Annotation of src/usr.bin/mandoc/mdoc.c, Revision 1.98
1.98 ! schwarze 1: /* $Id: mdoc.c,v 1.97 2013/12/30 00:52:18 schwarze Exp $ */
1.1 kristaps 2: /*
1.77 schwarze 3: * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
1.98 ! schwarze 4: * Copyright (c) 2010, 2012, 2013, 2014 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 *
1.93 schwarze 103: mdoc_node(const struct mdoc *mdoc)
1.1 kristaps 104: {
105:
1.93 schwarze 106: assert( ! (MDOC_HALT & mdoc->flags));
107: return(mdoc->first);
1.1 kristaps 108: }
109:
110:
111: const struct mdoc_meta *
1.93 schwarze 112: mdoc_meta(const struct mdoc *mdoc)
1.1 kristaps 113: {
114:
1.93 schwarze 115: assert( ! (MDOC_HALT & mdoc->flags));
116: return(&mdoc->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 *
1.98 ! schwarze 196: mdoc_alloc(struct roff *roff, struct mparse *parse,
! 197: char *defos, int quick)
1.1 kristaps 198: {
199: struct mdoc *p;
200:
1.32 schwarze 201: p = mandoc_calloc(1, sizeof(struct mdoc));
202:
1.83 schwarze 203: p->parse = parse;
1.87 schwarze 204: p->defos = defos;
1.98 ! schwarze 205: p->quick = quick;
1.84 schwarze 206: p->roff = roff;
1.1 kristaps 207:
1.32 schwarze 208: mdoc_hash_init();
209: mdoc_alloc1(p);
210: return(p);
1.1 kristaps 211: }
212:
213:
214: /*
215: * Climb back up the parse tree, validating open scopes. Mostly calls
1.8 schwarze 216: * through to macro_end() in macro.c.
1.1 kristaps 217: */
218: int
1.93 schwarze 219: mdoc_endparse(struct mdoc *mdoc)
1.1 kristaps 220: {
221:
1.93 schwarze 222: assert( ! (MDOC_HALT & mdoc->flags));
223: if (mdoc_macroend(mdoc))
1.1 kristaps 224: return(1);
1.93 schwarze 225: mdoc->flags |= MDOC_HALT;
1.1 kristaps 226: return(0);
1.81 schwarze 227: }
228:
229: int
1.93 schwarze 230: mdoc_addeqn(struct mdoc *mdoc, const struct eqn *ep)
1.81 schwarze 231: {
232: struct mdoc_node *n;
233:
1.93 schwarze 234: assert( ! (MDOC_HALT & mdoc->flags));
1.81 schwarze 235:
236: /* No text before an initial macro. */
237:
1.93 schwarze 238: if (SEC_NONE == mdoc->lastnamed) {
239: mdoc_pmsg(mdoc, ep->ln, ep->pos, MANDOCERR_NOTEXT);
1.81 schwarze 240: return(1);
241: }
242:
1.93 schwarze 243: n = node_alloc(mdoc, ep->ln, ep->pos, MDOC_MAX, MDOC_EQN);
1.81 schwarze 244: n->eqn = ep;
245:
1.93 schwarze 246: if ( ! node_append(mdoc, n))
1.81 schwarze 247: return(0);
248:
1.93 schwarze 249: mdoc->next = MDOC_NEXT_SIBLING;
1.81 schwarze 250: return(1);
1.1 kristaps 251: }
252:
1.77 schwarze 253: int
1.93 schwarze 254: mdoc_addspan(struct mdoc *mdoc, const struct tbl_span *sp)
1.77 schwarze 255: {
1.79 schwarze 256: struct mdoc_node *n;
1.77 schwarze 257:
1.93 schwarze 258: assert( ! (MDOC_HALT & mdoc->flags));
1.77 schwarze 259:
260: /* No text before an initial macro. */
261:
1.93 schwarze 262: if (SEC_NONE == mdoc->lastnamed) {
263: mdoc_pmsg(mdoc, sp->line, 0, MANDOCERR_NOTEXT);
1.77 schwarze 264: return(1);
265: }
266:
1.93 schwarze 267: n = node_alloc(mdoc, sp->line, 0, MDOC_MAX, MDOC_TBL);
1.79 schwarze 268: n->span = sp;
269:
1.93 schwarze 270: if ( ! node_append(mdoc, n))
1.79 schwarze 271: return(0);
272:
1.93 schwarze 273: mdoc->next = MDOC_NEXT_SIBLING;
1.79 schwarze 274: return(1);
1.77 schwarze 275: }
276:
1.1 kristaps 277:
278: /*
279: * Main parse routine. Parses a single line -- really just hands off to
1.45 schwarze 280: * the macro (mdoc_pmacro()) or text parser (mdoc_ptext()).
1.1 kristaps 281: */
282: int
1.93 schwarze 283: mdoc_parseln(struct mdoc *mdoc, int ln, char *buf, int offs)
1.1 kristaps 284: {
285:
1.93 schwarze 286: assert( ! (MDOC_HALT & mdoc->flags));
1.1 kristaps 287:
1.93 schwarze 288: mdoc->flags |= MDOC_NEWLINE;
1.60 schwarze 289:
290: /*
291: * Let the roff nS register switch SYNOPSIS mode early,
292: * such that the parser knows at all times
293: * whether this mode is on or off.
294: * Note that this mode is also switched by the Sh macro.
295: */
1.94 schwarze 296: if (roff_getreg(mdoc->roff, "nS"))
297: mdoc->flags |= MDOC_SYNOPSIS;
298: else
299: mdoc->flags &= ~MDOC_SYNOPSIS;
1.60 schwarze 300:
1.93 schwarze 301: return(roff_getcontrol(mdoc->roff, buf, &offs) ?
302: mdoc_pmacro(mdoc, ln, buf, offs) :
303: mdoc_ptext(mdoc, 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 &&
1.93 schwarze 314: MDOC_PBODY & mdoc->flags) {
315: mdoc_pmsg(mdoc, line, ppos, MANDOCERR_BADBODY);
1.76 schwarze 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) &&
1.93 schwarze 322: ! (MDOC_PBODY & mdoc->flags)) {
323: mdoc_pmsg(mdoc, line, ppos, MANDOCERR_BADPROLOG);
324: if (NULL == mdoc->meta.msec)
325: mdoc->meta.msec = mandoc_strdup("1");
326: if (NULL == mdoc->meta.title)
327: mdoc->meta.title = mandoc_strdup("UNKNOWN");
328: if (NULL == mdoc->meta.vol)
329: mdoc->meta.vol = mandoc_strdup("LOCAL");
330: if (NULL == mdoc->meta.os)
331: mdoc->meta.os = mandoc_strdup("LOCAL");
332: if (NULL == mdoc->meta.date)
333: mdoc->meta.date = mandoc_normdate
334: (mdoc->parse, NULL, line, ppos);
335: mdoc->flags |= MDOC_PBODY;
1.39 schwarze 336: }
1.1 kristaps 337:
1.93 schwarze 338: return((*mdoc_macros[tok].fp)(mdoc, 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):
1.92 schwarze 374: if (ENDBODY_NOT != p->end)
375: break;
1.75 schwarze 376: /* FALLTHROUGH */
377: case (MDOC_TAIL):
378: /* FALLTHROUGH */
379: case (MDOC_HEAD):
380: p->norm = p->parent->norm;
381: break;
382: default:
383: break;
384: }
385:
1.1 kristaps 386: if ( ! mdoc_valid_pre(mdoc, p))
387: return(0);
388:
389: switch (p->type) {
390: case (MDOC_HEAD):
391: assert(MDOC_BLOCK == p->parent->type);
392: p->parent->head = p;
393: break;
394: case (MDOC_TAIL):
395: assert(MDOC_BLOCK == p->parent->type);
396: p->parent->tail = p;
397: break;
398: case (MDOC_BODY):
1.59 schwarze 399: if (p->end)
400: break;
1.1 kristaps 401: assert(MDOC_BLOCK == p->parent->type);
402: p->parent->body = p;
403: break;
404: default:
405: break;
406: }
407:
408: mdoc->last = p;
409:
410: switch (p->type) {
1.77 schwarze 411: case (MDOC_TBL):
412: /* FALLTHROUGH */
1.1 kristaps 413: case (MDOC_TEXT):
414: if ( ! mdoc_valid_post(mdoc))
415: return(0);
416: break;
417: default:
418: break;
419: }
420:
421: return(1);
422: }
423:
424:
425: static struct mdoc_node *
1.93 schwarze 426: node_alloc(struct mdoc *mdoc, int line, int pos,
1.37 schwarze 427: enum mdoct tok, enum mdoc_type type)
1.1 kristaps 428: {
429: struct mdoc_node *p;
430:
1.32 schwarze 431: p = mandoc_calloc(1, sizeof(struct mdoc_node));
1.93 schwarze 432: p->sec = mdoc->lastsec;
1.1 kristaps 433: p->line = line;
434: p->pos = pos;
1.96 schwarze 435: p->lastline = line;
1.1 kristaps 436: p->tok = tok;
1.37 schwarze 437: p->type = type;
1.58 schwarze 438:
439: /* Flag analysis. */
440:
1.93 schwarze 441: if (MDOC_SYNOPSIS & mdoc->flags)
1.60 schwarze 442: p->flags |= MDOC_SYNPRETTY;
443: else
444: p->flags &= ~MDOC_SYNPRETTY;
1.93 schwarze 445: if (MDOC_NEWLINE & mdoc->flags)
1.47 schwarze 446: p->flags |= MDOC_LINE;
1.93 schwarze 447: mdoc->flags &= ~MDOC_NEWLINE;
1.58 schwarze 448:
1.1 kristaps 449: return(p);
450: }
451:
452:
453: int
1.93 schwarze 454: mdoc_tail_alloc(struct mdoc *mdoc, int line, int pos, enum mdoct tok)
1.1 kristaps 455: {
456: struct mdoc_node *p;
457:
1.93 schwarze 458: p = node_alloc(mdoc, line, pos, tok, MDOC_TAIL);
459: if ( ! node_append(mdoc, p))
1.25 schwarze 460: return(0);
1.93 schwarze 461: mdoc->next = MDOC_NEXT_CHILD;
1.25 schwarze 462: return(1);
1.1 kristaps 463: }
464:
465:
466: int
1.93 schwarze 467: mdoc_head_alloc(struct mdoc *mdoc, int line, int pos, enum mdoct tok)
1.1 kristaps 468: {
469: struct mdoc_node *p;
470:
1.93 schwarze 471: assert(mdoc->first);
472: assert(mdoc->last);
1.1 kristaps 473:
1.93 schwarze 474: p = node_alloc(mdoc, line, pos, tok, MDOC_HEAD);
475: if ( ! node_append(mdoc, p))
1.25 schwarze 476: return(0);
1.93 schwarze 477: mdoc->next = MDOC_NEXT_CHILD;
1.25 schwarze 478: return(1);
1.1 kristaps 479: }
480:
481:
482: int
1.93 schwarze 483: mdoc_body_alloc(struct mdoc *mdoc, int line, int pos, enum mdoct tok)
1.1 kristaps 484: {
485: struct mdoc_node *p;
486:
1.93 schwarze 487: p = node_alloc(mdoc, line, pos, tok, MDOC_BODY);
488: if ( ! node_append(mdoc, p))
1.25 schwarze 489: return(0);
1.93 schwarze 490: mdoc->next = MDOC_NEXT_CHILD;
1.59 schwarze 491: return(1);
492: }
493:
494:
495: int
1.93 schwarze 496: mdoc_endbody_alloc(struct mdoc *mdoc, int line, int pos, enum mdoct tok,
1.59 schwarze 497: struct mdoc_node *body, enum mdoc_endbody end)
498: {
499: struct mdoc_node *p;
500:
1.93 schwarze 501: p = node_alloc(mdoc, line, pos, tok, MDOC_BODY);
1.59 schwarze 502: p->pending = body;
1.92 schwarze 503: p->norm = body->norm;
1.59 schwarze 504: p->end = end;
1.93 schwarze 505: if ( ! node_append(mdoc, p))
1.59 schwarze 506: return(0);
1.93 schwarze 507: mdoc->next = MDOC_NEXT_SIBLING;
1.25 schwarze 508: return(1);
1.1 kristaps 509: }
510:
511:
512: int
1.93 schwarze 513: mdoc_block_alloc(struct mdoc *mdoc, int line, int pos,
1.37 schwarze 514: enum mdoct tok, struct mdoc_arg *args)
1.1 kristaps 515: {
516: struct mdoc_node *p;
517:
1.93 schwarze 518: p = node_alloc(mdoc, line, pos, tok, MDOC_BLOCK);
1.4 schwarze 519: p->args = args;
520: if (p->args)
1.1 kristaps 521: (args->refcnt)++;
1.75 schwarze 522:
523: switch (tok) {
524: case (MDOC_Bd):
525: /* FALLTHROUGH */
526: case (MDOC_Bf):
527: /* FALLTHROUGH */
528: case (MDOC_Bl):
529: /* FALLTHROUGH */
530: case (MDOC_Rs):
531: p->norm = mandoc_calloc(1, sizeof(union mdoc_data));
532: break;
533: default:
534: break;
535: }
536:
1.93 schwarze 537: if ( ! node_append(mdoc, p))
1.25 schwarze 538: return(0);
1.93 schwarze 539: mdoc->next = MDOC_NEXT_CHILD;
1.25 schwarze 540: return(1);
1.1 kristaps 541: }
542:
543:
544: int
1.93 schwarze 545: mdoc_elem_alloc(struct mdoc *mdoc, int line, int pos,
1.37 schwarze 546: enum mdoct tok, struct mdoc_arg *args)
1.1 kristaps 547: {
548: struct mdoc_node *p;
549:
1.93 schwarze 550: p = node_alloc(mdoc, line, pos, tok, MDOC_ELEM);
1.4 schwarze 551: p->args = args;
552: if (p->args)
1.1 kristaps 553: (args->refcnt)++;
1.75 schwarze 554:
555: switch (tok) {
556: case (MDOC_An):
557: p->norm = mandoc_calloc(1, sizeof(union mdoc_data));
558: break;
559: default:
560: break;
561: }
562:
1.93 schwarze 563: if ( ! node_append(mdoc, p))
1.25 schwarze 564: return(0);
1.93 schwarze 565: mdoc->next = MDOC_NEXT_CHILD;
1.25 schwarze 566: return(1);
1.1 kristaps 567: }
568:
1.46 schwarze 569: int
1.93 schwarze 570: mdoc_word_alloc(struct mdoc *mdoc, int line, int pos, const char *p)
1.1 kristaps 571: {
1.19 schwarze 572: struct mdoc_node *n;
1.1 kristaps 573:
1.93 schwarze 574: n = node_alloc(mdoc, line, pos, MDOC_MAX, MDOC_TEXT);
575: n->string = roff_strdup(mdoc->roff, p);
1.19 schwarze 576:
1.93 schwarze 577: if ( ! node_append(mdoc, n))
1.25 schwarze 578: return(0);
1.46 schwarze 579:
1.93 schwarze 580: mdoc->next = MDOC_NEXT_SIBLING;
1.25 schwarze 581: return(1);
1.19 schwarze 582: }
583:
1.95 schwarze 584: void
585: mdoc_word_append(struct mdoc *mdoc, const char *p)
586: {
587: struct mdoc_node *n;
588: char *addstr, *newstr;
589:
590: n = mdoc->last;
591: addstr = roff_strdup(mdoc->roff, p);
592: if (-1 == asprintf(&newstr, "%s %s", n->string, addstr)) {
593: perror(NULL);
594: exit((int)MANDOCLEVEL_SYSERR);
595: }
596: free(addstr);
597: free(n->string);
598: n->string = newstr;
599: mdoc->next = MDOC_NEXT_SIBLING;
600: }
1.19 schwarze 601:
1.61 schwarze 602: static void
1.1 kristaps 603: mdoc_node_free(struct mdoc_node *p)
604: {
1.61 schwarze 605:
1.75 schwarze 606: if (MDOC_BLOCK == p->type || MDOC_ELEM == p->type)
607: free(p->norm);
1.1 kristaps 608: if (p->string)
609: free(p->string);
610: if (p->args)
611: mdoc_argv_free(p->args);
612: free(p);
613: }
614:
615:
1.40 schwarze 616: static void
1.93 schwarze 617: mdoc_node_unlink(struct mdoc *mdoc, struct mdoc_node *n)
1.40 schwarze 618: {
619:
620: /* Adjust siblings. */
621:
622: if (n->prev)
623: n->prev->next = n->next;
624: if (n->next)
625: n->next->prev = n->prev;
626:
627: /* Adjust parent. */
628:
629: if (n->parent) {
630: n->parent->nchild--;
631: if (n->parent->child == n)
632: n->parent->child = n->prev ? n->prev : n->next;
1.72 schwarze 633: if (n->parent->last == n)
634: n->parent->last = n->prev ? n->prev : NULL;
1.40 schwarze 635: }
636:
637: /* Adjust parse point, if applicable. */
638:
1.93 schwarze 639: if (mdoc && mdoc->last == n) {
1.40 schwarze 640: if (n->prev) {
1.93 schwarze 641: mdoc->last = n->prev;
642: mdoc->next = MDOC_NEXT_SIBLING;
1.40 schwarze 643: } else {
1.93 schwarze 644: mdoc->last = n->parent;
645: mdoc->next = MDOC_NEXT_CHILD;
1.40 schwarze 646: }
647: }
648:
1.93 schwarze 649: if (mdoc && mdoc->first == n)
650: mdoc->first = NULL;
1.40 schwarze 651: }
652:
653:
1.1 kristaps 654: void
1.93 schwarze 655: mdoc_node_delete(struct mdoc *mdoc, struct mdoc_node *p)
1.1 kristaps 656: {
657:
1.40 schwarze 658: while (p->child) {
659: assert(p->nchild);
1.93 schwarze 660: mdoc_node_delete(mdoc, p->child);
1.40 schwarze 661: }
662: assert(0 == p->nchild);
1.1 kristaps 663:
1.93 schwarze 664: mdoc_node_unlink(mdoc, p);
1.1 kristaps 665: mdoc_node_free(p);
1.91 schwarze 666: }
667:
668: int
1.93 schwarze 669: mdoc_node_relink(struct mdoc *mdoc, struct mdoc_node *p)
1.91 schwarze 670: {
671:
1.93 schwarze 672: mdoc_node_unlink(mdoc, p);
673: return(node_append(mdoc, p));
1.1 kristaps 674: }
675:
1.84 schwarze 676: #if 0
677: /*
678: * Pre-treat a text line.
679: * Text lines can consist of equations, which must be handled apart from
680: * the regular text.
681: * Thus, use this function to step through a line checking if it has any
682: * equations embedded in it.
683: * This must handle multiple equations AND equations that do not end at
684: * the end-of-line, i.e., will re-enter in the next roff parse.
685: */
686: static int
1.93 schwarze 687: mdoc_preptext(struct mdoc *mdoc, int line, char *buf, int offs)
1.84 schwarze 688: {
689: char *start, *end;
690: char delim;
691:
692: while ('\0' != buf[offs]) {
693: /* Mark starting position if eqn is set. */
694: start = NULL;
1.93 schwarze 695: if ('\0' != (delim = roff_eqndelim(mdoc->roff)))
1.84 schwarze 696: if (NULL != (start = strchr(buf + offs, delim)))
697: *start++ = '\0';
698:
699: /* Parse text as normal. */
1.93 schwarze 700: if ( ! mdoc_ptext(mdoc, line, buf, offs))
1.84 schwarze 701: return(0);
702:
703: /* Continue only if an equation exists. */
704: if (NULL == start)
705: break;
706:
707: /* Read past the end of the equation. */
708: offs += start - (buf + offs);
709: assert(start == &buf[offs]);
710: if (NULL != (end = strchr(buf + offs, delim))) {
711: *end++ = '\0';
712: while (' ' == *end)
713: end++;
714: }
715:
716: /* Parse the equation itself. */
1.93 schwarze 717: roff_openeqn(mdoc->roff, NULL, line, offs, buf);
1.84 schwarze 718:
719: /* Process a finished equation? */
1.93 schwarze 720: if (roff_closeeqn(mdoc->roff))
721: if ( ! mdoc_addeqn(mdoc, roff_eqn(mdoc->roff)))
1.84 schwarze 722: return(0);
723: offs += (end - (buf + offs));
724: }
725:
726: return(1);
727: }
728: #endif
1.1 kristaps 729:
730: /*
731: * Parse free-form text, that is, a line that does not begin with the
732: * control character.
733: */
734: static int
1.93 schwarze 735: mdoc_ptext(struct mdoc *mdoc, int line, char *buf, int offs)
1.1 kristaps 736: {
1.56 schwarze 737: char *c, *ws, *end;
738: struct mdoc_node *n;
1.44 schwarze 739:
1.46 schwarze 740: /* No text before an initial macro. */
741:
1.93 schwarze 742: if (SEC_NONE == mdoc->lastnamed) {
743: mdoc_pmsg(mdoc, line, offs, MANDOCERR_NOTEXT);
1.76 schwarze 744: return(1);
745: }
1.46 schwarze 746:
1.93 schwarze 747: assert(mdoc->last);
748: n = mdoc->last;
1.56 schwarze 749:
750: /*
751: * Divert directly to list processing if we're encountering a
752: * columnar MDOC_BLOCK with or without a prior MDOC_BLOCK entry
753: * (a MDOC_BODY means it's already open, in which case we should
754: * process within its context in the normal way).
755: */
756:
757: if (MDOC_Bl == n->tok && MDOC_BODY == n->type &&
1.75 schwarze 758: LIST_column == n->norm->Bl.type) {
1.56 schwarze 759: /* `Bl' is open without any children. */
1.93 schwarze 760: mdoc->flags |= MDOC_FREECOL;
761: return(mdoc_macro(mdoc, MDOC_It, line, offs, &offs, buf));
1.56 schwarze 762: }
763:
764: if (MDOC_It == n->tok && MDOC_BLOCK == n->type &&
765: NULL != n->parent &&
766: MDOC_Bl == n->parent->tok &&
1.75 schwarze 767: LIST_column == n->parent->norm->Bl.type) {
1.56 schwarze 768: /* `Bl' has block-level `It' children. */
1.93 schwarze 769: mdoc->flags |= MDOC_FREECOL;
770: return(mdoc_macro(mdoc, MDOC_It, line, offs, &offs, buf));
1.56 schwarze 771: }
772:
1.52 schwarze 773: /*
774: * Search for the beginning of unescaped trailing whitespace (ws)
775: * and for the first character not to be output (end).
776: */
1.56 schwarze 777:
778: /* FIXME: replace with strcspn(). */
1.52 schwarze 779: ws = NULL;
1.53 schwarze 780: for (c = end = buf + offs; *c; c++) {
1.52 schwarze 781: switch (*c) {
782: case ' ':
783: if (NULL == ws)
784: ws = c;
785: continue;
786: case '\t':
787: /*
788: * Always warn about trailing tabs,
789: * even outside literal context,
790: * where they should be put on the next line.
791: */
792: if (NULL == ws)
793: ws = c;
794: /*
795: * Strip trailing tabs in literal context only;
796: * outside, they affect the next line.
797: */
1.93 schwarze 798: if (MDOC_LITERAL & mdoc->flags)
1.52 schwarze 799: continue;
800: break;
801: case '\\':
802: /* Skip the escaped character, too, if any. */
803: if (c[1])
804: c++;
805: /* FALLTHROUGH */
806: default:
807: ws = NULL;
808: break;
809: }
810: end = c + 1;
811: }
812: *end = '\0';
1.19 schwarze 813:
1.52 schwarze 814: if (ws)
1.93 schwarze 815: mdoc_pmsg(mdoc, line, (int)(ws-buf), MANDOCERR_EOLNSPACE);
1.34 schwarze 816:
1.93 schwarze 817: if ('\0' == buf[offs] && ! (MDOC_LITERAL & mdoc->flags)) {
818: mdoc_pmsg(mdoc, line, (int)(c-buf), MANDOCERR_NOBLANKLN);
1.46 schwarze 819:
1.40 schwarze 820: /*
1.66 schwarze 821: * Insert a `sp' in the case of a blank line. Technically,
1.46 schwarze 822: * blank lines aren't allowed, but enough manuals assume this
823: * behaviour that we want to work around it.
1.40 schwarze 824: */
1.93 schwarze 825: if ( ! mdoc_elem_alloc(mdoc, line, offs, MDOC_sp, NULL))
1.42 schwarze 826: return(0);
1.46 schwarze 827:
1.93 schwarze 828: mdoc->next = MDOC_NEXT_SIBLING;
1.89 schwarze 829:
1.93 schwarze 830: return(mdoc_valid_post(mdoc));
1.38 schwarze 831: }
1.1 kristaps 832:
1.93 schwarze 833: if ( ! mdoc_word_alloc(mdoc, line, offs, buf+offs))
1.1 kristaps 834: return(0);
835:
1.93 schwarze 836: if (MDOC_LITERAL & mdoc->flags)
1.52 schwarze 837: return(1);
838:
1.35 schwarze 839: /*
1.48 schwarze 840: * End-of-sentence check. If the last character is an unescaped
841: * EOS character, then flag the node as being the end of a
842: * sentence. The front-end will know how to interpret this.
1.35 schwarze 843: */
844:
1.52 schwarze 845: assert(buf < end);
1.48 schwarze 846:
1.97 schwarze 847: if (mandoc_eos(buf+offs, (size_t)(end-buf-offs)))
1.93 schwarze 848: mdoc->last->flags |= MDOC_EOS;
1.48 schwarze 849:
1.1 kristaps 850: return(1);
851: }
1.19 schwarze 852:
1.1 kristaps 853:
854: /*
855: * Parse a macro line, that is, a line beginning with the control
856: * character.
857: */
1.61 schwarze 858: static int
1.93 schwarze 859: mdoc_pmacro(struct mdoc *mdoc, int ln, char *buf, int offs)
1.1 kristaps 860: {
1.56 schwarze 861: enum mdoct tok;
1.83 schwarze 862: int i, sv;
1.56 schwarze 863: char mac[5];
864: struct mdoc_node *n;
1.1 kristaps 865:
1.83 schwarze 866: /* Empty post-control lines are ignored. */
1.1 kristaps 867:
1.83 schwarze 868: if ('"' == buf[offs]) {
1.93 schwarze 869: mdoc_pmsg(mdoc, ln, offs, MANDOCERR_BADCOMMENT);
1.83 schwarze 870: return(1);
871: } else if ('\0' == buf[offs])
1.1 kristaps 872: return(1);
873:
1.83 schwarze 874: sv = offs;
1.47 schwarze 875:
1.63 schwarze 876: /*
1.64 schwarze 877: * Copy the first word into a nil-terminated buffer.
878: * Stop copying when a tab, space, or eoln is encountered.
1.63 schwarze 879: */
1.1 kristaps 880:
1.83 schwarze 881: i = 0;
882: while (i < 4 && '\0' != buf[offs] &&
883: ' ' != buf[offs] && '\t' != buf[offs])
884: mac[i++] = buf[offs++];
885:
886: mac[i] = '\0';
887:
888: tok = (i > 1 || i < 4) ? mdoc_hash_find(mac) : MDOC_MAX;
1.1 kristaps 889:
1.65 schwarze 890: if (MDOC_MAX == tok) {
1.93 schwarze 891: mandoc_vmsg(MANDOCERR_MACRO, mdoc->parse,
1.83 schwarze 892: ln, sv, "%s", buf + sv - 1);
1.1 kristaps 893: return(1);
894: }
895:
1.63 schwarze 896: /* Disregard the first trailing tab, if applicable. */
897:
1.83 schwarze 898: if ('\t' == buf[offs])
899: offs++;
1.63 schwarze 900:
901: /* Jump to the next non-whitespace word. */
1.1 kristaps 902:
1.83 schwarze 903: while (buf[offs] && ' ' == buf[offs])
904: offs++;
1.34 schwarze 905:
1.46 schwarze 906: /*
907: * Trailing whitespace. Note that tabs are allowed to be passed
908: * into the parser as "text", so we only warn about spaces here.
909: */
1.34 schwarze 910:
1.83 schwarze 911: if ('\0' == buf[offs] && ' ' == buf[offs - 1])
1.93 schwarze 912: mdoc_pmsg(mdoc, ln, offs - 1, MANDOCERR_EOLNSPACE);
1.1 kristaps 913:
1.56 schwarze 914: /*
915: * If an initial macro or a list invocation, divert directly
916: * into macro processing.
917: */
918:
1.93 schwarze 919: if (NULL == mdoc->last || MDOC_It == tok || MDOC_El == tok) {
920: if ( ! mdoc_macro(mdoc, tok, ln, sv, &offs, buf))
1.56 schwarze 921: goto err;
922: return(1);
923: }
924:
1.93 schwarze 925: n = mdoc->last;
926: assert(mdoc->last);
1.56 schwarze 927:
928: /*
929: * If the first macro of a `Bl -column', open an `It' block
930: * context around the parsed macro.
931: */
932:
933: if (MDOC_Bl == n->tok && MDOC_BODY == n->type &&
1.75 schwarze 934: LIST_column == n->norm->Bl.type) {
1.93 schwarze 935: mdoc->flags |= MDOC_FREECOL;
936: if ( ! mdoc_macro(mdoc, MDOC_It, ln, sv, &sv, buf))
1.56 schwarze 937: goto err;
938: return(1);
939: }
940:
941: /*
942: * If we're following a block-level `It' within a `Bl -column'
943: * context (perhaps opened in the above block or in ptext()),
944: * then open an `It' block context around the parsed macro.
1.24 schwarze 945: */
1.56 schwarze 946:
947: if (MDOC_It == n->tok && MDOC_BLOCK == n->type &&
948: NULL != n->parent &&
949: MDOC_Bl == n->parent->tok &&
1.75 schwarze 950: LIST_column == n->parent->norm->Bl.type) {
1.93 schwarze 951: mdoc->flags |= MDOC_FREECOL;
952: if ( ! mdoc_macro(mdoc, MDOC_It, ln, sv, &sv, buf))
1.56 schwarze 953: goto err;
954: return(1);
955: }
956:
957: /* Normal processing of a macro. */
958:
1.93 schwarze 959: if ( ! mdoc_macro(mdoc, tok, ln, sv, &offs, buf))
1.1 kristaps 960: goto err;
1.98 ! schwarze 961:
! 962: /* In quick mode (for mandocdb), abort after the NAME section. */
! 963:
! 964: if (mdoc->quick && MDOC_Sh == tok &&
! 965: SEC_NAME != mdoc->last->sec)
! 966: return(2);
1.1 kristaps 967:
968: return(1);
969:
970: err: /* Error out. */
971:
1.93 schwarze 972: mdoc->flags |= MDOC_HALT;
1.1 kristaps 973: return(0);
974: }
1.24 schwarze 975:
1.83 schwarze 976: enum mdelim
977: mdoc_isdelim(const char *p)
978: {
979:
980: if ('\0' == p[0])
981: return(DELIM_NONE);
982:
983: if ('\0' == p[1])
984: switch (p[0]) {
985: case('('):
986: /* FALLTHROUGH */
987: case('['):
988: return(DELIM_OPEN);
989: case('|'):
990: return(DELIM_MIDDLE);
991: case('.'):
992: /* FALLTHROUGH */
993: case(','):
994: /* FALLTHROUGH */
995: case(';'):
996: /* FALLTHROUGH */
997: case(':'):
998: /* FALLTHROUGH */
999: case('?'):
1000: /* FALLTHROUGH */
1001: case('!'):
1002: /* FALLTHROUGH */
1003: case(')'):
1004: /* FALLTHROUGH */
1005: case(']'):
1006: return(DELIM_CLOSE);
1007: default:
1008: return(DELIM_NONE);
1009: }
1010:
1011: if ('\\' != p[0])
1012: return(DELIM_NONE);
1.24 schwarze 1013:
1.83 schwarze 1014: if (0 == strcmp(p + 1, "."))
1015: return(DELIM_CLOSE);
1.90 schwarze 1016: if (0 == strcmp(p + 1, "fR|\\fP"))
1.83 schwarze 1017: return(DELIM_MIDDLE);
1018:
1019: return(DELIM_NONE);
1020: }
![[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