Annotation of src/usr.bin/mandoc/mdoc.c, Revision 1.68
1.68 ! schwarze 1: /* $Id: mdoc.c,v 1.67 2010/10/16 13:38:29 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: if ( ! mdoc_action_pre(mdoc, p))
344: return(0);
345:
346: switch (p->type) {
347: case (MDOC_HEAD):
348: assert(MDOC_BLOCK == p->parent->type);
349: p->parent->head = p;
350: break;
351: case (MDOC_TAIL):
352: assert(MDOC_BLOCK == p->parent->type);
353: p->parent->tail = p;
354: break;
355: case (MDOC_BODY):
1.59 schwarze 356: if (p->end)
357: break;
1.1 kristaps 358: assert(MDOC_BLOCK == p->parent->type);
359: p->parent->body = p;
360: break;
361: default:
362: break;
363: }
364:
365: mdoc->last = p;
366:
367: switch (p->type) {
368: case (MDOC_TEXT):
369: if ( ! mdoc_valid_post(mdoc))
370: return(0);
371: if ( ! mdoc_action_post(mdoc))
372: return(0);
373: break;
374: default:
375: break;
376: }
377:
378: return(1);
379: }
380:
381:
382: static struct mdoc_node *
1.37 schwarze 383: node_alloc(struct mdoc *m, int line, int pos,
384: enum mdoct tok, enum mdoc_type type)
1.1 kristaps 385: {
386: struct mdoc_node *p;
387:
1.32 schwarze 388: p = mandoc_calloc(1, sizeof(struct mdoc_node));
1.19 schwarze 389: p->sec = m->lastsec;
1.1 kristaps 390: p->line = line;
391: p->pos = pos;
392: p->tok = tok;
1.37 schwarze 393: p->type = type;
1.58 schwarze 394:
395: /* Flag analysis. */
396:
1.60 schwarze 397: if (MDOC_SYNOPSIS & m->flags)
398: p->flags |= MDOC_SYNPRETTY;
399: else
400: p->flags &= ~MDOC_SYNPRETTY;
1.47 schwarze 401: if (MDOC_NEWLINE & m->flags)
402: p->flags |= MDOC_LINE;
403: m->flags &= ~MDOC_NEWLINE;
1.58 schwarze 404:
1.1 kristaps 405: return(p);
406: }
407:
408:
409: int
1.37 schwarze 410: mdoc_tail_alloc(struct mdoc *m, int line, int pos, enum mdoct tok)
1.1 kristaps 411: {
412: struct mdoc_node *p;
413:
1.19 schwarze 414: p = node_alloc(m, line, pos, tok, MDOC_TAIL);
1.25 schwarze 415: if ( ! node_append(m, p))
416: return(0);
417: m->next = MDOC_NEXT_CHILD;
418: return(1);
1.1 kristaps 419: }
420:
421:
422: int
1.37 schwarze 423: mdoc_head_alloc(struct mdoc *m, int line, int pos, enum mdoct tok)
1.1 kristaps 424: {
425: struct mdoc_node *p;
426:
1.19 schwarze 427: assert(m->first);
428: assert(m->last);
1.1 kristaps 429:
1.19 schwarze 430: p = node_alloc(m, line, pos, tok, MDOC_HEAD);
1.25 schwarze 431: if ( ! node_append(m, p))
432: return(0);
433: m->next = MDOC_NEXT_CHILD;
434: return(1);
1.1 kristaps 435: }
436:
437:
438: int
1.37 schwarze 439: mdoc_body_alloc(struct mdoc *m, int line, int pos, enum mdoct tok)
1.1 kristaps 440: {
