NULL != pp && '\0' == *pp) {
462: *t = mktime(&tm);
463: return(1);
464: }
465:
466: return(0);
467: }
468:
1.37 schwarze 469: static char *
470: time2a(time_t t)
471: {
1.56 kristaps 472: struct tm *tm;
1.38 schwarze 473: char *buf, *p;
474: size_t ssz;
1.37 schwarze 475: int isz;
476:
1.56 kristaps 477: tm = localtime(&t);
1.37 schwarze 478:
1.38 schwarze 479: /*
480: * Reserve space:
481: * up to 9 characters for the month (September) + blank
482: * up to 2 characters for the day + comma + blank
483: * 4 characters for the year and a terminating '\0'
484: */
485: p = buf = mandoc_malloc(10 + 4 + 4 + 1);
486:
1.56 kristaps 487: if (0 == (ssz = strftime(p, 10 + 1, "%B ", tm)))
1.38 schwarze 488: goto fail;
489: p += (int)ssz;
1.37 schwarze 490:
1.56 kristaps 491: if (-1 == (isz = snprintf(p, 4 + 1, "%d, ", tm->tm_mday)))
1.38 schwarze 492: goto fail;
1.37 schwarze 493: p += isz;
494:
1.56 kristaps 495: if (0 == strftime(p, 4 + 1, "%Y", tm))
1.38 schwarze 496: goto fail;
497: return(buf);
498:
499: fail:
500: free(buf);
501: return(NULL);
1.37 schwarze 502: }
503:
504: char *
1.42 kristaps 505: mandoc_normdate(struct mparse *parse, char *in, int ln, int pos)
1.7 kristaps 506: {
1.37 schwarze 507: char *out;
1.7 kristaps 508: time_t t;
509:
1.37 schwarze 510: if (NULL == in || '\0' == *in ||
511: 0 == strcmp(in, "$" "Mdocdate$")) {
1.80 schwarze 512: mandoc_msg(MANDOCERR_DATE_MISSING, parse, ln, pos, NULL);
1.37 schwarze 513: time(&t);
514: }
1.62 schwarze 515: else if (a2time(&t, "%Y-%m-%d", in))
516: t = 0;
1.37 schwarze 517: else if (!a2time(&t, "$" "Mdocdate: %b %d %Y $", in) &&
1.62 schwarze 518: !a2time(&t, "%b %d, %Y", in)) {
1.81 schwarze 519: mandoc_msg(MANDOCERR_DATE_BAD, parse, ln, pos, in);
1.37 schwarze 520: t = 0;
1.7 kristaps 521: }
1.37 schwarze 522: out = t ? time2a(t) : NULL;
1.38 schwarze 523: return(out ? out : mandoc_strdup(in));
1.7 kristaps 524: }
525:
1.12 kristaps 526: int
1.75 schwarze 527: mandoc_eos(const char *p, size_t sz)
1.12 kristaps 528: {
1.75 schwarze 529: const char *q;
530: int enclosed, found;
1.12 kristaps 531:
1.13 kristaps 532: if (0 == sz)
533: return(0);
1.12 kristaps 534:
1.14 kristaps 535: /*
536: * End-of-sentence recognition must include situations where
537: * some symbols, such as `)', allow prior EOS punctuation to
1.49 kristaps 538: * propagate outward.
1.14 kristaps 539: */
540:
1.75 schwarze 541: enclosed = found = 0;
1.25 kristaps 542: for (q = p + (int)sz - 1; q >= p; q--) {
1.23 schwarze 543: switch (*q) {
1.79 schwarze 544: case '\"':
1.14 kristaps 545: /* FALLTHROUGH */
1.79 schwarze 546: case '\'':
1.15 kristaps 547: /* FALLTHROUGH */
1.79 schwarze 548: case ']':
1.14 kristaps 549: /* FALLTHROUGH */
1.79 schwarze 550: case ')':
1.23 schwarze 551: if (0 == found)
552: enclosed = 1;
1.14 kristaps 553: break;
1.79 schwarze 554: case '.':
1.14 kristaps 555: /* FALLTHROUGH */
1.79 schwarze 556: case '!':
1.14 kristaps 557: /* FALLTHROUGH */
1.79 schwarze 558: case '?':
1.23 schwarze 559: found = 1;
560: break;
1.14 kristaps 561: default:
1.27 joerg 562: return(found && (!enclosed || isalnum((unsigned char)*q)));
1.14 kristaps 563: }
1.12 kristaps 564: }
565:
1.23 schwarze 566: return(found && !enclosed);
1.44 kristaps 567: }
1.50 kristaps 568:
569: /*
570: * Convert a string to a long that may not be <0.
571: * If the string is invalid, or is less than 0, return -1.
572: */
573: int
1.54 kristaps 574: mandoc_strntoi(const char *p, size_t sz, int base)
1.50 kristaps 575: {
576: char buf[32];
577: char *ep;
578: long v;
579:
580: if (sz > 31)
581: return(-1);
582:
583: memcpy(buf, p, sz);
1.51 kristaps 584: buf[(int)sz] = '\0';
1.50 kristaps 585:
586: errno = 0;
587: v = strtol(buf, &ep, base);
588:
589: if (buf[0] == '\0' || *ep != '\0')
590: return(-1);
591:
1.54 kristaps 592: if (v > INT_MAX)
593: v = INT_MAX;
594: if (v < INT_MIN)
595: v = INT_MIN;
1.50 kristaps 596:
597: return((int)v);
598: }
CVSweb