Annotation of src/usr.bin/mandoc/mdoc.c, Revision 1.103
1.103 ! schwarze 1: /* $Id: mdoc.c,v 1.102 2014/03/30 19:47:32 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>
1.101 schwarze 21: #include <ctype.h>
1.1 kristaps 22: #include <stdarg.h>
23: #include <stdio.h>
24: #include <stdlib.h>
25: #include <string.h>
1.39 schwarze 26: #include <time.h>
1.1 kristaps 27:
1.83 schwarze 28: #include "mdoc.h"
1.54 schwarze 29: #include "mandoc.h"
1.99 schwarze 30: #include "mandoc_aux.h"
1.1 kristaps 31: #include "libmdoc.h"
1.32 schwarze 32: #include "libmandoc.h"
1.1 kristaps 33:
1.103 ! schwarze 34: const char *const __mdoc_macronames[MDOC_MAX] = {
1.7 schwarze 35: "Ap", "Dd", "Dt", "Os",
1.1 kristaps 36: "Sh", "Ss", "Pp", "D1",
37: "Dl", "Bd", "Ed", "Bl",
38: "El", "It", "Ad", "An",
39: "Ar", "Cd", "Cm", "Dv",
40: "Er", "Ev", "Ex", "Fa",
41: "Fd", "Fl", "Fn", "Ft",
42: "Ic", "In", "Li", "Nd",
43: "Nm", "Op", "Ot", "Pa",
44: "Rv", "St", "Va", "Vt",
1.33 schwarze 45: "Xr", "%A", "%B", "%D",
46: "%I", "%J", "%N", "%O",
47: "%P", "%R", "%T", "%V",
1.1 kristaps 48: "Ac", "Ao", "Aq", "At",
49: "Bc", "Bf", "Bo", "Bq",
50: "Bsx", "Bx", "Db", "Dc",
51: "Do", "Dq", "Ec", "Ef",
52: "Em", "Eo", "Fx", "Ms",
53: "No", "Ns", "Nx", "Ox",
54: "Pc", "Pf", "Po", "Pq",
55: "Qc", "Ql", "Qo", "Qq",
56: "Re", "Rs", "Sc", "So",
57: "Sq", "Sm", "Sx", "Sy",
58: "Tn", "Ux", "Xc", "Xo",
59: "Fo", "Fc", "Oo", "Oc",
60: "Bk", "Ek", "Bt", "Hf",
1.7 schwarze 61: "Fr", "Ud", "Lb", "Lp",
62: "Lk", "Mt", "Brq", "Bro",
1.33 schwarze 63: "Brc", "%C", "Es", "En",
64: "Dx", "%Q", "br", "sp",
1.102 schwarze 65: "%U", "Ta", "ll",
1.1 kristaps 66: };
67:
1.103 ! schwarze 68: const char *const __mdoc_argnames[MDOC_ARG_MAX] = {
1.1 kristaps 69: "split", "nosplit", "ragged",
1.103 ! schwarze 70: "unfilled", "literal", "file",
! 71: "offset", "bullet", "dash",
! 72: "hyphen", "item", "enum",
! 73: "tag", "diag", "hang",
! 74: "ohang", "inset", "column",
! 75: "width", "compact", "std",
1.1 kristaps 76: "filled", "words", "emphasis",
1.30 schwarze 77: "symbolic", "nested", "centered"
1.1 kristaps 78: };
79:
80: const char * const *mdoc_macronames = __mdoc_macronames;
81: const char * const *mdoc_argnames = __mdoc_argnames;
82:
1.40 schwarze 83: static void mdoc_node_free(struct mdoc_node *);
1.103 ! schwarze 84: static void mdoc_node_unlink(struct mdoc *,
1.40 schwarze 85: struct mdoc_node *);
1.1 kristaps 86: static void mdoc_free1(struct mdoc *);
1.32 schwarze 87: static void mdoc_alloc1(struct mdoc *);
1.103 ! schwarze 88: static struct mdoc_node *node_alloc(struct mdoc *, int, int,
1.37 schwarze 89: enum mdoct, enum mdoc_type);
1.103 ! schwarze 90: static int node_append(struct mdoc *,
1.1 kristaps 91: struct mdoc_node *);
1.84 schwarze 92: #if 0
93: static int mdoc_preptext(struct mdoc *, int, char *, int);
94: #endif
1.53 schwarze 95: static int mdoc_ptext(struct mdoc *, int, char *, int);
96: static int mdoc_pmacro(struct mdoc *, int, char *, int);
1.1 kristaps 97:
1.103 ! schwarze 98:
1.1 kristaps 99: const struct mdoc_node *
1.93 schwarze 100: mdoc_node(const struct mdoc *mdoc)
1.1 kristaps 101: {
102:
1.93 schwarze 103: assert( ! (MDOC_HALT & mdoc->flags));
104: return(mdoc->first);
1.1 kristaps 105: }
106:
107: const struct mdoc_meta *
1.93 schwarze 108: mdoc_meta(const struct mdoc *mdoc)
1.1 kristaps 109: {
110:
1.93 schwarze 111: assert( ! (MDOC_HALT & mdoc->flags));
112: return(&mdoc->meta);
1.1 kristaps 113: }
114:
1.8 schwarze 115: /*
116: * Frees volatile resources (parse tree, meta-data, fields).
117: */
1.1 kristaps 118: static void
119: mdoc_free1(struct mdoc *mdoc)
120: {
121:
122: if (mdoc->first)
1.40 schwarze 123: mdoc_node_delete(mdoc, mdoc->first);
1.1 kristaps 124: if (mdoc->meta.title)
125: free(mdoc->meta.title);
126: if (mdoc->meta.os)
127: free(mdoc->meta.os);
128: if (mdoc->meta.name)
129: free(mdoc->meta.name);
130: if (mdoc->meta.arch)
131: free(mdoc->meta.arch);
132: if (mdoc->meta.vol)
133: free(mdoc->meta.vol);
1.50 schwarze 134: if (mdoc->meta.msec)
135: free(mdoc->meta.msec);
1.82 schwarze 136: if (mdoc->meta.date)
137: free(mdoc->meta.date);
1.1 kristaps 138: }
139:
1.8 schwarze 140: /*
141: * Allocate all volatile resources (parse tree, meta-data, fields).
142: */
1.32 schwarze 143: static void
1.1 kristaps 144: mdoc_alloc1(struct mdoc *mdoc)
145: {
146:
1.32 schwarze 147: memset(&mdoc->meta, 0, sizeof(struct mdoc_meta));
1.1 kristaps 148: mdoc->flags = 0;
1.9 schwarze 149: mdoc->lastnamed = mdoc->lastsec = SEC_NONE;
1.32 schwarze 150: mdoc->last = mandoc_calloc(1, sizeof(struct mdoc_node));
1.1 kristaps 151: mdoc->first = mdoc->last;
152: mdoc->last->type = MDOC_ROOT;
1.86 schwarze 153: mdoc->last->tok = MDOC_MAX;
1.1 kristaps 154: mdoc->next = MDOC_NEXT_CHILD;
155: }
156:
157: /*
1.8 schwarze 158: * Free up volatile resources (see mdoc_free1()) then re-initialises the
159: * data with mdoc_alloc1(). After invocation, parse data has been reset
160: * and the parser is ready for re-invocation on a new tree; however,
161: * cross-parse non-volatile data is kept intact.
1.1 kristaps 162: */
1.32 schwarze 163: void
1.1 kristaps 164: mdoc_reset(struct mdoc *mdoc)
165: {
166:
167: mdoc_free1(mdoc);
1.32 schwarze 168: mdoc_alloc1(mdoc);
1.1 kristaps 169: }
170:
171: /*
1.8 schwarze 172: * Completely free up all volatile and non-volatile parse resources.
173: * After invocation, the pointer is no longer usable.
1.1 kristaps 174: */
175: void
176: mdoc_free(struct mdoc *mdoc)
177: {
178:
179: mdoc_free1(mdoc);
180: free(mdoc);
181: }
182:
1.8 schwarze 183: /*
1.103 ! schwarze 184: * Allocate volatile and non-volatile parse resources.
1.8 schwarze 185: */
1.1 kristaps 186: struct mdoc *
1.98 schwarze 187: mdoc_alloc(struct roff *roff, struct mparse *parse,
188: char *defos, int quick)
1.1 kristaps 189: {
190: struct mdoc *p;
191:
1.32 schwarze 192: p = mandoc_calloc(1, sizeof(struct mdoc));
193:
1.83 schwarze 194: p->parse = parse;
1.87 schwarze 195: p->defos = defos;
1.98 schwarze 196: p->quick = quick;
1.84 schwarze 197: p->roff = roff;
1.1 kristaps 198:
1.32 schwarze 199: mdoc_hash_init();
200: mdoc_alloc1(p);
201: return(p);
1.1 kristaps 202: }
203:
204: /*
205: * Climb back up the parse tree, validating open scopes. Mostly calls
1.8 schwarze 206: * through to macro_end() in macro.c.
1.1 kristaps 207: */
208: int
1.93 schwarze 209: mdoc_endparse(struct mdoc *mdoc)
1.1 kristaps 210: {
211:
1.93 schwarze 212: assert( ! (MDOC_HALT & mdoc->flags));
213: if (mdoc_macroend(mdoc))
1.1 kristaps 214: return(1);
1.93 schwarze 215: mdoc->flags |= MDOC_HALT;
1.1 kristaps 216: return(0);
1.81 schwarze 217: }
218:
219: int
1.93 schwarze 220: mdoc_addeqn(struct mdoc *mdoc, const struct eqn *ep)
1.81 schwarze 221: {
222: struct mdoc_node *n;
223:
1.93 schwarze 224: assert( ! (MDOC_HALT & mdoc->flags));
1.81 schwarze 225:
226: /* No text before an initial macro. */
227:
1.93 schwarze 228: if (SEC_NONE == mdoc->lastnamed) {
229: mdoc_pmsg(mdoc, ep->ln, ep->pos, MANDOCERR_NOTEXT);
1.81 schwarze 230: return(1);
231: }
232:
1.93 schwarze 233: n = node_alloc(mdoc, ep->ln, ep->pos, MDOC_MAX, MDOC_EQN);
1.81 schwarze 234: n->eqn = ep;
235:
1.93 schwarze 236: if ( ! node_append(mdoc, n))
1.81 schwarze 237: return(0);
238:
1.93 schwarze 239: mdoc->next = MDOC_NEXT_SIBLING;
1.81 schwarze 240: return(1);
1.1 kristaps 241: }
242:
1.77 schwarze 243: int
1.93 schwarze 244: mdoc_addspan(struct mdoc *mdoc, const struct tbl_span *sp)
1.77 schwarze 245: {
1.79 schwarze 246: struct mdoc_node *n;
1.77 schwarze 247:
1.93 schwarze 248: assert( ! (MDOC_HALT & mdoc->flags));
1.77 schwarze 249:
250: /* No text before an initial macro. */
251:
1.93 schwarze 252: if (SEC_NONE == mdoc->lastnamed) {
253: mdoc_pmsg(mdoc, sp->line, 0, MANDOCERR_NOTEXT);
1.77 schwarze 254: return(1);
255: }
256:
1.93 schwarze 257: n = node_alloc(mdoc, sp->line, 0, MDOC_MAX, MDOC_TBL);
1.79 schwarze 258: n->span = sp;
259:
1.93 schwarze 260: if ( ! node_append(mdoc, n))
1.79 schwarze 261: return(0);
262:
1.93 schwarze 263: mdoc->next = MDOC_NEXT_SIBLING;
1.79 schwarze 264: return(1);
1.77 schwarze 265: }
266:
1.1 kristaps 267: /*
268: * Main parse routine. Parses a single line -- really just hands off to
1.45 schwarze 269: * the macro (mdoc_pmacro()) or text parser (mdoc_ptext()).
1.1 kristaps 270: */
271: int
1.93 schwarze 272: mdoc_parseln(struct mdoc *mdoc, int ln, char *buf, int offs)
1.1 kristaps 273: {
274:
1.93 schwarze 275: assert( ! (MDOC_HALT & mdoc->flags));
1.1 kristaps 276:
1.93 schwarze 277: mdoc->flags |= MDOC_NEWLINE;
1.60 schwarze 278:
279: /*
280: * Let the roff nS register switch SYNOPSIS mode early,
281: * such that the parser knows at all times
282: * whether this mode is on or off.
283: * Note that this mode is also switched by the Sh macro.
284: */
1.94 schwarze 285: if (roff_getreg(mdoc->roff, "nS"))
286: mdoc->flags |= MDOC_SYNOPSIS;
287: else
288: mdoc->flags &= ~MDOC_SYNOPSIS;
1.60 schwarze 289:
1.93 schwarze 290: return(roff_getcontrol(mdoc->roff, buf, &offs) ?
1.103 ! schwarze 291: mdoc_pmacro(mdoc, ln, buf, offs) :
! 292: mdoc_ptext(mdoc, ln, buf, offs));
1.1 kristaps 293: }
294:
1.5 schwarze 295: int
1.58 schwarze 296: mdoc_macro(MACRO_PROT_ARGS)
1.1 kristaps 297: {
1.46 schwarze 298: assert(tok < MDOC_MAX);
299:
300: /* If we're in the body, deny prologue calls. */
1.37 schwarze 301:
1.103 ! schwarze 302: if (MDOC_PROLOGUE & mdoc_macros[tok].flags &&
! 303: MDOC_PBODY & mdoc->flags) {
1.93 schwarze 304: mdoc_pmsg(mdoc, line, ppos, MANDOCERR_BADBODY);
1.76 schwarze 305: return(1);
306: }
1.46 schwarze 307:
308: /* If we're in the prologue, deny "body" macros. */
309:
1.103 ! schwarze 310: if ( ! (MDOC_PROLOGUE & mdoc_macros[tok].flags) &&
! 311: ! (MDOC_PBODY & mdoc->flags)) {
1.93 schwarze 312: mdoc_pmsg(mdoc, line, ppos, MANDOCERR_BADPROLOG);
313: if (NULL == mdoc->meta.msec)
314: mdoc->meta.msec = mandoc_strdup("1");
315: if (NULL == mdoc->meta.title)
316: mdoc->meta.title = mandoc_strdup("UNKNOWN");
317: if (NULL == mdoc->meta.vol)
318: mdoc->meta.vol = mandoc_strdup("LOCAL");
319: if (NULL == mdoc->meta.os)
320: mdoc->meta.os = mandoc_strdup("LOCAL");
321: if (NULL == mdoc->meta.date)
322: mdoc->meta.date = mandoc_normdate
323: (mdoc->parse, NULL, line, ppos);
324: mdoc->flags |= MDOC_PBODY;
1.39 schwarze 325: }
1.1 kristaps 326:
1.93 schwarze 327: return((*mdoc_macros[tok].fp)(mdoc, tok, line, ppos, pos, buf));
1.1 kristaps 328: }
329:
330:
331: static int
332: node_append(struct mdoc *mdoc, struct mdoc_node *p)
333: {
334:
335: assert(mdoc->last);
336: assert(mdoc->first);
337: assert(MDOC_ROOT != p->type);
338:
339: switch (mdoc->next) {
1.103 ! schwarze 340: case MDOC_NEXT_SIBLING:
1.1 kristaps 341: mdoc->last->next = p;
342: p->prev = mdoc->last;
343: p->parent = mdoc->last->parent;
344: break;
1.103 ! schwarze 345: case MDOC_NEXT_CHILD:
1.1 kristaps 346: mdoc->last->child = p;
347: p->parent = mdoc->last;
348: break;
349: default:
350: abort();
351: /* NOTREACHED */
352: }
353:
1.10 schwarze 354: p->parent->nchild++;
355:
1.75 schwarze 356: /*
357: * Copy over the normalised-data pointer of our parent. Not
358: * everybody has one, but copying a null pointer is fine.
359: */
360:
361: switch (p->type) {
1.103 ! schwarze 362: case MDOC_BODY:
1.92 schwarze 363: if (ENDBODY_NOT != p->end)
364: break;
1.75 schwarze 365: /* FALLTHROUGH */
1.103 ! schwarze 366: case MDOC_TAIL:
1.75 schwarze 367: /* FALLTHROUGH */
1.103 ! schwarze 368: case MDOC_HEAD:
1.75 schwarze 369: p->norm = p->parent->norm;
370: break;
371: default:
372: break;
373: }
374:
1.1 kristaps 375: if ( ! mdoc_valid_pre(mdoc, p))
376: return(0);
377:
378: switch (p->type) {
1.103 ! schwarze 379: case MDOC_HEAD:
1.1 kristaps 380: assert(MDOC_BLOCK == p->parent->type);
381: p->parent->head = p;
382: break;
1.103 ! schwarze 383: case MDOC_TAIL:
1.1 kristaps 384: assert(MDOC_BLOCK == p->parent->type);
385: p->parent->tail = p;
386: break;
1.103 ! schwarze 387: case MDOC_BODY:
1.59 schwarze 388: if (p->end)
389: break;
1.1 kristaps 390: assert(MDOC_BLOCK == p->parent->type);
391: p->parent->body = p;
392: break;
393: default:
394: break;
395: }
396:
397: mdoc->last = p;
398:
399: switch (p->type) {
1.103 ! schwarze 400: case MDOC_TBL:
1.77 schwarze 401: /* FALLTHROUGH */
1.103 ! schwarze 402: case MDOC_TEXT:
1.1 kristaps 403: if ( ! mdoc_valid_post(mdoc))
404: return(0);
405: break;
406: default:
407: break;
408: }
409:
410: return(1);
411: }
412:
413: static struct mdoc_node *
1.103 ! schwarze 414: node_alloc(struct mdoc *mdoc, int line, int pos,
1.37 schwarze 415: enum mdoct tok, enum mdoc_type type)
1.1 kristaps 416: {
417: struct mdoc_node *p;
418:
1.32 schwarze 419: p = mandoc_calloc(1, sizeof(struct mdoc_node));
1.93 schwarze 420: p->sec = mdoc->lastsec;
1.1 kristaps 421: p->line = line;
422: p->pos = pos;
1.96 schwarze 423: p->lastline = line;
1.1 kristaps 424: p->tok = tok;
1.37 schwarze 425: p->type = type;
1.58 schwarze 426:
427: /* Flag analysis. */
428:
1.93 schwarze 429: if (MDOC_SYNOPSIS & mdoc->flags)
1.60 schwarze 430: p->flags |= MDOC_SYNPRETTY;
431: else
432: p->flags &= ~MDOC_SYNPRETTY;
1.93 schwarze 433: if (MDOC_NEWLINE & mdoc->flags)
1.47 schwarze 434: p->flags |= MDOC_LINE;
1.93 schwarze 435: mdoc->flags &= ~MDOC_NEWLINE;
1.58 schwarze 436:
1.1 kristaps 437: return(p);
438: }
439:
440: int
1.93 schwarze 441: mdoc_tail_alloc(struct mdoc *mdoc, int line, int pos, enum mdoct tok)
1.1 kristaps 442: {
443: struct mdoc_node *p;
444:
1.93 schwarze 445: p = node_alloc(mdoc, line, pos, tok, MDOC_TAIL);
446: if ( ! node_append(mdoc, p))
1.25 schwarze 447: return(0);
1.93 schwarze 448: mdoc->next = MDOC_NEXT_CHILD;
1.25 schwarze 449: return(1);
1.1 kristaps 450: }
451:
452: int
1.93 schwarze 453: mdoc_head_alloc(struct mdoc *mdoc, int line, int pos, enum mdoct tok)
1.1 kristaps 454: {
455: struct mdoc_node *p;
456:
1.93 schwarze 457: assert(mdoc->first);
458: assert(mdoc->last);
1.1 kristaps 459:
1.93 schwarze 460: p = node_alloc(mdoc, line, pos, tok, MDOC_HEAD);
461: if ( ! node_append(mdoc, p))
1.25 schwarze 462: return(0);
1.93 schwarze 463: mdoc->next = MDOC_NEXT_CHILD;
1.25 schwarze 464: return(1);
1.1 kristaps 465: }
466:
467: int
1.93 schwarze 468: mdoc_body_alloc(struct mdoc *mdoc, int line, int pos, enum mdoct tok)
1.1 kristaps 469: {
470: struct mdoc_node *p;
471:
1.93 schwarze 472: p = node_alloc(mdoc, line, pos, tok, MDOC_BODY);
473: if ( ! node_append(mdoc, p))
1.25 schwarze 474: return(0);
1.93 schwarze 475: mdoc->next = MDOC_NEXT_CHILD;
1.59 schwarze 476: return(1);
477: }
478:
479: int
1.93 schwarze 480: mdoc_endbody_alloc(struct mdoc *mdoc, int line, int pos, enum mdoct tok,
1.59 schwarze 481: struct mdoc_node *body, enum mdoc_endbody end)
482: {
483: struct mdoc_node *p;
484:
1.93 schwarze 485: p = node_alloc(mdoc, line, pos, tok, MDOC_BODY);
1.59 schwarze 486: p->pending = body;
1.92 schwarze 487: p->norm = body->norm;
1.59 schwarze 488: p->end = end;
1.93 schwarze 489: if ( ! node_append(mdoc, p))
1.59 schwarze 490: return(0);
1.93 schwarze 491: mdoc->next = MDOC_NEXT_SIBLING;
1.25 schwarze 492: return(1);
1.1 kristaps 493: }
494:
495: int
1.103 ! schwarze 496: mdoc_block_alloc(struct mdoc *mdoc, int line, int pos,
1.37 schwarze 497: enum mdoct tok, struct mdoc_arg *args)
1.1 kristaps 498: {
499: struct mdoc_node *p;
500:
1.93 schwarze 501: p = node_alloc(mdoc, line, pos, tok, MDOC_BLOCK);
1.4 schwarze 502: p->args = args;
503: if (p->args)
1.1 kristaps 504: (args->refcnt)++;
1.75 schwarze 505:
506: switch (tok) {
1.103 ! schwarze 507: case MDOC_Bd:
1.75 schwarze 508: /* FALLTHROUGH */
1.103 ! schwarze 509: case MDOC_Bf:
1.75 schwarze 510: /* FALLTHROUGH */
1.103 ! schwarze 511: case MDOC_Bl:
1.75 schwarze 512: /* FALLTHROUGH */
1.103 ! schwarze 513: case MDOC_Rs:
1.75 schwarze 514: p->norm = mandoc_calloc(1, sizeof(union mdoc_data));
515: break;
516: default:
517: break;
518: }
519:
1.93 schwarze 520: if ( ! node_append(mdoc, p))
1.25 schwarze 521: return(0);
1.93 schwarze 522: mdoc->next = MDOC_NEXT_CHILD;
1.25 schwarze 523: return(1);
1.1 kristaps 524: }
525:
526: int
1.103 ! schwarze 527: mdoc_elem_alloc(struct mdoc *mdoc, int line, int pos,
1.37 schwarze 528: enum mdoct tok, struct mdoc_arg *args)
1.1 kristaps 529: {
530: struct mdoc_node *p;
531:
1.93 schwarze 532: p = node_alloc(mdoc, line, pos, tok, MDOC_ELEM);
1.4 schwarze 533: p->args = args;
534: if (p->args)
1.1 kristaps 535: (args->refcnt)++;
1.75 schwarze 536:
537: switch (tok) {
1.103 ! schwarze 538: case MDOC_An:
1.75 schwarze 539: p->norm = mandoc_calloc(1, sizeof(union mdoc_data));
540: break;
541: default:
542: break;
543: }
544:
1.93 schwarze 545: if ( ! node_append(mdoc, p))
1.25 schwarze 546: return(0);
1.93 schwarze 547: mdoc->next = MDOC_NEXT_CHILD;
1.25 schwarze 548: return(1);
1.1 kristaps 549: }
550:
1.46 schwarze 551: int
1.93 schwarze 552: mdoc_word_alloc(struct mdoc *mdoc, int line, int pos, const char *p)
1.1 kristaps 553: {
1.19 schwarze 554: struct mdoc_node *n;
1.1 kristaps 555:
1.93 schwarze 556: n = node_alloc(mdoc, line, pos, MDOC_MAX, MDOC_TEXT);
557: n->string = roff_strdup(mdoc->roff, p);
1.19 schwarze 558:
1.93 schwarze 559: if ( ! node_append(mdoc, n))
1.25 schwarze 560: return(0);
1.46 schwarze 561:
1.93 schwarze 562: mdoc->next = MDOC_NEXT_SIBLING;
1.25 schwarze 563: return(1);
1.19 schwarze 564: }
565:
1.95 schwarze 566: void
567: mdoc_word_append(struct mdoc *mdoc, const char *p)
568: {
569: struct mdoc_node *n;
570: char *addstr, *newstr;
571:
572: n = mdoc->last;
573: addstr = roff_strdup(mdoc->roff, p);
1.100 schwarze 574: mandoc_asprintf(&newstr, "%s %s", n->string, addstr);
1.95 schwarze 575: free(addstr);
576: free(n->string);
577: n->string = newstr;
578: mdoc->next = MDOC_NEXT_SIBLING;
579: }
1.19 schwarze 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:
1.40 schwarze 594: static void
1.93 schwarze 595: mdoc_node_unlink(struct mdoc *mdoc, struct mdoc_node *n)
1.40 schwarze 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:
1.93 schwarze 617: if (mdoc && mdoc->last == n) {
1.40 schwarze 618: if (n->prev) {
1.93 schwarze 619: mdoc->last = n->prev;
620: mdoc->next = MDOC_NEXT_SIBLING;
1.40 schwarze 621: } else {
1.93 schwarze 622: mdoc->last = n->parent;
623: mdoc->next = MDOC_NEXT_CHILD;
1.40 schwarze 624: }
625: }
626:
1.93 schwarze 627: if (mdoc && mdoc->first == n)
628: mdoc->first = NULL;
1.40 schwarze 629: }
630:
1.1 kristaps 631: void
1.93 schwarze 632: mdoc_node_delete(struct mdoc *mdoc, struct mdoc_node *p)
1.1 kristaps 633: {
634:
1.40 schwarze 635: while (p->child) {
636: assert(p->nchild);
1.93 schwarze 637: mdoc_node_delete(mdoc, p->child);
1.40 schwarze 638: }
639: assert(0 == p->nchild);
1.1 kristaps 640:
1.93 schwarze 641: mdoc_node_unlink(mdoc, p);
1.1 kristaps 642: mdoc_node_free(p);
1.91 schwarze 643: }
644:
645: int
1.93 schwarze 646: mdoc_node_relink(struct mdoc *mdoc, struct mdoc_node *p)
1.91 schwarze 647: {
648:
1.93 schwarze 649: mdoc_node_unlink(mdoc, p);
650: return(node_append(mdoc, p));
1.1 kristaps 651: }
652:
1.84 schwarze 653: #if 0
654: /*
655: * Pre-treat a text line.
656: * Text lines can consist of equations, which must be handled apart from
657: * the regular text.
658: * Thus, use this function to step through a line checking if it has any
659: * equations embedded in it.
660: * This must handle multiple equations AND equations that do not end at
661: * the end-of-line, i.e., will re-enter in the next roff parse.
662: */
663: static int
1.93 schwarze 664: mdoc_preptext(struct mdoc *mdoc, int line, char *buf, int offs)
1.84 schwarze 665: {
666: char *start, *end;
667: char delim;
668:
669: while ('\0' != buf[offs]) {
670: /* Mark starting position if eqn is set. */
671: start = NULL;
1.93 schwarze 672: if ('\0' != (delim = roff_eqndelim(mdoc->roff)))
1.84 schwarze 673: if (NULL != (start = strchr(buf + offs, delim)))
674: *start++ = '\0';
675:
676: /* Parse text as normal. */
1.93 schwarze 677: if ( ! mdoc_ptext(mdoc, line, buf, offs))
1.84 schwarze 678: return(0);
679:
680: /* Continue only if an equation exists. */
681: if (NULL == start)
682: break;
683:
684: /* Read past the end of the equation. */
685: offs += start - (buf + offs);
686: assert(start == &buf[offs]);
687: if (NULL != (end = strchr(buf + offs, delim))) {
688: *end++ = '\0';
689: while (' ' == *end)
690: end++;
691: }
692:
693: /* Parse the equation itself. */
1.93 schwarze 694: roff_openeqn(mdoc->roff, NULL, line, offs, buf);
1.84 schwarze 695:
696: /* Process a finished equation? */
1.93 schwarze 697: if (roff_closeeqn(mdoc->roff))
698: if ( ! mdoc_addeqn(mdoc, roff_eqn(mdoc->roff)))
1.84 schwarze 699: return(0);
700: offs += (end - (buf + offs));
1.103 ! schwarze 701: }
1.84 schwarze 702:
703: return(1);
704: }
705: #endif
1.1 kristaps 706:
707: /*
708: * Parse free-form text, that is, a line that does not begin with the
709: * control character.
710: */
711: static int
1.93 schwarze 712: mdoc_ptext(struct mdoc *mdoc, int line, char *buf, int offs)
1.1 kristaps 713: {
1.56 schwarze 714: char *c, *ws, *end;
715: struct mdoc_node *n;
1.44 schwarze 716:
1.46 schwarze 717: /* No text before an initial macro. */
718:
1.93 schwarze 719: if (SEC_NONE == mdoc->lastnamed) {
720: mdoc_pmsg(mdoc, line, offs, MANDOCERR_NOTEXT);
1.76 schwarze 721: return(1);
722: }
1.46 schwarze 723:
1.93 schwarze 724: assert(mdoc->last);
725: n = mdoc->last;
1.56 schwarze 726:
727: /*
728: * Divert directly to list processing if we're encountering a
729: * columnar MDOC_BLOCK with or without a prior MDOC_BLOCK entry
730: * (a MDOC_BODY means it's already open, in which case we should
731: * process within its context in the normal way).
732: */
733:
734: if (MDOC_Bl == n->tok && MDOC_BODY == n->type &&
1.103 ! schwarze 735: LIST_column == n->norm->Bl.type) {
1.56 schwarze 736: /* `Bl' is open without any children. */
1.93 schwarze 737: mdoc->flags |= MDOC_FREECOL;
738: return(mdoc_macro(mdoc, MDOC_It, line, offs, &offs, buf));
1.56 schwarze 739: }
740:
741: if (MDOC_It == n->tok && MDOC_BLOCK == n->type &&
1.103 ! schwarze 742: NULL != n->parent &&
! 743: MDOC_Bl == n->parent->tok &&
! 744: LIST_column == n->parent->norm->Bl.type) {
1.56 schwarze 745: /* `Bl' has block-level `It' children. */
1.93 schwarze 746: mdoc->flags |= MDOC_FREECOL;
747: return(mdoc_macro(mdoc, MDOC_It, line, offs, &offs, buf));
1.56 schwarze 748: }
749:
1.52 schwarze 750: /*
751: * Search for the beginning of unescaped trailing whitespace (ws)
752: * and for the first character not to be output (end).
753: */
1.56 schwarze 754:
755: /* FIXME: replace with strcspn(). */
1.52 schwarze 756: ws = NULL;
1.53 schwarze 757: for (c = end = buf + offs; *c; c++) {
1.52 schwarze 758: switch (*c) {
759: case ' ':
760: if (NULL == ws)
761: ws = c;
762: continue;
763: case '\t':
764: /*
765: * Always warn about trailing tabs,
766: * even outside literal context,
767: * where they should be put on the next line.
768: */
769: if (NULL == ws)
770: ws = c;
771: /*
772: * Strip trailing tabs in literal context only;
773: * outside, they affect the next line.
774: */
1.93 schwarze 775: if (MDOC_LITERAL & mdoc->flags)
1.52 schwarze 776: continue;
777: break;
778: case '\\':
779: /* Skip the escaped character, too, if any. */
780: if (c[1])
781: c++;
782: /* FALLTHROUGH */
783: default:
784: ws = NULL;
785: break;
786: }
787: end = c + 1;
788: }
789: *end = '\0';
1.19 schwarze 790:
1.52 schwarze 791: if (ws)
1.93 schwarze 792: mdoc_pmsg(mdoc, line, (int)(ws-buf), MANDOCERR_EOLNSPACE);
1.34 schwarze 793:
1.93 schwarze 794: if ('\0' == buf[offs] && ! (MDOC_LITERAL & mdoc->flags)) {
795: mdoc_pmsg(mdoc, line, (int)(c-buf), MANDOCERR_NOBLANKLN);
1.46 schwarze 796:
1.40 schwarze 797: /*
1.66 schwarze 798: * Insert a `sp' in the case of a blank line. Technically,
1.46 schwarze 799: * blank lines aren't allowed, but enough manuals assume this
800: * behaviour that we want to work around it.
1.40 schwarze 801: */
1.93 schwarze 802: if ( ! mdoc_elem_alloc(mdoc, line, offs, MDOC_sp, NULL))
1.42 schwarze 803: return(0);
1.46 schwarze 804:
1.93 schwarze 805: mdoc->next = MDOC_NEXT_SIBLING;
1.89 schwarze 806:
1.93 schwarze 807: return(mdoc_valid_post(mdoc));
1.38 schwarze 808: }
1.1 kristaps 809:
1.93 schwarze 810: if ( ! mdoc_word_alloc(mdoc, line, offs, buf+offs))
1.1 kristaps 811: return(0);
812:
1.93 schwarze 813: if (MDOC_LITERAL & mdoc->flags)
1.52 schwarze 814: return(1);
815:
1.35 schwarze 816: /*
1.48 schwarze 817: * End-of-sentence check. If the last character is an unescaped
818: * EOS character, then flag the node as being the end of a
819: * sentence. The front-end will know how to interpret this.
1.35 schwarze 820: */
821:
1.52 schwarze 822: assert(buf < end);
1.48 schwarze 823:
1.97 schwarze 824: if (mandoc_eos(buf+offs, (size_t)(end-buf-offs)))
1.93 schwarze 825: mdoc->last->flags |= MDOC_EOS;
1.48 schwarze 826:
1.1 kristaps 827: return(1);
828: }
1.19 schwarze 829:
1.1 kristaps 830: /*
831: * Parse a macro line, that is, a line beginning with the control
832: * character.
833: */
1.61 schwarze 834: static int
1.93 schwarze 835: mdoc_pmacro(struct mdoc *mdoc, int ln, char *buf, int offs)
1.1 kristaps 836: {
1.56 schwarze 837: enum mdoct tok;
1.83 schwarze 838: int i, sv;
1.56 schwarze 839: char mac[5];
840: struct mdoc_node *n;
1.1 kristaps 841:
1.83 schwarze 842: /* Empty post-control lines are ignored. */
1.1 kristaps 843:
1.83 schwarze 844: if ('"' == buf[offs]) {
1.93 schwarze 845: mdoc_pmsg(mdoc, ln, offs, MANDOCERR_BADCOMMENT);
1.83 schwarze 846: return(1);
847: } else if ('\0' == buf[offs])
1.1 kristaps 848: return(1);
849:
1.83 schwarze 850: sv = offs;
1.47 schwarze 851:
1.103 ! schwarze 852: /*
1.64 schwarze 853: * Copy the first word into a nil-terminated buffer.
854: * Stop copying when a tab, space, or eoln is encountered.
1.63 schwarze 855: */
1.1 kristaps 856:
1.83 schwarze 857: i = 0;
1.103 ! schwarze 858: while (i < 4 && '\0' != buf[offs] && ' ' != buf[offs] &&
! 859: '\t' != buf[offs])
1.83 schwarze 860: mac[i++] = buf[offs++];
861:
862: mac[i] = '\0';
863:
864: tok = (i > 1 || i < 4) ? mdoc_hash_find(mac) : MDOC_MAX;
1.1 kristaps 865:
1.65 schwarze 866: if (MDOC_MAX == tok) {
1.103 ! schwarze 867: mandoc_vmsg(MANDOCERR_MACRO, mdoc->parse,
! 868: ln, sv, "%s", buf + sv - 1);
1.1 kristaps 869: return(1);
870: }
871:
1.63 schwarze 872: /* Disregard the first trailing tab, if applicable. */
873:
1.83 schwarze 874: if ('\t' == buf[offs])
875: offs++;
1.63 schwarze 876:
877: /* Jump to the next non-whitespace word. */
1.1 kristaps 878:
1.83 schwarze 879: while (buf[offs] && ' ' == buf[offs])
880: offs++;
1.34 schwarze 881:
1.103 ! schwarze 882: /*
1.46 schwarze 883: * Trailing whitespace. Note that tabs are allowed to be passed
884: * into the parser as "text", so we only warn about spaces here.
885: */
1.34 schwarze 886:
1.83 schwarze 887: if ('\0' == buf[offs] && ' ' == buf[offs - 1])
1.93 schwarze 888: mdoc_pmsg(mdoc, ln, offs - 1, MANDOCERR_EOLNSPACE);
1.1 kristaps 889:
1.56 schwarze 890: /*
891: * If an initial macro or a list invocation, divert directly
892: * into macro processing.
893: */
894:
1.93 schwarze 895: if (NULL == mdoc->last || MDOC_It == tok || MDOC_El == tok) {
1.103 ! schwarze 896: if ( ! mdoc_macro(mdoc, tok, ln, sv, &offs, buf))
1.56 schwarze 897: goto err;
898: return(1);
899: }
900:
1.93 schwarze 901: n = mdoc->last;
902: assert(mdoc->last);
1.56 schwarze 903:
904: /*
905: * If the first macro of a `Bl -column', open an `It' block
906: * context around the parsed macro.
907: */
908:
909: if (MDOC_Bl == n->tok && MDOC_BODY == n->type &&
1.103 ! schwarze 910: LIST_column == n->norm->Bl.type) {
1.93 schwarze 911: mdoc->flags |= MDOC_FREECOL;
912: if ( ! mdoc_macro(mdoc, MDOC_It, ln, sv, &sv, buf))
1.56 schwarze 913: goto err;
914: return(1);
915: }
916:
917: /*
918: * If we're following a block-level `It' within a `Bl -column'
919: * context (perhaps opened in the above block or in ptext()),
920: * then open an `It' block context around the parsed macro.
1.24 schwarze 921: */
1.56 schwarze 922:
923: if (MDOC_It == n->tok && MDOC_BLOCK == n->type &&
1.103 ! schwarze 924: NULL != n->parent &&
! 925: MDOC_Bl == n->parent->tok &&
! 926: LIST_column == n->parent->norm->Bl.type) {
1.93 schwarze 927: mdoc->flags |= MDOC_FREECOL;
1.103 ! schwarze 928: if ( ! mdoc_macro(mdoc, MDOC_It, ln, sv, &sv, buf))
1.56 schwarze 929: goto err;
930: return(1);
931: }
932:
933: /* Normal processing of a macro. */
934:
1.103 ! schwarze 935: if ( ! mdoc_macro(mdoc, tok, ln, sv, &offs, buf))
1.1 kristaps 936: goto err;
1.98 schwarze 937:
938: /* In quick mode (for mandocdb), abort after the NAME section. */
939:
940: if (mdoc->quick && MDOC_Sh == tok &&
941: SEC_NAME != mdoc->last->sec)
942: return(2);
1.1 kristaps 943:
944: return(1);
945:
946: err: /* Error out. */
947:
1.93 schwarze 948: mdoc->flags |= MDOC_HALT;
1.1 kristaps 949: return(0);
950: }
1.24 schwarze 951:
1.83 schwarze 952: enum mdelim
953: mdoc_isdelim(const char *p)
954: {
955:
956: if ('\0' == p[0])
957: return(DELIM_NONE);
958:
959: if ('\0' == p[1])
960: switch (p[0]) {
1.103 ! schwarze 961: case '(':
1.83 schwarze 962: /* FALLTHROUGH */
1.103 ! schwarze 963: case '[':
1.83 schwarze 964: return(DELIM_OPEN);
1.103 ! schwarze 965: case '|':
1.83 schwarze 966: return(DELIM_MIDDLE);
1.103 ! schwarze 967: case '.':
1.83 schwarze 968: /* FALLTHROUGH */
1.103 ! schwarze 969: case ',':
1.83 schwarze 970: /* FALLTHROUGH */
1.103 ! schwarze 971: case ';':
1.83 schwarze 972: /* FALLTHROUGH */
1.103 ! schwarze 973: case ':':
1.83 schwarze 974: /* FALLTHROUGH */
1.103 ! schwarze 975: case '?':
1.83 schwarze 976: /* FALLTHROUGH */
1.103 ! schwarze 977: case '!':
1.83 schwarze 978: /* FALLTHROUGH */
1.103 ! schwarze 979: case ')':
1.83 schwarze 980: /* FALLTHROUGH */
1.103 ! schwarze 981: case ']':
1.83 schwarze 982: return(DELIM_CLOSE);
983: default:
984: return(DELIM_NONE);
985: }
986:
987: if ('\\' != p[0])
988: return(DELIM_NONE);
1.24 schwarze 989:
1.83 schwarze 990: if (0 == strcmp(p + 1, "."))
991: return(DELIM_CLOSE);
1.90 schwarze 992: if (0 == strcmp(p + 1, "fR|\\fP"))
1.83 schwarze 993: return(DELIM_MIDDLE);
994:
995: return(DELIM_NONE);
1.101 schwarze 996: }
997:
998: void
999: mdoc_deroff(char **dest, const struct mdoc_node *n)
1000: {
1001: char *cp;
1002: size_t sz;
1003:
1004: if (MDOC_TEXT != n->type) {
1005: for (n = n->child; n; n = n->next)
1006: mdoc_deroff(dest, n);
1007: return;
1008: }
1009:
1010: /* Skip leading whitespace. */
1011:
1012: for (cp = n->string; '\0' != *cp; cp++)
1013: if (0 == isspace((unsigned char)*cp))
1014: break;
1015:
1016: /* Skip trailing whitespace. */
1017:
1018: for (sz = strlen(cp); sz; sz--)
1019: if (0 == isspace((unsigned char)cp[sz-1]))
1020: break;
1021:
1022: /* Skip empty strings. */
1023:
1024: if (0 == sz)
1025: return;
1026:
1027: if (NULL == *dest) {
1028: *dest = mandoc_strndup(cp, sz);
1029: return;
1030: }
1031:
1032: mandoc_asprintf(&cp, "%s %*s", *dest, (int)sz, cp);
1033: free(*dest);
1034: *dest = cp;
1.83 schwarze 1035: }
![[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