441: struct mdoc_node *p;
442:
1.19 schwarze 443: p = node_alloc(m, line, pos, tok, MDOC_BODY);
1.25 schwarze 444: if ( ! node_append(m, p))
445: return(0);
446: m->next = MDOC_NEXT_CHILD;
1.59 schwarze 447: return(1);
448: }
449:
450:
451: int
452: mdoc_endbody_alloc(struct mdoc *m, int line, int pos, enum mdoct tok,
453: struct mdoc_node *body, enum mdoc_endbody end)
454: {
455: struct mdoc_node *p;
456:
457: p = node_alloc(m, line, pos, tok, MDOC_BODY);
458: p->pending = body;
459: p->end = end;
460: if ( ! node_append(m, p))
461: return(0);
462: m->next = MDOC_NEXT_SIBLING;
1.25 schwarze 463: return(1);
1.1 kristaps 464: }
465:
466:
467: int
1.19 schwarze 468: mdoc_block_alloc(struct mdoc *m, int line, int pos,
1.37 schwarze 469: enum mdoct tok, struct mdoc_arg *args)
1.1 kristaps 470: {
471: struct mdoc_node *p;
472:
1.19 schwarze 473: p = node_alloc(m, line, pos, tok, MDOC_BLOCK);
1.4 schwarze 474: p->args = args;
475: if (p->args)
1.1 kristaps 476: (args->refcnt)++;
1.25 schwarze 477: if ( ! node_append(m, p))
478: return(0);
479: m->next = MDOC_NEXT_CHILD;
480: return(1);
1.1 kristaps 481: }
482:
483:
484: int
1.19 schwarze 485: mdoc_elem_alloc(struct mdoc *m, int line, int pos,
1.37 schwarze 486: enum mdoct tok, struct mdoc_arg *args)
1.1 kristaps 487: {
488: struct mdoc_node *p;
489:
1.19 schwarze 490: p = node_alloc(m, line, pos, tok, MDOC_ELEM);
1.4 schwarze 491: p->args = args;
492: if (p->args)
1.1 kristaps 493: (args->refcnt)++;
1.25 schwarze 494: if ( ! node_append(m, p))
495: return(0);
496: m->next = MDOC_NEXT_CHILD;
497: return(1);
1.1 kristaps 498: }
499:
500:
1.46 schwarze 501: int
502: mdoc_word_alloc(struct mdoc *m, int line, int pos, const char *p)
1.1 kristaps 503: {
1.19 schwarze 504: struct mdoc_node *n;
1.46 schwarze 505: size_t sv, len;
506:
507: len = strlen(p);
1.1 kristaps 508:
1.46 schwarze 509: n = node_alloc(m, line, pos, MDOC_MAX, MDOC_TEXT);
1.32 schwarze 510: n->string = mandoc_malloc(len + 1);
1.19 schwarze 511: sv = strlcpy(n->string, p, len + 1);
512:
513: /* Prohibit truncation. */
514: assert(sv < len + 1);
515:
1.25 schwarze 516: if ( ! node_append(m, n))
517: return(0);
1.46 schwarze 518:
1.25 schwarze 519: m->next = MDOC_NEXT_SIBLING;
520: return(1);
1.19 schwarze 521: }
522:
523:
1.61 schwarze 524: static void
1.1 kristaps 525: mdoc_node_free(struct mdoc_node *p)
526: {
527:
1.61 schwarze 528: /*
529: * XXX: if these end up being problematic in terms of memory
530: * management and dereferencing freed blocks, then make them
531: * into reference-counted double-pointers.
