Annotation of src/usr.bin/mandoc/mdoc.c, Revision 1.88
1.88 ! schwarze 1: /* $Id: mdoc.c,v 1.87 2012/05/24 23:33:23 schwarze Exp $ */
1.1 kristaps 2: /*
1.77 schwarze 3: * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
1.61 schwarze 4: * Copyright (c) 2010 Ingo Schwarze <schwarze@openbsd.org>
1.1 kristaps 5: *
6: * Permission to use, copy, modify, and distribute this software for any
1.3 schwarze 7: * purpose with or without fee is hereby granted, provided that the above
8: * copyright notice and this permission notice appear in all copies.
1.1 kristaps 9: *
1.3 schwarze 10: * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
11: * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
12: * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
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.
1.1 kristaps 17: */
1.28 schwarze 18: #include <sys/types.h>
19:
1.1 kristaps 20: #include <assert.h>
21: #include <stdarg.h>
22: #include <stdio.h>
23: #include <stdlib.h>
24: #include <string.h>
1.39 schwarze 25: #include <time.h>
1.1 kristaps 26:
1.83 schwarze 27: #include "mdoc.h"
1.54 schwarze 28: #include "mandoc.h"
1.1 kristaps 29: #include "libmdoc.h"
1.32 schwarze 30: #include "libmandoc.h"
1.1 kristaps 31:
32: const char *const __mdoc_macronames[MDOC_MAX] = {
1.7 schwarze 33: "Ap", "Dd", "Dt", "Os",
1.1 kristaps 34: "Sh", "Ss", "Pp", "D1",
35: "Dl", "Bd", "Ed", "Bl",
36: "El", "It", "Ad", "An",
37: "Ar", "Cd", "Cm", "Dv",
38: "Er", "Ev", "Ex", "Fa",
39: "Fd", "Fl", "Fn", "Ft",
40: "Ic", "In", "Li", "Nd",
41: "Nm", "Op", "Ot", "Pa",
42: "Rv", "St", "Va", "Vt",
43: /* LINTED */
1.33 schwarze 44: "Xr", "%A", "%B", "%D",
1.1 kristaps 45: /* LINTED */
1.33 schwarze 46: "%I", "%J", "%N", "%O",
1.1 kristaps 47: /* LINTED */
1.33 schwarze 48: "%P", "%R", "%T", "%V",
1.1 kristaps 49: "Ac", "Ao", "Aq", "At",
50: "Bc", "Bf", "Bo", "Bq",
51: "Bsx", "Bx", "Db", "Dc",
52: "Do", "Dq", "Ec", "Ef",
53: "Em", "Eo", "Fx", "Ms",
54: "No", "Ns", "Nx", "Ox",
55: "Pc", "Pf", "Po", "Pq",
56: "Qc", "Ql", "Qo", "Qq",
57: "Re", "Rs", "Sc", "So",
58: "Sq", "Sm", "Sx", "Sy",
59: "Tn", "Ux", "Xc", "Xo",
60: "Fo", "Fc", "Oo", "Oc",
61: "Bk", "Ek", "Bt", "Hf",
1.7 schwarze 62: "Fr", "Ud", "Lb", "Lp",
63: "Lk", "Mt", "Brq", "Bro",
1.1 kristaps 64: /* LINTED */
1.33 schwarze 65: "Brc", "%C", "Es", "En",
1.1 kristaps 66: /* LINTED */
1.33 schwarze 67: "Dx", "%Q", "br", "sp",
1.31 schwarze 68: /* LINTED */
1.77 schwarze 69: "%U", "Ta"
1.1 kristaps 70: };
71:
72: const char *const __mdoc_argnames[MDOC_ARG_MAX] = {
73: "split", "nosplit", "ragged",
74: "unfilled", "literal", "file",
75: "offset", "bullet", "dash",
76: "hyphen", "item", "enum",
77: "tag", "diag", "hang",
78: "ohang", "inset", "column",
79: "width", "compact", "std",
80: "filled", "words", "emphasis",
1.30 schwarze 81: "symbolic", "nested", "centered"
1.1 kristaps 82: };
83:
84: const char * const *mdoc_macronames = __mdoc_macronames;
85: const char * const *mdoc_argnames = __mdoc_argnames;
86:
1.40 schwarze 87: static void mdoc_node_free(struct mdoc_node *);
88: static void mdoc_node_unlink(struct mdoc *,
89: struct mdoc_node *);
1.1 kristaps 90: static void mdoc_free1(struct mdoc *);
1.32 schwarze 91: static void mdoc_alloc1(struct mdoc *);
1.1 kristaps 92: static struct mdoc_node *node_alloc(struct mdoc *, int, int,
1.37 schwarze 93: enum mdoct, enum mdoc_type);
1.1 kristaps 94: static int node_append(struct mdoc *,
95: struct mdoc_node *);
1.84 schwarze 96: #if 0
97: static int mdoc_preptext(struct mdoc *, int, char *, int);
98: #endif
1.53 schwarze 99: static int mdoc_ptext(struct mdoc *, int, char *, int);
100: static int mdoc_pmacro(struct mdoc *, int, char *, int);
1.1 kristaps 101:
102: const struct mdoc_node *
103: mdoc_node(const struct mdoc *m)
104: {
105:
1.78 schwarze 106: assert( ! (MDOC_HALT & m->flags));
107: return(m->first);
1.1 kristaps 108: }
109:
110:
111: const struct mdoc_meta *
112: mdoc_meta(const struct mdoc *m)
113: {
114:
1.78 schwarze 115: assert( ! (MDOC_HALT & m->flags));
116: return(&m->meta);
1.1 kristaps 117: }
118:
119:
1.8 schwarze 120: /*
121: * Frees volatile resources (parse tree, meta-data, fields).
