Annotation of src/usr.bin/mandoc/mandoc.c, Revision 1.56
1.56 ! schwarze 1: /* $OpenBSD$ */
1.1 schwarze 2: /*
1.24 schwarze 3: * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
1.47 schwarze 4: * Copyright (c) 2011, 2012, 2013, 2014 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:
1.14 schwarze 29: #include "mandoc.h"
1.45 schwarze 30: #include "mandoc_aux.h"
1.1 schwarze 31: #include "libmandoc.h"
32:
1.22 schwarze 33: #define DATESIZE 32
34:
1.14 schwarze 35: static int a2time(time_t *, const char *, const char *);
1.22 schwarze 36: static char *time2a(time_t);
1.5 schwarze 37:
1.26 schwarze 38:
39: enum mandoc_esc
1.44 schwarze 40: mandoc_escape(const char **end, const char **start, int *sz)
1.26 schwarze 41: {
1.34 schwarze 42: const char *local_start;
43: int local_sz;
44: char term;
1.48 schwarze 45: enum mandoc_esc gly;
1.26 schwarze 46:
1.34 schwarze 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:
1.26 schwarze 63: gly = ESCAPE_ERROR;
1.34 schwarze 64: *start = ++*end;
65: *sz = 0;
1.33 schwarze 66: term = '\0';
1.26 schwarze 67:
1.34 schwarze 68: switch ((*start)[-1]) {
1.26 schwarze 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: */
1.48 schwarze 74: case '(':
1.26 schwarze 75: gly = ESCAPE_SPECIAL;
1.34 schwarze 76: *sz = 2;
1.26 schwarze 77: break;
1.48 schwarze 78: case '[':
1.26 schwarze 79: gly = ESCAPE_SPECIAL;
80: term = ']';
81: break;
1.48 schwarze 82: case 'C':
1.34 schwarze 83: if ('\'' != **start)
1.26 schwarze 84: return(ESCAPE_ERROR);
1.34 schwarze 85: *start = ++*end;
1.54 schwarze 86: gly = ESCAPE_SPECIAL;
1.26 schwarze 87: term = '\'';
88: break;
1.41 schwarze 89:
90: /*
91: * Escapes taking no arguments at all.
92: */
1.48 schwarze 93: case 'd':
1.41 schwarze 94: /* FALLTHROUGH */
1.48 schwarze 95: case 'u':
1.41 schwarze 96: return(ESCAPE_IGNORE);
1.32 schwarze 97:
98: /*
99: * The \z escape is supposed to output the following
1.48 schwarze 100: * character without advancing the cursor position.
1.32 schwarze 101: * Since we are mostly dealing with terminal mode,
102: * let us just skip the next character.
103: */
1.48 schwarze 104: case 'z':
1.32 schwarze 105: return(ESCAPE_SKIPCHAR);
1.1 schwarze 106:
1.26 schwarze 107: /*
108: * Handle all triggers matching \X(xy, \Xx, and \X[xxxx], where
109: * 'X' is the trigger. These have opaque sub-strings.
110: */
1.48 schwarze 111: case 'F':
1.16 schwarze 112: /* FALLTHROUGH */
1.48 schwarze 113: case 'g':
1.16 schwarze 114: /* FALLTHROUGH */
1.48 schwarze 115: case 'k':
1.1 schwarze 116: /* FALLTHROUGH */
1.48 schwarze 117: case 'M':
1.14 schwarze 118: /* FALLTHROUGH */
1.48 schwarze 119: case 'm':
1.1 schwarze 120: /* FALLTHROUGH */
1.48 schwarze 121: case 'n':
1.1 schwarze 122: /* FALLTHROUGH */
1.48 schwarze 123: case 'V':
1.1 schwarze 124: /* FALLTHROUGH */
1.48 schwarze 125: case 'Y':
1.29 schwarze 126: gly = ESCAPE_IGNORE;
1.1 schwarze 127: /* FALLTHROUGH */
1.48 schwarze 128: case 'f':
1.26 schwarze 129: if (ESCAPE_ERROR == gly)
130: gly = ESCAPE_FONT;
1.34 schwarze 131: switch (**start) {
1.48 schwarze 132: case '(':
1.34 schwarze 133: *start = ++*end;
134: *sz = 2;
1.26 schwarze 135: break;
1.48 schwarze 136: case '[':
1.34 schwarze 137: *start = ++*end;
1.26 schwarze 138: term = ']';
139: break;
140: default:
1.34 schwarze 141: *sz = 1;
1.26 schwarze 142: break;
143: }
144: break;
145:
146: /*
147: * These escapes are of the form \X'Y', where 'X' is the trigger
148: * and 'Y' is any string. These have opaque sub-strings.
1.47 schwarze 149: * The \B and \w escapes are handled in roff.c, roff_res().
1.26 schwarze 150: */
1.48 schwarze 151: case 'A':
1.13 schwarze 152: /* FALLTHROUGH */
1.48 schwarze 153: case 'b':
1.1 schwarze 154: /* FALLTHROUGH */
1.48 schwarze 155: case 'D':
1.1 schwarze 156: /* FALLTHROUGH */
1.48 schwarze 157: case 'o':
1.1 schwarze 158: /* FALLTHROUGH */
1.48 schwarze 159: case 'R':
1.42 schwarze 160: /* FALLTHROUGH */
1.48 schwarze 161: case 'X':
1.1 schwarze 162: /* FALLTHROUGH */
1.48 schwarze 163: case 'Z':
1.46 schwarze 164: if ('\0' == **start)
1.26 schwarze 165: return(ESCAPE_ERROR);
166: gly = ESCAPE_IGNORE;
1.46 schwarze 167: term = **start;
1.34 schwarze 168: *start = ++*end;
1.16 schwarze 169: break;
1.26 schwarze 170:
171: /*
172: * These escapes are of the form \X'N', where 'X' is the trigger
173: * and 'N' resolves to a numerical expression.
174: */
1.48 schwarze 175: case 'h':
1.17 schwarze 176: /* FALLTHROUGH */
1.48 schwarze 177: case 'H':
1.26 schwarze 178: /* FALLTHROUGH */
1.48 schwarze 179: case 'L':
1.26 schwarze 180: /* FALLTHROUGH */
1.48 schwarze 181: case 'l':
1.26 schwarze 182: /* FALLTHROUGH */
1.48 schwarze 183: case 'S':
1.26 schwarze 184: /* FALLTHROUGH */
1.48 schwarze 185: case 'v':
1.17 schwarze 186: /* FALLTHROUGH */
1.48 schwarze 187: case 'x':
1.51 schwarze 188: if (strchr(" %&()*+-./0123456789:<=>", **start)) {
1.53 schwarze 189: if ('\0' != **start)
190: ++*end;
1.34 schwarze 191: return(ESCAPE_ERROR);
1.51 schwarze 192: }
1.42 schwarze 193: gly = ESCAPE_IGNORE;
1.46 schwarze 194: term = **start;
1.34 schwarze 195: *start = ++*end;
1.26 schwarze 196: break;
1.29 schwarze 197:
198: /*
199: * Special handling for the numbered character escape.
200: * XXX Do any other escapes need similar handling?
201: */
1.48 schwarze 202: case 'N':
1.34 schwarze 203: if ('\0' == **start)
1.29 schwarze 204: return(ESCAPE_ERROR);
1.34 schwarze 205: (*end)++;
206: if (isdigit((unsigned char)**start)) {
207: *sz = 1;
1.29 schwarze 208: return(ESCAPE_IGNORE);
1.34 schwarze 209: }
210: (*start)++;
1.29 schwarze 211: while (isdigit((unsigned char)**end))
212: (*end)++;
1.34 schwarze 213: *sz = *end - *start;
1.29 schwarze 214: if ('\0' != **end)
215: (*end)++;
216: return(ESCAPE_NUMBERED);
1.26 schwarze 217:
1.48 schwarze 218: /*
1.26 schwarze 219: * Sizes get a special category of their own.
220: */
1.48 schwarze 221: case 's':
1.26 schwarze 222: gly = ESCAPE_IGNORE;
1.17 schwarze 223:
1.26 schwarze 224: /* See +/- counts as a sign. */
1.34 schwarze 225: if ('+' == **end || '-' == **end || ASCII_HYPH == **end)
226: (*end)++;
1.6 schwarze 227:
1.34 schwarze 228: switch (**end) {
1.48 schwarze 229: case '(':
1.34 schwarze 230: *start = ++*end;
231: *sz = 2;
1.16 schwarze 232: break;
1.48 schwarze 233: case '[':
1.34 schwarze 234: *start = ++*end;
1.33 schwarze 235: term = ']';
1.16 schwarze 236: break;
1.48 schwarze 237: case '\'':
1.34 schwarze 238: *start = ++*end;
1.33 schwarze 239: term = '\'';
1.16 schwarze 240: break;
241: default:
1.34 schwarze 242: *sz = 1;
1.16 schwarze 243: break;
1.6 schwarze 244: }
245:
1.26 schwarze 246: break;
247:
248: /*
249: * Anything else is assumed to be a glyph.
1.34 schwarze 250: * In this case, pass back the character after the backslash.
1.26 schwarze 251: */
252: default:
253: gly = ESCAPE_SPECIAL;
1.34 schwarze 254: *start = --*end;
255: *sz = 1;
1.26 schwarze 256: break;
257: }
258:
259: assert(ESCAPE_ERROR != gly);
260:
261: /*
1.33 schwarze 262: * Read up to the terminating character,
263: * paying attention to nested escapes.
1.26 schwarze 264: */
265:
266: if ('\0' != term) {
1.33 schwarze 267: while (**end != term) {
268: switch (**end) {
1.48 schwarze 269: case '\0':
1.33 schwarze 270: return(ESCAPE_ERROR);
1.48 schwarze 271: case '\\':
1.33 schwarze 272: (*end)++;
273: if (ESCAPE_ERROR ==
274: mandoc_escape(end, NULL, NULL))
275: return(ESCAPE_ERROR);
276: break;
277: default:
278: (*end)++;
279: break;
280: }
281: }
1.34 schwarze 282: *sz = (*end)++ - *start;
1.33 schwarze 283: } else {
1.34 schwarze 284: assert(*sz > 0);
285: if ((size_t)*sz > strlen(*start))
1.26 schwarze 286: return(ESCAPE_ERROR);
1.34 schwarze 287: *end += *sz;
1.26 schwarze 288: }
1.19 schwarze 289:
1.26 schwarze 290: /* Run post-processors. */
1.19 schwarze 291:
1.26 schwarze 292: switch (gly) {
1.48 schwarze 293: case ESCAPE_FONT:
1.37 schwarze 294: if (2 == *sz) {
295: if ('C' == **start) {
296: /*
297: * Treat constant-width font modes
298: * just like regular font modes.
299: */
300: (*start)++;
301: (*sz)--;
302: } else {
303: if ('B' == (*start)[0] && 'I' == (*start)[1])
304: gly = ESCAPE_FONTBI;
305: break;
306: }
1.34 schwarze 307: } else if (1 != *sz)
1.26 schwarze 308: break;
1.30 schwarze 309:
1.34 schwarze 310: switch (**start) {
1.48 schwarze 311: case '3':
1.26 schwarze 312: /* FALLTHROUGH */
1.48 schwarze 313: case 'B':
1.26 schwarze 314: gly = ESCAPE_FONTBOLD;
315: break;
1.48 schwarze 316: case '2':
1.26 schwarze 317: /* FALLTHROUGH */
1.48 schwarze 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':
1.26 schwarze 325: /* FALLTHROUGH */
1.48 schwarze 326: case 'R':
1.26 schwarze 327: gly = ESCAPE_FONTROMAN;
1.1 schwarze 328: break;
329: }
1.16 schwarze 330: break;
1.48 schwarze 331: case ESCAPE_SPECIAL:
1.34 schwarze 332: if (1 == *sz && 'c' == **start)
1.26 schwarze 333: gly = ESCAPE_NOSPACE;
1.54 schwarze 334: /*
1.55 schwarze 335: * Unicode escapes are defined in groff as \[u0000]
1.54 schwarze 336: * to \[u10FFFF], where the contained value must be
337: * a valid Unicode codepoint. Here, however, only
1.55 schwarze 338: * check the length and range.
1.54 schwarze 339: */
1.55 schwarze 340: if (**start != 'u' || *sz < 5 || *sz > 7)
341: break;
342: if (*sz == 7 && ((*start)[1] != '1' || (*start)[2] != '0'))
343: break;
344: if (*sz == 6 && (*start)[1] == '0')
345: break;
346: if ((int)strspn(*start + 1, "0123456789ABCDEFabcdef")
1.54 schwarze 347: + 1 == *sz)
348: gly = ESCAPE_UNICODE;
1.16 schwarze 349: break;
1.1 schwarze 350: default:
1.16 schwarze 351: break;
1.1 schwarze 352: }
353:
1.26 schwarze 354: return(gly);
1.21 schwarze 355: }
356:
357: /*
358: * Parse a quoted or unquoted roff-style request or macro argument.
359: * Return a pointer to the parsed argument, which is either the original
360: * pointer or advanced by one byte in case the argument is quoted.
1.40 schwarze 361: * NUL-terminate the argument in place.
1.21 schwarze 362: * Collapse pairs of quotes inside quoted arguments.
363: * Advance the argument pointer to the next argument,
1.40 schwarze 364: * or to the NUL byte terminating the argument line.
1.21 schwarze 365: */
366: char *
1.25 schwarze 367: mandoc_getarg(struct mparse *parse, char **cpp, int ln, int *pos)
1.21 schwarze 368: {
369: char *start, *cp;
370: int quoted, pairs, white;
371:
372: /* Quoting can only start with a new word. */
373: start = *cpp;
1.26 schwarze 374: quoted = 0;
1.21 schwarze 375: if ('"' == *start) {
376: quoted = 1;
377: start++;
1.48 schwarze 378: }
1.21 schwarze 379:
380: pairs = 0;
381: white = 0;
382: for (cp = start; '\0' != *cp; cp++) {
1.36 schwarze 383:
384: /*
385: * Move the following text left
386: * after quoted quotes and after "\\" and "\t".
387: */
1.21 schwarze 388: if (pairs)
389: cp[-pairs] = cp[0];
1.36 schwarze 390:
1.21 schwarze 391: if ('\\' == cp[0]) {
1.36 schwarze 392: /*
393: * In copy mode, translate double to single
394: * backslashes and backslash-t to literal tabs.
395: */
396: switch (cp[1]) {
1.48 schwarze 397: case 't':
1.36 schwarze 398: cp[0] = '\t';
399: /* FALLTHROUGH */
1.48 schwarze 400: case '\\':
1.21 schwarze 401: pairs++;
402: cp++;
1.36 schwarze 403: break;
1.48 schwarze 404: case ' ':
1.21 schwarze 405: /* Skip escaped blanks. */
1.36 schwarze 406: if (0 == quoted)
407: cp++;
408: break;
409: default:
410: break;
411: }
1.21 schwarze 412: } else if (0 == quoted) {
413: if (' ' == cp[0]) {
414: /* Unescaped blanks end unquoted args. */
415: white = 1;
416: break;
417: }
418: } else if ('"' == cp[0]) {
419: if ('"' == cp[1]) {
420: /* Quoted quotes collapse. */
421: pairs++;
422: cp++;
423: } else {
424: /* Unquoted quotes end quoted args. */
425: quoted = 2;
426: break;
427: }
428: }
429: }
430:
431: /* Quoted argument without a closing quote. */
1.25 schwarze 432: if (1 == quoted)
1.52 schwarze 433: mandoc_msg(MANDOCERR_ARG_QUOTE, parse, ln, *pos, NULL);
1.21 schwarze 434:
1.40 schwarze 435: /* NUL-terminate this argument and move to the next one. */
1.21 schwarze 436: if (pairs)
437: cp[-pairs] = '\0';
438: if ('\0' != *cp) {
439: *cp++ = '\0';
440: while (' ' == *cp)
441: cp++;
442: }
1.24 schwarze 443: *pos += (int)(cp - start) + (quoted ? 1 : 0);
1.21 schwarze 444: *cpp = cp;
445:
1.25 schwarze 446: if ('\0' == *cp && (white || ' ' == cp[-1]))
1.52 schwarze 447: mandoc_msg(MANDOCERR_SPACE_EOL, parse, ln, *pos, NULL);
1.21 schwarze 448:
449: return(start);
1.4 schwarze 450: }
1.5 schwarze 451:
452: static int
453: a2time(time_t *t, const char *fmt, const char *p)
454: {
455: struct tm tm;
456: char *pp;
457:
458: memset(&tm, 0, sizeof(struct tm));
459:
460: pp = strptime(p, fmt, &tm);
461: if (NULL != pp && '\0' == *pp) {
462: *t = mktime(&tm);
463: return(1);
464: }
465:
466: return(0);
467: }
468:
1.22 schwarze 469: static char *
470: time2a(time_t t)
471: {
1.28 schwarze 472: struct tm *tm;
1.23 schwarze 473: char *buf, *p;
474: size_t ssz;
1.22 schwarze 475: int isz;
476:
1.28 schwarze 477: tm = localtime(&t);
1.22 schwarze 478:
1.23 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:
1.28 schwarze 487: if (0 == (ssz = strftime(p, 10 + 1, "%B ", tm)))
1.23 schwarze 488: goto fail;
489: p += (int)ssz;
1.22 schwarze 490:
1.28 schwarze 491: if (-1 == (isz = snprintf(p, 4 + 1, "%d, ", tm->tm_mday)))
1.23 schwarze 492: goto fail;
1.22 schwarze 493: p += isz;
494:
1.28 schwarze 495: if (0 == strftime(p, 4 + 1, "%Y", tm))
1.23 schwarze 496: goto fail;
497: return(buf);
498:
499: fail:
500: free(buf);
501: return(NULL);
1.22 schwarze 502: }
503:
504: char *
1.25 schwarze 505: mandoc_normdate(struct mparse *parse, char *in, int ln, int pos)
1.5 schwarze 506: {
1.22 schwarze 507: char *out;
1.5 schwarze 508: time_t t;
509:
1.22 schwarze 510: if (NULL == in || '\0' == *in ||
511: 0 == strcmp(in, "$" "Mdocdate$")) {
1.49 schwarze 512: mandoc_msg(MANDOCERR_DATE_MISSING, parse, ln, pos, NULL);
1.22 schwarze 513: time(&t);
514: }
1.31 schwarze 515: else if (a2time(&t, "%Y-%m-%d", in))
516: t = 0;
1.22 schwarze 517: else if (!a2time(&t, "$" "Mdocdate: %b %d %Y $", in) &&
1.31 schwarze 518: !a2time(&t, "%b %d, %Y", in)) {
1.50 schwarze 519: mandoc_msg(MANDOCERR_DATE_BAD, parse, ln, pos, in);
1.22 schwarze 520: t = 0;
1.5 schwarze 521: }
1.22 schwarze 522: out = t ? time2a(t) : NULL;
1.23 schwarze 523: return(out ? out : mandoc_strdup(in));
1.5 schwarze 524: }
525:
1.9 schwarze 526: int
1.43 schwarze 527: mandoc_eos(const char *p, size_t sz)
1.9 schwarze 528: {
1.43 schwarze 529: const char *q;
530: int enclosed, found;
1.9 schwarze 531:
1.10 schwarze 532: if (0 == sz)
533: return(0);
1.9 schwarze 534:
1.11 schwarze 535: /*
536: * End-of-sentence recognition must include situations where
537: * some symbols, such as `)', allow prior EOS punctuation to
1.26 schwarze 538: * propagate outward.
1.11 schwarze 539: */
540:
1.43 schwarze 541: enclosed = found = 0;
1.16 schwarze 542: for (q = p + (int)sz - 1; q >= p; q--) {
1.15 schwarze 543: switch (*q) {
1.48 schwarze 544: case '\"':
1.11 schwarze 545: /* FALLTHROUGH */
1.48 schwarze 546: case '\'':
1.11 schwarze 547: /* FALLTHROUGH */
1.48 schwarze 548: case ']':
1.11 schwarze 549: /* FALLTHROUGH */
1.48 schwarze 550: case ')':
1.15 schwarze 551: if (0 == found)
552: enclosed = 1;
1.11 schwarze 553: break;
1.48 schwarze 554: case '.':
1.11 schwarze 555: /* FALLTHROUGH */
1.48 schwarze 556: case '!':
1.11 schwarze 557: /* FALLTHROUGH */
1.48 schwarze 558: case '?':
1.15 schwarze 559: found = 1;
560: break;
1.11 schwarze 561: default:
1.20 schwarze 562: return(found && (!enclosed || isalnum((unsigned char)*q)));
1.11 schwarze 563: }
1.9 schwarze 564: }
565:
1.15 schwarze 566: return(found && !enclosed);
1.9 schwarze 567: }
1.26 schwarze 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
1.27 schwarze 574: mandoc_strntoi(const char *p, size_t sz, int base)
1.26 schwarze 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);
584: buf[(int)sz] = '\0';
585:
586: errno = 0;
587: v = strtol(buf, &ep, base);
588:
589: if (buf[0] == '\0' || *ep != '\0')
590: return(-1);
591:
1.27 schwarze 592: if (v > INT_MAX)
593: v = INT_MAX;
594: if (v < INT_MIN)
595: v = INT_MIN;
1.26 schwarze 596:
597: return((int)v);
598: }