532: */
533:
534: if (MDOC_Bd == p->tok && MDOC_BLOCK == p->type)
535: if (p->data.Bd)
536: free(p->data.Bd);
537: if (MDOC_Bl == p->tok && MDOC_BLOCK == p->type)
538: if (p->data.Bl)
539: free(p->data.Bl);
540: if (MDOC_Bf == p->tok && MDOC_HEAD == p->type)
541: if (p->data.Bf)
542: free(p->data.Bf);
1.67 schwarze 543: if (MDOC_TS == p->tok && MDOC_BLOCK == p->type)
544: if (p->data.TS)
545: tbl_free(p->data.TS);
1.61 schwarze 546:
1.1 kristaps 547: if (p->string)
548: free(p->string);
549: if (p->args)
550: mdoc_argv_free(p->args);
551: free(p);
552: }
553:
554:
1.40 schwarze 555: static void
556: mdoc_node_unlink(struct mdoc *m, struct mdoc_node *n)
557: {
558:
559: /* Adjust siblings. */
560:
561: if (n->prev)
562: n->prev->next = n->next;
563: if (n->next)
564: n->next->prev = n->prev;
565:
566: /* Adjust parent. */
567:
568: if (n->parent) {
569: n->parent->nchild--;
570: if (n->parent->child == n)
571: n->parent->child = n->prev ? n->prev : n->next;
572: }
573:
574: /* Adjust parse point, if applicable. */
575:
576: if (m && m->last == n) {
577: if (n->prev) {
578: m->last = n->prev;
579: m->next = MDOC_NEXT_SIBLING;
580: } else {
581: m->last = n->parent;
582: m->next = MDOC_NEXT_CHILD;
583: }
584: }
585:
586: if (m && m->first == n)
587: m->first = NULL;
588: }
589:
590:
1.1 kristaps 591: void
1.40 schwarze 592: mdoc_node_delete(struct mdoc *m, struct mdoc_node *p)
1.1 kristaps 593: {
594:
1.40 schwarze 595: while (p->child) {
596: assert(p->nchild);
597: mdoc_node_delete(m, p->child);
598: }
599: assert(0 == p->nchild);
1.1 kristaps 600:
1.40 schwarze 601: mdoc_node_unlink(m, p);
1.1 kristaps 602: mdoc_node_free(p);
603: }
604:
605:
606: /*
607: * Parse free-form text, that is, a line that does not begin with the
608: * control character.
609: */
610: static int
1.53 schwarze 611: mdoc_ptext(struct mdoc *m, int line, char *buf, int offs)
1.1 kristaps 612: {
1.56 schwarze 613: char *c, *ws, *end;
614: struct mdoc_node *n;
1.44 schwarze 615:
616: /* Ignore bogus comments. */
617:
1.53 schwarze 618: if ('\\' == buf[offs] &&
619: '.' == buf[offs + 1] &&
620: '"' == buf[offs + 2])
1.54 schwarze 621: return(mdoc_pmsg(m, line, offs, MANDOCERR_BADCOMMENT));
1.1 kristaps 622:
1.46 schwarze 623: /* No text before an initial macro. */
624:
1.9 schwarze 625: if (SEC_NONE == m->lastnamed)
1.54 schwarze 626: return(mdoc_pmsg(m, line, offs, MANDOCERR_NOTEXT));
1.46 schwarze 627:
1.56 schwarze 628: assert(m->last);
629: n = m->last;
630:
631: /*
632: * Divert directly to list processing if we're encountering a
633: * columnar MDOC_BLOCK with or without a prior MDOC_BLOCK entry
634: * (a MDOC_BODY means it's already open, in which case we should
635: * process within its context in the normal way).
636: */
637:
638: if (MDOC_Bl == n->tok && MDOC_BODY == n->type &&
1.61 schwarze 639: LIST_column == n->data.Bl->type) {
1.56 schwarze 640: /* `Bl' is open without any children. */
641: m->flags |= MDOC_FREECOL;
642: return(mdoc_macro(m, MDOC_It, line, offs, &offs, buf));
643: }
644:
645: if (MDOC_It == n->tok && MDOC_BLOCK == n->type &&
646: NULL != n->parent &&
647: MDOC_Bl == n->parent->tok &&
1.61 schwarze 648: LIST_column == n->parent->data.Bl->type) {
1.56 schwarze 649: /* `Bl' has block-level `It' children. */
650: m->flags |= MDOC_FREECOL;
651: return(mdoc_macro(m, MDOC_It, line, offs, &offs, buf));
652: }
653:
1.52 schwarze 654: /*
655: * Search for the beginning of unescaped trailing whitespace (ws)
656: * and for the first character not to be output (end).