122: */
1.1 kristaps 123: static void
124: mdoc_free1(struct mdoc *mdoc)
125: {
126:
127: if (mdoc->first)
1.40 schwarze 128: mdoc_node_delete(mdoc, mdoc->first);
1.1 kristaps 129: if (mdoc->meta.title)
130: free(mdoc->meta.title);
131: if (mdoc->meta.os)
132: free(mdoc->meta.os);
133: if (mdoc->meta.name)
134: free(mdoc->meta.name);
135: if (mdoc->meta.arch)
136: free(mdoc->meta.arch);
137: if (mdoc->meta.vol)
138: free(mdoc->meta.vol);
1.50 schwarze 139: if (mdoc->meta.msec)
140: free(mdoc->meta.msec);
1.82 schwarze 141: if (mdoc->meta.date)
142: free(mdoc->meta.date);
1.1 kristaps 143: }
144:
145:
1.8 schwarze 146: /*
147: * Allocate all volatile resources (parse tree, meta-data, fields).
148: */
1.32 schwarze 149: static void
1.1 kristaps 150: mdoc_alloc1(struct mdoc *mdoc)
151: {
152:
1.32 schwarze 153: memset(&mdoc->meta, 0, sizeof(struct mdoc_meta));
1.1 kristaps 154: mdoc->flags = 0;
1.9 schwarze 155: mdoc->lastnamed = mdoc->lastsec = SEC_NONE;
1.32 schwarze 156: mdoc->last = mandoc_calloc(1, sizeof(struct mdoc_node));
1.1 kristaps 157: mdoc->first = mdoc->last;
158: mdoc->last->type = MDOC_ROOT;
1.86 schwarze 159: mdoc->last->tok = MDOC_MAX;
1.1 kristaps 160: mdoc->next = MDOC_NEXT_CHILD;
161: }
162:
163:
164: /*
1.8 schwarze 165: * Free up volatile resources (see mdoc_free1()) then re-initialises the
166: * data with mdoc_alloc1(). After invocation, parse data has been reset
167: * and the parser is ready for re-invocation on a new tree; however,
168: * cross-parse non-volatile data is kept intact.
1.1 kristaps 169: */
1.32 schwarze 170: void
1.1 kristaps 171: mdoc_reset(struct mdoc *mdoc)
172: {
173:
174: mdoc_free1(mdoc);
1.32 schwarze 175: mdoc_alloc1(mdoc);
1.1 kristaps 176: }
177:
178:
179: /*
1.8 schwarze 180: * Completely free up all volatile and non-volatile parse resources.
181: * After invocation, the pointer is no longer usable.
1.1 kristaps 182: */
183: void
184: mdoc_free(struct mdoc *mdoc)
185: {
186:
187: mdoc_free1(mdoc);
188: free(mdoc);
189: }
190:
191:
1.8 schwarze 192: /*
193: * Allocate volatile and non-volatile parse resources.
194: */
1.1 kristaps 195: struct mdoc *
1.87 schwarze 196: mdoc_alloc(struct roff *roff, struct mparse *parse, char *defos)
1.1 kristaps 197: {
198: struct mdoc *p;
199:
1.32 schwarze 200: p = mandoc_calloc(1, sizeof(struct mdoc));
201:
1.83 schwarze 202: p->parse = parse;
1.87 schwarze 203: p->defos = defos;
1.84 schwarze 204: p->roff = roff;
1.1 kristaps 205:
1.32 schwarze 206: mdoc_hash_init();
207: mdoc_alloc1(p);
208: return(p);
1.1 kristaps 209: }
210:
211:
212: /*
213: * Climb back up the parse tree, validating open scopes. Mostly calls
1.8 schwarze 214: * through to macro_end() in macro.c.
1.1 kristaps 215: */
216: int
217: mdoc_endparse(struct mdoc *m)
218: {
219:
1.78 schwarze 220: assert( ! (MDOC_HALT & m->flags));
221: if (mdoc_macroend(m))
1.1 kristaps 222: return(1);
223: m->flags |= MDOC_HALT;
224: return(0);
1.81 schwarze 225: }
226:
227: int
228: mdoc_addeqn(struct mdoc *m, const struct eqn *ep)
229: {
230: struct mdoc_node *n;
231:
232: assert( ! (MDOC_HALT & m->flags));
233:
234: /* No text before an initial macro. */
235:
236: if (SEC_NONE == m->lastnamed) {
1.84 schwarze 237: mdoc_pmsg(m, ep->ln, ep->pos, MANDOCERR_NOTEXT);
1.81 schwarze 238: return(1);
239: }
240:
1.84 schwarze 241: n = node_alloc(m, ep->ln, ep->pos, MDOC_MAX, MDOC_EQN);
1.81 schwarze 242: n->eqn = ep;
243:
244: if ( ! node_append(m, n))
245: return(0);
246:
247: m->next = MDOC_NEXT_SIBLING;
248: return(1);
1.1 kristaps 249: }
250:
1.77 schwarze 251: int
252: mdoc_addspan(struct mdoc *m, const struct tbl_span *sp)
253: {
1.79 schwarze 254: struct mdoc_node *n;
1.77 schwarze 255:
1.78 schwarze 256: assert( ! (MDOC_HALT & m->flags));
1.77 schwarze 257:
258: /* No text before an initial macro. */
259:
260: if (SEC_NONE == m->lastnamed) {
1.79 schwarze 261: mdoc_pmsg(m, sp->line, 0, MANDOCERR_NOTEXT);
1.77 schwarze 262: return(1);
263: }
264:
1.79 schwarze 265: n = node_alloc(m, sp->line, 0, MDOC_MAX, MDOC_TBL);
266: n->span = sp;
267:
268: if ( ! node_append(m, n))
269: return(0);
270:
271: m->next = MDOC_NEXT_SIBLING;
272: return(1);
1.77 schwarze 273: }
274:
1.1 kristaps 275:
276: /*
277: * Main parse routine. Parses a single line -- really just hands off to
1.45 schwarze 278: * the macro (mdoc_pmacro()) or text parser (mdoc_ptext()).
1.1 kristaps 279: */
280: int
1.53 schwarze 281: mdoc_parseln(struct mdoc *m, int ln, char *buf, int offs)
1.1 kristaps 282: {
283:
1.78 schwarze 284: assert( ! (MDOC_HALT & m->flags));
1.1 kristaps 285:
1.47 schwarze 286: m->flags |= MDOC_NEWLINE;
1.60 schwarze 287:
288: /*
289: * Let the roff nS register switch SYNOPSIS mode early,
290: * such that the parser knows at all times
291: * whether this mode is on or off.
292: * Note that this mode is also switched by the Sh macro.
293: */
1.84 schwarze 294: if (roff_regisset(m->roff, REG_nS)) {
295: if (roff_regget(m->roff, REG_nS))
1.60 schwarze 296: m->flags |= MDOC_SYNOPSIS;
297: else
298: m->flags &= ~MDOC_SYNOPSIS;
299: }
300:
1.88 ! schwarze 301: return(roff_getcontrol(m->roff, buf, &offs) ?
1.53 schwarze 302: mdoc_pmacro(m, ln, buf, offs) :
303: mdoc_ptext(m, ln, buf, offs));
1.1 kristaps 304: }
305:
1.5 schwarze 306: int
1.58 schwarze 307: mdoc_macro(MACRO_PROT_ARGS)
1.1 kristaps 308: {
1.46 schwarze 309: assert(tok < MDOC_MAX);
310:
311: /* If we're in the body, deny prologue calls. */
1.37 schwarze 312:
1.1 kristaps 313: if (MDOC_PROLOGUE & mdoc_macros[tok].flags &&
1.76 schwarze 314: MDOC_PBODY & m->flags) {
315: mdoc_pmsg(m, line, ppos, MANDOCERR_BADBODY);
316: return(1);
317: }
1.46 schwarze 318:
319: /* If we're in the prologue, deny "body" macros. */
320:
1.1 kristaps 321: if ( ! (MDOC_PROLOGUE & mdoc_macros[tok].flags) &&
1.39 schwarze 322: ! (MDOC_PBODY & m->flags)) {
1.76 schwarze 323: mdoc_pmsg(m, line, ppos, MANDOCERR_BADPROLOG);
1.71 schwarze 324: if (NULL == m->meta.msec)
325: m->meta.msec = mandoc_strdup("1");
1.39 schwarze 326: if (NULL == m->meta.title)
1.56 schwarze 327: m->meta.title = mandoc_strdup("UNKNOWN");
1.39 schwarze 328: if (NULL == m->meta.vol)
1.56 schwarze 329: m->meta.vol = mandoc_strdup("LOCAL");
1.39 schwarze 330: if (NULL == m->meta.os)
1.56 schwarze 331: m->meta.os = mandoc_strdup("LOCAL");
1.80 schwarze 332: if (NULL == m->meta.date)
1.83 schwarze 333: m->meta.date = mandoc_normdate
334: (m->parse, NULL, line, ppos);
1.39 schwarze 335: m->flags |= MDOC_PBODY;
336: }
1.1 kristaps 337:
1.58 schwarze 338: return((*mdoc_macros[tok].fp)(m, tok, line, ppos, pos, buf));
1.1 kristaps 339: }
340:
341:
342: static int
343: node_append(struct mdoc *mdoc, struct mdoc_node *p)
344: {
345:
346: assert(mdoc->last);
347: assert(mdoc->first);
348: assert(MDOC_ROOT != p->type);
349:
350: switch (mdoc->next) {
351: case (MDOC_NEXT_SIBLING):
352: mdoc->last->next = p;
353: p->prev = mdoc->last;
354: p->parent = mdoc->last->parent;
355: break;
356: case (MDOC_NEXT_CHILD):
357: mdoc->last->child = p;
358: p->parent = mdoc->last;
359: break;
360: default:
361: abort();
362: /* NOTREACHED */
363: }
364:
1.10 schwarze 365: p->parent->nchild++;
366:
1.75 schwarze 367: /*
368: * Copy over the normalised-data pointer of our parent. Not
369: * everybody has one, but copying a null pointer is fine.
370: */
371:
372: switch (p->type) {
373: case (MDOC_BODY):
374: /* FALLTHROUGH */
375: case (MDOC_TAIL):
376: /* FALLTHROUGH */
377: case (MDOC_HEAD):
378: p->norm = p->parent->norm;
379: break;
380: default:
381: break;
382: }
383:
1.1 kristaps 384: if ( ! mdoc_valid_pre(mdoc, p))
385: return(0);
386:
387: switch (p->type) {
388: case (MDOC_HEAD):
389: assert(MDOC_BLOCK == p->parent->type);
390: p->parent->head = p;
391: break;
392: case (MDOC_TAIL):
393: assert(MDOC_BLOCK == p->parent->type);
394: p->parent->tail = p;
395: break;
396: case (MDOC_BODY):
1.59 schwarze 397: if (p->end)
398: break;
1.1 kristaps 399: assert(MDOC_BLOCK == p->parent->type);
400: p->parent->body = p;
401: break;
402: default:
403: break;
404: }
405:
406: mdoc->last = p;
407:
408: switch (p->type) {
1.77 schwarze 409: case (MDOC_TBL):
410: /* FALLTHROUGH */
1.1 kristaps 411: case (MDOC_TEXT):
412: if ( ! mdoc_valid_post(mdoc))
413: return(0);
414: break;
415: default:
416: break;
417: }
418:
419: return(1);
420: }
421:
422:
423: static struct mdoc_node *
1.37 schwarze 424: node_alloc(struct mdoc *m, int line, int pos,
425: enum mdoct tok, enum mdoc_type type)
1.1 kristaps 426: {
427: struct mdoc_node *p;
428:
1.32 schwarze 429: p = mandoc_calloc(1, sizeof(struct mdoc_node));
1.19 schwarze 430: p->sec = m->lastsec;
1.1 kristaps 431: p->line = line;
432: p->pos = pos;
433: p->tok = tok;
1.37 schwarze 434: p->type = type;
1.58 schwarze 435:
436: /* Flag analysis. */
437:
1.60 schwarze 438: if (MDOC_SYNOPSIS & m->flags)
439: p->flags |= MDOC_SYNPRETTY;
440: else
441: p->flags &= ~MDOC_SYNPRETTY;
1.47 schwarze 442: if (MDOC_NEWLINE & m->flags)
443: p->flags |= MDOC_LINE;
444: m->flags &= ~MDOC_NEWLINE;
1.58 schwarze 445:
1.1 kristaps 446: return(p);
447: }
448:
449:
450: int
1.37 schwarze 451: mdoc_tail_alloc(struct mdoc *m, int line, int pos, enum mdoct tok)
1.1 kristaps 452: {
453: struct mdoc_node *p;
454:
1.19 schwarze 455: p = node_alloc(m, line, pos, tok, MDOC_TAIL);
1.25 schwarze 456: if ( ! node_append(m, p))
457: return(0);
458: m->next = MDOC_NEXT_CHILD;
459: return(1);
1.1 kristaps 460: }
461:
462:
463: int
1.37 schwarze 464: mdoc_head_alloc(struct mdoc *m, int line, int pos, enum mdoct tok)
1.1 kristaps 465: {
466: struct mdoc_node *p;
467:
1.19 schwarze 468: assert(m->first);
469: assert(m->last);
1.1 kristaps 470:
1.19 schwarze 471: p = node_alloc(m, line, pos, tok, MDOC_HEAD);
1.25 schwarze 472: if ( ! node_append(m, p))
473: return(0);
474: m->next = MDOC_NEXT_CHILD;
475: return(1);
1.1 kristaps 476: }
477:
478:
479: int
1.37 schwarze 480: mdoc_body_alloc(struct mdoc *m, int line, int pos, enum mdoct tok)
1.1 kristaps 481: {
482: struct mdoc_node *p;
483:
1.19 schwarze 484: p = node_alloc(m, line, pos, tok, MDOC_BODY);
1.25 schwarze 485: if ( ! node_append(m, p))
486: return(0);
487: m->next = MDOC_NEXT_CHILD;
1.59 schwarze 488: return(1);
489: }
490:
491:
492: int
493: mdoc_endbody_alloc(struct mdoc *m, int line, int pos, enum mdoct tok,
494: struct mdoc_node *body, enum mdoc_endbody end)
495: {
496: struct mdoc_node *p;
497:
498: p = node_alloc(m, line, pos, tok, MDOC_BODY);
499: p->pending = body;
500: p->end = end;
501: if ( ! node_append(m, p))
502: return(0);
503: m->next = MDOC_NEXT_SIBLING;
1.25 schwarze 504: return(1);
1.1 kristaps 505: }
506:
507:
508: int
1.19 schwarze 509: mdoc_block_alloc(struct mdoc *m, int line, int pos,
1.37 schwarze 510: enum mdoct tok, struct mdoc_arg *args)
1.1 kristaps 511: {
512: struct mdoc_node *p;
513:
1.19 schwarze 514: p = node_alloc(m, line, pos, tok, MDOC_BLOCK);
1.4 schwarze 515: p->args = args;
516: if (p->args)
1.1 kristaps 517: (args->refcnt)++;
1.75 schwarze 518:
519: switch (tok) {
520: case (MDOC_Bd):
521: /* FALLTHROUGH */
522: case (MDOC_Bf):
523: /* FALLTHROUGH */
524: case (MDOC_Bl):
525: /* FALLTHROUGH */
526: case (MDOC_Rs):
527: p->norm = mandoc_calloc(1, sizeof(union mdoc_data));
528: break;
529: default:
530: break;
531: }
532:
1.25 schwarze 533: if ( ! node_append(m, p))
534: return(0);
535: m->next = MDOC_NEXT_CHILD;
536: return(1);
1.1 kristaps 537: }
538:
539:
540: int
1.19 schwarze 541: mdoc_elem_alloc(struct mdoc *m, int line, int pos,
1.37 schwarze 542: enum mdoct tok, struct mdoc_arg *args)
1.1 kristaps 543: {
544: struct mdoc_node *p;
545:
1.19 schwarze 546: p = node_alloc(m, line, pos, tok, MDOC_ELEM);
1.4 schwarze 547: p->args = args;
548: if (p->args)
1.1 kristaps 549: (args->refcnt)++;
1.75 schwarze 550:
551: switch (tok) {
552: case (MDOC_An):
553: p->norm = mandoc_calloc(1, sizeof(union mdoc_data));
554: break;
555: default:
556: break;
557: }
558:
1.25 schwarze 559: if ( ! node_append(m, p))
560: return(0);
561: m->next = MDOC_NEXT_CHILD;
562: return(1);
1.1 kristaps 563: }
564:
1.46 schwarze 565: int
566: mdoc_word_alloc(struct mdoc *m, int line, int pos, const char *p)
1.1 kristaps 567: {
1.19 schwarze 568: struct mdoc_node *n;
1.1 kristaps 569:
1.46 schwarze 570: n = node_alloc(m, line, pos, MDOC_MAX, MDOC_TEXT);
1.85 schwarze 571: n->string = roff_strdup(m->roff, p);
1.19 schwarze 572:
1.25 schwarze 573: if ( ! node_append(m, n))
574: return(0);
1.46 schwarze 575:
1.25 schwarze 576: m->next = MDOC_NEXT_SIBLING;
577: return(1);
1.19 schwarze 578: }
579:
580:
1.61 schwarze 581: static void
1.1 kristaps 582: mdoc_node_free(struct mdoc_node *p)
583: {
1.61 schwarze 584:
1.75 schwarze 585: if (MDOC_BLOCK == p->type || MDOC_ELEM == p->type)
586: free(p->norm);
1.1 kristaps 587: if (p->string)
588: free(p->string);
589: if (p->args)
590: mdoc_argv_free(p->args);
591: free(p);
592: }
593:
594:
1.40 schwarze 595: static void
596: mdoc_node_unlink(struct mdoc *m, struct mdoc_node *n)
597: {
598:
599: /* Adjust siblings. */
600:
601: if (n->prev)
602: n->prev->next = n->next;
603: if (n->next)
604: n->next->prev = n->prev;
605:
606: /* Adjust parent. */
607:
608: if (n->parent) {
609: n->parent->nchild--;
610: if (n->parent->child == n)
611: n->parent->child = n->prev ? n->prev : n->next;
1.72 schwarze 612: if (n->parent->last == n)
613: n->parent->last = n->prev ? n->prev : NULL;
1.40 schwarze 614: }
615:
616: /* Adjust parse point, if applicable. */
617:
618: if (m && m->last == n) {
619: if (n->prev) {
620: m->last = n->prev;
621: m->next = MDOC_NEXT_SIBLING;
622: } else {
623: m->last = n->parent;
624: m->next = MDOC_NEXT_CHILD;
625: }
626: }
627:
628: if (m && m->first == n)
629: m->first = NULL;
630: }
631:
632:
1.1 kristaps 633: void
1.40 schwarze 634: mdoc_node_delete(struct mdoc *m, struct mdoc_node *p)
1.1 kristaps 635: {
636:
1.40 schwarze 637: while (p->child) {
638: assert(p->nchild);
639: mdoc_node_delete(m, p->child);
640: }
641: assert(0 == p->nchild);
1.1 kristaps 642:
1.40 schwarze 643: mdoc_node_unlink(m, p);
1.1 kristaps 644: mdoc_node_free(p);
645: }
646:
1.84 schwarze 647: #if 0
648: /*
649: * Pre-treat a text line.
650: * Text lines can consist of equations, which must be handled apart from
651: * the regular text.
652: * Thus, use this function to step through a line checking if it has any
653: * equations embedded in it.
654: * This must handle multiple equations AND equations that do not end at
655: * the end-of-line, i.e., will re-enter in the next roff parse.
656: */
657: static int
658: mdoc_preptext(struct mdoc *m, int line, char *buf, int offs)
659: {
660: char *start, *end;
661: char delim;
662:
663: while ('\0' != buf[offs]) {
664: /* Mark starting position if eqn is set. */
665: start = NULL;
666: if ('\0' != (delim = roff_eqndelim(m->roff)))
667: if (NULL != (start = strchr(buf + offs, delim)))
668: *start++ = '\0';
669:
670: /* Parse text as normal. */
671: if ( ! mdoc_ptext(m, line, buf, offs))
672: return(0);
673:
674: /* Continue only if an equation exists. */
675: if (NULL == start)
676: break;
677:
678: /* Read past the end of the equation. */
679: offs += start - (buf + offs);
680: assert(start == &buf[offs]);
681: if (NULL != (end = strchr(buf + offs, delim))) {
682: *end++ = '\0';
683: while (' ' == *end)
684: end++;
685: }
686:
687: /* Parse the equation itself. */
688: roff_openeqn(m->roff, NULL, line, offs, buf);
689:
690: /* Process a finished equation? */
691: if (roff_closeeqn(m->roff))
692: if ( ! mdoc_addeqn(m, roff_eqn(m->roff)))
693: return(0);
694: offs += (end - (buf + offs));
695: }
696:
697: return(1);
698: }
699: #endif
1.1 kristaps 700:
701: /*
702: * Parse free-form text, that is, a line that does not begin with the
703: * control character.
704: */
705: static int
1.53 schwarze 706: mdoc_ptext(struct mdoc *m, int line, char *buf, int offs)
1.1 kristaps 707: {
1.56 schwarze 708: char *c, *ws, *end;
709: struct mdoc_node *n;
1.44 schwarze 710:
1.46 schwarze 711: /* No text before an initial macro. */
712:
1.76 schwarze 713: if (SEC_NONE == m->lastnamed) {
714: mdoc_pmsg(m, line, offs, MANDOCERR_NOTEXT);
715: return(1);
716: }
1.46 schwarze 717:
1.56 schwarze 718: assert(m->last);
719: n = m->last;
720:
721: /*
722: * Divert directly to list processing if we're encountering a
723: * columnar MDOC_BLOCK with or without a prior MDOC_BLOCK entry
724: * (a MDOC_BODY means it's already open, in which case we should
725: * process within its context in the normal way).
726: */
727:
728: if (MDOC_Bl == n->tok && MDOC_BODY == n->type &&
1.75 schwarze 729: LIST_column == n->norm->Bl.type) {
1.56 schwarze 730: /* `Bl' is open without any children. */
731: m->flags |= MDOC_FREECOL;
732: return(mdoc_macro(m, MDOC_It, line, offs, &offs, buf));
733: }
734:
735: if (MDOC_It == n->tok && MDOC_BLOCK == n->type &&
736: NULL != n->parent &&
737: MDOC_Bl == n->parent->tok &&
1.75 schwarze 738: LIST_column == n->parent->norm->Bl.type) {
1.56 schwarze 739: /* `Bl' has block-level `It' children. */
740: m->flags |= MDOC_FREECOL;
741: return(mdoc_macro(m, MDOC_It, line, offs, &offs, buf));
742: }
743:
1.52 schwarze 744: /*
745: * Search for the beginning of unescaped trailing whitespace (ws)
746: * and for the first character not to be output (end).
747: */
1.56 schwarze 748:
749: /* FIXME: replace with strcspn(). */
1.52 schwarze 750: ws = NULL;
1.53 schwarze 751: for (c = end = buf + offs; *c; c++) {
1.52 schwarze 752: switch (*c) {
753: case ' ':
754: if (NULL == ws)
755: ws = c;
756: continue;
757: case '\t':
758: /*
759: * Always warn about trailing tabs,
760: * even outside literal context,
761: * where they should be put on the next line.
762: */
763: if (NULL == ws)
764: ws = c;
765: /*
766: * Strip trailing tabs in literal context only;
767: * outside, they affect the next line.
768: */
769: if (MDOC_LITERAL & m->flags)
770: continue;
771: break;
772: case '\\':
773: /* Skip the escaped character, too, if any. */
774: if (c[1])
775: c++;
776: /* FALLTHROUGH */
777: default:
778: ws = NULL;
779: break;
780: }
781: end = c + 1;
782: }
783: *end = '\0';
1.19 schwarze 784:
1.52 schwarze 785: if (ws)
1.76 schwarze 786: mdoc_pmsg(m, line, (int)(ws-buf), MANDOCERR_EOLNSPACE);
1.34 schwarze 787:
1.53 schwarze 788: if ('\0' == buf[offs] && ! (MDOC_LITERAL & m->flags)) {
1.76 schwarze 789: mdoc_pmsg(m, line, (int)(c-buf), MANDOCERR_NOBLANKLN);
1.46 schwarze 790:
1.40 schwarze 791: /*
1.66 schwarze 792: * Insert a `sp' in the case of a blank line. Technically,
1.46 schwarze 793: * blank lines aren't allowed, but enough manuals assume this
794: * behaviour that we want to work around it.
1.40 schwarze 795: */
1.66 schwarze 796: if ( ! mdoc_elem_alloc(m, line, offs, MDOC_sp, NULL))
1.42 schwarze 797: return(0);
1.46 schwarze 798:
1.42 schwarze 799: m->next = MDOC_NEXT_SIBLING;
800: return(1);
1.38 schwarze 801: }
1.1 kristaps 802:
1.53 schwarze 803: if ( ! mdoc_word_alloc(m, line, offs, buf+offs))
1.1 kristaps 804: return(0);
805:
1.52 schwarze 806: if (MDOC_LITERAL & m->flags)
807: return(1);
808:
1.35 schwarze 809: /*
1.48 schwarze 810: * End-of-sentence check. If the last character is an unescaped
811: * EOS character, then flag the node as being the end of a
812: * sentence. The front-end will know how to interpret this.
1.35 schwarze 813: */
814:
1.52 schwarze 815: assert(buf < end);
1.48 schwarze 816:
1.62 schwarze 817: if (mandoc_eos(buf+offs, (size_t)(end-buf-offs), 0))
1.48 schwarze 818: m->last->flags |= MDOC_EOS;
819:
1.1 kristaps 820: return(1);
821: }
1.19 schwarze 822:
1.1 kristaps 823:
824: /*
825: * Parse a macro line, that is, a line beginning with the control
826: * character.
827: */
1.61 schwarze 828: static int
1.53 schwarze 829: mdoc_pmacro(struct mdoc *m, int ln, char *buf, int offs)
1.1 kristaps 830: {
1.56 schwarze 831: enum mdoct tok;
1.83 schwarze 832: int i, sv;
1.56 schwarze 833: char mac[5];
834: struct mdoc_node *n;
1.1 kristaps 835:
1.83 schwarze 836: /* Empty post-control lines are ignored. */
1.1 kristaps 837:
1.83 schwarze 838: if ('"' == buf[offs]) {
839: mdoc_pmsg(m, ln, offs, MANDOCERR_BADCOMMENT);
840: return(1);
841: } else if ('\0' == buf[offs])
1.1 kristaps 842: return(1);
843:
1.83 schwarze 844: sv = offs;
1.47 schwarze 845:
1.63 schwarze 846: /*
1.64 schwarze 847: * Copy the first word into a nil-terminated buffer.
848: * Stop copying when a tab, space, or eoln is encountered.
1.63 schwarze 849: */
1.1 kristaps 850:
1.83 schwarze 851: i = 0;
852: while (i < 4 && '\0' != buf[offs] &&
853: ' ' != buf[offs] && '\t' != buf[offs])
854: mac[i++] = buf[offs++];
855:
856: mac[i] = '\0';
857:
858: tok = (i > 1 || i < 4) ? mdoc_hash_find(mac) : MDOC_MAX;
1.1 kristaps 859:
1.65 schwarze 860: if (MDOC_MAX == tok) {
1.83 schwarze 861: mandoc_vmsg(MANDOCERR_MACRO, m->parse,
862: ln, sv, "%s", buf + sv - 1);
1.1 kristaps 863: return(1);
864: }
865:
1.63 schwarze 866: /* Disregard the first trailing tab, if applicable. */
867:
1.83 schwarze 868: if ('\t' == buf[offs])
869: offs++;
1.63 schwarze 870:
871: /* Jump to the next non-whitespace word. */
1.1 kristaps 872:
1.83 schwarze 873: while (buf[offs] && ' ' == buf[offs])
874: offs++;
1.34 schwarze 875:
1.46 schwarze 876: /*
877: * Trailing whitespace. Note that tabs are allowed to be passed
878: * into the parser as "text", so we only warn about spaces here.
879: */
1.34 schwarze 880:
1.83 schwarze 881: if ('\0' == buf[offs] && ' ' == buf[offs - 1])
882: mdoc_pmsg(m, ln, offs - 1, MANDOCERR_EOLNSPACE);
1.1 kristaps 883:
1.56 schwarze 884: /*
885: * If an initial macro or a list invocation, divert directly
886: * into macro processing.
887: */
888:
889: if (NULL == m->last || MDOC_It == tok || MDOC_El == tok) {
1.83 schwarze 890: if ( ! mdoc_macro(m, tok, ln, sv, &offs, buf))
1.56 schwarze 891: goto err;
892: return(1);
893: }
894:
895: n = m->last;
896: assert(m->last);
897:
898: /*
899: * If the first macro of a `Bl -column', open an `It' block
900: * context around the parsed macro.
901: */
902:
903: if (MDOC_Bl == n->tok && MDOC_BODY == n->type &&
1.75 schwarze 904: LIST_column == n->norm->Bl.type) {
1.56 schwarze 905: m->flags |= MDOC_FREECOL;
1.58 schwarze 906: if ( ! mdoc_macro(m, MDOC_It, ln, sv, &sv, buf))
1.56 schwarze 907: goto err;
908: return(1);
909: }
910:
911: /*
912: * If we're following a block-level `It' within a `Bl -column'
913: * context (perhaps opened in the above block or in ptext()),
914: * then open an `It' block context around the parsed macro.
1.24 schwarze 915: */
1.56 schwarze 916:
917: if (MDOC_It == n->tok && MDOC_BLOCK == n->type &&
918: NULL != n->parent &&
919: MDOC_Bl == n->parent->tok &&
1.75 schwarze 920: LIST_column == n->parent->norm->Bl.type) {
1.56 schwarze 921: m->flags |= MDOC_FREECOL;
922: if ( ! mdoc_macro(m, MDOC_It, ln, sv, &sv, buf))
923: goto err;
924: return(1);
925: }
926:
927: /* Normal processing of a macro. */
928:
1.83 schwarze 929: if ( ! mdoc_macro(m, tok, ln, sv, &offs, buf))
1.1 kristaps 930: goto err;
931:
932: return(1);
933:
934: err: /* Error out. */
935:
936: m->flags |= MDOC_HALT;
937: return(0);
938: }
1.24 schwarze 939:
1.83 schwarze 940: enum mdelim
941: mdoc_isdelim(const char *p)
942: {
943:
944: if ('\0' == p[0])
945: return(DELIM_NONE);
946:
947: if ('\0' == p[1])
948: switch (p[0]) {
949: case('('):
950: /* FALLTHROUGH */
951: case('['):
952: return(DELIM_OPEN);
953: case('|'):
954: return(DELIM_MIDDLE);
955: case('.'):
956: /* FALLTHROUGH */
957: case(','):
958: /* FALLTHROUGH */
959: case(';'):
960: /* FALLTHROUGH */
961: case(':'):
962: /* FALLTHROUGH */
963: case('?'):
964: /* FALLTHROUGH */
965: case('!'):
966: /* FALLTHROUGH */
967: case(')'):
968: /* FALLTHROUGH */
969: case(']'):
970: return(DELIM_CLOSE);
971: default:
972: return(DELIM_NONE);
973: }
974:
975: if ('\\' != p[0])
976: return(DELIM_NONE);
1.24 schwarze 977:
1.83 schwarze 978: if (0 == strcmp(p + 1, "."))
979: return(DELIM_CLOSE);
980: if (0 == strcmp(p + 1, "*(Ba"))
981: return(DELIM_MIDDLE);
982:
983: return(DELIM_NONE);
984: }
![[BACK]](/icons/back.gif){kind=icon}
Return to mdoc.c CVS log ![[TXT]](/icons/text.gif){kind=icon}
![[DIR]](/icons/dir.gif){kind=icon}
Up to [local] / src / usr.bin / mandoc