Annotation of src/usr.bin/mandoc/mdoc.c, Revision 1.70
1.70 ! schwarze 1: /* $Id: mdoc.c,v 1.69 2010/11/28 19:35:33 schwarze Exp $ */
1.1 kristaps 2: /*
1.61 schwarze 3: * Copyright (c) 2008, 2009, 2010 Kristaps Dzonsons <kristaps@bsd.lv>
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:
1.54 schwarze 27: #include "mandoc.h"
1.1 kristaps 28: #include "libmdoc.h"
1.32 schwarze 29: #include "libmandoc.h"
1.1 kristaps 30:
1.67 schwarze 31: #include "out.h"
32: #include "term.h"
33: #include "tbl.h"
34:
1.1 kristaps 35: const char *const __mdoc_macronames[MDOC_MAX] = {
1.7 schwarze 36: "Ap", "Dd", "Dt", "Os",
1.1 kristaps 37: "Sh", "Ss", "Pp", "D1",
38: "Dl", "Bd", "Ed", "Bl",
39: "El", "It", "Ad", "An",
40: "Ar", "Cd", "Cm", "Dv",
41: "Er", "Ev", "Ex", "Fa",
42: "Fd", "Fl", "Fn", "Ft",
43: "Ic", "In", "Li", "Nd",
44: "Nm", "Op", "Ot", "Pa",
45: "Rv", "St", "Va", "Vt",
46: /* LINTED */
1.33 schwarze 47: "Xr", "%A", "%B", "%D",
1.1 kristaps 48: /* LINTED */
1.33 schwarze 49: "%I", "%J", "%N", "%O",
1.1 kristaps 50: /* LINTED */
1.33 schwarze 51: "%P", "%R", "%T", "%V",
1.1 kristaps 52: "Ac", "Ao", "Aq", "At",
53: "Bc", "Bf", "Bo", "Bq",
54: "Bsx", "Bx", "Db", "Dc",
55: "Do", "Dq", "Ec", "Ef",
56: "Em", "Eo", "Fx", "Ms",
57: "No", "Ns", "Nx", "Ox",
58: "Pc", "Pf", "Po", "Pq",
59: "Qc", "Ql", "Qo", "Qq",
60: "Re", "Rs", "Sc", "So",
61: "Sq", "Sm", "Sx", "Sy",
62: "Tn", "Ux", "Xc", "Xo",
63: "Fo", "Fc", "Oo", "Oc",
64: "Bk", "Ek", "Bt", "Hf",
1.7 schwarze 65: "Fr", "Ud", "Lb", "Lp",
66: "Lk", "Mt", "Brq", "Bro",
1.1 kristaps 67: /* LINTED */
1.33 schwarze 68: "Brc", "%C", "Es", "En",
1.1 kristaps 69: /* LINTED */
1.33 schwarze 70: "Dx", "%Q", "br", "sp",
1.31 schwarze 71: /* LINTED */
1.67 schwarze 72: "%U", "Ta", "TS", "TE"
1.1 kristaps 73: };
74:
75: const char *const __mdoc_argnames[MDOC_ARG_MAX] = {
76: "split", "nosplit", "ragged",
77: "unfilled", "literal", "file",
78: "offset", "bullet", "dash",
79: "hyphen", "item", "enum",
80: "tag", "diag", "hang",
81: "ohang", "inset", "column",
82: "width", "compact", "std",
83: "filled", "words", "emphasis",
1.30 schwarze 84: "symbolic", "nested", "centered"
1.1 kristaps 85: };
86:
87: const char * const *mdoc_macronames = __mdoc_macronames;
88: const char * const *mdoc_argnames = __mdoc_argnames;
89:
1.40 schwarze 90: static void mdoc_node_free(struct mdoc_node *);
91: static void mdoc_node_unlink(struct mdoc *,
92: struct mdoc_node *);
1.1 kristaps 93: static void mdoc_free1(struct mdoc *);
1.32 schwarze 94: static void mdoc_alloc1(struct mdoc *);
1.1 kristaps 95: static struct mdoc_node *node_alloc(struct mdoc *, int, int,
1.37 schwarze 96: enum mdoct, enum mdoc_type);
1.1 kristaps 97: static int node_append(struct mdoc *,
98: struct mdoc_node *);
1.53 schwarze 99: static int mdoc_ptext(struct mdoc *, int, char *, int);
100: static int mdoc_pmacro(struct mdoc *, int, char *, int);
1.46 schwarze 101:
1.1 kristaps 102:
103: const struct mdoc_node *
104: mdoc_node(const struct mdoc *m)
105: {
106:
107: return(MDOC_HALT & m->flags ? NULL : m->first);
108: }
109:
110:
111: const struct mdoc_meta *
112: mdoc_meta(const struct mdoc *m)
113: {
114:
115: return(MDOC_HALT & m->flags ? NULL : &m->meta);
116: }
117:
118:
1.8 schwarze 119: /*
120: * Frees volatile resources (parse tree, meta-data, fields).