657: */
1.56 schwarze 658:
659: /* FIXME: replace with strcspn(). */
1.52 schwarze 660: ws = NULL;
1.53 schwarze 661: for (c = end = buf + offs; *c; c++) {
1.52 schwarze 662: switch (*c) {
1.55 schwarze 663: case '-':
664: if (mandoc_hyph(buf + offs, c))
665: *c = ASCII_HYPH;
1.56 schwarze 666: ws = NULL;
1.55 schwarze 667: break;
1.52 schwarze 668: case ' ':
669: if (NULL == ws)
670: ws = c;
671: continue;
672: case '\t':
673: /*
674: * Always warn about trailing tabs,
675: * even outside literal context,
676: * where they should be put on the next line.
677: */
678: if (NULL == ws)
679: ws = c;
680: /*
681: * Strip trailing tabs in literal context only;
682: * outside, they affect the next line.
683: */
684: if (MDOC_LITERAL & m->flags)
685: continue;
686: break;
687: case '\\':
688: /* Skip the escaped character, too, if any. */
689: if (c[1])
690: c++;
691: /* FALLTHROUGH */
692: default:
693: ws = NULL;
694: break;
695: }
696: end = c + 1;
697: }
698: *end = '\0';
1.19 schwarze 699:
1.52 schwarze 700: if (ws)
1.54 schwarze 701: if ( ! mdoc_pmsg(m, line, (int)(ws-buf), MANDOCERR_EOLNSPACE))
1.52 schwarze 702: return(0);
1.34 schwarze 703:
1.53 schwarze 704: if ('\0' == buf[offs] && ! (MDOC_LITERAL & m->flags)) {
1.54 schwarze 705: if ( ! mdoc_pmsg(m, line, (int)(c-buf), MANDOCERR_NOBLANKLN))
1.38 schwarze 706: return(0);
1.46 schwarze 707:
1.40 schwarze 708: /*
1.66 schwarze 709: * Insert a `sp' in the case of a blank line. Technically,
1.46 schwarze 710: * blank lines aren't allowed, but enough manuals assume this
711: * behaviour that we want to work around it.
1.40 schwarze 712: */
1.66 schwarze 713: if ( ! mdoc_elem_alloc(m, line, offs, MDOC_sp, NULL))
1.42 schwarze 714: return(0);
1.46 schwarze 715:
1.42 schwarze 716: m->next = MDOC_NEXT_SIBLING;
717: return(1);
1.38 schwarze 718: }
1.1 kristaps 719:
1.53 schwarze 720: if ( ! mdoc_word_alloc(m, line, offs, buf+offs))
1.1 kristaps 721: return(0);
722:
1.52 schwarze 723: if (MDOC_LITERAL & m->flags)
724: return(1);
725:
1.35 schwarze 726: /*
1.48 schwarze 727: * End-of-sentence check. If the last character is an unescaped
728: * EOS character, then flag the node as being the end of a
729: * sentence. The front-end will know how to interpret this.
1.35 schwarze 730: */
731:
1.52 schwarze 732: assert(buf < end);
1.48 schwarze 733:
1.62 schwarze 734: if (mandoc_eos(buf+offs, (size_t)(end-buf-offs), 0))
1.48 schwarze 735: m->last->flags |= MDOC_EOS;
736:
1.1 kristaps 737: return(1);
738: }
1.19 schwarze 739:
1.1 kristaps 740:
741: /*
742: * Parse a macro line, that is, a line beginning with the control
743: * character.
744: */
1.61 schwarze 745: static int
1.53 schwarze 746: mdoc_pmacro(struct mdoc *m, int ln, char *buf, int offs)
1.1 kristaps 747: {
1.56 schwarze 748: enum mdoct tok;
749: int i, j, sv;
750: char mac[5];
751: struct mdoc_node *n;
1.1 kristaps 752:
1.7 schwarze 753: /* Empty lines are ignored. */
1.1 kristaps 754:
1.53 schwarze 755: offs++;
756:
757: if ('\0' == buf[offs])
1.1 kristaps 758: return(1);
759:
1.53 schwarze 760: i = offs;
1.24 schwarze 761:
1.63 schwarze 762: /* Accept tabs/whitespace after the initial control char. */
1.24 schwarze 763:
1.63 schwarze 764: if (' ' == buf[i] || '\t' == buf[i]) {
1.24 schwarze 765: i++;
1.63 schwarze 766: while (buf[i] && (' ' == buf[i] || '\t' == buf[i]))
1.1 kristaps 767: i++;
1.34 schwarze 768: if ('\0' == buf[i])
1.1 kristaps 769: return(1);
770: }
771:
1.47 schwarze 772: sv = i;
773:
1.63 schwarze 774: /*
1.64 schwarze 775: * Copy the first word into a nil-terminated buffer.
