Annotation of src/usr.bin/mandoc/man.c, Revision 1.60
1.60 ! schwarze 1: /* $Id: man.c,v 1.59 2011/04/24 16:22:02 schwarze Exp $ */
1.1 kristaps 2: /*
1.52 schwarze 3: * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
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 <stdarg.h>
21: #include <stdlib.h>
22: #include <stdio.h>
23: #include <string.h>
24:
1.59 schwarze 25: #include "man.h"
1.33 schwarze 26: #include "mandoc.h"
1.1 kristaps 27: #include "libman.h"
1.16 schwarze 28: #include "libmandoc.h"
1.1 kristaps 29:
30: const char *const __man_macronames[MAN_MAX] = {
1.3 schwarze 31: "br", "TH", "SH", "SS",
1.1 kristaps 32: "TP", "LP", "PP", "P",
33: "IP", "HP", "SM", "SB",
34: "BI", "IB", "BR", "RB",
35: "R", "B", "I", "IR",
1.50 schwarze 36: "RI", "na", "sp", "nf",
37: "fi", "RE", "RS", "DT",
38: "UC", "PD", "AT", "in",
1.52 schwarze 39: "ft"
1.1 kristaps 40: };
41:
42: const char * const *man_macronames = __man_macronames;
43:
1.54 schwarze 44: static struct man_node *man_node_alloc(struct man *, int, int,
1.22 schwarze 45: enum man_type, enum mant);
1.1 kristaps 46: static int man_node_append(struct man *,
47: struct man_node *);
1.22 schwarze 48: static void man_node_free(struct man_node *);
49: static void man_node_unlink(struct man *,
50: struct man_node *);
1.31 schwarze 51: static int man_ptext(struct man *, int, char *, int);
52: static int man_pmacro(struct man *, int, char *, int);
1.1 kristaps 53: static void man_free1(struct man *);
1.16 schwarze 54: static void man_alloc1(struct man *);
1.52 schwarze 55: static int man_descope(struct man *, int, int);
1.1 kristaps 56:
57:
58: const struct man_node *
59: man_node(const struct man *m)
60: {
61:
1.53 schwarze 62: assert( ! (MAN_HALT & m->flags));
63: return(m->first);
1.1 kristaps 64: }
65:
66:
67: const struct man_meta *
68: man_meta(const struct man *m)
69: {
70:
1.53 schwarze 71: assert( ! (MAN_HALT & m->flags));
72: return(&m->meta);
1.1 kristaps 73: }
74:
75:
1.16 schwarze 76: void
1.1 kristaps 77: man_reset(struct man *man)
78: {
79:
80: man_free1(man);
1.16 schwarze 81: man_alloc1(man);
1.1 kristaps 82: }
83:
84:
85: void
86: man_free(struct man *man)
87: {
88:
89: man_free1(man);
90: free(man);
91: }
92:
93:
94: struct man *
1.60 ! schwarze 95: man_alloc(struct roff *roff, struct mparse *parse)
1.1 kristaps 96: {
97: struct man *p;
98:
1.16 schwarze 99: p = mandoc_calloc(1, sizeof(struct man));
1.1 kristaps 100:
1.13 schwarze 101: man_hash_init();
1.59 schwarze 102: p->parse = parse;
1.60 ! schwarze 103: p->roff = roff;
1.16 schwarze 104:
105: man_alloc1(p);
1.1 kristaps 106: return(p);
107: }
108:
109:
110: int
111: man_endparse(struct man *m)
112: {
113:
1.53 schwarze 114: assert( ! (MAN_HALT & m->flags));
115: if (man_macroend(m))
1.1 kristaps 116: return(1);
117: m->flags |= MAN_HALT;
118: return(0);
119: }
120:
121:
122: int
1.31 schwarze 123: man_parseln(struct man *m, int ln, char *buf, int offs)
1.1 kristaps 124: {
1.25 schwarze 125:
1.54 schwarze 126: m->flags |= MAN_NEWLINE;
127:
1.53 schwarze 128: assert( ! (MAN_HALT & m->flags));
1.59 schwarze 129:
130: return (mandoc_getcontrol(buf, &offs) ?