121: */
1.1 kristaps 122: static void
123: mdoc_free1(struct mdoc *mdoc)
124: {
125:
126: if (mdoc->first)
1.40 schwarze 127: mdoc_node_delete(mdoc, mdoc->first);
1.1 kristaps 128: if (mdoc->meta.title)
129: free(mdoc->meta.title);
130: if (mdoc->meta.os)
131: free(mdoc->meta.os);
132: if (mdoc->meta.name)
133: free(mdoc->meta.name);
134: if (mdoc->meta.arch)
135: free(mdoc->meta.arch);
136: if (mdoc->meta.vol)
137: free(mdoc->meta.vol);
1.50 schwarze 138: if (mdoc->meta.msec)
139: free(mdoc->meta.msec);
1.1 kristaps 140: }
141:
142:
1.8 schwarze 143: /*
144: * Allocate all volatile resources (parse tree, meta-data, fields).
145: */
1.32 schwarze 146: static void
1.1 kristaps 147: mdoc_alloc1(struct mdoc *mdoc)
148: {
149:
1.32 schwarze 150: memset(&mdoc->meta, 0, sizeof(struct mdoc_meta));
1.1 kristaps 151: mdoc->flags = 0;
1.9 schwarze 152: mdoc->lastnamed = mdoc->lastsec = SEC_NONE;
1.32 schwarze 153: mdoc->last = mandoc_calloc(1, sizeof(struct mdoc_node));
1.1 kristaps 154: mdoc->first = mdoc->last;
155: mdoc->last->type = MDOC_ROOT;
156: mdoc->next = MDOC_NEXT_CHILD;
157: }
158:
159:
160: /*
1.8 schwarze 161: * Free up volatile resources (see mdoc_free1()) then re-initialises the
162: * data with mdoc_alloc1(). After invocation, parse data has been reset
163: * and the parser is ready for re-invocation on a new tree; however,
164: * cross-parse non-volatile data is kept intact.
1.1 kristaps 165: */
1.32 schwarze 166: void
1.1 kristaps 167: mdoc_reset(struct mdoc *mdoc)
168: {
169:
170: mdoc_free1(mdoc);
1.32 schwarze 171: mdoc_alloc1(mdoc);
1.1 kristaps 172: }
173:
174:
175: /*
1.8 schwarze 176: * Completely free up all volatile and non-volatile parse resources.
177: * After invocation, the pointer is no longer usable.
1.1 kristaps 178: */
179: void
180: mdoc_free(struct mdoc *mdoc)
181: {
182:
183: mdoc_free1(mdoc);
184: free(mdoc);
185: }
186:
187:
1.8 schwarze 188: /*
189: * Allocate volatile and non-volatile parse resources.
190: */
1.1 kristaps 191: struct mdoc *
1.65 schwarze 192: mdoc_alloc(struct regset *regs, void *data, mandocmsg msg)
1.1 kristaps 193: {
194: struct mdoc *p;
195:
1.32 schwarze 196: p = mandoc_calloc(1, sizeof(struct mdoc));
197:
1.54 schwarze 198: p->msg = msg;
1.1 kristaps 199: p->data = data;
1.58 schwarze 200: p->regs = regs;
1.1 kristaps 201:
1.32 schwarze 202: mdoc_hash_init();
203: mdoc_alloc1(p);
204: return(p);
1.1 kristaps 205: }
206:
207:
208: /*
209: * Climb back up the parse tree, validating open scopes. Mostly calls
1.8 schwarze 210: * through to macro_end() in macro.c.
