Annotation of mandoc/mandoc.c, Revision 1.84.2.1
1.84.2.1! 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;
1.52 kristaps 82: /*
83: * Unicode escapes are defined in groff as \[uXXXX] to
84: * \[u10FFFF], where the contained value must be a valid
85: * Unicode codepoint. Here, however, only check whether
86: * it's not a zero-width escape.
87: */
1.65 schwarze 88: if ('u' == (*start)[0] && ']' != (*start)[1])
1.52 kristaps 89: gly = ESCAPE_UNICODE;
1.45 kristaps 90: term = ']';
91: break;
1.79 schwarze 92: case 'C':
1.65 schwarze 93: if ('\'' != **start)
1.45 kristaps 94: return(ESCAPE_ERROR);
1.65 schwarze 95: *start = ++*end;
1.70 schwarze 96: if ('u' == (*start)[0] && '\'' != (*start)[1])
97: gly = ESCAPE_UNICODE;
98: else
99: gly = ESCAPE_SPECIAL;
1.45 kristaps 100: term = '\'';
101: break;
1.72 schwarze 102:
103: /*
104: * Escapes taking no arguments at all.
105: */
1.79 schwarze 106: case 'd':
1.72 schwarze 107: /* FALLTHROUGH */
1.79 schwarze 108: case 'u':
1.72 schwarze 109: return(ESCAPE_IGNORE);
1.63 schwarze 110:
111: /*
112: * The \z escape is supposed to output the following
1.79 schwarze 113: * character without advancing the cursor position.
1.63 schwarze 114: * Since we are mostly dealing with terminal mode,
115: * let us just skip the next character.
116: */
1.79 schwarze 117: case 'z':
1.63 schwarze 118: return(ESCAPE_SKIPCHAR);
1.1 kristaps 119:
1.45 kristaps 120: /*
121: * Handle all triggers matching \X(xy, \Xx, and \X[xxxx], where
122: * 'X' is the trigger. These have opaque sub-strings.
123: */
1.79 schwarze 124: case 'F':
1.45 kristaps 125: /* FALLTHROUGH */
1.79 schwarze 126: case 'g':
1.24 kristaps 127: /* FALLTHROUGH */
1.79 schwarze 128: case 'k':
1.24 kristaps 129: /* FALLTHROUGH */
1.79 schwarze 130: case 'M':
1.24 kristaps 131: /* FALLTHROUGH */
1.79 schwarze 132: case 'm':
1.24 kristaps 133: /* FALLTHROUGH */
1.79 schwarze 134: case 'n':
1.24 kristaps 135: /* FALLTHROUGH */
1.79 schwarze 136: case 'V':
1.24 kristaps 137: /* FALLTHROUGH */
1.79 schwarze 138: case 'Y':
1.60 schwarze 139: gly = ESCAPE_IGNORE;
1.24 kristaps 140: /* FALLTHROUGH */
1.79 schwarze 141: case 'f':
1.45 kristaps 142: if (ESCAPE_ERROR == gly)
143: gly = ESCAPE_FONT;
1.65 schwarze 144: switch (**start) {
1.79 schwarze 145: case '(':
1.65 schwarze 146: *start = ++*end;
147: *sz = 2;
1.45 kristaps 148: break;
1.79 schwarze 149: case '[':
1.65 schwarze 150: *start = ++*end;
1.45 kristaps 151: term = ']';
152: break;
153: default:
1.65 schwarze 154: *sz = 1;
1.45 kristaps 155: break;
156: }
157: break;
158:
159: /*
160: * These escapes are of the form \X'Y', where 'X' is the trigger
161: * and 'Y' is any string. These have opaque sub-strings.
1.78 schwarze 162: * The \B and \w escapes are handled in roff.c, roff_res().
1.45 kristaps 163: */
1.79 schwarze 164: case 'A':
1.24 kristaps 165: /* FALLTHROUGH */
1.79 schwarze 166: case 'b':
1.24 kristaps 167: /* FALLTHROUGH */
1.79 schwarze 168: case 'D':
1.24 kristaps 169: /* FALLTHROUGH */
1.79 schwarze 170: case 'o':
1.24 kristaps 171: /* FALLTHROUGH */
1.79 schwarze 172: case 'R':
1.73 schwarze 173: /* FALLTHROUGH */
1.79 schwarze 174: case 'X':
1.24 kristaps 175: /* FALLTHROUGH */
1.79 schwarze 176: case 'Z':
1.77 schwarze 177: if ('\0' == **start)
1.45 kristaps 178: return(ESCAPE_ERROR);
179: gly = ESCAPE_IGNORE;
1.77 schwarze 180: term = **start;
1.65 schwarze 181: *start = ++*end;
1.24 kristaps 182: break;
1.45 kristaps 183:
184: /*
185: * These escapes are of the form \X'N', where 'X' is the trigger
186: * and 'N' resolves to a numerical expression.
187: */
1.79 schwarze 188: case 'h':
1.28 kristaps 189: /* FALLTHROUGH */
1.79 schwarze 190: case 'H':
1.45 kristaps 191: /* FALLTHROUGH */
1.79 schwarze 192: case 'L':
1.45 kristaps 193: /* FALLTHROUGH */
1.79 schwarze 194: case 'l':
1.45 kristaps 195: /* FALLTHROUGH */
1.79 schwarze 196: case 'S':
1.45 kristaps 197: /* FALLTHROUGH */
1.79 schwarze 198: case 'v':
1.28 kristaps 199: /* FALLTHROUGH */
1.79 schwarze 200: case 'x':
1.82 schwarze 201: if (strchr(" %&()*+-./0123456789:<=>", **start)) {
1.84.2.1! schwarze 202: if ('\0' != **start)
! 203: ++*end;
1.65 schwarze 204: return(ESCAPE_ERROR);
1.82 schwarze 205: }
1.73 schwarze 206: gly = ESCAPE_IGNORE;
1.77 schwarze 207: term = **start;
1.65 schwarze 208: *start = ++*end;
1.45 kristaps 209: break;
1.60 schwarze 210:
211: /*
212: * Special handling for the numbered character escape.
213: * XXX Do any other escapes need similar handling?
214: */
1.79 schwarze 215: case 'N':
1.65 schwarze 216: if ('\0' == **start)
1.60 schwarze 217: return(ESCAPE_ERROR);
1.65 schwarze 218: (*end)++;
219: if (isdigit((unsigned char)**start)) {
220: *sz = 1;
1.60 schwarze 221: return(ESCAPE_IGNORE);
1.65 schwarze 222: }
223: (*start)++;
1.60 schwarze 224: while (isdigit((unsigned char)**end))
225: (*end)++;
1.65 schwarze 226: *sz = *end - *start;
1.60 schwarze 227: if ('\0' != **end)
228: (*end)++;
229: return(ESCAPE_NUMBERED);
1.45 kristaps 230:
1.79 schwarze 231: /*
1.45 kristaps 232: * Sizes get a special category of their own.
233: */
1.79 schwarze 234: case 's':
1.45 kristaps 235: gly = ESCAPE_IGNORE;
1.28 kristaps 236:
1.45 kristaps 237: /* See +/- counts as a sign. */
1.65 schwarze 238: if ('+' == **end || '-' == **end || ASCII_HYPH == **end)
239: (*end)++;
1.8 kristaps 240:
1.65 schwarze 241: switch (**end) {
1.79 schwarze 242: case '(':
1.65 schwarze 243: *start = ++*end;
244: *sz = 2;
1.22 kristaps 245: break;
1.79 schwarze 246: case '[':
1.65 schwarze 247: *start = ++*end;
1.64 schwarze 248: term = ']';
1.22 kristaps 249: break;
1.79 schwarze 250: case '\'':
1.65 schwarze 251: *start = ++*end;
1.64 schwarze 252: term = '\'';
1.22 kristaps 253: break;
254: default:
1.65 schwarze 255: *sz = 1;
1.22 kristaps 256: break;
1.8 kristaps 257: }
258:
1.45 kristaps 259: break;
1.33 kristaps 260:
1.45 kristaps 261: /*
262: * Anything else is assumed to be a glyph.
1.65 schwarze 263: * In this case, pass back the character after the backslash.
1.45 kristaps 264: */
265: default:
266: gly = ESCAPE_SPECIAL;
1.65 schwarze 267: *start = --*end;
268: *sz = 1;
1.22 kristaps 269: break;
1.45 kristaps 270: }
271:
272: assert(ESCAPE_ERROR != gly);
273:
274: /*
1.64 schwarze 275: * Read up to the terminating character,
276: * paying attention to nested escapes.
1.45 kristaps 277: */
278:
279: if ('\0' != term) {
1.64 schwarze 280: while (**end != term) {
281: switch (**end) {
1.79 schwarze 282: case '\0':
1.64 schwarze 283: return(ESCAPE_ERROR);
1.79 schwarze 284: case '\\':
1.64 schwarze 285: (*end)++;
286: if (ESCAPE_ERROR ==
287: mandoc_escape(end, NULL, NULL))
288: return(ESCAPE_ERROR);
289: break;
290: default:
291: (*end)++;
292: break;
293: }
294: }
1.65 schwarze 295: *sz = (*end)++ - *start;
1.64 schwarze 296: } else {
1.65 schwarze 297: assert(*sz > 0);
298: if ((size_t)*sz > strlen(*start))
1.45 kristaps 299: return(ESCAPE_ERROR);
1.65 schwarze 300: *end += *sz;
1.45 kristaps 301: }
302:
303: /* Run post-processors. */
304:
305: switch (gly) {
1.79 schwarze 306: case ESCAPE_FONT:
1.68 schwarze 307: if (2 == *sz) {
308: if ('C' == **start) {
309: /*
310: * Treat constant-width font modes
311: * just like regular font modes.
312: */
313: (*start)++;
314: (*sz)--;
315: } else {
316: if ('B' == (*start)[0] && 'I' == (*start)[1])
317: gly = ESCAPE_FONTBI;
318: break;
319: }
1.65 schwarze 320: } else if (1 != *sz)
1.45 kristaps 321: break;
1.61 kristaps 322:
1.65 schwarze 323: switch (**start) {
1.79 schwarze 324: case '3':
1.45 kristaps 325: /* FALLTHROUGH */
1.79 schwarze 326: case 'B':
1.45 kristaps 327: gly = ESCAPE_FONTBOLD;
328: break;
1.79 schwarze 329: case '2':
1.45 kristaps 330: /* FALLTHROUGH */
1.79 schwarze 331: case 'I':
1.45 kristaps 332: gly = ESCAPE_FONTITALIC;
1.22 kristaps 333: break;
1.79 schwarze 334: case 'P':
1.45 kristaps 335: gly = ESCAPE_FONTPREV;
1.22 kristaps 336: break;
1.79 schwarze 337: case '1':
1.45 kristaps 338: /* FALLTHROUGH */
1.79 schwarze 339: case 'R':
1.45 kristaps 340: gly = ESCAPE_FONTROMAN;
1.1 kristaps 341: break;
342: }
1.46 kristaps 343: break;
1.79 schwarze 344: case ESCAPE_SPECIAL:
1.65 schwarze 345: if (1 == *sz && 'c' == **start)
1.45 kristaps 346: gly = ESCAPE_NOSPACE;
1.22 kristaps 347: break;
1.1 kristaps 348: default:
1.22 kristaps 349: break;
1.1 kristaps 350: }
351:
1.45 kristaps 352: return(gly);
1.36 schwarze 353: }
354:
355: /*
356: * Parse a quoted or unquoted roff-style request or macro argument.
357: * Return a pointer to the parsed argument, which is either the original
358: * pointer or advanced by one byte in case the argument is quoted.
1.71 schwarze 359: * NUL-terminate the argument in place.
1.36 schwarze 360: * Collapse pairs of quotes inside quoted arguments.
361: * Advance the argument pointer to the next argument,
1.71 schwarze 362: * or to the NUL byte terminating the argument line.
1.36 schwarze 363: */
364: char *
1.48 kristaps 365: mandoc_getarg(struct mparse *parse, char **cpp, int ln, int *pos)
1.36 schwarze 366: {
367: char *start, *cp;
368: int quoted, pairs, white;
369:
370: /* Quoting can only start with a new word. */
371: start = *cpp;
1.47 kristaps 372: quoted = 0;
1.36 schwarze 373: if ('"' == *start) {
374: quoted = 1;
375: start++;
1.79 schwarze 376: }
1.36 schwarze 377:
378: pairs = 0;
379: white = 0;
380: for (cp = start; '\0' != *cp; cp++) {
1.67 schwarze 381:
382: /*
383: * Move the following text left
384: * after quoted quotes and after "\\" and "\t".
385: */
1.36 schwarze 386: if (pairs)
387: cp[-pairs] = cp[0];
1.67 schwarze 388:
1.36 schwarze 389: if ('\\' == cp[0]) {
1.67 schwarze 390: /*
391: * In copy mode, translate double to single
392: * backslashes and backslash-t to literal tabs.
393: */
394: switch (cp[1]) {
1.79 schwarze 395: case 't':
1.67 schwarze 396: cp[0] = '\t';
397: /* FALLTHROUGH */
1.79 schwarze 398: case '\\':
1.36 schwarze 399: pairs++;
400: cp++;
1.67 schwarze 401: break;
1.79 schwarze 402: case ' ':
1.36 schwarze 403: /* Skip escaped blanks. */
1.67 schwarze 404: if (0 == quoted)
405: cp++;
406: break;
407: default:
408: break;
409: }
1.36 schwarze 410: } else if (0 == quoted) {
411: if (' ' == cp[0]) {
412: /* Unescaped blanks end unquoted args. */
413: white = 1;
414: break;
415: }
416: } else if ('"' == cp[0]) {
417: if ('"' == cp[1]) {
418: /* Quoted quotes collapse. */
419: pairs++;
420: cp++;
421: } else {
422: /* Unquoted quotes end quoted args. */
423: quoted = 2;
424: break;
425: }
426: }
427: }
428:
429: /* Quoted argument without a closing quote. */
1.48 kristaps 430: if (1 == quoted)
1.83 schwarze 431: mandoc_msg(MANDOCERR_ARG_QUOTE, parse, ln, *pos, NULL);
1.36 schwarze 432:
1.71 schwarze 433: /* NUL-terminate this argument and move to the next one. */
1.36 schwarze 434: if (pairs)
435: cp[-pairs] = '\0';
436: if ('\0' != *cp) {
437: *cp++ = '\0';
438: while (' ' == *cp)
439: cp++;
440: }
1.39 kristaps 441: *pos += (int)(cp - start) + (quoted ? 1 : 0);
1.36 schwarze 442: *cpp = cp;
443:
1.48 kristaps 444: if ('\0' == *cp && (white || ' ' == cp[-1]))
1.83 schwarze 445: mandoc_msg(MANDOCERR_SPACE_EOL, parse, ln, *pos, NULL);
1.36 schwarze 446:
447: return(start);
1.4 kristaps 448: }
1.7 kristaps 449:
450: static int
451: a2time(time_t *t, const char *fmt, const char *p)
452: {
453: struct tm tm;
454: char *pp;
455:
456: memset(&tm, 0, sizeof(struct tm));
457:
1.56 kristaps 458: pp = NULL;
459: #ifdef HAVE_STRPTIME
1.7 kristaps 460: pp = strptime(p, fmt, &tm);
1.56 kristaps 461: #endif
1.7 kristaps 462: if (NULL != pp && '\0' == *pp) {
463: *t = mktime(&tm);
464: return(1);
465: }
466:
467: return(0);
468: }
469:
1.37 schwarze 470: static char *
471: time2a(time_t t)
472: {
1.56 kristaps 473: struct tm *tm;
1.38 schwarze 474: char *buf, *p;
475: size_t ssz;
1.37 schwarze 476: int isz;
477:
1.56 kristaps 478: tm = localtime(&t);
1.37 schwarze 479:
1.38 schwarze 480: /*
481: * Reserve space:
482: * up to 9 characters for the month (September) + blank
483: * up to 2 characters for the day + comma + blank
484: * 4 characters for the year and a terminating '\0'
485: */
486: p = buf = mandoc_malloc(10 + 4 + 4 + 1);
487:
1.56 kristaps 488: if (0 == (ssz = strftime(p, 10 + 1, "%B ", tm)))
1.38 schwarze 489: goto fail;
490: p += (int)ssz;
1.37 schwarze 491:
1.56 kristaps 492: if (-1 == (isz = snprintf(p, 4 + 1, "%d, ", tm->tm_mday)))
1.38 schwarze 493: goto fail;
1.37 schwarze 494: p += isz;
495:
1.56 kristaps 496: if (0 == strftime(p, 4 + 1, "%Y", tm))
1.38 schwarze 497: goto fail;
498: return(buf);
499:
500: fail:
501: free(buf);
502: return(NULL);
1.37 schwarze 503: }
504:
505: char *
1.42 kristaps 506: mandoc_normdate(struct mparse *parse, char *in, int ln, int pos)
1.7 kristaps 507: {
1.37 schwarze 508: char *out;
1.7 kristaps 509: time_t t;
510:
1.37 schwarze 511: if (NULL == in || '\0' == *in ||
512: 0 == strcmp(in, "$" "Mdocdate$")) {
1.80 schwarze 513: mandoc_msg(MANDOCERR_DATE_MISSING, parse, ln, pos, NULL);
1.37 schwarze 514: time(&t);
515: }
1.62 schwarze 516: else if (a2time(&t, "%Y-%m-%d", in))
517: t = 0;
1.37 schwarze 518: else if (!a2time(&t, "$" "Mdocdate: %b %d %Y $", in) &&
1.62 schwarze 519: !a2time(&t, "%b %d, %Y", in)) {
1.81 schwarze 520: mandoc_msg(MANDOCERR_DATE_BAD, parse, ln, pos, in);
1.37 schwarze 521: t = 0;
1.7 kristaps 522: }
1.37 schwarze 523: out = t ? time2a(t) : NULL;
1.38 schwarze 524: return(out ? out : mandoc_strdup(in));
1.7 kristaps 525: }
526:
1.12 kristaps 527: int
1.75 schwarze 528: mandoc_eos(const char *p, size_t sz)
1.12 kristaps 529: {
1.75 schwarze 530: const char *q;
531: int enclosed, found;
1.12 kristaps 532:
1.13 kristaps 533: if (0 == sz)
534: return(0);
1.12 kristaps 535:
1.14 kristaps 536: /*
537: * End-of-sentence recognition must include situations where
538: * some symbols, such as `)', allow prior EOS punctuation to
1.49 kristaps 539: * propagate outward.
1.14 kristaps 540: */
541:
1.75 schwarze 542: enclosed = found = 0;
1.25 kristaps 543: for (q = p + (int)sz - 1; q >= p; q--) {
1.23 schwarze 544: switch (*q) {
1.79 schwarze 545: case '\"':
1.14 kristaps 546: /* FALLTHROUGH */
1.79 schwarze 547: case '\'':
1.15 kristaps 548: /* FALLTHROUGH */
1.79 schwarze 549: case ']':
1.14 kristaps 550: /* FALLTHROUGH */
1.79 schwarze 551: case ')':
1.23 schwarze 552: if (0 == found)
553: enclosed = 1;
1.14 kristaps 554: break;
1.79 schwarze 555: case '.':
1.14 kristaps 556: /* FALLTHROUGH */
1.79 schwarze 557: case '!':
1.14 kristaps 558: /* FALLTHROUGH */
1.79 schwarze 559: case '?':
1.23 schwarze 560: found = 1;
561: break;
1.14 kristaps 562: default:
1.27 joerg 563: return(found && (!enclosed || isalnum((unsigned char)*q)));
1.14 kristaps 564: }
1.12 kristaps 565: }
566:
1.23 schwarze 567: return(found && !enclosed);
1.44 kristaps 568: }
1.50 kristaps 569:
570: /*
571: * Convert a string to a long that may not be <0.
572: * If the string is invalid, or is less than 0, return -1.
573: */
574: int
1.54 kristaps 575: mandoc_strntoi(const char *p, size_t sz, int base)
1.50 kristaps 576: {
577: char buf[32];
578: char *ep;
579: long v;
580:
581: if (sz > 31)
582: return(-1);
583:
584: memcpy(buf, p, sz);
1.51 kristaps 585: buf[(int)sz] = '\0';
1.50 kristaps 586:
587: errno = 0;
588: v = strtol(buf, &ep, base);
589:
590: if (buf[0] == '\0' || *ep != '\0')
591: return(-1);
592:
1.54 kristaps 593: if (v > INT_MAX)
594: v = INT_MAX;
595: if (v < INT_MIN)
596: v = INT_MIN;
1.50 kristaps 597:
598: return((int)v);
599: }
CVSweb