1.31 schwarze 131: man_pmacro(m, ln, buf, offs) :
132: man_ptext(m, ln, buf, offs));
1.1 kristaps 133: }
134:
135:
136: static void
137: man_free1(struct man *man)
138: {
139:
140: if (man->first)
1.22 schwarze 141: man_node_delete(man, man->first);
1.1 kristaps 142: if (man->meta.title)
143: free(man->meta.title);
144: if (man->meta.source)
145: free(man->meta.source);
1.56 schwarze 146: if (man->meta.date)
147: free(man->meta.date);
1.1 kristaps 148: if (man->meta.vol)
149: free(man->meta.vol);
1.29 schwarze 150: if (man->meta.msec)
151: free(man->meta.msec);
1.1 kristaps 152: }
153:
154:
1.16 schwarze 155: static void
1.1 kristaps 156: man_alloc1(struct man *m)
157: {
158:
1.16 schwarze 159: memset(&m->meta, 0, sizeof(struct man_meta));
1.1 kristaps 160: m->flags = 0;
1.16 schwarze 161: m->last = mandoc_calloc(1, sizeof(struct man_node));
1.1 kristaps 162: m->first = m->last;
163: m->last->type = MAN_ROOT;
1.22 schwarze 164: m->last->tok = MAN_MAX;
1.1 kristaps 165: m->next = MAN_NEXT_CHILD;
166: }
167:
168:
169: static int
170: man_node_append(struct man *man, struct man_node *p)
171: {
172:
173: assert(man->last);
174: assert(man->first);
175: assert(MAN_ROOT != p->type);
176:
177: switch (man->next) {
178: case (MAN_NEXT_SIBLING):
179: man->last->next = p;
180: p->prev = man->last;
181: p->parent = man->last->parent;
182: break;
183: case (MAN_NEXT_CHILD):
184: man->last->child = p;
185: p->parent = man->last;
186: break;
187: default:
188: abort();
189: /* NOTREACHED */
190: }
1.5 schwarze 191:
1.22 schwarze 192: assert(p->parent);
1.5 schwarze 193: p->parent->nchild++;
1.1 kristaps 194:
1.9 schwarze 195: if ( ! man_valid_pre(man, p))
196: return(0);
197:
198: switch (p->type) {
199: case (MAN_HEAD):
200: assert(MAN_BLOCK == p->parent->type);
201: p->parent->head = p;
202: break;
1.59 schwarze 203: case (MAN_TAIL):
204: assert(MAN_BLOCK == p->parent->type);
205: p->parent->tail = p;
206: break;
1.9 schwarze 207: case (MAN_BODY):
208: assert(MAN_BLOCK == p->parent->type);
209: p->parent->body = p;
210: break;
211: default:
212: break;
213: }
214:
1.1 kristaps 215: man->last = p;
216:
217: switch (p->type) {
1.52 schwarze 218: case (MAN_TBL):
219: /* FALLTHROUGH */
1.1 kristaps 220: case (MAN_TEXT):
221: if ( ! man_valid_post(man))
222: return(0);
223: break;
224: default:
225: break;
226: }
227:
228: return(1);
229: }
230:
231:
232: static struct man_node *
1.54 schwarze 233: man_node_alloc(struct man *m, int line, int pos,
234: enum man_type type, enum mant tok)
1.1 kristaps 235: {
236: struct man_node *p;
237:
1.16 schwarze 238: p = mandoc_calloc(1, sizeof(struct man_node));
1.1 kristaps 239: p->line = line;
240: p->pos = pos;