1.1 kristaps 211: */
212: int
213: mdoc_endparse(struct mdoc *m)
214: {
215:
216: if (MDOC_HALT & m->flags)
217: return(0);
218: else if (mdoc_macroend(m))
219: return(1);
220: m->flags |= MDOC_HALT;
221: return(0);
222: }
223:
224:
225: /*
226: * Main parse routine. Parses a single line -- really just hands off to
1.45 schwarze 227: * the macro (mdoc_pmacro()) or text parser (mdoc_ptext()).
1.1 kristaps 228: */
229: int
1.53 schwarze 230: mdoc_parseln(struct mdoc *m, int ln, char *buf, int offs)
1.1 kristaps 231: {
1.67 schwarze 232: struct mdoc_node *n;
1.1 kristaps 233:
234: if (MDOC_HALT & m->flags)
235: return(0);
236:
1.47 schwarze 237: m->flags |= MDOC_NEWLINE;
1.60 schwarze 238:
1.67 schwarze 239: n = m->last;
240:
241: if (n && MDOC_TS == n->tok && MDOC_BODY == n->type &&
242: strncmp(buf+offs, ".TE", 3)) {
243: n = n->parent;
1.68 schwarze 244: if ( ! tbl_read(n->data.TS, "mdoc tbl parser",
245: ln, buf+offs, strlen(buf+offs)))
246: mdoc_nmsg(m, n, MANDOCERR_TBL);
247: return(1);
1.67 schwarze 248: }
249:
1.60 schwarze 250: /*
251: * Let the roff nS register switch SYNOPSIS mode early,
252: * such that the parser knows at all times
253: * whether this mode is on or off.
254: * Note that this mode is also switched by the Sh macro.
255: */
256: if (m->regs->regs[(int)REG_nS].set) {
257: if (m->regs->regs[(int)REG_nS].v.u)
258: m->flags |= MDOC_SYNOPSIS;
259: else
260: m->flags &= ~MDOC_SYNOPSIS;
261: }
262:
1.53 schwarze 263: return(('.' == buf[offs] || '\'' == buf[offs]) ?
264: mdoc_pmacro(m, ln, buf, offs) :
265: mdoc_ptext(m, ln, buf, offs));
1.1 kristaps 266: }
267:
268:
269: int
1.54 schwarze 270: mdoc_vmsg(struct mdoc *mdoc, enum mandocerr t,
271: int ln, int pos, const char *fmt, ...)
1.1 kristaps 272: {
273: char buf[256];
274: va_list ap;
275:
276: va_start(ap, fmt);
1.54 schwarze 277: vsnprintf(buf, sizeof(buf) - 1, fmt, ap);
1.1 kristaps 278: va_end(ap);
1.17 schwarze 279:
1.54 schwarze 280: return((*mdoc->msg)(t, mdoc->data, ln, pos, buf));
1.5 schwarze 281: }
282:
283:
284: int
1.58 schwarze 285: mdoc_macro(MACRO_PROT_ARGS)
1.1 kristaps 286: {
1.46 schwarze 287: assert(tok < MDOC_MAX);
288:
289: /* If we're in the body, deny prologue calls. */
1.37 schwarze 290:
1.1 kristaps 291: if (MDOC_PROLOGUE & mdoc_macros[tok].flags &&
1.9 schwarze 292: MDOC_PBODY & m->flags)
1.58 schwarze 293: return(mdoc_pmsg(m, line, ppos, MANDOCERR_BADBODY));
1.46 schwarze 294:
295: /* If we're in the prologue, deny "body" macros. */
296:
1.1 kristaps 297: if ( ! (MDOC_PROLOGUE & mdoc_macros[tok].flags) &&
1.39 schwarze 298: ! (MDOC_PBODY & m->flags)) {
1.58 schwarze 299: if ( ! mdoc_pmsg(m, line, ppos, MANDOCERR_BADPROLOG))
1.39 schwarze 300: return(0);
301: if (NULL == m->meta.title)
1.56 schwarze 302: m->meta.title = mandoc_strdup("UNKNOWN");
1.39 schwarze 303: if (NULL == m->meta.vol)
1.56 schwarze 304: m->meta.vol = mandoc_strdup("LOCAL");
