Annotation of src/usr.bin/mandoc/man.c, Revision 1.19
1.19 ! schwarze 1: /* $Id: man.c,v 1.18 2010/02/18 02:11:26 schwarze Exp $ */
1.1 kristaps 2: /*
1.2 schwarze 3: * Copyright (c) 2008, 2009 Kristaps Dzonsons <kristaps@kth.se>
1.1 kristaps 4: *
5: * Permission to use, copy, modify, and distribute this software for any
1.2 schwarze 6: * purpose with or without fee is hereby granted, provided that the above
7: * copyright notice and this permission notice appear in all copies.
1.1 kristaps 8: *
1.2 schwarze 9: * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
10: * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
11: * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
12: * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
13: * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
14: * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
15: * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
1.1 kristaps 16: */
1.14 schwarze 17: #include <sys/types.h>
18:
1.1 kristaps 19: #include <assert.h>
20: #include <ctype.h>
21: #include <stdarg.h>
22: #include <stdlib.h>
23: #include <stdio.h>
24: #include <string.h>
25:
26: #include "libman.h"
1.16 schwarze 27: #include "libmandoc.h"
1.1 kristaps 28:
1.7 schwarze 29: const char *const __man_merrnames[WERRMAX] = {
30: "invalid character", /* WNPRINT */
31: "invalid manual section", /* WMSEC */
32: "invalid date format", /* WDATE */
33: "scope of prior line violated", /* WLNSCOPE */
34: "trailing whitespace", /* WTSPACE */
35: "unterminated quoted parameter", /* WTQUOTE */
36: "document has no body", /* WNODATA */
37: "document has no title/section", /* WNOTITLE */
38: "invalid escape sequence", /* WESCAPE */
1.8 schwarze 39: "invalid number format", /* WNUMFMT */
1.9 schwarze 40: "expected block head arguments", /* WHEADARGS */
41: "expected block body arguments", /* WBODYARGS */
42: "expected empty block head", /* WNHEADARGS */
43: "ill-formed macro", /* WMACROFORM */
1.10 schwarze 44: "scope open on exit", /* WEXITSCOPE */
45: "no scope context", /* WNOSCOPE */
46: "literal context already open", /* WOLITERAL */
47: "no literal context open" /* WNLITERAL */
1.7 schwarze 48: };
49:
1.1 kristaps 50: const char *const __man_macronames[MAN_MAX] = {
1.3 schwarze 51: "br", "TH", "SH", "SS",
1.1 kristaps 52: "TP", "LP", "PP", "P",
53: "IP", "HP", "SM", "SB",
54: "BI", "IB", "BR", "RB",
55: "R", "B", "I", "IR",
1.9 schwarze 56: "RI", "na", "i", "sp",
1.10 schwarze 57: "nf", "fi", "r", "RE",
1.15 schwarze 58: "RS", "DT", "UC", "PD"
1.1 kristaps 59: };
60:
61: const char * const *man_macronames = __man_macronames;
62:
63: static struct man_node *man_node_alloc(int, int,
64: enum man_type, int);
65: static int man_node_append(struct man *,
66: struct man_node *);
67: static int man_ptext(struct man *, int, char *);
68: static int man_pmacro(struct man *, int, char *);
69: static void man_free1(struct man *);
1.16 schwarze 70: static void man_alloc1(struct man *);
1.10 schwarze 71: static int pstring(struct man *, int, int,
72: const char *, size_t);
1.15 schwarze 73: static int macrowarn(struct man *, int, const char *);
1.1 kristaps 74:
75:
76: const struct man_node *
77: man_node(const struct man *m)
78: {
79:
80: return(MAN_HALT & m->flags ? NULL : m->first);
81: }
82:
83:
84: const struct man_meta *
85: man_meta(const struct man *m)
86: {
87:
88: return(MAN_HALT & m->flags ? NULL : &m->meta);
89: }
90:
91:
1.16 schwarze 92: void
1.1 kristaps 93: man_reset(struct man *man)
94: {
95:
96: man_free1(man);
1.16 schwarze 97: man_alloc1(man);
1.1 kristaps 98: }
99:
100:
101: void
102: man_free(struct man *man)
103: {
104:
105: man_free1(man);
106: free(man);
107: }
108:
109:
110: struct man *
111: man_alloc(void *data, int pflags, const struct man_cb *cb)
112: {
113: struct man *p;
114:
1.16 schwarze 115: p = mandoc_calloc(1, sizeof(struct man));
1.1 kristaps 116:
1.16 schwarze 117: if (cb)
118: memcpy(&p->cb, cb, sizeof(struct man_cb));
1.1 kristaps 119:
1.13 schwarze 120: man_hash_init();
1.1 kristaps 121: p->data = data;
122: p->pflags = pflags;
1.16 schwarze 123:
124: man_alloc1(p);
1.1 kristaps 125: return(p);
126: }
127:
128:
129: int
130: man_endparse(struct man *m)
131: {
132:
133: if (MAN_HALT & m->flags)
134: return(0);
135: else if (man_macroend(m))
136: return(1);
137: m->flags |= MAN_HALT;
138: return(0);
139: }
140:
141:
142: int
143: man_parseln(struct man *m, int ln, char *buf)
144: {
145:
146: return('.' == *buf ?