241: p->type = type;
242: p->tok = tok;
1.54 schwarze 243:
244: if (MAN_NEWLINE & m->flags)
245: p->flags |= MAN_LINE;
246: m->flags &= ~MAN_NEWLINE;
1.1 kristaps 247: return(p);
248: }
249:
250:
251: int
1.22 schwarze 252: man_elem_alloc(struct man *m, int line, int pos, enum mant tok)
1.1 kristaps 253: {
254: struct man_node *p;
255:
1.54 schwarze 256: p = man_node_alloc(m, line, pos, MAN_ELEM, tok);
1.10 schwarze 257: if ( ! man_node_append(m, p))
258: return(0);
259: m->next = MAN_NEXT_CHILD;
260: return(1);
1.1 kristaps 261: }
262:
263:
264: int
1.59 schwarze 265: man_tail_alloc(struct man *m, int line, int pos, enum mant tok)
266: {
267: struct man_node *p;
268:
269: p = man_node_alloc(m, line, pos, MAN_TAIL, tok);
270: if ( ! man_node_append(m, p))
271: return(0);
272: m->next = MAN_NEXT_CHILD;
273: return(1);
274: }
275:
276:
277: int
1.22 schwarze 278: man_head_alloc(struct man *m, int line, int pos, enum mant tok)
1.9 schwarze 279: {
280: struct man_node *p;
281:
1.54 schwarze 282: p = man_node_alloc(m, line, pos, MAN_HEAD, tok);
1.9 schwarze 283: if ( ! man_node_append(m, p))
284: return(0);
285: m->next = MAN_NEXT_CHILD;
286: return(1);
287: }
288:
289:
290: int
1.22 schwarze 291: man_body_alloc(struct man *m, int line, int pos, enum mant tok)
1.9 schwarze 292: {
293: struct man_node *p;
294:
1.54 schwarze 295: p = man_node_alloc(m, line, pos, MAN_BODY, tok);
1.9 schwarze 296: if ( ! man_node_append(m, p))
297: return(0);
298: m->next = MAN_NEXT_CHILD;
299: return(1);
300: }
301:
302:
303: int
1.22 schwarze 304: man_block_alloc(struct man *m, int line, int pos, enum mant tok)
1.9 schwarze 305: {
306: struct man_node *p;
307:
1.54 schwarze 308: p = man_node_alloc(m, line, pos, MAN_BLOCK, tok);
1.9 schwarze 309: if ( ! man_node_append(m, p))
310: return(0);
311: m->next = MAN_NEXT_CHILD;
312: return(1);
313: }
314:
1.27 schwarze 315: int
316: man_word_alloc(struct man *m, int line, int pos, const char *word)
1.1 kristaps 317: {
1.10 schwarze 318: struct man_node *n;
1.27 schwarze 319: size_t sv, len;
320:
321: len = strlen(word);
1.1 kristaps 322:
1.54 schwarze 323: n = man_node_alloc(m, line, pos, MAN_TEXT, MAN_MAX);
1.16 schwarze 324: n->string = mandoc_malloc(len + 1);
1.27 schwarze 325: sv = strlcpy(n->string, word, len + 1);
1.10 schwarze 326:
327: /* Prohibit truncation. */
328: assert(sv < len + 1);
329:
330: if ( ! man_node_append(m, n))
1.1 kristaps 331: return(0);
1.27 schwarze 332:
1.10 schwarze 333: m->next = MAN_NEXT_SIBLING;
334: return(1);
335: }
336:
337:
1.22 schwarze 338: /*
339: * Free all of the resources held by a node. This does NOT unlink a
340: * node from its context; for that, see man_node_unlink().
341: */
342: static void
1.1 kristaps 343: man_node_free(struct man_node *p)
344: {
345:
346: if (p->string)
347: free(p->string);
348: free(p);
349: }
350:
351:
352: void
1.22 schwarze 353: man_node_delete(struct man *m, struct man_node *p)
1.1 kristaps 354: {
355:
1.22 schwarze 356: while (p->child)
357: man_node_delete(m, p->child);
358:
359: man_node_unlink(m, p);
1.1 kristaps 360: man_node_free(p);
361: }
362:
1.57 schwarze 363: int
364: man_addeqn(struct man *m, const struct eqn *ep)
365: {
366: struct man_node *n;
367:
368: assert( ! (MAN_HALT & m->flags));
369:
1.60 ! schwarze 370: n = man_node_alloc(m, ep->ln, ep->pos, MAN_EQN, MAN_MAX);
1.57 schwarze 371: n->eqn = ep;
372:
373: if ( ! man_node_append(m, n))
374: return(0);
375:
376: m->next = MAN_NEXT_SIBLING;
1.60 ! schwarze 377: return(man_descope(m, ep->ln, ep->pos));
1.57 schwarze 378: }
1.1 kristaps 379:
1.52 schwarze 380: int
381: man_addspan(struct man *m, const struct tbl_span *sp)
382: {
1.55 schwarze 383: struct man_node *n;
1.52 schwarze 384:
1.53 schwarze 385: assert( ! (MAN_HALT & m->flags));
1.55 schwarze 386:
387: n = man_node_alloc(m, sp->line, 0, MAN_TBL, MAN_MAX);
388: n->span = sp;
389:
390: if ( ! man_node_append(m, n))
1.52 schwarze 391: return(0);
1.55 schwarze 392:
393: m->next = MAN_NEXT_SIBLING;
394: return(man_descope(m, sp->line, 0));
1.52 schwarze 395: }
396:
397: static int
398: man_descope(struct man *m, int line, int offs)
399: {
400: /*
401: * Co-ordinate what happens with having a next-line scope open:
402: * first close out the element scope (if applicable), then close
403: * out the block scope (also if applicable).
404: */
405:
406: if (MAN_ELINE & m->flags) {
407: m->flags &= ~MAN_ELINE;
408: if ( ! man_unscope(m, m->last->parent, MANDOCERR_MAX))
409: return(0);
410: }
411:
412: if ( ! (MAN_BLINE & m->flags))
413: return(1);
414: m->flags &= ~MAN_BLINE;
415:
416: if ( ! man_unscope(m, m->last->parent, MANDOCERR_MAX))
417: return(0);
418: return(man_body_alloc(m, line, offs, m->last->tok));
419: }
420:
1.1 kristaps 421: static int
1.31 schwarze 422: man_ptext(struct man *m, int line, char *buf, int offs)
1.1 kristaps 423: {
1.27 schwarze 424: int i;
1.26 schwarze 425:
1.10 schwarze 426: /* Literal free-form text whitespace is preserved. */
427:
428: if (MAN_LITERAL & m->flags) {
1.31 schwarze 429: if ( ! man_word_alloc(m, line, offs, buf + offs))
1.10 schwarze 430: return(0);
1.52 schwarze 431: return(man_descope(m, line, offs));
1.10 schwarze 432: }
433:
1.27 schwarze 434: /* Pump blank lines directly into the backend. */
1.10 schwarze 435:
1.31 schwarze 436: for (i = offs; ' ' == buf[i]; i++)
1.10 schwarze 437: /* Skip leading whitespace. */ ;
1.18 schwarze 438:
439: if ('\0' == buf[i]) {
1.27 schwarze 440: /* Allocate a blank entry. */
1.31 schwarze 441: if ( ! man_word_alloc(m, line, offs, ""))
1.10 schwarze 442: return(0);
1.52 schwarze 443: return(man_descope(m, line, offs));
1.10 schwarze 444: }
1.1 kristaps 445:
1.27 schwarze 446: /*
447: * Warn if the last un-escaped character is whitespace. Then
448: * strip away the remaining spaces (tabs stay!).
449: */
1.18 schwarze 450:
1.27 schwarze 451: i = (int)strlen(buf);
452: assert(i);
1.18 schwarze 453:
1.27 schwarze 454: if (' ' == buf[i - 1] || '\t' == buf[i - 1]) {
455: if (i > 1 && '\\' != buf[i - 2])
1.51 schwarze 456: man_pmsg(m, line, i - 1, MANDOCERR_EOLNSPACE);
1.18 schwarze 457:
1.27 schwarze 458: for (--i; i && ' ' == buf[i]; i--)
459: /* Spin back to non-space. */ ;
1.10 schwarze 460:
1.27 schwarze 461: /* Jump ahead of escaped whitespace. */
462: i += '\\' == buf[i] ? 2 : 1;
1.18 schwarze 463:
1.27 schwarze 464: buf[i] = '\0';
1.10 schwarze 465: }
1.9 schwarze 466:
1.31 schwarze 467: if ( ! man_word_alloc(m, line, offs, buf + offs))
1.1 kristaps 468: return(0);
1.28 schwarze 469:
470: /*
471: * End-of-sentence check. If the last character is an unescaped
472: * EOS character, then flag the node as being the end of a
473: * sentence. The front-end will know how to interpret this.
474: */
475:
476: assert(i);
1.37 schwarze 477: if (mandoc_eos(buf, (size_t)i, 0))
1.28 schwarze 478: m->last->flags |= MAN_EOS;
1.10 schwarze 479:
1.52 schwarze 480: return(man_descope(m, line, offs));
1.1 kristaps 481: }
482:
1.53 schwarze 483: static int
1.31 schwarze 484: man_pmacro(struct man *m, int ln, char *buf, int offs)
1.1 kristaps 485: {
1.59 schwarze 486: int i, ppos;
1.22 schwarze 487: enum mant tok;
1.10 schwarze 488: char mac[5];
489: struct man_node *n;
1.1 kristaps 490:
1.59 schwarze 491: if ('"' == buf[offs]) {
492: man_pmsg(m, ln, offs, MANDOCERR_BADCOMMENT);
493: return(1);
494: } else if ('\0' == buf[offs])
1.17 schwarze 495: return(1);
1.1 kristaps 496:
1.59 schwarze 497: ppos = offs;
1.1 kristaps 498:
1.23 schwarze 499: /*
1.59 schwarze 500: * Copy the first word into a nil-terminated buffer.
501: * Stop copying when a tab, space, or eoln is encountered.
1.23 schwarze 502: */
1.27 schwarze 503:
1.59 schwarze 504: i = 0;
505: while (i < 4 && '\0' != buf[offs] &&
506: ' ' != buf[offs] && '\t' != buf[offs])
507: mac[i++] = buf[offs++];
1.1 kristaps 508:
1.59 schwarze 509: mac[i] = '\0';
1.1 kristaps 510:
1.59 schwarze 511: tok = (i > 0 && i < 4) ? man_hash_find(mac) : MAN_MAX;
1.1 kristaps 512:
1.40 schwarze 513: if (MAN_MAX == tok) {
1.59 schwarze 514: mandoc_vmsg(MANDOCERR_MACRO, m->parse, ln,
515: ppos, "%s", buf + ppos - 1);
1.1 kristaps 516: return(1);
517: }
518:
519: /* The macro is sane. Jump to the next word. */
520:
1.59 schwarze 521: while (buf[offs] && ' ' == buf[offs])
522: offs++;
1.18 schwarze 523:
1.27 schwarze 524: /*
525: * Trailing whitespace. Note that tabs are allowed to be passed
526: * into the parser as "text", so we only warn about spaces here.
527: */
1.18 schwarze 528:
1.59 schwarze 529: if ('\0' == buf[offs] && ' ' == buf[offs - 1])
530: man_pmsg(m, ln, offs - 1, MANDOCERR_EOLNSPACE);
1.1 kristaps 531:
1.21 schwarze 532: /*
1.46 schwarze 533: * Remove prior ELINE macro, as it's being clobbered by a new
1.21 schwarze 534: * macro. Note that NSCOPED macros do not close out ELINE
535: * macros---they don't print text---so we let those slip by.
536: */
537:
1.22 schwarze 538: if ( ! (MAN_NSCOPED & man_macros[tok].flags) &&
1.21 schwarze 539: m->flags & MAN_ELINE) {
1.10 schwarze 540: n = m->last;
1.46 schwarze 541: assert(MAN_TEXT != n->type);
1.21 schwarze 542:
1.49 schwarze 543: /* Remove repeated NSCOPED macros causing ELINE. */
544:
1.46 schwarze 545: if (MAN_NSCOPED & man_macros[n->tok].flags)
546: n = n->parent;
1.21 schwarze 547:
1.59 schwarze 548: mandoc_vmsg(MANDOCERR_LINESCOPE, m->parse, n->line,
549: n->pos, "%s", man_macronames[n->tok]);
1.10 schwarze 550:
1.22 schwarze 551: man_node_delete(m, n);
1.10 schwarze 552: m->flags &= ~MAN_ELINE;
553: }
554:
1.24 schwarze 555: /*
556: * Save the fact that we're in the next-line for a block. In
557: * this way, embedded roff instructions can "remember" state
558: * when they exit.
559: */
560:
561: if (MAN_BLINE & m->flags)
562: m->flags |= MAN_BPLINE;
563:
564: /* Call to handler... */
1.1 kristaps 565:
1.22 schwarze 566: assert(man_macros[tok].fp);
1.59 schwarze 567: if ( ! (*man_macros[tok].fp)(m, tok, ln, ppos, &offs, buf))
1.1 kristaps 568: goto err;
569:
1.21 schwarze 570: /*
571: * We weren't in a block-line scope when entering the
572: * above-parsed macro, so return.
573: */
574:
1.24 schwarze 575: if ( ! (MAN_BPLINE & m->flags)) {
1.21 schwarze 576: m->flags &= ~MAN_ILINE;
1.9 schwarze 577: return(1);
1.21 schwarze 578: }
1.24 schwarze 579: m->flags &= ~MAN_BPLINE;
1.21 schwarze 580:
581: /*
582: * If we're in a block scope, then allow this macro to slip by
583: * without closing scope around it.
584: */
585:
586: if (MAN_ILINE & m->flags) {
587: m->flags &= ~MAN_ILINE;
588: return(1);
589: }
1.9 schwarze 590:
591: /*
592: * If we've opened a new next-line element scope, then return
593: * now, as the next line will close out the block scope.
594: */
595:
596: if (MAN_ELINE & m->flags)
597: return(1);
598:
599: /* Close out the block scope opened in the prior line. */
1.1 kristaps 600:
1.9 schwarze 601: assert(MAN_BLINE & m->flags);
602: m->flags &= ~MAN_BLINE;
1.1 kristaps 603:
1.33 schwarze 604: if ( ! man_unscope(m, m->last->parent, MANDOCERR_MAX))
1.9 schwarze 605: return(0);
1.59 schwarze 606: return(man_body_alloc(m, ln, ppos, m->last->tok));
1.1 kristaps 607:
608: err: /* Error out. */
609:
610: m->flags |= MAN_HALT;
611: return(0);
612: }
1.21 schwarze 613:
1.22 schwarze 614: /*
615: * Unlink a node from its context. If "m" is provided, the last parse
616: * point will also be adjusted accordingly.
617: */
618: static void
1.21 schwarze 619: man_node_unlink(struct man *m, struct man_node *n)
620: {
621:
1.22 schwarze 622: /* Adjust siblings. */
623:
624: if (n->prev)
1.21 schwarze 625: n->prev->next = n->next;
1.22 schwarze 626: if (n->next)
627: n->next->prev = n->prev;
628:
629: /* Adjust parent. */
630:
631: if (n->parent) {
632: n->parent->nchild--;
633: if (n->parent->child == n)
634: n->parent->child = n->prev ? n->prev : n->next;
635: }
636:
637: /* Adjust parse point, if applicable. */
638:
639: if (m && m->last == n) {
640: /*XXX: this can occur when bailing from validation. */
641: /*assert(NULL == n->next);*/
642: if (n->prev) {
1.21 schwarze 643: m->last = n->prev;
644: m->next = MAN_NEXT_SIBLING;
1.22 schwarze 645: } else {
1.21 schwarze 646: m->last = n->parent;
647: m->next = MAN_NEXT_CHILD;
648: }
649: }
650:
1.22 schwarze 651: if (m && m->first == n)
652: m->first = NULL;
1.4 schwarze 653: }