776: * Stop copying when a tab, space, or eoln is encountered.
1.63 schwarze 777: */
1.1 kristaps 778:
1.64 schwarze 779: j = 0;
780: while (j < 4 && '\0' != buf[i] && ' ' != buf[i] && '\t' != buf[i])
781: mac[j++] = buf[i++];
1.53 schwarze 782: mac[j] = '\0';
1.1 kristaps 783:
1.65 schwarze 784: tok = (j > 1 || j < 4) ? mdoc_hash_find(mac) : MDOC_MAX;
785: if (MDOC_MAX == tok) {
786: mdoc_vmsg(m, MANDOCERR_MACRO, ln, sv,
787: "unknown macro: %s%s",
788: buf, strlen(buf) > 3 ? "..." : "");
1.1 kristaps 789: return(1);
790: }
791:
1.63 schwarze 792: /* Disregard the first trailing tab, if applicable. */
793:
794: if ('\t' == buf[i])
795: i++;
796:
797: /* Jump to the next non-whitespace word. */
1.1 kristaps 798:
799: while (buf[i] && ' ' == buf[i])
800: i++;
1.34 schwarze 801:
1.46 schwarze 802: /*
803: * Trailing whitespace. Note that tabs are allowed to be passed
804: * into the parser as "text", so we only warn about spaces here.
805: */
1.34 schwarze 806:
807: if ('\0' == buf[i] && ' ' == buf[i - 1])
1.54 schwarze 808: if ( ! mdoc_pmsg(m, ln, i - 1, MANDOCERR_EOLNSPACE))
1.34 schwarze 809: goto err;
1.1 kristaps 810:
1.56 schwarze 811: /*
812: * If an initial macro or a list invocation, divert directly
813: * into macro processing.
814: */
815:
816: if (NULL == m->last || MDOC_It == tok || MDOC_El == tok) {
817: if ( ! mdoc_macro(m, tok, ln, sv, &i, buf))
818: goto err;
819: return(1);
820: }
821:
822: n = m->last;
823: assert(m->last);
824:
825: /*
826: * If the first macro of a `Bl -column', open an `It' block
827: * context around the parsed macro.
828: */
829:
830: if (MDOC_Bl == n->tok && MDOC_BODY == n->type &&
1.61 schwarze 831: LIST_column == n->data.Bl->type) {
1.56 schwarze 832: m->flags |= MDOC_FREECOL;
1.58 schwarze 833: if ( ! mdoc_macro(m, MDOC_It, ln, sv, &sv, buf))
1.56 schwarze 834: goto err;
835: return(1);
836: }
837:
838: /*
839: * If we're following a block-level `It' within a `Bl -column'
840: * context (perhaps opened in the above block or in ptext()),
841: * then open an `It' block context around the parsed macro.
1.24 schwarze 842: */
1.56 schwarze 843:
844: if (MDOC_It == n->tok && MDOC_BLOCK == n->type &&
845: NULL != n->parent &&
846: MDOC_Bl == n->parent->tok &&
1.61 schwarze 847: LIST_column == n->parent->data.Bl->type) {
1.56 schwarze 848: m->flags |= MDOC_FREECOL;
849: if ( ! mdoc_macro(m, MDOC_It, ln, sv, &sv, buf))
850: goto err;
851: return(1);
852: }
853:
854: /* Normal processing of a macro. */
855:
1.47 schwarze 856: if ( ! mdoc_macro(m, tok, ln, sv, &i, buf))
1.1 kristaps 857: goto err;
858:
859: return(1);
860:
861: err: /* Error out. */
862:
863: m->flags |= MDOC_HALT;
864: return(0);
865: }
1.24 schwarze 866:
867: