Annotation of src/usr.bin/mandoc/mdoc.c, Revision 1.86
1.86 ! schwarze 1: /* $Id: mdoc.c,v 1.85 2011/09/18 15:54:48 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 *
1.84 schwarze 196: mdoc_alloc(struct roff *roff, struct mparse *parse)
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;
1.84 schwarze 203: p->roff = roff;
1.1 kristaps 204:
1.32 schwarze 205: mdoc_hash_init();
206: mdoc_alloc1(p);
207: return(p);
1.1 kristaps 208: }
209:
210:
211: /*
212: * Climb back up the parse tree, validating open scopes. Mostly calls
1.8 schwarze 213: * through to macro_end() in macro.c.
1.1 kristaps 214: */
215: int
216: mdoc_endparse(struct mdoc *m)
217: {
218:
1.78 schwarze 219: assert( ! (MDOC_HALT & m->flags));
220: if (mdoc_macroend(m))
1.1 kristaps 221: return(1);
222: m->flags |= MDOC_HALT;
223: return(0);
1.81 schwarze 224: }
225:
226: int
227: mdoc_addeqn(struct mdoc *m, const struct eqn *ep)
228: {
229: struct mdoc_node *n;
230:
231: assert( ! (MDOC_HALT & m->flags));
232:
233: /* No text before an initial macro. */
234:
235: if (SEC_NONE == m->lastnamed) {
1.84 schwarze 236: mdoc_pmsg(m, ep->ln, ep->pos, MANDOCERR_NOTEXT);
1.81 schwarze 237: return(1);
238: }
239:
1.84 schwarze 240: n = node_alloc(m, ep->ln, ep->pos, MDOC_MAX, MDOC_EQN);
1.81 schwarze 241: n->eqn = ep;
242:
243: if ( ! node_append(m, n))
244: return(0);
245:
246: m->next = MDOC_NEXT_SIBLING;
247: return(1);
1.1 kristaps 248: }
249:
1.77 schwarze 250: int
251: mdoc_addspan(struct mdoc *m, const struct tbl_span *sp)
252: {
1.79 schwarze 253: struct mdoc_node *n;
1.77 schwarze 254:
1.78 schwarze 255: assert( ! (MDOC_HALT & m->flags));
1.77 schwarze 256:
257: /* No text before an initial macro. */
258:
259: if (SEC_NONE == m->lastnamed) {
1.79 schwarze 260: mdoc_pmsg(m, sp->line, 0, MANDOCERR_NOTEXT);
1.77 schwarze 261: return(1);
262: }
263:
1.79 schwarze 264: n = node_alloc(m, sp->line, 0, MDOC_MAX, MDOC_TBL);
265: n->span = sp;
266:
267: if ( ! node_append(m, n))
268: return(0);
269:
270: m->next = MDOC_NEXT_SIBLING;
271: return(1);
1.77 schwarze 272: }
273:
1.1 kristaps 274:
275: /*
276: * Main parse routine. Parses a single line -- really just hands off to
1.45 schwarze 277: * the macro (mdoc_pmacro()) or text parser (mdoc_ptext()).
1.1 kristaps 278: */
279: int
1.53 schwarze 280: mdoc_parseln(struct mdoc *m, int ln, char *buf, int offs)
1.1 kristaps 281: {
282:
1.78 schwarze 283: assert( ! (MDOC_HALT & m->flags));
1.1 kristaps 284:
1.47 schwarze 285: m->flags |= MDOC_NEWLINE;
1.60 schwarze 286:
287: /*
288: * Let the roff nS register switch SYNOPSIS mode early,
289: * such that the parser knows at all times
290: * whether this mode is on or off.
291: * Note that this mode is also switched by the Sh macro.
292: */
1.84 schwarze 293: if (roff_regisset(m->roff, REG_nS)) {
294: if (roff_regget(m->roff, REG_nS))
1.60 schwarze 295: m->flags |= MDOC_SYNOPSIS;
296: else
297: m->flags &= ~MDOC_SYNOPSIS;
298: }
299:
1.83 schwarze 300: return(mandoc_getcontrol(buf, &offs) ?
1.53 schwarze 301: mdoc_pmacro(m, ln, buf, offs) :
302: mdoc_ptext(m, ln, buf, offs));
1.1 kristaps 303: }
304:
1.5 schwarze 305: int
1.58 schwarze 306: mdoc_macro(MACRO_PROT_ARGS)
1.1 kristaps 307: {
1.46 schwarze 308: assert(tok < MDOC_MAX);
309:
310: /* If we're in the body, deny prologue calls. */
1.37 schwarze 311:
1.1 kristaps 312: if (MDOC_PROLOGUE & mdoc_macros[tok].flags &&
1.76 schwarze 313: MDOC_PBODY & m->flags) {
314: mdoc_pmsg(m, line, ppos, MANDOCERR_BADBODY);
315: return(1);
316: }
1.46 schwarze 317:
318: /* If we're in the prologue, deny "body" macros. */
319:
1.1 kristaps 320: if ( ! (MDOC_PROLOGUE & mdoc_macros[tok].flags) &&
1.39 schwarze 321: ! (MDOC_PBODY & m->flags)) {
1.76 schwarze 322: mdoc_pmsg(m, line, ppos, MANDOCERR_BADPROLOG);
1.71 schwarze 323: if (NULL == m->meta.msec)
324: m->meta.msec = mandoc_strdup("1");
1.39 schwarze 325: if (NULL == m->meta.title)
1.56 schwarze 326: m->meta.title = mandoc_strdup("UNKNOWN");
1.39 schwarze 327: if (NULL == m->meta.vol)
1.56 schwarze 328: m->meta.vol = mandoc_strdup("LOCAL");
1.39 schwarze 329: if (NULL == m->meta.os)
1.56 schwarze 330: m->meta.os = mandoc_strdup("LOCAL");
1.80 schwarze 331: if (NULL == m->meta.date)
1.83 schwarze 332: m->meta.date = mandoc_normdate
333: (m->parse, NULL, line, ppos);
1.39 schwarze 334: m->flags |= MDOC_PBODY;
335: }
1.1 kristaps 336:
1.58 schwarze 337: return((*mdoc_macros[tok].fp)(m, tok, line, ppos, pos, buf));
1.1 kristaps 338: }
339:
340:
341: static int
342: node_append(struct mdoc *mdoc, struct mdoc_node *p)
343: {
344:
345: assert(mdoc->last);
346: assert(mdoc->first);
347: assert(MDOC_ROOT != p->type);
348:
349: switch (mdoc->next) {
350: case (MDOC_NEXT_SIBLING):
351: mdoc->last->next = p;
352: p->prev = mdoc->last;
353: p->parent = mdoc->last->parent;
354: break;
355: case (MDOC_NEXT_CHILD):
356: mdoc->last->child = p;
357: p->parent = mdoc->last;
358: break;
359: default:
360: abort();
361: /* NOTREACHED */
362: }
363:
1.10 schwarze 364: p->parent->nchild++;
365:
1.75 schwarze 366: /*
367: * Copy over the normalised-data pointer of our parent. Not
368: * everybody has one, but copying a null pointer is fine.
369: */
370:
371: switch (p->type) {
372: case (MDOC_BODY):
373: /* FALLTHROUGH */
374: case (MDOC_TAIL):
375: /* FALLTHROUGH */
376: case (MDOC_HEAD):
377: p->norm = p->parent->norm;
378: break;
379: default:
380: break;
381: }
382:
1.1 kristaps 383: if ( ! mdoc_valid_pre(mdoc, p))
384: return(0);
385:
386: switch (p->type) {
387: case (MDOC_HEAD):
388: assert(MDOC_BLOCK == p->parent->type);
389: p->parent->head = p;
390: break;
391: case (MDOC_TAIL):
392: assert(MDOC_BLOCK == p->parent->type);
393: p->parent->tail = p;
394: break;
395: case (MDOC_BODY):
1.59 schwarze 396: if (p->end)
397: break;
1.1 kristaps 398: assert(MDOC_BLOCK == p->parent->type);
399: p->parent->body = p;
400: break;
401: default:
402: break;
403: }
404:
405: mdoc->last = p;
406:
407: switch (p->type) {
1.77 schwarze 408: case (MDOC_TBL):
409: /* FALLTHROUGH */
1.1 kristaps 410: case (MDOC_TEXT):
411: if ( ! mdoc_valid_post(mdoc))
412: return(0);
413: break;
414: default:
415: break;
416: }
417:
418: return(1);
419: }
420:
421:
422: static struct mdoc_node *
1.37 schwarze 423: node_alloc(struct mdoc *m, int line, int pos,
424: enum mdoct tok, enum mdoc_type type)
1.1 kristaps 425: {
426: struct mdoc_node *p;
427:
1.32 schwarze 428: p = mandoc_calloc(1, sizeof(struct mdoc_node));
1.19 schwarze 429: p->sec = m->lastsec;
1.1 kristaps 430: p->line = line;
431: p->pos = pos;
432: p->tok = tok;
1.37 schwarze 433: p->type = type;
1.58 schwarze 434:
435: /* Flag analysis. */
436:
1.60 schwarze 437: if (MDOC_SYNOPSIS & m->flags)
438: p->flags |= MDOC_SYNPRETTY;
439: else
440: p->flags &= ~MDOC_SYNPRETTY;
1.47 schwarze 441: if (MDOC_NEWLINE & m->flags)
442: p->flags |= MDOC_LINE;
443: m->flags &= ~MDOC_NEWLINE;
1.58 schwarze 444:
1.1 kristaps 445: return(p);
446: }
447:
448:
449: int
1.37 schwarze 450: mdoc_tail_alloc(struct mdoc *m, int line, int pos, enum mdoct tok)
1.1 kristaps 451: {
452: struct mdoc_node *p;
453:
1.19 schwarze 454: p = node_alloc(m, line, pos, tok, MDOC_TAIL);
1.25 schwarze 455: if ( ! node_append(m, p))
456: return(0);
457: m->next = MDOC_NEXT_CHILD;
458: return(1);
1.1 kristaps 459: }
460:
461:
462: int
1.37 schwarze 463: mdoc_head_alloc(struct mdoc *m, int line, int pos, enum mdoct tok)
1.1 kristaps 464: {
465: struct mdoc_node *p;
466:
1.19 schwarze 467: assert(m->first);
468: assert(m->last);
1.1 kristaps 469:
1.19 schwarze 470: p = node_alloc(m, line, pos, tok, MDOC_HEAD);
1.25 schwarze 471: if ( ! node_append(m, p))
472: return(0);
473: m->next = MDOC_NEXT_CHILD;
474: return(1);
1.1 kristaps 475: }
476:
477:
478: int
1.37 schwarze 479: mdoc_body_alloc(struct mdoc *m, int line, int pos, enum mdoct tok)
1.1 kristaps 480: {
481: struct mdoc_node *p;
482:
1.19 schwarze 483: p = node_alloc(m, line, pos, tok, MDOC_BODY);
1.25 schwarze 484: if ( ! node_append(m, p))
485: return(0);
486: m->next = MDOC_NEXT_CHILD;
1.59 schwarze 487: return(1);
488: }
489:
490:
491: int
492: mdoc_endbody_alloc(struct mdoc *m, int line, int pos, enum mdoct tok,
493: struct mdoc_node *body, enum mdoc_endbody end)
494: {
495: struct mdoc_node *p;
496:
497: p = node_alloc(m, line, pos, tok, MDOC_BODY);
498: p->pending = body;
499: p->end = end;
500: if ( ! node_append(m, p))
501: return(0);
502: m->next = MDOC_NEXT_SIBLING;
1.25 schwarze 503: return(1);
1.1 kristaps 504: }
505:
506:
507: int
1.19 schwarze 508: mdoc_block_alloc(struct mdoc *m, int line, int pos,
1.37 schwarze 509: enum mdoct tok, struct mdoc_arg *args)
1.1 kristaps 510: {
511: struct mdoc_node *p;
512:
1.19 schwarze 513: p = node_alloc(m, line, pos, tok, MDOC_BLOCK);
1.4 schwarze 514: p->args = args;
515: if (p->args)
1.1 kristaps 516: (args->refcnt)++;
1.75 schwarze 517:
518: switch (tok) {
519: case (MDOC_Bd):
520: /* FALLTHROUGH */
521: case (MDOC_Bf):
522: /* FALLTHROUGH */
523: case (MDOC_Bl):
524: /* FALLTHROUGH */
525: case (MDOC_Rs):
526: p->norm = mandoc_calloc(1, sizeof(union mdoc_data));
527: break;
528: default:
529: break;
530: }
531:
1.25 schwarze 532: if ( ! node_append(m, p))
533: return(0);
534: m->next = MDOC_NEXT_CHILD;
535: return(1);
1.1 kristaps 536: }
537:
538:
539: int
1.19 schwarze 540: mdoc_elem_alloc(struct mdoc *m, int line, int pos,
1.37 schwarze 541: enum mdoct tok, struct mdoc_arg *args)
1.1 kristaps 542: {
543: struct mdoc_node *p;
544:
1.19 schwarze 545: p = node_alloc(m, line, pos, tok, MDOC_ELEM);
1.4 schwarze 546: p->args = args;
547: if (p->args)
1.1 kristaps 548: (args->refcnt)++;
1.75 schwarze 549:
550: switch (tok) {
551: case (MDOC_An):
552: p->norm = mandoc_calloc(1, sizeof(union mdoc_data));
553: break;
554: default:
555: break;
556: }
557:
1.25 schwarze 558: if ( ! node_append(m, p))
559: return(0);
560: m->next = MDOC_NEXT_CHILD;
561: return(1);
1.1 kristaps 562: }
563:
1.46 schwarze 564: int
565: mdoc_word_alloc(struct mdoc *m, int line, int pos, const char *p)
1.1 kristaps 566: {
1.19 schwarze 567: struct mdoc_node *n;
1.1 kristaps 568:
1.46 schwarze 569: n = node_alloc(m, line, pos, MDOC_MAX, MDOC_TEXT);
1.85 schwarze 570: n->string = roff_strdup(m->roff, p);
1.19 schwarze 571:
1.25 schwarze 572: if ( ! node_append(m, n))
573: return(0);
1.46 schwarze 574:
1.25 schwarze 575: m->next = MDOC_NEXT_SIBLING;
576: return(1);
1.19 schwarze 577: }
578:
579:
1.61 schwarze 580: static void
1.1 kristaps 581: mdoc_node_free(struct mdoc_node *p)
582: {
1.61 schwarze 583:
1.75 schwarze 584: if (MDOC_BLOCK == p->type || MDOC_ELEM == p->type)
585: free(p->norm);
1.1 kristaps 586: if (p->string)
587: free(p->string);
588: if (p->args)
589: mdoc_argv_free(p->args);
590: free(p);
591: }
592:
593:
1.40 schwarze 594: static void
595: mdoc_node_unlink(struct mdoc *m, struct mdoc_node *n)
596: {
597:
598: /* Adjust siblings. */
599:
600: if (n->prev)
601: n->prev->next = n->next;
602: if (n->next)
603: n->next->prev = n->prev;
604:
605: /* Adjust parent. */
606:
607: if (n->parent) {
608: n->parent->nchild--;
609: if (n->parent->child == n)
610: n->parent->child = n->prev ? n->prev : n->next;
1.72 schwarze 611: if (n->parent->last == n)
612: n->parent->last = n->prev ? n->prev : NULL;
1.40 schwarze 613: }
614:
615: /* Adjust parse point, if applicable. */
616:
617: if (m && m->last == n) {
618: if (n->prev) {
619: m->last = n->prev;
620: m->next = MDOC_NEXT_SIBLING;
621: } else {
622: m->last = n->parent;
623: m->next = MDOC_NEXT_CHILD;
624: }
625: }
626:
627: if (m && m->first == n)
628: m->first = NULL;
629: }
630:
631:
1.1 kristaps 632: void
1.40 schwarze 633: mdoc_node_delete(struct mdoc *m, struct mdoc_node *p)
1.1 kristaps 634: {
635:
1.40 schwarze 636: while (p->child) {
637: assert(p->nchild);
638: mdoc_node_delete(m, p->child);
639: }
640: assert(0 == p->nchild);
1.1 kristaps 641:
1.40 schwarze 642: mdoc_node_unlink(m, p);
1.1 kristaps 643: mdoc_node_free(p);
644: }
645:
1.84 schwarze 646: #if 0
647: /*
648: * Pre-treat a text line.
649: * Text lines can consist of equations, which must be handled apart from
650: * the regular text.
651: * Thus, use this function to step through a line checking if it has any
652: * equations embedded in it.
653: * This must handle multiple equations AND equations that do not end at
654: * the end-of-line, i.e., will re-enter in the next roff parse.
655: */
656: static int
657: mdoc_preptext(struct mdoc *m, int line, char *buf, int offs)
658: {
659: char *start, *end;
660: char delim;
661:
662: while ('\0' != buf[offs]) {
663: /* Mark starting position if eqn is set. */
664: start = NULL;
665: if ('\0' != (delim = roff_eqndelim(m->roff)))
666: if (NULL != (start = strchr(buf + offs, delim)))
667: *start++ = '\0';
668:
669: /* Parse text as normal. */
670: if ( ! mdoc_ptext(m, line, buf, offs))
671: return(0);
672:
673: /* Continue only if an equation exists. */
674: if (NULL == start)
675: break;
676:
677: /* Read past the end of the equation. */
678: offs += start - (buf + offs);
679: assert(start == &buf[offs]);
680: if (NULL != (end = strchr(buf + offs, delim))) {
681: *end++ = '\0';
682: while (' ' == *end)
683: end++;
684: }
685:
686: /* Parse the equation itself. */
687: roff_openeqn(m->roff, NULL, line, offs, buf);
688:
689: /* Process a finished equation? */
690: if (roff_closeeqn(m->roff))
691: if ( ! mdoc_addeqn(m, roff_eqn(m->roff)))
692: return(0);
693: offs += (end - (buf + offs));
694: }
695:
696: return(1);
697: }
698: #endif
1.1 kristaps 699:
700: /*
701: * Parse free-form text, that is, a line that does not begin with the
702: * control character.
703: */
704: static int
1.53 schwarze 705: mdoc_ptext(struct mdoc *m, int line, char *buf, int offs)
1.1 kristaps 706: {
1.56 schwarze 707: char *c, *ws, *end;
708: struct mdoc_node *n;
1.44 schwarze 709:
1.46 schwarze 710: /* No text before an initial macro. */
711:
1.76 schwarze 712: if (SEC_NONE == m->lastnamed) {
713: mdoc_pmsg(m, line, offs, MANDOCERR_NOTEXT);
714: return(1);
715: }
1.46 schwarze 716:
1.56 schwarze 717: assert(m->last);
718: n = m->last;
719:
720: /*
721: * Divert directly to list processing if we're encountering a
722: * columnar MDOC_BLOCK with or without a prior MDOC_BLOCK entry
723: * (a MDOC_BODY means it's already open, in which case we should
724: * process within its context in the normal way).
725: */
726:
727: if (MDOC_Bl == n->tok && MDOC_BODY == n->type &&
1.75 schwarze 728: LIST_column == n->norm->Bl.type) {
1.56 schwarze 729: /* `Bl' is open without any children. */
730: m->flags |= MDOC_FREECOL;
731: return(mdoc_macro(m, MDOC_It, line, offs, &offs, buf));
732: }
733:
734: if (MDOC_It == n->tok && MDOC_BLOCK == n->type &&
735: NULL != n->parent &&
736: MDOC_Bl == n->parent->tok &&
1.75 schwarze 737: LIST_column == n->parent->norm->Bl.type) {
1.56 schwarze 738: /* `Bl' has block-level `It' children. */
739: m->flags |= MDOC_FREECOL;
740: return(mdoc_macro(m, MDOC_It, line, offs, &offs, buf));
741: }
742:
1.52 schwarze 743: /*
744: * Search for the beginning of unescaped trailing whitespace (ws)
745: * and for the first character not to be output (end).
746: */
1.56 schwarze 747:
748: /* FIXME: replace with strcspn(). */
1.52 schwarze 749: ws = NULL;
1.53 schwarze 750: for (c = end = buf + offs; *c; c++) {
1.52 schwarze 751: switch (*c) {
752: case ' ':
753: if (NULL == ws)
754: ws = c;
755: continue;
756: case '\t':
757: /*
758: * Always warn about trailing tabs,
759: * even outside literal context,
760: * where they should be put on the next line.
761: */
762: if (NULL == ws)
763: ws = c;
764: /*
765: * Strip trailing tabs in literal context only;
766: * outside, they affect the next line.
767: */
768: if (MDOC_LITERAL & m->flags)
769: continue;
770: break;
771: case '\\':
772: /* Skip the escaped character, too, if any. */
773: if (c[1])
774: c++;
775: /* FALLTHROUGH */
776: default:
777: ws = NULL;
778: break;
779: }
780: end = c + 1;
781: }
782: *end = '\0';
1.19 schwarze 783:
1.52 schwarze 784: if (ws)
1.76 schwarze 785: mdoc_pmsg(m, line, (int)(ws-buf), MANDOCERR_EOLNSPACE);
1.34 schwarze 786:
1.53 schwarze 787: if ('\0' == buf[offs] && ! (MDOC_LITERAL & m->flags)) {
1.76 schwarze 788: mdoc_pmsg(m, line, (int)(c-buf), MANDOCERR_NOBLANKLN);
1.46 schwarze 789:
1.40 schwarze 790: /*
1.66 schwarze 791: * Insert a `sp' in the case of a blank line. Technically,
1.46 schwarze 792: * blank lines aren't allowed, but enough manuals assume this
793: * behaviour that we want to work around it.
1.40 schwarze 794: */
1.66 schwarze 795: if ( ! mdoc_elem_alloc(m, line, offs, MDOC_sp, NULL))
1.42 schwarze 796: return(0);
1.46 schwarze 797:
1.42 schwarze 798: m->next = MDOC_NEXT_SIBLING;
799: return(1);
1.38 schwarze 800: }
1.1 kristaps 801:
1.53 schwarze 802: if ( ! mdoc_word_alloc(m, line, offs, buf+offs))
1.1 kristaps 803: return(0);
804:
1.52 schwarze 805: if (MDOC_LITERAL & m->flags)
806: return(1);
807:
1.35 schwarze 808: /*
1.48 schwarze 809: * End-of-sentence check. If the last character is an unescaped
810: * EOS character, then flag the node as being the end of a
811: * sentence. The front-end will know how to interpret this.
1.35 schwarze 812: */
813:
1.52 schwarze 814: assert(buf < end);
1.48 schwarze 815:
1.62 schwarze 816: if (mandoc_eos(buf+offs, (size_t)(end-buf-offs), 0))
1.48 schwarze 817: m->last->flags |= MDOC_EOS;
818:
1.1 kristaps 819: return(1);
820: }
1.19 schwarze 821:
1.1 kristaps 822:
823: /*
824: * Parse a macro line, that is, a line beginning with the control
825: * character.
826: */
1.61 schwarze 827: static int
1.53 schwarze 828: mdoc_pmacro(struct mdoc *m, int ln, char *buf, int offs)
1.1 kristaps 829: {
1.56 schwarze 830: enum mdoct tok;
1.83 schwarze 831: int i, sv;
1.56 schwarze 832: char mac[5];
833: struct mdoc_node *n;
1.1 kristaps 834:
1.83 schwarze 835: /* Empty post-control lines are ignored. */
1.1 kristaps 836:
1.83 schwarze 837: if ('"' == buf[offs]) {
838: mdoc_pmsg(m, ln, offs, MANDOCERR_BADCOMMENT);
839: return(1);
840: } else if ('\0' == buf[offs])
1.1 kristaps 841: return(1);
842:
1.83 schwarze 843: sv = offs;
1.47 schwarze 844:
1.63 schwarze 845: /*
1.64 schwarze 846: * Copy the first word into a nil-terminated buffer.
847: * Stop copying when a tab, space, or eoln is encountered.
1.63 schwarze 848: */
1.1 kristaps 849:
1.83 schwarze 850: i = 0;
851: while (i < 4 && '\0' != buf[offs] &&
852: ' ' != buf[offs] && '\t' != buf[offs])
853: mac[i++] = buf[offs++];
854:
855: mac[i] = '\0';
856:
857: tok = (i > 1 || i < 4) ? mdoc_hash_find(mac) : MDOC_MAX;
1.1 kristaps 858:
1.65 schwarze 859: if (MDOC_MAX == tok) {
1.83 schwarze 860: mandoc_vmsg(MANDOCERR_MACRO, m->parse,
861: ln, sv, "%s", buf + sv - 1);
1.1 kristaps 862: return(1);
863: }
864:
1.63 schwarze 865: /* Disregard the first trailing tab, if applicable. */
866:
1.83 schwarze 867: if ('\t' == buf[offs])
868: offs++;
1.63 schwarze 869:
870: /* Jump to the next non-whitespace word. */
1.1 kristaps 871:
1.83 schwarze 872: while (buf[offs] && ' ' == buf[offs])
873: offs++;
1.34 schwarze 874:
1.46 schwarze 875: /*
876: * Trailing whitespace. Note that tabs are allowed to be passed
877: * into the parser as "text", so we only warn about spaces here.
878: */
1.34 schwarze 879:
1.83 schwarze 880: if ('\0' == buf[offs] && ' ' == buf[offs - 1])
881: mdoc_pmsg(m, ln, offs - 1, MANDOCERR_EOLNSPACE);
1.1 kristaps 882:
1.56 schwarze 883: /*
884: * If an initial macro or a list invocation, divert directly
885: * into macro processing.
886: */
887:
888: if (NULL == m->last || MDOC_It == tok || MDOC_El == tok) {
1.83 schwarze 889: if ( ! mdoc_macro(m, tok, ln, sv, &offs, buf))
1.56 schwarze 890: goto err;
891: return(1);
892: }
893:
894: n = m->last;
895: assert(m->last);
896:
897: /*
898: * If the first macro of a `Bl -column', open an `It' block
899: * context around the parsed macro.
900: */
901:
902: if (MDOC_Bl == n->tok && MDOC_BODY == n->type &&
1.75 schwarze 903: LIST_column == n->norm->Bl.type) {
1.56 schwarze 904: m->flags |= MDOC_FREECOL;
1.58 schwarze 905: if ( ! mdoc_macro(m, MDOC_It, ln, sv, &sv, buf))
1.56 schwarze 906: goto err;
907: return(1);
908: }
909:
910: /*
911: * If we're following a block-level `It' within a `Bl -column'
912: * context (perhaps opened in the above block or in ptext()),
913: * then open an `It' block context around the parsed macro.
1.24 schwarze 914: */
1.56 schwarze 915:
916: if (MDOC_It == n->tok && MDOC_BLOCK == n->type &&
917: NULL != n->parent &&
918: MDOC_Bl == n->parent->tok &&
1.75 schwarze 919: LIST_column == n->parent->norm->Bl.type) {
1.56 schwarze 920: m->flags |= MDOC_FREECOL;
921: if ( ! mdoc_macro(m, MDOC_It, ln, sv, &sv, buf))
922: goto err;
923: return(1);
924: }
925:
926: /* Normal processing of a macro. */
927:
1.83 schwarze 928: if ( ! mdoc_macro(m, tok, ln, sv, &offs, buf))
1.1 kristaps 929: goto err;
930:
931: return(1);
932:
933: err: /* Error out. */
934:
935: m->flags |= MDOC_HALT;
936: return(0);
937: }
1.24 schwarze 938:
1.83 schwarze 939: enum mdelim
940: mdoc_isdelim(const char *p)
941: {
942:
943: if ('\0' == p[0])
944: return(DELIM_NONE);
945:
946: if ('\0' == p[1])
947: switch (p[0]) {
948: case('('):
949: /* FALLTHROUGH */
950: case('['):
951: return(DELIM_OPEN);
952: case('|'):
953: return(DELIM_MIDDLE);
954: case('.'):
955: /* FALLTHROUGH */
956: case(','):
957: /* FALLTHROUGH */
958: case(';'):
959: /* FALLTHROUGH */
960: case(':'):
961: /* FALLTHROUGH */
962: case('?'):
963: /* FALLTHROUGH */
964: case('!'):
965: /* FALLTHROUGH */
966: case(')'):
967: /* FALLTHROUGH */
968: case(']'):
969: return(DELIM_CLOSE);
970: default:
971: return(DELIM_NONE);
972: }
973:
974: if ('\\' != p[0])
975: return(DELIM_NONE);
1.24 schwarze 976:
1.83 schwarze 977: if (0 == strcmp(p + 1, "."))
978: return(DELIM_CLOSE);
979: if (0 == strcmp(p + 1, "*(Ba"))
980: return(DELIM_MIDDLE);
981:
982: return(DELIM_NONE);
983: }
![[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