Annotation of src/usr.bin/mandoc/mandoc.c, Revision 1.71
1.71 ! schwarze 1: /* $OpenBSD: mandoc.c,v 1.70 2017/06/14 01:31:19 schwarze Exp $ */
1.1 schwarze 2: /*
1.58 schwarze 3: * Copyright (c) 2008-2011, 2014 Kristaps Dzonsons <kristaps@bsd.lv>
1.68 schwarze 4: * Copyright (c) 2011-2015, 2017 Ingo Schwarze <schwarze@openbsd.org>
1.1 schwarze 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: *
1.21 schwarze 10: * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHORS DISCLAIM ALL WARRANTIES
1.1 schwarze 11: * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
1.21 schwarze 12: * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHORS BE LIABLE FOR
1.1 schwarze 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: */
1.2 schwarze 18: #include <sys/types.h>
19:
1.1 schwarze 20: #include <assert.h>
21: #include <ctype.h>
1.26 schwarze 22: #include <errno.h>
23: #include <limits.h>
1.1 schwarze 24: #include <stdlib.h>
1.4 schwarze 25: #include <stdio.h>
26: #include <string.h>
1.5 schwarze 27: #include <time.h>
1.1 schwarze 28:
1.69 schwarze 29: #include "mandoc_aux.h"
1.14 schwarze 30: #include "mandoc.h"
1.69 schwarze 31: #include "roff.h"
1.1 schwarze 32: #include "libmandoc.h"
1.22 schwarze 33:
1.14 schwarze 34: static int a2time(time_t *, const char *, const char *);
1.22 schwarze 35: static char *time2a(time_t);
1.5 schwarze 36:
1.26 schwarze 37:
38: enum mandoc_esc
1.44 schwarze 39: mandoc_escape(const char **end, const char **start, int *sz)
1.26 schwarze 40: {
1.34 schwarze 41: const char *local_start;
42: int local_sz;
43: char term;
1.48 schwarze 44: enum mandoc_esc gly;
1.26 schwarze 45:
1.34 schwarze 46: /*
47: * When the caller doesn't provide return storage,
48: * use local storage.
49: */
50:
51: if (NULL == start)
52: start = &local_start;
53: if (NULL == sz)
54: sz = &local_sz;
55:
56: /*
57: * Beyond the backslash, at least one input character
58: * is part of the escape sequence. With one exception
59: * (see below), that character won't be returned.
60: */
61:
1.26 schwarze 62: gly = ESCAPE_ERROR;
1.34 schwarze 63: *start = ++*end;
64: *sz = 0;
1.33 schwarze 65: term = '\0';
1.26 schwarze 66:
1.34 schwarze 67: switch ((*start)[-1]) {
1.26 schwarze 68: /*
69: * First the glyphs. There are several different forms of
70: * these, but each eventually returns a substring of the glyph
71: * name.
72: */
1.48 schwarze 73: case '(':
1.26 schwarze 74: gly = ESCAPE_SPECIAL;
1.34 schwarze 75: *sz = 2;
1.26 schwarze 76: break;
1.48 schwarze 77: case '[':
1.26 schwarze 78: gly = ESCAPE_SPECIAL;
79: term = ']';
80: break;
1.48 schwarze 81: case 'C':
1.34 schwarze 82: if ('\'' != **start)
1.62 schwarze 83: return ESCAPE_ERROR;
1.34 schwarze 84: *start = ++*end;
1.54 schwarze 85: gly = ESCAPE_SPECIAL;
1.26 schwarze 86: term = '\'';
87: break;
1.41 schwarze 88:
89: /*
90: * Escapes taking no arguments at all.
91: */
1.48 schwarze 92: case 'd':
93: case 'u':
1.61 schwarze 94: case ',':
95: case '/':
1.62 schwarze 96: return ESCAPE_IGNORE;
1.70 schwarze 97: case 'p':
98: return ESCAPE_BREAK;
1.32 schwarze 99:
100: /*
101: * The \z escape is supposed to output the following
1.48 schwarze 102: * character without advancing the cursor position.
1.32 schwarze 103: * Since we are mostly dealing with terminal mode,
104: * let us just skip the next character.
105: */
1.48 schwarze 106: case 'z':
1.62 schwarze 107: return ESCAPE_SKIPCHAR;
1.1 schwarze 108:
1.26 schwarze 109: /*
110: * Handle all triggers matching \X(xy, \Xx, and \X[xxxx], where
111: * 'X' is the trigger. These have opaque sub-strings.
112: */
1.48 schwarze 113: case 'F':
114: case 'g':
115: case 'k':
116: case 'M':
117: case 'm':
118: case 'n':
119: case 'V':
120: case 'Y':
1.29 schwarze 121: gly = ESCAPE_IGNORE;
1.1 schwarze 122: /* FALLTHROUGH */
1.48 schwarze 123: case 'f':
1.26 schwarze 124: if (ESCAPE_ERROR == gly)
125: gly = ESCAPE_FONT;
1.34 schwarze 126: switch (**start) {
1.48 schwarze 127: case '(':
1.34 schwarze 128: *start = ++*end;
129: *sz = 2;
1.26 schwarze 130: break;
1.48 schwarze 131: case '[':
1.34 schwarze 132: *start = ++*end;
1.26 schwarze 133: term = ']';
134: break;
135: default:
1.34 schwarze 136: *sz = 1;
1.26 schwarze 137: break;
138: }
139: break;
140:
141: /*
142: * These escapes are of the form \X'Y', where 'X' is the trigger
143: * and 'Y' is any string. These have opaque sub-strings.
1.47 schwarze 144: * The \B and \w escapes are handled in roff.c, roff_res().
1.26 schwarze 145: */
1.48 schwarze 146: case 'A':
147: case 'b':
148: case 'D':
149: case 'R':
150: case 'X':
151: case 'Z':
1.59 schwarze 152: gly = ESCAPE_IGNORE;
153: /* FALLTHROUGH */
154: case 'o':
155: if (**start == '\0')
1.62 schwarze 156: return ESCAPE_ERROR;
1.59 schwarze 157: if (gly == ESCAPE_ERROR)
158: gly = ESCAPE_OVERSTRIKE;
1.46 schwarze 159: term = **start;
1.34 schwarze 160: *start = ++*end;
1.16 schwarze 161: break;
1.26 schwarze 162:
163: /*
164: * These escapes are of the form \X'N', where 'X' is the trigger
165: * and 'N' resolves to a numerical expression.
166: */
1.48 schwarze 167: case 'h':
168: case 'H':
169: case 'L':
170: case 'l':
171: case 'S':
172: case 'v':
173: case 'x':
1.51 schwarze 174: if (strchr(" %&()*+-./0123456789:<=>", **start)) {
1.53 schwarze 175: if ('\0' != **start)
176: ++*end;
1.62 schwarze 177: return ESCAPE_ERROR;
1.51 schwarze 178: }
1.68 schwarze 179: switch ((*start)[-1]) {
180: case 'h':
181: gly = ESCAPE_HORIZ;
182: break;
183: case 'l':
184: gly = ESCAPE_HLINE;
185: break;
186: default:
187: gly = ESCAPE_IGNORE;
188: break;
189: }
1.46 schwarze 190: term = **start;
1.34 schwarze 191: *start = ++*end;
1.26 schwarze 192: break;
1.29 schwarze 193:
194: /*
195: * Special handling for the numbered character escape.
196: * XXX Do any other escapes need similar handling?
197: */
1.48 schwarze 198: case 'N':
1.34 schwarze 199: if ('\0' == **start)
1.62 schwarze 200: return ESCAPE_ERROR;
1.34 schwarze 201: (*end)++;
202: if (isdigit((unsigned char)**start)) {
203: *sz = 1;
1.62 schwarze 204: return ESCAPE_IGNORE;
1.34 schwarze 205: }
206: (*start)++;
1.29 schwarze 207: while (isdigit((unsigned char)**end))
208: (*end)++;
1.34 schwarze 209: *sz = *end - *start;
1.29 schwarze 210: if ('\0' != **end)
211: (*end)++;
1.62 schwarze 212: return ESCAPE_NUMBERED;
1.26 schwarze 213:
1.48 schwarze 214: /*
1.26 schwarze 215: * Sizes get a special category of their own.
216: */
1.48 schwarze 217: case 's':
1.26 schwarze 218: gly = ESCAPE_IGNORE;
1.17 schwarze 219:
1.26 schwarze 220: /* See +/- counts as a sign. */
1.34 schwarze 221: if ('+' == **end || '-' == **end || ASCII_HYPH == **end)
1.58 schwarze 222: *start = ++*end;
1.6 schwarze 223:
1.34 schwarze 224: switch (**end) {
1.48 schwarze 225: case '(':
1.34 schwarze 226: *start = ++*end;
227: *sz = 2;
1.16 schwarze 228: break;
1.48 schwarze 229: case '[':
1.34 schwarze 230: *start = ++*end;
1.33 schwarze 231: term = ']';
1.16 schwarze 232: break;
1.48 schwarze 233: case '\'':
1.34 schwarze 234: *start = ++*end;
1.33 schwarze 235: term = '\'';
1.60 schwarze 236: break;
237: case '3':
238: case '2':
239: case '1':
240: *sz = (*end)[-1] == 's' &&
241: isdigit((unsigned char)(*end)[1]) ? 2 : 1;
1.16 schwarze 242: break;
243: default:
1.34 schwarze 244: *sz = 1;
1.16 schwarze 245: break;
1.6 schwarze 246: }
247:
1.26 schwarze 248: break;
249:
250: /*
251: * Anything else is assumed to be a glyph.
1.34 schwarze 252: * In this case, pass back the character after the backslash.
1.26 schwarze 253: */
254: default:
255: gly = ESCAPE_SPECIAL;
1.34 schwarze 256: *start = --*end;
257: *sz = 1;
1.26 schwarze 258: break;
259: }
260:
261: assert(ESCAPE_ERROR != gly);
262:
263: /*
1.33 schwarze 264: * Read up to the terminating character,
265: * paying attention to nested escapes.
1.26 schwarze 266: */
267:
268: if ('\0' != term) {
1.33 schwarze 269: while (**end != term) {
270: switch (**end) {
1.48 schwarze 271: case '\0':
1.62 schwarze 272: return ESCAPE_ERROR;
1.48 schwarze 273: case '\\':
1.33 schwarze 274: (*end)++;
275: if (ESCAPE_ERROR ==
276: mandoc_escape(end, NULL, NULL))
1.62 schwarze 277: return ESCAPE_ERROR;
1.33 schwarze 278: break;
279: default:
280: (*end)++;
281: break;
282: }
283: }
1.34 schwarze 284: *sz = (*end)++ - *start;
1.33 schwarze 285: } else {
1.34 schwarze 286: assert(*sz > 0);
287: if ((size_t)*sz > strlen(*start))
1.62 schwarze 288: return ESCAPE_ERROR;
1.34 schwarze 289: *end += *sz;
1.26 schwarze 290: }
1.19 schwarze 291:
1.26 schwarze 292: /* Run post-processors. */
1.19 schwarze 293:
1.26 schwarze 294: switch (gly) {
1.48 schwarze 295: case ESCAPE_FONT:
1.37 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: }
1.34 schwarze 309: } else if (1 != *sz)
1.26 schwarze 310: break;
1.30 schwarze 311:
1.34 schwarze 312: switch (**start) {
1.48 schwarze 313: case '3':
314: case 'B':
1.26 schwarze 315: gly = ESCAPE_FONTBOLD;
316: break;
1.48 schwarze 317: case '2':
318: case 'I':
1.26 schwarze 319: gly = ESCAPE_FONTITALIC;
1.16 schwarze 320: break;
1.48 schwarze 321: case 'P':
1.26 schwarze 322: gly = ESCAPE_FONTPREV;
1.16 schwarze 323: break;
1.48 schwarze 324: case '1':
325: case 'R':
1.26 schwarze 326: gly = ESCAPE_FONTROMAN;
1.1 schwarze 327: break;
328: }
1.16 schwarze 329: break;
1.48 schwarze 330: case ESCAPE_SPECIAL:
1.34 schwarze 331: if (1 == *sz && 'c' == **start)
1.26 schwarze 332: gly = ESCAPE_NOSPACE;
1.54 schwarze 333: /*
1.55 schwarze 334: * Unicode escapes are defined in groff as \[u0000]
1.54 schwarze 335: * to \[u10FFFF], where the contained value must be
336: * a valid Unicode codepoint. Here, however, only
1.55 schwarze 337: * check the length and range.
1.54 schwarze 338: */
1.55 schwarze 339: if (**start != 'u' || *sz < 5 || *sz > 7)
340: break;
341: if (*sz == 7 && ((*start)[1] != '1' || (*start)[2] != '0'))
342: break;
343: if (*sz == 6 && (*start)[1] == '0')
1.64 schwarze 344: break;
345: if (*sz == 5 && (*start)[1] == 'D' &&
346: strchr("89ABCDEF", (*start)[2]) != NULL)
1.55 schwarze 347: break;
348: if ((int)strspn(*start + 1, "0123456789ABCDEFabcdef")
1.54 schwarze 349: + 1 == *sz)
350: gly = ESCAPE_UNICODE;
1.16 schwarze 351: break;
1.1 schwarze 352: default:
1.16 schwarze 353: break;
1.1 schwarze 354: }
355:
1.62 schwarze 356: return gly;
1.21 schwarze 357: }
358:
359: /*
360: * Parse a quoted or unquoted roff-style request or macro argument.
361: * Return a pointer to the parsed argument, which is either the original
362: * pointer or advanced by one byte in case the argument is quoted.
1.40 schwarze 363: * NUL-terminate the argument in place.
1.21 schwarze 364: * Collapse pairs of quotes inside quoted arguments.
365: * Advance the argument pointer to the next argument,
1.40 schwarze 366: * or to the NUL byte terminating the argument line.
1.21 schwarze 367: */
368: char *
1.25 schwarze 369: mandoc_getarg(struct mparse *parse, char **cpp, int ln, int *pos)
1.21 schwarze 370: {
371: char *start, *cp;
372: int quoted, pairs, white;
373:
374: /* Quoting can only start with a new word. */
375: start = *cpp;
1.26 schwarze 376: quoted = 0;
1.21 schwarze 377: if ('"' == *start) {
378: quoted = 1;
379: start++;
1.48 schwarze 380: }
1.21 schwarze 381:
382: pairs = 0;
383: white = 0;
384: for (cp = start; '\0' != *cp; cp++) {
1.36 schwarze 385:
386: /*
387: * Move the following text left
388: * after quoted quotes and after "\\" and "\t".
389: */
1.21 schwarze 390: if (pairs)
391: cp[-pairs] = cp[0];
1.36 schwarze 392:
1.21 schwarze 393: if ('\\' == cp[0]) {
1.36 schwarze 394: /*
395: * In copy mode, translate double to single
396: * backslashes and backslash-t to literal tabs.
397: */
398: switch (cp[1]) {
1.48 schwarze 399: case 't':
1.36 schwarze 400: cp[0] = '\t';
401: /* FALLTHROUGH */
1.48 schwarze 402: case '\\':
1.21 schwarze 403: pairs++;
404: cp++;
1.36 schwarze 405: break;
1.48 schwarze 406: case ' ':
1.21 schwarze 407: /* Skip escaped blanks. */
1.36 schwarze 408: if (0 == quoted)
409: cp++;
410: break;
411: default:
412: break;
413: }
1.21 schwarze 414: } else if (0 == quoted) {
415: if (' ' == cp[0]) {
416: /* Unescaped blanks end unquoted args. */
417: white = 1;
418: break;
419: }
420: } else if ('"' == cp[0]) {
421: if ('"' == cp[1]) {
422: /* Quoted quotes collapse. */
423: pairs++;
424: cp++;
425: } else {
426: /* Unquoted quotes end quoted args. */
427: quoted = 2;
428: break;
429: }
430: }
431: }
432:
433: /* Quoted argument without a closing quote. */
1.25 schwarze 434: if (1 == quoted)
1.52 schwarze 435: mandoc_msg(MANDOCERR_ARG_QUOTE, parse, ln, *pos, NULL);
1.21 schwarze 436:
1.40 schwarze 437: /* NUL-terminate this argument and move to the next one. */
1.21 schwarze 438: if (pairs)
439: cp[-pairs] = '\0';
440: if ('\0' != *cp) {
441: *cp++ = '\0';
442: while (' ' == *cp)
443: cp++;
444: }
1.24 schwarze 445: *pos += (int)(cp - start) + (quoted ? 1 : 0);
1.21 schwarze 446: *cpp = cp;
447:
1.25 schwarze 448: if ('\0' == *cp && (white || ' ' == cp[-1]))
1.52 schwarze 449: mandoc_msg(MANDOCERR_SPACE_EOL, parse, ln, *pos, NULL);
1.21 schwarze 450:
1.62 schwarze 451: return start;
1.4 schwarze 452: }
1.5 schwarze 453:
454: static int
455: a2time(time_t *t, const char *fmt, const char *p)
456: {
457: struct tm tm;
458: char *pp;
459:
460: memset(&tm, 0, sizeof(struct tm));
461:
462: pp = strptime(p, fmt, &tm);
463: if (NULL != pp && '\0' == *pp) {
464: *t = mktime(&tm);
1.62 schwarze 465: return 1;
1.5 schwarze 466: }
467:
1.62 schwarze 468: return 0;
1.5 schwarze 469: }
470:
1.22 schwarze 471: static char *
472: time2a(time_t t)
473: {
1.28 schwarze 474: struct tm *tm;
1.23 schwarze 475: char *buf, *p;
476: size_t ssz;
1.22 schwarze 477: int isz;
478:
1.28 schwarze 479: tm = localtime(&t);
1.57 schwarze 480: if (tm == NULL)
1.62 schwarze 481: return NULL;
1.22 schwarze 482:
1.23 schwarze 483: /*
484: * Reserve space:
485: * up to 9 characters for the month (September) + blank
486: * up to 2 characters for the day + comma + blank
487: * 4 characters for the year and a terminating '\0'
488: */
1.66 schwarze 489:
1.23 schwarze 490: p = buf = mandoc_malloc(10 + 4 + 4 + 1);
491:
1.66 schwarze 492: if ((ssz = strftime(p, 10 + 1, "%B ", tm)) == 0)
1.23 schwarze 493: goto fail;
494: p += (int)ssz;
1.22 schwarze 495:
1.66 schwarze 496: /*
497: * The output format is just "%d" here, not "%2d" or "%02d".
498: * That's also the reason why we can't just format the
499: * date as a whole with "%B %e, %Y" or "%B %d, %Y".
500: * Besides, the present approach is less prone to buffer
501: * overflows, in case anybody should ever introduce the bug
502: * of looking at LC_TIME.
503: */
504:
505: if ((isz = snprintf(p, 4 + 1, "%d, ", tm->tm_mday)) == -1)
1.23 schwarze 506: goto fail;
1.22 schwarze 507: p += isz;
508:
1.66 schwarze 509: if (strftime(p, 4 + 1, "%Y", tm) == 0)
1.23 schwarze 510: goto fail;
1.62 schwarze 511: return buf;
1.23 schwarze 512:
513: fail:
514: free(buf);
1.62 schwarze 515: return NULL;
1.22 schwarze 516: }
517:
518: char *
1.69 schwarze 519: mandoc_normdate(struct roff_man *man, char *in, int ln, int pos)
1.5 schwarze 520: {
1.71 ! schwarze 521: char *cp;
1.5 schwarze 522: time_t t;
523:
1.66 schwarze 524: /* No date specified: use today's date. */
525:
526: if (in == NULL || *in == '\0' || strcmp(in, "$" "Mdocdate$") == 0) {
1.69 schwarze 527: mandoc_msg(MANDOCERR_DATE_MISSING, man->parse, ln, pos, NULL);
1.66 schwarze 528: return time2a(time(NULL));
1.22 schwarze 529: }
1.66 schwarze 530:
531: /* Valid mdoc(7) date format. */
532:
533: if (a2time(&t, "$" "Mdocdate: %b %d %Y $", in) ||
1.71 ! schwarze 534: a2time(&t, "%b %d, %Y", in)) {
! 535: cp = time2a(t);
! 536: if (t > time(NULL) + 86400)
! 537: mandoc_msg(MANDOCERR_DATE_FUTURE, man->parse,
! 538: ln, pos, cp);
! 539: return cp;
! 540: }
1.66 schwarze 541:
1.69 schwarze 542: /* In man(7), do not warn about the legacy format. */
1.66 schwarze 543:
1.69 schwarze 544: if (a2time(&t, "%Y-%m-%d", in) == 0)
545: mandoc_msg(MANDOCERR_DATE_BAD, man->parse, ln, pos, in);
1.71 ! schwarze 546: else if (t > time(NULL) + 86400)
! 547: mandoc_msg(MANDOCERR_DATE_FUTURE, man->parse, ln, pos, in);
1.69 schwarze 548: else if (man->macroset == MACROSET_MDOC)
549: mandoc_vmsg(MANDOCERR_DATE_LEGACY, man->parse,
550: ln, pos, "Dd %s", in);
1.66 schwarze 551:
552: /* Use any non-mdoc(7) date verbatim. */
553:
554: return mandoc_strdup(in);
1.5 schwarze 555: }
556:
1.9 schwarze 557: int
1.43 schwarze 558: mandoc_eos(const char *p, size_t sz)
1.9 schwarze 559: {
1.43 schwarze 560: const char *q;
561: int enclosed, found;
1.9 schwarze 562:
1.10 schwarze 563: if (0 == sz)
1.62 schwarze 564: return 0;
1.9 schwarze 565:
1.11 schwarze 566: /*
567: * End-of-sentence recognition must include situations where
568: * some symbols, such as `)', allow prior EOS punctuation to
1.26 schwarze 569: * propagate outward.
1.11 schwarze 570: */
571:
1.43 schwarze 572: enclosed = found = 0;
1.16 schwarze 573: for (q = p + (int)sz - 1; q >= p; q--) {
1.15 schwarze 574: switch (*q) {
1.48 schwarze 575: case '\"':
576: case '\'':
577: case ']':
578: case ')':
1.15 schwarze 579: if (0 == found)
580: enclosed = 1;
1.11 schwarze 581: break;
1.48 schwarze 582: case '.':
583: case '!':
584: case '?':
1.15 schwarze 585: found = 1;
586: break;
1.11 schwarze 587: default:
1.62 schwarze 588: return found &&
589: (!enclosed || isalnum((unsigned char)*q));
1.11 schwarze 590: }
1.9 schwarze 591: }
592:
1.62 schwarze 593: return found && !enclosed;
1.9 schwarze 594: }
1.26 schwarze 595:
596: /*
597: * Convert a string to a long that may not be <0.
598: * If the string is invalid, or is less than 0, return -1.
599: */
600: int
1.27 schwarze 601: mandoc_strntoi(const char *p, size_t sz, int base)
1.26 schwarze 602: {
603: char buf[32];
604: char *ep;
605: long v;
606:
607: if (sz > 31)
1.62 schwarze 608: return -1;
1.26 schwarze 609:
610: memcpy(buf, p, sz);
611: buf[(int)sz] = '\0';
612:
613: errno = 0;
614: v = strtol(buf, &ep, base);
615:
616: if (buf[0] == '\0' || *ep != '\0')
1.62 schwarze 617: return -1;
1.26 schwarze 618:
1.27 schwarze 619: if (v > INT_MAX)
620: v = INT_MAX;
621: if (v < INT_MIN)
622: v = INT_MIN;
1.26 schwarze 623:
1.62 schwarze 624: return (int)v;
1.26 schwarze 625: }