1.39 schwarze 305: if (NULL == m->meta.os)
1.56 schwarze 306: m->meta.os = mandoc_strdup("LOCAL");
1.39 schwarze 307: if (0 == m->meta.date)
308: m->meta.date = time(NULL);
309: m->flags |= MDOC_PBODY;
310: }
1.1 kristaps 311:
1.58 schwarze 312: return((*mdoc_macros[tok].fp)(m, tok, line, ppos, pos, buf));
1.1 kristaps 313: }
314:
315:
316: static int
317: node_append(struct mdoc *mdoc, struct mdoc_node *p)
318: {
319:
320: assert(mdoc->last);
321: assert(mdoc->first);
322: assert(MDOC_ROOT != p->type);
323:
324: switch (mdoc->next) {
325: case (MDOC_NEXT_SIBLING):
326: mdoc->last->next = p;
327: p->prev = mdoc->last;
328: p->parent = mdoc->last->parent;
329: break;
330: case (MDOC_NEXT_CHILD):
331: mdoc->last->child = p;
332: p->parent = mdoc->last;
333: break;
334: default:
335: abort();
336: /* NOTREACHED */
337: }
338:
1.10 schwarze 339: p->parent->nchild++;
340:
1.1 kristaps 341: if ( ! mdoc_valid_pre(mdoc, p))
342: return(0);
343:
344: switch (p->type) {
345: case (MDOC_HEAD):
346: assert(MDOC_BLOCK == p->parent->type);
347: p->parent->head = p;
348: break;
349: case (MDOC_TAIL):
350: assert(MDOC_BLOCK == p->parent->type);
351: p->parent->tail = p;
352: break;
353: case (MDOC_BODY):
1.59 schwarze 354: if (p->end)
355: break;
1.1 kristaps 356: assert(MDOC_BLOCK == p->parent->type);
357: p->parent->body = p;
358: break;
359: default:
360: break;
361: }
362:
363: mdoc->last = p;
364:
365: switch (p->type) {
366: case (MDOC_TEXT):
367: if ( ! mdoc_valid_post(mdoc))
368: return(0);
369: break;
370: default:
371: break;
372: }
373:
374: return(1);
375: }
376:
377:
378: static struct mdoc_node *
1.37 schwarze 379: node_alloc(struct mdoc *m, int line, int pos,
380: enum mdoct tok, enum mdoc_type type)
1.1 kristaps 381: {
382: struct mdoc_node *p;
383:
1.32 schwarze 384: p = mandoc_calloc(1, sizeof(struct mdoc_node));
1.19 schwarze 385: p->sec = m->lastsec;
1.1 kristaps 386: p->line = line;
387: p->pos = pos;
388: p->tok = tok;
1.37 schwarze 389: p->type = type;
1.58 schwarze 390:
391: /* Flag analysis. */
392:
1.60 schwarze 393: if (MDOC_SYNOPSIS & m->flags)
394: p->flags |= MDOC_SYNPRETTY;
395: else
396: p->flags &= ~MDOC_SYNPRETTY;
1.47 schwarze 397: if (MDOC_NEWLINE & m->flags)
398: p->flags |= MDOC_LINE;
399: m->flags &= ~MDOC_NEWLINE;
1.58 schwarze 400:
1.1 kristaps 401: return(p);
402: }
403:
404:
405: int
1.37 schwarze 406: mdoc_tail_alloc(struct mdoc *m, int line, int pos, enum mdoct tok)
1.1 kristaps 407: {
408: struct mdoc_node *p;
409:
1.19 schwarze 410: p = node_alloc(m, line, pos, tok, MDOC_TAIL);
1.25 schwarze 411: if ( ! node_append(m, p))
412: return(0);
413: m->next = MDOC_NEXT_CHILD;
414: return(1);
1.1 kristaps 415: }
416:
417:
418: int
1.37 schwarze 419: mdoc_head_alloc(struct mdoc *m, int line, int pos, enum mdoct tok)
1.1 kristaps 420: {
421: struct mdoc_node *p;
422:
1.19 schwarze 423: assert(m->first);
424: assert(m->last);
1.1 kristaps 425:
1.19 schwarze 426: p = node_alloc(m, line, pos, tok, MDOC_HEAD);
1.25 schwarze 427: if ( ! node_append(m, p))
428: return(0);
429: m->next = MDOC_NEXT_CHILD;
430: return(1);
1.1 kristaps 431: }
432:
433:
434: int
1.37 schwarze 435: mdoc_body_alloc(struct mdoc *m, int line, int pos, enum mdoct tok)
1.1 kristaps 436: {
437: struct mdoc_node *p;
438:
1.19 schwarze 439: p = node_alloc(m, line, pos, tok, MDOC_BODY);
1.25 schwarze 440: if ( ! node_append(m, p))
441: return(0);
442: m->next = MDOC_NEXT_CHILD;
1.59 schwarze 443: return(1);
444: }
445:
446:
447: int
448: mdoc_endbody_alloc(struct mdoc *m, int line, int pos, enum mdoct tok,
449: struct mdoc_node *body, enum mdoc_endbody end)
450: {
451: struct mdoc_node *p;
452:
453: p = node_alloc(m, line, pos, tok, MDOC_BODY);
454: p->pending = body;
455: p->end = end;
456: if ( ! node_append(m, p))
457: return(0);
458: m->next = MDOC_NEXT_SIBLING;
1.25 schwarze 459: return(1);
1.1 kristaps 460: }
461:
462:
463: int
1.19 schwarze 464: mdoc_block_alloc(struct mdoc *m, int line, int pos,
1.37 schwarze 465: enum mdoct tok, struct mdoc_arg *args)
1.1 kristaps 466: {
467: struct mdoc_node *p;
468:
1.19 schwarze 469: p = node_alloc(m, line, pos, tok, MDOC_BLOCK);
1.4 schwarze 470: p->args = args;
471: if (p->args)
1.1 kristaps 472: (args->refcnt)++;
1.25 schwarze 473: if ( ! node_append(m, p))
474: return(0);
475: m->next = MDOC_NEXT_CHILD;
476: return(1);
1.1 kristaps 477: }
478:
479:
480: int
1.19 schwarze 481: mdoc_elem_alloc(struct mdoc *m, int line, int pos,
1.37 schwarze 482: enum mdoct tok, struct mdoc_arg *args)
1.1 kristaps 483: {
484: struct mdoc_node *p;
485:
1.19 schwarze 486: p = node_alloc(m, line, pos, tok, MDOC_ELEM);
1.4 schwarze 487: p->args = args;
488: if (p->args)
1.1 kristaps 489: (args->refcnt)++;
1.25 schwarze 490: if ( ! node_append(m, p))
491: return(0);
492: m->next = MDOC_NEXT_CHILD;
493: return(1);
1.1 kristaps 494: }
495:
496:
1.46 schwarze 497: int
498: mdoc_word_alloc(struct mdoc *m, int line, int pos, const char *p)
1.1 kristaps 499: {
1.19 schwarze 500: struct mdoc_node *n;
1.46 schwarze 501: size_t sv, len;
502:
503: len = strlen(p);
1.1 kristaps 504:
1.46 schwarze 505: n = node_alloc(m, line, pos, MDOC_MAX, MDOC_TEXT);
1.32 schwarze 506: n->string = mandoc_malloc(len + 1);
1.19 schwarze 507: sv = strlcpy(n->string, p, len + 1);
508:
509: /* Prohibit truncation. */
510: assert(sv < len + 1);
511:
1.25 schwarze 512: if ( ! node_append(m, n))
513: return(0);
1.46 schwarze 514:
1.25 schwarze 515: m->next = MDOC_NEXT_SIBLING;
516: return(1);
1.19 schwarze 517: }
518:
519:
1.61 schwarze 520: static void
1.1 kristaps 521: mdoc_node_free(struct mdoc_node *p)
522: {
523:
1.61 schwarze 524: /*
525: * XXX: if these end up being problematic in terms of memory
526: * management and dereferencing freed blocks, then make them
527: * into reference-counted double-pointers.
528: */
529:
530: if (MDOC_Bd == p->tok && MDOC_BLOCK == p->type)
531: if (p->data.Bd)
532: free(p->data.Bd);
533: if (MDOC_Bl == p->tok && MDOC_BLOCK == p->type)
534: if (p->data.Bl)
535: free(p->data.Bl);
536: if (MDOC_Bf == p->tok && MDOC_HEAD == p->type)
537: if (p->data.Bf)
538: free(p->data.Bf);
1.67 schwarze 539: if (MDOC_TS == p->tok && MDOC_BLOCK == p->type)
540: if (p->data.TS)
541: tbl_free(p->data.TS);
1.61 schwarze 542:
1.1 kristaps 543: if (p->string)
544: free(p->string);
545: if (p->args)
546: mdoc_argv_free(p->args);
547: free(p);
548: }
549:
550:
1.40 schwarze 551: static void
552: mdoc_node_unlink(struct mdoc *m, struct mdoc_node *n)
553: {
554:
555: /* Adjust siblings. */
556:
557: if (n->prev)
558: n->prev->next = n->next;
559: if (n->next)
560: n->next->prev = n->prev;
561:
562: /* Adjust parent. */
563:
564: if (n->parent) {
565: n->parent->nchild--;
566: if (n->parent->child == n)
567: n->parent->child = n->prev ? n->prev : n->next;
568: }
569:
570: /* Adjust parse point, if applicable. */
571:
572: if (m && m->last == n) {
573: if (n->prev) {
574: m->last = n->prev;
575: m->next = MDOC_NEXT_SIBLING;
576: } else {
577: m->last = n->parent;
578: m->next = MDOC_NEXT_CHILD;
579: }
580: }
581:
582: if (m && m->first == n)
583: m->first = NULL;
584: }
585:
586:
1.1 kristaps 587: void
1.40 schwarze 588: mdoc_node_delete(struct mdoc *m, struct mdoc_node *p)
1.1 kristaps 589: {
590:
1.40 schwarze 591: while (p->child) {
592: assert(p->nchild);
593: mdoc_node_delete(m, p->child);
594: }
595: assert(0 == p->nchild);
1.1 kristaps 596:
1.40 schwarze 597: mdoc_node_unlink(m, p);
1.1 kristaps 598: mdoc_node_free(p);
599: }
600:
601:
602: /*
603: * Parse free-form text, that is, a line that does not begin with the
604: * control character.
605: */
606: static int
1.53 schwarze 607: mdoc_ptext(struct mdoc *m, int line, char *buf, int offs)
1.1 kristaps 608: {
1.56 schwarze 609: char *c, *ws, *end;
610: struct mdoc_node *n;
1.44 schwarze 611:
612: /* Ignore bogus comments. */
613:
1.53 schwarze 614: if ('\\' == buf[offs] &&
615: '.' == buf[offs + 1] &&
616: '"' == buf[offs + 2])
1.54 schwarze 617: return(mdoc_pmsg(m, line, offs, MANDOCERR_BADCOMMENT));
1.1 kristaps 618:
1.46 schwarze 619: /* No text before an initial macro. */
620:
1.9 schwarze 621: if (SEC_NONE == m->lastnamed)
1.54 schwarze 622: return(mdoc_pmsg(m, line, offs, MANDOCERR_NOTEXT));
1.46 schwarze 623:
1.56 schwarze 624: assert(m->last);
625: n = m->last;
626:
627: /*
628: * Divert directly to list processing if we're encountering a
629: * columnar MDOC_BLOCK with or without a prior MDOC_BLOCK entry
630: * (a MDOC_BODY means it's already open, in which case we should
631: * process within its context in the normal way).
632: */
633:
634: if (MDOC_Bl == n->tok && MDOC_BODY == n->type &&
1.61 schwarze 635: LIST_column == n->data.Bl->type) {
1.56 schwarze 636: /* `Bl' is open without any children. */
637: m->flags |= MDOC_FREECOL;
638: return(mdoc_macro(m, MDOC_It, line, offs, &offs, buf));
639: }
640:
641: if (MDOC_It == n->tok && MDOC_BLOCK == n->type &&
642: NULL != n->parent &&
643: MDOC_Bl == n->parent->tok &&
1.61 schwarze 644: LIST_column == n->parent->data.Bl->type) {
1.56 schwarze 645: /* `Bl' has block-level `It' children. */
646: m->flags |= MDOC_FREECOL;
647: return(mdoc_macro(m, MDOC_It, line, offs, &offs, buf));
648: }
649:
1.52 schwarze 650: /*
651: * Search for the beginning of unescaped trailing whitespace (ws)
652: * and for the first character not to be output (end).
653: */
1.56 schwarze 654:
655: /* FIXME: replace with strcspn(). */
1.52 schwarze 656: ws = NULL;
1.53 schwarze 657: for (c = end = buf + offs; *c; c++) {
1.52 schwarze 658: switch (*c) {
1.55 schwarze 659: case '-':
660: if (mandoc_hyph(buf + offs, c))
661: *c = ASCII_HYPH;
1.56 schwarze 662: ws = NULL;
1.55 schwarze 663: break;
1.52 schwarze 664: case ' ':
665: if (NULL == ws)
666: ws = c;
667: continue;
668: case '\t':
669: /*
670: * Always warn about trailing tabs,
671: * even outside literal context,
672: * where they should be put on the next line.
673: */
674: if (NULL == ws)
675: ws = c;
676: /*
677: * Strip trailing tabs in literal context only;
678: * outside, they affect the next line.
679: */
680: if (MDOC_LITERAL & m->flags)
681: continue;
682: break;
683: case '\\':
684: /* Skip the escaped character, too, if any. */
685: if (c[1])
686: c++;
687: /* FALLTHROUGH */
688: default:
689: ws = NULL;
690: break;
691: }
692: end = c + 1;
693: }
694: *end = '\0';
1.19 schwarze 695:
1.52 schwarze 696: if (ws)
1.54 schwarze 697: if ( ! mdoc_pmsg(m, line, (int)(ws-buf), MANDOCERR_EOLNSPACE))
1.52 schwarze 698: return(0);
1.34 schwarze 699:
1.53 schwarze 700: if ('\0' == buf[offs] && ! (MDOC_LITERAL & m->flags)) {
1.54 schwarze 701: if ( ! mdoc_pmsg(m, line, (int)(c-buf), MANDOCERR_NOBLANKLN))
1.38 schwarze 702: return(0);
1.46 schwarze 703:
1.40 schwarze 704: /*
1.66 schwarze 705: * Insert a `sp' in the case of a blank line. Technically,
1.46 schwarze 706: * blank lines aren't allowed, but enough manuals assume this
707: * behaviour that we want to work around it.
1.40 schwarze 708: */
1.66 schwarze 709: if ( ! mdoc_elem_alloc(m, line, offs, MDOC_sp, NULL))
1.42 schwarze 710: return(0);
1.46 schwarze 711:
1.42 schwarze 712: m->next = MDOC_NEXT_SIBLING;
713: return(1);
1.38 schwarze 714: }
1.1 kristaps 715:
1.53 schwarze 716: if ( ! mdoc_word_alloc(m, line, offs, buf+offs))
1.1 kristaps 717: return(0);
718:
1.52 schwarze 719: if (MDOC_LITERAL & m->flags)
720: return(1);
721:
1.35 schwarze 722: /*
1.48 schwarze 723: * End-of-sentence check. If the last character is an unescaped
724: * EOS character, then flag the node as being the end of a
725: * sentence. The front-end will know how to interpret this.
1.35 schwarze 726: */
727:
1.52 schwarze 728: assert(buf < end);
1.48 schwarze 729:
1.62 schwarze 730: if (mandoc_eos(buf+offs, (size_t)(end-buf-offs), 0))
1.48 schwarze 731: m->last->flags |= MDOC_EOS;
732:
1.1 kristaps 733: return(1);
734: }
1.19 schwarze 735:
1.1 kristaps 736:
737: /*
738: * Parse a macro line, that is, a line beginning with the control
739: * character.
740: */
1.61 schwarze 741: static int
1.53 schwarze 742: mdoc_pmacro(struct mdoc *m, int ln, char *buf, int offs)
1.1 kristaps 743: {
1.56 schwarze 744: enum mdoct tok;
745: int i, j, sv;
746: char mac[5];
747: struct mdoc_node *n;
1.1 kristaps 748:
1.7 schwarze 749: /* Empty lines are ignored. */
1.1 kristaps 750:
1.53 schwarze 751: offs++;
752:
753: if ('\0' == buf[offs])
1.1 kristaps 754: return(1);
755:
1.53 schwarze 756: i = offs;
1.24 schwarze 757:
1.63 schwarze 758: /* Accept tabs/whitespace after the initial control char. */
1.24 schwarze 759:
1.63 schwarze 760: if (' ' == buf[i] || '\t' == buf[i]) {
1.24 schwarze 761: i++;
1.63 schwarze 762: while (buf[i] && (' ' == buf[i] || '\t' == buf[i]))
1.1 kristaps 763: i++;
1.34 schwarze 764: if ('\0' == buf[i])
1.1 kristaps 765: return(1);
766: }
767:
1.47 schwarze 768: sv = i;
769:
1.63 schwarze 770: /*
1.64 schwarze 771: * Copy the first word into a nil-terminated buffer.