147: man_pmacro(m, ln, buf) :
148: man_ptext(m, ln, buf));
149: }
150:
151:
152: static void
153: man_free1(struct man *man)
154: {
155:
156: if (man->first)
157: man_node_freelist(man->first);
158: if (man->meta.title)
159: free(man->meta.title);
160: if (man->meta.source)
161: free(man->meta.source);
162: if (man->meta.vol)
163: free(man->meta.vol);
164: }
165:
166:
1.16 schwarze 167: static void
1.1 kristaps 168: man_alloc1(struct man *m)
169: {
170:
1.16 schwarze 171: memset(&m->meta, 0, sizeof(struct man_meta));
1.1 kristaps 172: m->flags = 0;
1.16 schwarze 173: m->last = mandoc_calloc(1, sizeof(struct man_node));
1.1 kristaps 174: m->first = m->last;
175: m->last->type = MAN_ROOT;
176: m->next = MAN_NEXT_CHILD;
177: }
178:
179:
180: static int
181: man_node_append(struct man *man, struct man_node *p)
182: {
183:
184: assert(man->last);
185: assert(man->first);
186: assert(MAN_ROOT != p->type);
187:
188: switch (man->next) {
189: case (MAN_NEXT_SIBLING):
190: man->last->next = p;
191: p->prev = man->last;
192: p->parent = man->last->parent;
193: break;
194: case (MAN_NEXT_CHILD):
195: man->last->child = p;
196: p->parent = man->last;
197: break;
198: default:
199: abort();
200: /* NOTREACHED */
201: }
1.5 schwarze 202:
203: p->parent->nchild++;
1.1 kristaps 204:
1.9 schwarze 205: if ( ! man_valid_pre(man, p))
206: return(0);
207:
208: switch (p->type) {
209: case (MAN_HEAD):
210: assert(MAN_BLOCK == p->parent->type);
211: p->parent->head = p;
212: break;
213: case (MAN_BODY):
214: assert(MAN_BLOCK == p->parent->type);
215: p->parent->body = p;
216: break;
217: default:
218: break;
219: }
220:
1.1 kristaps 221: man->last = p;
222:
223: switch (p->type) {
224: case (MAN_TEXT):
225: if ( ! man_valid_post(man))
226: return(0);
227: if ( ! man_action_post(man))
228: return(0);
229: break;
230: default:
231: break;
232: }
233:
234: return(1);
235: }
236:
237:
238: static struct man_node *
239: man_node_alloc(int line, int pos, enum man_type type, int tok)
240: {
241: struct man_node *p;
242:
1.16 schwarze 243: p = mandoc_calloc(1, sizeof(struct man_node));
1.1 kristaps 244: p->line = line;
245: p->pos = pos;
246: p->type = type;
247: p->tok = tok;
248: return(p);
249: }
250:
251:
252: int
1.10 schwarze 253: man_elem_alloc(struct man *m, int line, int pos, int tok)
1.1 kristaps 254: {
255: struct man_node *p;
256:
257: p = man_node_alloc(line, pos, MAN_ELEM, tok);
1.10 schwarze 258: if ( ! man_node_append(m, p))
259: return(0);
260: m->next = MAN_NEXT_CHILD;
261: return(1);
1.1 kristaps 262: }
263:
264:
265: int
1.9 schwarze 266: man_head_alloc(struct man *m, int line, int pos, int tok)
267: {
268: struct man_node *p;
269:
270: p = man_node_alloc(line, pos, MAN_HEAD, tok);
271: if ( ! man_node_append(m, p))
272: return(0);
273: m->next = MAN_NEXT_CHILD;
274: return(1);
275: }
276:
277:
278: int
279: man_body_alloc(struct man *m, int line, int pos, int tok)
280: {
281: struct man_node *p;
282:
283: p = man_node_alloc(line, pos, MAN_BODY, tok);
284: if ( ! man_node_append(m, p))
285: return(0);
286: m->next = MAN_NEXT_CHILD;
287: return(1);
288: }
289:
290:
291: int
292: man_block_alloc(struct man *m, int line, int pos, int tok)
293: {
294: struct man_node *p;
295:
296: p = man_node_alloc(line, pos, MAN_BLOCK, tok);
297: if ( ! man_node_append(m, p))
298: return(0);
299: m->next = MAN_NEXT_CHILD;
300: return(1);
301: }
302:
303:
1.10 schwarze 304: static int
305: pstring(struct man *m, int line, int pos,
306: const char *p, size_t len)
1.1 kristaps 307: {
1.10 schwarze 308: struct man_node *n;
309: size_t sv;
1.1 kristaps 310:
1.10 schwarze 311: n = man_node_alloc(line, pos, MAN_TEXT, -1);
1.16 schwarze 312: n->string = mandoc_malloc(len + 1);
1.10 schwarze 313: sv = strlcpy(n->string, p, len + 1);
314:
315: /* Prohibit truncation. */
316: assert(sv < len + 1);
317:
318: if ( ! man_node_append(m, n))
1.1 kristaps 319: return(0);
1.10 schwarze 320: m->next = MAN_NEXT_SIBLING;
321: return(1);
322: }
323:
324:
325: int
326: man_word_alloc(struct man *m, int line, int pos, const char *word)
327: {
328:
329: return(pstring(m, line, pos, word, strlen(word)));
1.1 kristaps 330: }
331:
332:
333: void
334: man_node_free(struct man_node *p)
335: {
336:
337: if (p->string)
338: free(p->string);
1.5 schwarze 339: if (p->parent)
340: p->parent->nchild--;
1.1 kristaps 341: free(p);
342: }
343:
344:
345: void
346: man_node_freelist(struct man_node *p)
347: {
1.10 schwarze 348: struct man_node *n;
1.1 kristaps 349:
350: if (p->child)
351: man_node_freelist(p->child);
1.5 schwarze 352: assert(0 == p->nchild);
1.10 schwarze 353: n = p->next;
1.1 kristaps 354: man_node_free(p);
1.10 schwarze 355: if (n)
356: man_node_freelist(n);
1.1 kristaps 357: }
358:
359:
360: static int
361: man_ptext(struct man *m, int line, char *buf)
362: {
1.10 schwarze 363: int i, j;
1.18 schwarze 364: char sv;
1.10 schwarze 365:
366: /* Literal free-form text whitespace is preserved. */
367:
368: if (MAN_LITERAL & m->flags) {
369: if ( ! man_word_alloc(m, line, 0, buf))
370: return(0);
371: goto descope;
372: }
373:
374: /* First de-chunk and allocate words. */
375:
376: for (i = 0; ' ' == buf[i]; i++)
377: /* Skip leading whitespace. */ ;
1.18 schwarze 378:
379: if ('\0' == buf[i]) {
380: /* Trailing whitespace? */
381: if (i && ' ' == buf[i - 1])
382: if ( ! man_pwarn(m, line, i - 1, WTSPACE))
383: return(0);
1.10 schwarze 384: if ( ! pstring(m, line, 0, &buf[i], 0))
385: return(0);
386: goto descope;
387: }
1.1 kristaps 388:
1.10 schwarze 389: for (j = i; buf[i]; i++) {
390: if (' ' != buf[i])
391: continue;
392:
393: /* Escaped whitespace. */
394: if (i && ' ' == buf[i] && '\\' == buf[i - 1])
395: continue;
396:
1.18 schwarze 397: sv = buf[i];
398: buf[i++] = '\0';
399:
1.10 schwarze 400: if ( ! pstring(m, line, j, &buf[j], (size_t)(i - j)))
401: return(0);
402:
1.18 schwarze 403: /* Trailing whitespace? Check at overwritten byte. */
404:
405: if (' ' == sv && '\0' == buf[i])
406: if ( ! man_pwarn(m, line, i - 1, WTSPACE))
407: return(0);
408:
1.10 schwarze 409: for ( ; ' ' == buf[i]; i++)
410: /* Skip trailing whitespace. */ ;
411:
412: j = i;
1.18 schwarze 413:
414: /* Trailing whitespace? */
415:
416: if (' ' == buf[i - 1] && '\0' == buf[i])
417: if ( ! man_pwarn(m, line, i - 1, WTSPACE))
418: return(0);
419:
420: if ('\0' == buf[i])
1.10 schwarze 421: break;
422: }
1.9 schwarze 423:
1.10 schwarze 424: if (j != i && ! pstring(m, line, j, &buf[j], (size_t)(i - j)))
1.1 kristaps 425: return(0);
1.10 schwarze 426:
427: descope:
1.1 kristaps 428:
429: /*
1.9 schwarze 430: * Co-ordinate what happens with having a next-line scope open:
431: * first close out the element scope (if applicable), then close
432: * out the block scope (also if applicable).
1.1 kristaps 433: */
434:
1.9 schwarze 435: if (MAN_ELINE & m->flags) {
436: m->flags &= ~MAN_ELINE;
437: if ( ! man_unscope(m, m->last->parent))
438: return(0);
439: }
440:
441: if ( ! (MAN_BLINE & m->flags))
1.1 kristaps 442: return(1);
1.9 schwarze 443: m->flags &= ~MAN_BLINE;
1.1 kristaps 444:
1.9 schwarze 445: if ( ! man_unscope(m, m->last->parent))
1.1 kristaps 446: return(0);
1.9 schwarze 447: return(man_body_alloc(m, line, 0, m->last->tok));
1.1 kristaps 448: }
449:
450:
1.15 schwarze 451: static int
452: macrowarn(struct man *m, int ln, const char *buf)
453: {
454: if ( ! (MAN_IGN_MACRO & m->pflags))
455: return(man_verr(m, ln, 0,
456: "unknown macro: %s%s",
457: buf, strlen(buf) > 3 ? "..." : ""));
458: return(man_vwarn(m, ln, 0, "unknown macro: %s%s",
459: buf, strlen(buf) > 3 ? "..." : ""));
460: }
461:
462:
1.1 kristaps 463: int
464: man_pmacro(struct man *m, int ln, char *buf)
465: {
1.10 schwarze 466: int i, j, c, ppos, fl;
467: char mac[5];
468: struct man_node *n;
1.1 kristaps 469:
470: /* Comments and empties are quickly ignored. */
471:
1.9 schwarze 472: fl = m->flags;
1.1 kristaps 473:
1.17 schwarze 474: if ('\0' == buf[1])
475: return(1);
1.1 kristaps 476:
477: i = 1;
478:
479: if (' ' == buf[i]) {
480: i++;
481: while (buf[i] && ' ' == buf[i])
482: i++;
1.18 schwarze 483: if ('\0' == buf[i])
1.1 kristaps 484: goto out;
485: }
486:
487: ppos = i;
488:
489: /* Copy the first word into a nil-terminated buffer. */
490:
491: for (j = 0; j < 4; j++, i++) {
1.18 schwarze 492: if ('\0' == (mac[j] = buf[i]))
1.1 kristaps 493: break;
494: else if (' ' == buf[i])
495: break;
1.11 schwarze 496:
497: /* Check for invalid characters. */
498:
499: if (isgraph((u_char)buf[i]))
500: continue;
501: return(man_perr(m, ln, i, WNPRINT));
1.1 kristaps 502: }
503:
1.17 schwarze 504: mac[j] = '\0';
1.1 kristaps 505:
506: if (j == 4 || j < 1) {
507: if ( ! (MAN_IGN_MACRO & m->pflags)) {
1.9 schwarze 508: (void)man_perr(m, ln, ppos, WMACROFORM);
1.1 kristaps 509: goto err;
510: }
1.9 schwarze 511: if ( ! man_pwarn(m, ln, ppos, WMACROFORM))
1.1 kristaps 512: goto err;
513: return(1);
514: }
515:
1.13 schwarze 516: if (MAN_MAX == (c = man_hash_find(mac))) {
1.15 schwarze 517: if ( ! macrowarn(m, ln, mac))
1.1 kristaps 518: goto err;
519: return(1);
520: }
521:
522: /* The macro is sane. Jump to the next word. */
523:
524: while (buf[i] && ' ' == buf[i])
525: i++;
1.18 schwarze 526:
527: /* Trailing whitespace? */
528:
529: if ('\0' == buf[i] && ' ' == buf[i - 1])
530: if ( ! man_pwarn(m, ln, i - 1, WTSPACE))
531: goto err;
1.1 kristaps 532:
1.10 schwarze 533: /* Remove prior ELINE macro, if applicable. */
534:
535: if (m->flags & MAN_ELINE) {
536: n = m->last;
537: assert(NULL == n->child);
538: assert(0 == n->nchild);
539: if ( ! man_nwarn(m, n, WLNSCOPE))
540: return(0);
541:
542: if (n->prev) {
543: assert(n != n->parent->child);
544: assert(n == n->prev->next);
545: n->prev->next = NULL;
546: m->last = n->prev;
547: m->next = MAN_NEXT_SIBLING;
548: } else {
549: assert(n == n->parent->child);
550: n->parent->child = NULL;
551: m->last = n->parent;
552: m->next = MAN_NEXT_CHILD;
553: }
554:
555: man_node_free(n);
556: m->flags &= ~MAN_ELINE;
557: }
558:
1.1 kristaps 559: /* Begin recursive parse sequence. */
560:
1.9 schwarze 561: assert(man_macros[c].fp);
562:
563: if ( ! (*man_macros[c].fp)(m, c, ln, ppos, &i, buf))
1.1 kristaps 564: goto err;
565:
566: out:
1.19 ! schwarze 567: if ( ! (MAN_BLINE & fl) || (MAN_TEXT != m->last->type &&
! 568: (NULL == m->last->child || MAN_TEXT != m->last->child->type)))
1.9 schwarze 569: return(1);
570:
571: /*
572: * If we've opened a new next-line element scope, then return
573: * now, as the next line will close out the block scope.
574: */
575:
576: if (MAN_ELINE & m->flags)
577: return(1);
578:
579: /* Close out the block scope opened in the prior line. */
1.1 kristaps 580:
1.9 schwarze 581: assert(MAN_BLINE & m->flags);
582: m->flags &= ~MAN_BLINE;
1.1 kristaps 583:
1.9 schwarze 584: if ( ! man_unscope(m, m->last->parent))
585: return(0);
586: return(man_body_alloc(m, ln, 0, m->last->tok));
1.1 kristaps 587:
588: err: /* Error out. */
589:
590: m->flags |= MAN_HALT;
591: return(0);
592: }
593:
594:
595: int
596: man_verr(struct man *man, int ln, int pos, const char *fmt, ...)
597: {
598: char buf[256];
599: va_list ap;
600:
601: if (NULL == man->cb.man_err)
602: return(0);
603:
604: va_start(ap, fmt);
605: (void)vsnprintf(buf, sizeof(buf) - 1, fmt, ap);
606: va_end(ap);
607: return((*man->cb.man_err)(man->data, ln, pos, buf));
608: }
609:
610:
611: int
612: man_vwarn(struct man *man, int ln, int pos, const char *fmt, ...)
613: {
614: char buf[256];
615: va_list ap;
616:
617: if (NULL == man->cb.man_warn)
618: return(0);
619:
620: va_start(ap, fmt);
621: (void)vsnprintf(buf, sizeof(buf) - 1, fmt, ap);
622: va_end(ap);
623: return((*man->cb.man_warn)(man->data, ln, pos, buf));
624: }
625:
626:
1.4 schwarze 627: int
1.7 schwarze 628: man_err(struct man *m, int line, int pos, int iserr, enum merr type)
1.4 schwarze 629: {
630: const char *p;
631:
1.7 schwarze 632: p = __man_merrnames[(int)type];
1.4 schwarze 633: assert(p);
634:
635: if (iserr)
636: return(man_verr(m, line, pos, p));
637:
638: return(man_vwarn(m, line, pos, p));
639: }