772: * Stop copying when a tab, space, or eoln is encountered.
1.63 schwarze 773: */
1.1 kristaps 774:
1.64 schwarze 775: j = 0;
776: while (j < 4 && '\0' != buf[i] && ' ' != buf[i] && '\t' != buf[i])
777: mac[j++] = buf[i++];
1.53 schwarze 778: mac[j] = '\0';
1.1 kristaps 779:
1.65 schwarze 780: tok = (j > 1 || j < 4) ? mdoc_hash_find(mac) : MDOC_MAX;
781: if (MDOC_MAX == tok) {
1.69 schwarze 782: mdoc_vmsg(m, MANDOCERR_MACRO, ln, sv, "%s", buf + sv - 1);
1.1 kristaps 783: return(1);
784: }
785:
1.63 schwarze 786: /* Disregard the first trailing tab, if applicable. */
787:
788: if ('\t' == buf[i])
789: i++;
790:
791: /* Jump to the next non-whitespace word. */
1.1 kristaps 792:
793: while (buf[i] && ' ' == buf[i])
794: i++;
1.34 schwarze 795:
1.46 schwarze 796: /*
797: * Trailing whitespace. Note that tabs are allowed to be passed
798: * into the parser as "text", so we only warn about spaces here.
799: */
1.34 schwarze 800:
801: if ('\0' == buf[i] && ' ' == buf[i - 1])
1.54 schwarze 802: if ( ! mdoc_pmsg(m, ln, i - 1, MANDOCERR_EOLNSPACE))
1.34 schwarze 803: goto err;
1.1 kristaps 804:
1.56 schwarze 805: /*
806: * If an initial macro or a list invocation, divert directly
807: * into macro processing.
808: */
809:
810: if (NULL == m->last || MDOC_It == tok || MDOC_El == tok) {
811: if ( ! mdoc_macro(m, tok, ln, sv, &i, buf))
812: goto err;
813: return(1);
814: }
815:
816: n = m->last;
817: assert(m->last);
818:
819: /*
820: * If the first macro of a `Bl -column', open an `It' block
821: * context around the parsed macro.
822: */
823:
824: if (MDOC_Bl == n->tok && MDOC_BODY == n->type &&
1.61 schwarze 825: LIST_column == n->data.Bl->type) {
1.56 schwarze 826: m->flags |= MDOC_FREECOL;
1.58 schwarze 827: if ( ! mdoc_macro(m, MDOC_It, ln, sv, &sv, buf))
1.56 schwarze 828: goto err;
829: return(1);
830: }
831:
832: /*
833: * If we're following a block-level `It' within a `Bl -column'
834: * context (perhaps opened in the above block or in ptext()),
835: * then open an `It' block context around the parsed macro.
1.24 schwarze 836: */
1.56 schwarze 837:
838: if (MDOC_It == n->tok && MDOC_BLOCK == n->type &&
839: NULL != n->parent &&
840: MDOC_Bl == n->parent->tok &&
1.61 schwarze 841: LIST_column == n->parent->data.Bl->type) {
1.56 schwarze 842: m->flags |= MDOC_FREECOL;
843: if ( ! mdoc_macro(m, MDOC_It, ln, sv, &sv, buf))
844: goto err;
845: return(1);
846: }
847:
848: /* Normal processing of a macro. */
849:
1.47 schwarze 850: if ( ! mdoc_macro(m, tok, ln, sv, &i, buf))
1.1 kristaps 851: goto err;
852:
853: return(1);
854:
855: err: /* Error out. */
856:
857: m->flags |= MDOC_HALT;
858: return(0);
859: }
1.24 schwarze 860:
861: