]> git.cameronkatri.com Git - mandoc.git/blob - mdoc_macro.c
git.ckatri.com / mandoc.git / blob
? search:
summary | shortlog | log | commit | commitdiff | tree
history | raw | HEAD
* need a space before .No even if it starts with a closing delimiter
[mandoc.git] / mdoc_macro.c
1 /* $Id: mdoc_macro.c,v 1.95 2010/10/01 21:51:13 schwarze Exp $ */
2 /*
3 * Copyright (c) 2008, 2009, 2010 Kristaps Dzonsons <kristaps@bsd.lv>
4 * Copyright (c) 2010 Ingo Schwarze <schwarze@openbsd.org>
5 *
6 * Permission to use, copy, modify, and distribute this software for any
7 * purpose with or without fee is hereby granted, provided that the above
8 * copyright notice and this permission notice appear in all copies.
9 *
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.
17 */
18 #ifdef HAVE_CONFIG_H
19 #include "config.h"
20 #endif
21
22 #include <assert.h>
23 #include <ctype.h>
24 #include <stdlib.h>
25 #include <stdio.h>
26 #include <string.h>
27 #include <time.h>
28
29 #include "mandoc.h"
30 #include "libmdoc.h"
31 #include "libmandoc.h"
32
33 enum rew { /* see rew_dohalt() */
34 REWIND_NONE,
35 REWIND_THIS,
36 REWIND_MORE,
37 REWIND_LATER,
38 REWIND_ERROR
39 };
40
41 static int blk_full(MACRO_PROT_ARGS);
42 static int blk_exp_close(MACRO_PROT_ARGS);
43 static int blk_part_exp(MACRO_PROT_ARGS);
44 static int blk_part_imp(MACRO_PROT_ARGS);
45 static int ctx_synopsis(MACRO_PROT_ARGS);
46 static int in_line_eoln(MACRO_PROT_ARGS);
47 static int in_line_argn(MACRO_PROT_ARGS);
48 static int in_line(MACRO_PROT_ARGS);
49 static int obsolete(MACRO_PROT_ARGS);
50 static int phrase_ta(MACRO_PROT_ARGS);
51
52 static int append_delims(struct mdoc *,
53 int, int *, char *);
54 static enum mdoct lookup(enum mdoct, const char *);
55 static enum mdoct lookup_raw(const char *);
56 static int make_pending(struct mdoc_node *, enum mdoct,
57 struct mdoc *, int, int);
58 static int phrase(struct mdoc *, int, int, char *);
59 static enum mdoct rew_alt(enum mdoct);
60 static enum rew rew_dohalt(enum mdoct, enum mdoc_type,
61 const struct mdoc_node *);
62 static int rew_elem(struct mdoc *, enum mdoct);
63 static int rew_last(struct mdoc *,
64 const struct mdoc_node *);
65 static int rew_sub(enum mdoc_type, struct mdoc *,
66 enum mdoct, int, int);
67
68 const struct mdoc_macro __mdoc_macros[MDOC_MAX] = {
69 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* Ap */
70 { in_line_eoln, MDOC_PROLOGUE }, /* Dd */
71 { in_line_eoln, MDOC_PROLOGUE }, /* Dt */
72 { in_line_eoln, MDOC_PROLOGUE }, /* Os */
73 { blk_full, 0 }, /* Sh */
74 { blk_full, 0 }, /* Ss */
75 { in_line_eoln, 0 }, /* Pp */
76 { blk_part_imp, MDOC_PARSED }, /* D1 */
77 { blk_part_imp, MDOC_PARSED }, /* Dl */
78 { blk_full, MDOC_EXPLICIT }, /* Bd */
79 { blk_exp_close, MDOC_EXPLICIT }, /* Ed */
80 { blk_full, MDOC_EXPLICIT }, /* Bl */
81 { blk_exp_close, MDOC_EXPLICIT }, /* El */
82 { blk_full, MDOC_PARSED }, /* It */
83 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Ad */
84 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* An */
85 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Ar */
86 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Cd */
87 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Cm */
88 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Dv */
89 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Er */
90 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Ev */
91 { in_line_eoln, 0 }, /* Ex */
92 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Fa */
93 { in_line_eoln, 0 }, /* Fd */
94 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Fl */
95 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Fn */
96 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Ft */
97 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Ic */
98 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* In */
99 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Li */
100 { blk_full, 0 }, /* Nd */
101 { ctx_synopsis, MDOC_CALLABLE | MDOC_PARSED }, /* Nm */
102 { blk_part_imp, MDOC_CALLABLE | MDOC_PARSED }, /* Op */
103 { obsolete, 0 }, /* Ot */
104 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Pa */
105 { in_line_eoln, 0 }, /* Rv */
106 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* St */
107 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Va */
108 { ctx_synopsis, MDOC_CALLABLE | MDOC_PARSED }, /* Vt */
109 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* Xr */
110 { in_line_eoln, 0 }, /* %A */
111 { in_line_eoln, 0 }, /* %B */
112 { in_line_eoln, 0 }, /* %D */
113 { in_line_eoln, 0 }, /* %I */
114 { in_line_eoln, 0 }, /* %J */
115 { in_line_eoln, 0 }, /* %N */
116 { in_line_eoln, 0 }, /* %O */
117 { in_line_eoln, 0 }, /* %P */
118 { in_line_eoln, 0 }, /* %R */
119 { in_line_eoln, 0 }, /* %T */
120 { in_line_eoln, 0 }, /* %V */
121 { blk_exp_close, MDOC_EXPLICIT | MDOC_CALLABLE | MDOC_PARSED }, /* Ac */
122 { blk_part_exp, MDOC_CALLABLE | MDOC_PARSED | MDOC_EXPLICIT }, /* Ao */
123 { blk_part_imp, MDOC_CALLABLE | MDOC_PARSED }, /* Aq */
124 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* At */
125 { blk_exp_close, MDOC_EXPLICIT | MDOC_CALLABLE | MDOC_PARSED }, /* Bc */
126 { blk_full, MDOC_EXPLICIT }, /* Bf */
127 { blk_part_exp, MDOC_CALLABLE | MDOC_PARSED | MDOC_EXPLICIT }, /* Bo */
128 { blk_part_imp, MDOC_CALLABLE | MDOC_PARSED }, /* Bq */
129 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* Bsx */
130 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* Bx */
131 { in_line_eoln, 0 }, /* Db */
132 { blk_exp_close, MDOC_EXPLICIT | MDOC_CALLABLE | MDOC_PARSED }, /* Dc */
133 { blk_part_exp, MDOC_CALLABLE | MDOC_PARSED | MDOC_EXPLICIT }, /* Do */
134 { blk_part_imp, MDOC_CALLABLE | MDOC_PARSED }, /* Dq */
135 { blk_exp_close, MDOC_EXPLICIT | MDOC_CALLABLE | MDOC_PARSED }, /* Ec */
136 { blk_exp_close, MDOC_EXPLICIT }, /* Ef */
137 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Em */
138 { blk_part_exp, MDOC_CALLABLE | MDOC_PARSED | MDOC_EXPLICIT }, /* Eo */
139 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* Fx */
140 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Ms */
141 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED | MDOC_IGNDELIM }, /* No */
142 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED | MDOC_IGNDELIM }, /* Ns */
143 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* Nx */
144 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* Ox */
145 { blk_exp_close, MDOC_EXPLICIT | MDOC_CALLABLE | MDOC_PARSED }, /* Pc */
146 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED | MDOC_IGNDELIM }, /* Pf */
147 { blk_part_exp, MDOC_CALLABLE | MDOC_PARSED | MDOC_EXPLICIT }, /* Po */
148 { blk_part_imp, MDOC_CALLABLE | MDOC_PARSED }, /* Pq */
149 { blk_exp_close, MDOC_EXPLICIT | MDOC_CALLABLE | MDOC_PARSED }, /* Qc */
150 { blk_part_imp, MDOC_CALLABLE | MDOC_PARSED }, /* Ql */
151 { blk_part_exp, MDOC_CALLABLE | MDOC_PARSED | MDOC_EXPLICIT }, /* Qo */
152 { blk_part_imp, MDOC_CALLABLE | MDOC_PARSED }, /* Qq */
153 { blk_exp_close, MDOC_EXPLICIT }, /* Re */
154 { blk_full, MDOC_EXPLICIT }, /* Rs */
155 { blk_exp_close, MDOC_EXPLICIT | MDOC_CALLABLE | MDOC_PARSED }, /* Sc */
156 { blk_part_exp, MDOC_CALLABLE | MDOC_PARSED | MDOC_EXPLICIT }, /* So */
157 { blk_part_imp, MDOC_CALLABLE | MDOC_PARSED }, /* Sq */
158 { in_line_eoln, 0 }, /* Sm */
159 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Sx */
160 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Sy */
161 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Tn */
162 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* Ux */
163 { blk_exp_close, MDOC_EXPLICIT | MDOC_CALLABLE | MDOC_PARSED }, /* Xc */
164 { blk_part_exp, MDOC_CALLABLE | MDOC_PARSED | MDOC_EXPLICIT }, /* Xo */
165 { blk_full, MDOC_EXPLICIT | MDOC_CALLABLE }, /* Fo */
166 { blk_exp_close, MDOC_EXPLICIT | MDOC_CALLABLE | MDOC_PARSED }, /* Fc */
167 { blk_part_exp, MDOC_CALLABLE | MDOC_PARSED | MDOC_EXPLICIT }, /* Oo */
168 { blk_exp_close, MDOC_EXPLICIT | MDOC_CALLABLE | MDOC_PARSED }, /* Oc */
169 { blk_full, MDOC_EXPLICIT }, /* Bk */
170 { blk_exp_close, MDOC_EXPLICIT }, /* Ek */
171 { in_line_eoln, 0 }, /* Bt */
172 { in_line_eoln, 0 }, /* Hf */
173 { obsolete, 0 }, /* Fr */
174 { in_line_eoln, 0 }, /* Ud */
175 { in_line, 0 }, /* Lb */
176 { in_line_eoln, 0 }, /* Lp */
177 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Lk */
178 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Mt */
179 { blk_part_imp, MDOC_CALLABLE | MDOC_PARSED }, /* Brq */
180 { blk_part_exp, MDOC_CALLABLE | MDOC_PARSED | MDOC_EXPLICIT }, /* Bro */
181 { blk_exp_close, MDOC_EXPLICIT | MDOC_CALLABLE | MDOC_PARSED }, /* Brc */
182 { in_line_eoln, 0 }, /* %C */
183 { obsolete, 0 }, /* Es */
184 { obsolete, 0 }, /* En */
185 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* Dx */
186 { in_line_eoln, 0 }, /* %Q */
187 { in_line_eoln, 0 }, /* br */
188 { in_line_eoln, 0 }, /* sp */
189 { in_line_eoln, 0 }, /* %U */
190 { phrase_ta, MDOC_CALLABLE | MDOC_PARSED }, /* Ta */
191 };
192
193 const struct mdoc_macro * const mdoc_macros = __mdoc_macros;
194
195
196 /*
197 * This is called at the end of parsing. It must traverse up the tree,
198 * closing out open [implicit] scopes. Obviously, open explicit scopes
199 * are errors.
200 */
201 int
202 mdoc_macroend(struct mdoc *m)
203 {
204 struct mdoc_node *n;
205
206 /* Scan for open explicit scopes. */
207
208 n = MDOC_VALID & m->last->flags ? m->last->parent : m->last;
209
210 for ( ; n; n = n->parent)
211 if (MDOC_BLOCK == n->type &&
212 MDOC_EXPLICIT & mdoc_macros[n->tok].flags)
213 mdoc_nmsg(m, n, MANDOCERR_SCOPEEXIT);
214
215 /* Rewind to the first. */
216
217 return(rew_last(m, m->first));
218 }
219
220
221 /*
222 * Look up a macro from within a subsequent context.
223 */
224 static enum mdoct
225 lookup(enum mdoct from, const char *p)
226 {
227 /* FIXME: make -diag lists be un-PARSED. */
228
229 if ( ! (MDOC_PARSED & mdoc_macros[from].flags))
230 return(MDOC_MAX);
231 return(lookup_raw(p));
232 }
233
234
235 /*
236 * Lookup a macro following the initial line macro.
237 */
238 static enum mdoct
239 lookup_raw(const char *p)
240 {
241 enum mdoct res;
242
243 if (MDOC_MAX == (res = mdoc_hash_find(p)))
244 return(MDOC_MAX);
245 if (MDOC_CALLABLE & mdoc_macros[res].flags)
246 return(res);
247 return(MDOC_MAX);
248 }
249
250
251 static int
252 rew_last(struct mdoc *mdoc, const struct mdoc_node *to)
253 {
254
255 assert(to);
256 mdoc->next = MDOC_NEXT_SIBLING;
257
258 /* LINTED */
259 while (mdoc->last != to) {
260 if ( ! mdoc_valid_post(mdoc))
261 return(0);
262 if ( ! mdoc_action_post(mdoc))
263 return(0);
264 mdoc->last = mdoc->last->parent;
265 assert(mdoc->last);
266 }
267
268 if ( ! mdoc_valid_post(mdoc))
269 return(0);
270 return(mdoc_action_post(mdoc));
271 }
272
273
274 /*
275 * For a block closing macro, return the corresponding opening one.
276 * Otherwise, return the macro itself.
277 */
278 static enum mdoct
279 rew_alt(enum mdoct tok)
280 {
281 switch (tok) {
282 case (MDOC_Ac):
283 return(MDOC_Ao);
284 case (MDOC_Bc):
285 return(MDOC_Bo);
286 case (MDOC_Brc):
287 return(MDOC_Bro);
288 case (MDOC_Dc):
289 return(MDOC_Do);
290 case (MDOC_Ec):
291 return(MDOC_Eo);
292 case (MDOC_Ed):
293 return(MDOC_Bd);
294 case (MDOC_Ef):
295 return(MDOC_Bf);
296 case (MDOC_Ek):
297 return(MDOC_Bk);
298 case (MDOC_El):
299 return(MDOC_Bl);
300 case (MDOC_Fc):
301 return(MDOC_Fo);
302 case (MDOC_Oc):
303 return(MDOC_Oo);
304 case (MDOC_Pc):
305 return(MDOC_Po);
306 case (MDOC_Qc):
307 return(MDOC_Qo);
308 case (MDOC_Re):
309 return(MDOC_Rs);
310 case (MDOC_Sc):
311 return(MDOC_So);
312 case (MDOC_Xc):
313 return(MDOC_Xo);
314 default:
315 return(tok);
316 }
317 /* NOTREACHED */
318 }
319
320
321 /*
322 * Rewinding to tok, how do we have to handle *p?
323 * REWIND_NONE: *p would delimit tok, but no tok scope is open
324 * inside *p, so there is no need to rewind anything at all.
325 * REWIND_THIS: *p matches tok, so rewind *p and nothing else.
326 * REWIND_MORE: *p is implicit, rewind it and keep searching for tok.
327 * REWIND_LATER: *p is explicit and still open, postpone rewinding.
328 * REWIND_ERROR: No tok block is open at all.
329 */
330 static enum rew
331 rew_dohalt(enum mdoct tok, enum mdoc_type type,
332 const struct mdoc_node *p)
333 {
334
335 /*
336 * No matching token, no delimiting block, no broken block.
337 * This can happen when full implicit macros are called for
338 * the first time but try to rewind their previous
339 * instance anyway.
340 */
341 if (MDOC_ROOT == p->type)
342 return(MDOC_BLOCK == type &&
343 MDOC_EXPLICIT & mdoc_macros[tok].flags ?
344 REWIND_ERROR : REWIND_NONE);
345
346 /*
347 * When starting to rewind, skip plain text
348 * and nodes that have already been rewound.
349 */
350 if (MDOC_TEXT == p->type || MDOC_VALID & p->flags)
351 return(REWIND_MORE);
352
353 /*
354 * The easiest case: Found a matching token.
355 * This applies to both blocks and elements.
356 */
357 tok = rew_alt(tok);
358 if (tok == p->tok)
359 return(p->end ? REWIND_NONE :
360 type == p->type ? REWIND_THIS : REWIND_MORE);
361
362 /*
363 * While elements do require rewinding for themselves,
364 * they never affect rewinding of other nodes.
365 */
366 if (MDOC_ELEM == p->type)
367 return(REWIND_MORE);
368
369 /*
370 * Blocks delimited by our target token get REWIND_MORE.
371 * Blocks delimiting our target token get REWIND_NONE.
372 */
373 switch (tok) {
374 case (MDOC_Bl):
375 if (MDOC_It == p->tok)
376 return(REWIND_MORE);
377 break;
378 case (MDOC_It):
379 if (MDOC_BODY == p->type && MDOC_Bl == p->tok)
380 return(REWIND_NONE);
381 break;
382 /*
383 * XXX Badly nested block handling still fails badly
384 * when one block is breaking two blocks of the same type.
385 * This is an incomplete and extremely ugly workaround,
386 * required to let the OpenBSD tree build.
387 */
388 case (MDOC_Oo):
389 if (MDOC_Op == p->tok)
390 return(REWIND_MORE);
391 break;
392 case (MDOC_Nm):
393 return(REWIND_NONE);
394 case (MDOC_Nd):
395 /* FALLTHROUGH */
396 case (MDOC_Ss):
397 if (MDOC_BODY == p->type && MDOC_Sh == p->tok)
398 return(REWIND_NONE);
399 /* FALLTHROUGH */
400 case (MDOC_Sh):
401 if (MDOC_Nd == p->tok || MDOC_Ss == p->tok ||
402 MDOC_Sh == p->tok)
403 return(REWIND_MORE);
404 break;
405 default:
406 break;
407 }
408
409 /*
410 * Default block rewinding rules.
411 * In particular, always skip block end markers,
412 * and let all blocks rewind Nm children.
413 */
414 if (ENDBODY_NOT != p->end || MDOC_Nm == p->tok ||
415 (MDOC_BLOCK == p->type &&
416 ! (MDOC_EXPLICIT & mdoc_macros[tok].flags)))
417 return(REWIND_MORE);
418
419 /*
420 * Partial blocks allow delayed rewinding by default.
421 */
422 if (&blk_full != mdoc_macros[tok].fp)
423 return (REWIND_LATER);
424
425 /*
426 * Full blocks can only be rewound when matching
427 * or when there is an explicit rule.
428 */
429 return(REWIND_ERROR);
430 }
431
432
433 static int
434 rew_elem(struct mdoc *mdoc, enum mdoct tok)
435 {
436 struct mdoc_node *n;
437
438 n = mdoc->last;
439 if (MDOC_ELEM != n->type)
440 n = n->parent;
441 assert(MDOC_ELEM == n->type);
442 assert(tok == n->tok);
443
444 return(rew_last(mdoc, n));
445 }
446
447
448 /*
449 * We are trying to close a block identified by tok,
450 * but the child block *broken is still open.
451 * Thus, postpone closing the tok block
452 * until the rew_sub call closing *broken.
453 */
454 static int
455 make_pending(struct mdoc_node *broken, enum mdoct tok,
456 struct mdoc *m, int line, int ppos)
457 {
458 struct mdoc_node *breaker;
459
460 /*
461 * Iterate backwards, searching for the block matching tok,
462 * that is, the block breaking the *broken block.
463 */
464 for (breaker = broken->parent; breaker; breaker = breaker->parent) {
465
466 /*
467 * If the *broken block had already been broken before
468 * and we encounter its breaker, make the tok block
469 * pending on the inner breaker.
470 * Graphically, "[A breaker=[B broken=[C->B B] tok=A] C]"
471 * becomes "[A broken=[B [C->B B] tok=A] C]"
472 * and finally "[A [B->A [C->B B] A] C]".
473 */
474 if (breaker == broken->pending) {
475 broken = breaker;
476 continue;
477 }
478
479 if (REWIND_THIS != rew_dohalt(tok, MDOC_BLOCK, breaker))
480 continue;
481 if (MDOC_BODY == broken->type)
482 broken = broken->parent;
483
484 /*
485 * Found the breaker.
486 * If another, outer breaker is already pending on
487 * the *broken block, we must not clobber the link
488 * to the outer breaker, but make it pending on the
489 * new, now inner breaker.
490 * Graphically, "[A breaker=[B broken=[C->A A] tok=B] C]"
491 * becomes "[A breaker=[B->A broken=[C A] tok=B] C]"
492 * and finally "[A [B->A [C->B A] B] C]".
493 */
494 if (broken->pending) {
495 struct mdoc_node *taker;
496
497 /*
498 * If the breaker had also been broken before,
499 * it cannot take on the outer breaker itself,
500 * but must hand it on to its own breakers.
501 * Graphically, this is the following situation:
502 * "[A [B breaker=[C->B B] broken=[D->A A] tok=C] D]"
503 * "[A taker=[B->A breaker=[C->B B] [D->C A] C] D]"
504 */
505 taker = breaker;
506 while (taker->pending)
507 taker = taker->pending;
508 taker->pending = broken->pending;
509 }
510 broken->pending = breaker;
511 mdoc_vmsg(m, MANDOCERR_SCOPENEST, line, ppos,
512 "%s breaks %s", mdoc_macronames[tok],
513 mdoc_macronames[broken->tok]);
514 return(1);
515 }
516
517 /*
518 * Found no matching block for tok.
519 * Are you trying to close a block that is not open?
520 * XXX Make this non-fatal.
521 */
522 mdoc_pmsg(m, line, ppos, MANDOCERR_SYNTNOSCOPE);
523 return(0);
524 }
525
526
527 static int
528 rew_sub(enum mdoc_type t, struct mdoc *m,
529 enum mdoct tok, int line, int ppos)
530 {
531 struct mdoc_node *n;
532
533 n = m->last;
534 while (n) {
535 switch (rew_dohalt(tok, t, n)) {
536 case (REWIND_NONE):
537 return(1);
538 case (REWIND_THIS):
539 break;
540 case (REWIND_MORE):
541 n = n->parent;
542 continue;
543 case (REWIND_LATER):
544 return(make_pending(n, tok, m, line, ppos));
545 case (REWIND_ERROR):
546 /* XXX Make this non-fatal. */
547 mdoc_vmsg(m, MANDOCERR_SCOPEFATAL, line, ppos,
548 "%s cannot break %s", mdoc_macronames[tok],
549 mdoc_macronames[n->tok]);
550 return 0;
551 }
552 break;
553 }
554
555 assert(n);
556 if ( ! rew_last(m, n))
557 return(0);
558
559 /*
560 * The current block extends an enclosing block.
561 * Now that the current block ends, close the enclosing block, too.
562 */
563 while (NULL != (n = n->pending)) {
564 if ( ! rew_last(m, n))
565 return(0);
566 if (MDOC_HEAD == n->type &&
567 ! mdoc_body_alloc(m, n->line, n->pos, n->tok))
568 return(0);
569 }
570
571 return(1);
572 }
573
574
575 static int
576 append_delims(struct mdoc *m, int line, int *pos, char *buf)
577 {
578 int la;
579 enum margserr ac;
580 char *p;
581
582 if ('\0' == buf[*pos])
583 return(1);
584
585 for (;;) {
586 la = *pos;
587 ac = mdoc_zargs(m, line, pos, buf, ARGS_NOWARN, &p);
588
589 if (ARGS_ERROR == ac)
590 return(0);
591 else if (ARGS_EOLN == ac)
592 break;
593
594 assert(DELIM_NONE != mdoc_isdelim(p));
595 if ( ! mdoc_word_alloc(m, line, la, p))
596 return(0);
597
598 /*
599 * If we encounter end-of-sentence symbols, then trigger
600 * the double-space.
601 *
602 * XXX: it's easy to allow this to propogate outward to
603 * the last symbol, such that `. )' will cause the
604 * correct double-spacing. However, (1) groff isn't
605 * smart enough to do this and (2) it would require
606 * knowing which symbols break this behaviour, for
607 * example, `. ;' shouldn't propogate the double-space.
608 */
609 if (mandoc_eos(p, strlen(p), 0))
610 m->last->flags |= MDOC_EOS;
611 }
612
613 return(1);
614 }
615
616
617 /*
618 * Close out block partial/full explicit.
619 */
620 static int
621 blk_exp_close(MACRO_PROT_ARGS)
622 {
623 struct mdoc_node *body; /* Our own body. */
624 struct mdoc_node *later; /* A sub-block starting later. */
625 struct mdoc_node *n; /* For searching backwards. */
626
627 int j, lastarg, maxargs, flushed, nl;
628 enum margserr ac;
629 enum mdoct atok, ntok;
630 char *p;
631
632 nl = MDOC_NEWLINE & m->flags;
633
634 switch (tok) {
635 case (MDOC_Ec):
636 maxargs = 1;
637 break;
638 default:
639 maxargs = 0;
640 break;
641 }
642
643 /*
644 * Search backwards for beginnings of blocks,
645 * both of our own and of pending sub-blocks.
646 */
647 atok = rew_alt(tok);
648 body = later = NULL;
649 for (n = m->last; n; n = n->parent) {
650 if (MDOC_VALID & n->flags)
651 continue;
652
653 /* Remember the start of our own body. */
654 if (MDOC_BODY == n->type && atok == n->tok) {
655 if (ENDBODY_NOT == n->end)
656 body = n;
657 continue;
658 }
659
660 if (MDOC_BLOCK != n->type || MDOC_Nm == n->tok)
661 continue;
662 if (atok == n->tok) {
663 assert(body);
664
665 /*
666 * Found the start of our own block.
667 * When there is no pending sub block,
668 * just proceed to closing out.
669 */
670 if (NULL == later)
671 break;
672
673 /*
674 * When there is a pending sub block,
675 * postpone closing out the current block
676 * until the rew_sub() closing out the sub-block.
677 */
678 if ( ! make_pending(later, tok, m, line, ppos))
679 return(0);
680
681 /*
682 * Mark the place where the formatting - but not
683 * the scope - of the current block ends.
684 */
685 if ( ! mdoc_endbody_alloc(m, line, ppos,
686 atok, body, ENDBODY_SPACE))
687 return(0);
688 break;
689 }
690
691 /*
692 * When finding an open sub block, remember the last
693 * open explicit block, or, in case there are only
694 * implicit ones, the first open implicit block.
695 */
696 if (later &&
697 MDOC_EXPLICIT & mdoc_macros[later->tok].flags)
698 continue;
699 if (MDOC_CALLABLE & mdoc_macros[n->tok].flags) {
700 assert( ! (MDOC_ACTED & n->flags));
701 later = n;
702 }
703 }
704
705 if ( ! (MDOC_CALLABLE & mdoc_macros[tok].flags)) {
706 /* FIXME: do this in validate */
707 if (buf[*pos])
708 if ( ! mdoc_pmsg(m, line, ppos, MANDOCERR_ARGSLOST))
709 return(0);
710
711 if ( ! rew_sub(MDOC_BODY, m, tok, line, ppos))
712 return(0);
713 return(rew_sub(MDOC_BLOCK, m, tok, line, ppos));
714 }
715
716 if ( ! rew_sub(MDOC_BODY, m, tok, line, ppos))
717 return(0);
718
719 if (NULL == later && maxargs > 0)
720 if ( ! mdoc_tail_alloc(m, line, ppos, rew_alt(tok)))
721 return(0);
722
723 for (flushed = j = 0; ; j++) {
724 lastarg = *pos;
725
726 if (j == maxargs && ! flushed) {
727 if ( ! rew_sub(MDOC_BLOCK, m, tok, line, ppos))
728 return(0);
729 flushed = 1;
730 }
731
732 ac = mdoc_args(m, line, pos, buf, tok, &p);
733
734 if (ARGS_ERROR == ac)
735 return(0);
736 if (ARGS_PUNCT == ac)
737 break;
738 if (ARGS_EOLN == ac)
739 break;
740
741 ntok = ARGS_QWORD == ac ? MDOC_MAX : lookup(tok, p);
742
743 if (MDOC_MAX == ntok) {
744 if ( ! mdoc_word_alloc(m, line, lastarg, p))
745 return(0);
746 continue;
747 }
748
749 if ( ! flushed) {
750 if ( ! rew_sub(MDOC_BLOCK, m, tok, line, ppos))
751 return(0);
752 flushed = 1;
753 }
754 if ( ! mdoc_macro(m, ntok, line, lastarg, pos, buf))
755 return(0);
756 break;
757 }
758
759 if ( ! flushed && ! rew_sub(MDOC_BLOCK, m, tok, line, ppos))
760 return(0);
761
762 if ( ! nl)
763 return(1);
764 return(append_delims(m, line, pos, buf));
765 }
766
767
768 static int
769 in_line(MACRO_PROT_ARGS)
770 {
771 int la, scope, cnt, nc, nl;
772 enum margverr av;
773 enum mdoct ntok;
774 enum margserr ac;
775 enum mdelim d;
776 struct mdoc_arg *arg;
777 char *p;
778
779 nl = MDOC_NEWLINE & m->flags;
780
781 /*
782 * Whether we allow ignored elements (those without content,
783 * usually because of reserved words) to squeak by.
784 */
785
786 switch (tok) {
787 case (MDOC_An):
788 /* FALLTHROUGH */
789 case (MDOC_Ar):
790 /* FALLTHROUGH */
791 case (MDOC_Fl):
792 /* FALLTHROUGH */
793 case (MDOC_Lk):
794 /* FALLTHROUGH */
795 case (MDOC_Nm):
796 /* FALLTHROUGH */
797 case (MDOC_Pa):
798 nc = 1;
799 break;
800 default:
801 nc = 0;
802 break;
803 }
804
805 for (arg = NULL;; ) {
806 la = *pos;
807 av = mdoc_argv(m, line, tok, &arg, pos, buf);
808
809 if (ARGV_WORD == av) {
810 *pos = la;
811 break;
812 }
813 if (ARGV_EOLN == av)
814 break;
815 if (ARGV_ARG == av)
816 continue;
817
818 mdoc_argv_free(arg);
819 return(0);
820 }
821
822 for (cnt = scope = 0;; ) {
823 la = *pos;
824 ac = mdoc_args(m, line, pos, buf, tok, &p);
825
826 if (ARGS_ERROR == ac)
827 return(0);
828 if (ARGS_EOLN == ac)
829 break;
830 if (ARGS_PUNCT == ac)
831 break;
832
833 ntok = ARGS_QWORD == ac ? MDOC_MAX : lookup(tok, p);
834
835 /*
836 * In this case, we've located a submacro and must
837 * execute it. Close out scope, if open. If no
838 * elements have been generated, either create one (nc)
839 * or raise a warning.
840 */
841
842 if (MDOC_MAX != ntok) {
843 if (scope && ! rew_elem(m, tok))
844 return(0);
845 if (nc && 0 == cnt) {
846 if ( ! mdoc_elem_alloc(m, line, ppos, tok, arg))
847 return(0);
848 if ( ! rew_last(m, m->last))
849 return(0);
850 } else if ( ! nc && 0 == cnt) {
851 mdoc_argv_free(arg);
852 if ( ! mdoc_pmsg(m, line, ppos, MANDOCERR_MACROEMPTY))
853 return(0);
854 }
855 if ( ! mdoc_macro(m, ntok, line, la, pos, buf))
856 return(0);
857 if ( ! nl)
858 return(1);
859 return(append_delims(m, line, pos, buf));
860 }
861
862 /*
863 * Non-quote-enclosed punctuation. Set up our scope, if
864 * a word; rewind the scope, if a delimiter; then append
865 * the word.
866 */
867
868 d = ARGS_QWORD == ac ? DELIM_NONE : mdoc_isdelim(p);
869
870 if (DELIM_NONE != d) {
871 /*
872 * If we encounter closing punctuation, no word
873 * has been omitted, no scope is open, and we're
874 * allowed to have an empty element, then start
875 * a new scope. `Ar', `Fl', and `Li', only do
876 * this once per invocation. There may be more
877 * of these (all of them?).
878 */
879 if (0 == cnt && (nc || MDOC_Li == tok) &&
880 DELIM_CLOSE == d && ! scope) {
881 if ( ! mdoc_elem_alloc(m, line, ppos, tok, arg))
882 return(0);
883 if (MDOC_Ar == tok || MDOC_Li == tok ||
884 MDOC_Fl == tok)
885 cnt++;
886 scope = 1;
887 }
888 /*
889 * Close out our scope, if one is open, before
890 * any punctuation.
891 */
892 if (scope && ! rew_elem(m, tok))
893 return(0);
894 scope = 0;
895 } else if ( ! scope) {
896 if ( ! mdoc_elem_alloc(m, line, ppos, tok, arg))
897 return(0);
898 scope = 1;
899 }
900
901 if (DELIM_NONE == d)
902 cnt++;
903 if ( ! mdoc_word_alloc(m, line, la, p))
904 return(0);
905
906 /*
907 * `Fl' macros have their scope re-opened with each new
908 * word so that the `-' can be added to each one without
909 * having to parse out spaces.
910 */
911 if (scope && MDOC_Fl == tok) {
912 if ( ! rew_elem(m, tok))
913 return(0);
914 scope = 0;
915 }
916 }
917
918 if (scope && ! rew_elem(m, tok))
919 return(0);
920
921 /*
922 * If no elements have been collected and we're allowed to have
923 * empties (nc), open a scope and close it out. Otherwise,
924 * raise a warning.
925 */
926
927 if (nc && 0 == cnt) {
928 if ( ! mdoc_elem_alloc(m, line, ppos, tok, arg))
929 return(0);
930 if ( ! rew_last(m, m->last))
931 return(0);
932 } else if ( ! nc && 0 == cnt) {
933 mdoc_argv_free(arg);
934 if ( ! mdoc_pmsg(m, line, ppos, MANDOCERR_MACROEMPTY))
935 return(0);
936 }
937
938 if ( ! nl)
939 return(1);
940 return(append_delims(m, line, pos, buf));
941 }
942
943
944 static int
945 blk_full(MACRO_PROT_ARGS)
946 {
947 int la, nl;
948 struct mdoc_arg *arg;
949 struct mdoc_node *head; /* save of head macro */
950 struct mdoc_node *body; /* save of body macro */
951 struct mdoc_node *n;
952 enum mdoc_type mtt;
953 enum mdoct ntok;
954 enum margserr ac, lac;
955 enum margverr av;
956 char *p;
957
958 nl = MDOC_NEWLINE & m->flags;
959
960 /* Close out prior implicit scope. */
961
962 if ( ! (MDOC_EXPLICIT & mdoc_macros[tok].flags)) {
963 if ( ! rew_sub(MDOC_BODY, m, tok, line, ppos))
964 return(0);
965 if ( ! rew_sub(MDOC_BLOCK, m, tok, line, ppos))
966 return(0);
967 }
968
969 /*
970 * This routine accomodates implicitly- and explicitly-scoped
971 * macro openings. Implicit ones first close out prior scope
972 * (seen above). Delay opening the head until necessary to
973 * allow leading punctuation to print. Special consideration
974 * for `It -column', which has phrase-part syntax instead of
975 * regular child nodes.
976 */
977
978 for (arg = NULL;; ) {
979 la = *pos;
980 av = mdoc_argv(m, line, tok, &arg, pos, buf);
981
982 if (ARGV_WORD == av) {
983 *pos = la;
984 break;
985 }
986
987 if (ARGV_EOLN == av)
988 break;
989 if (ARGV_ARG == av)
990 continue;
991
992 mdoc_argv_free(arg);
993 return(0);
994 }
995
996 if ( ! mdoc_block_alloc(m, line, ppos, tok, arg))
997 return(0);
998
999 head = body = NULL;
1000
1001 /*
1002 * The `Nd' macro has all arguments in its body: it's a hybrid
1003 * of block partial-explicit and full-implicit. Stupid.
1004 */
1005
1006 if (MDOC_Nd == tok) {
1007 if ( ! mdoc_head_alloc(m, line, ppos, tok))
1008 return(0);
1009 head = m->last;
1010 if ( ! rew_sub(MDOC_HEAD, m, tok, line, ppos))
1011 return(0);
1012 if ( ! mdoc_body_alloc(m, line, ppos, tok))
1013 return(0);
1014 body = m->last;
1015 }
1016
1017 ac = ARGS_ERROR;
1018
1019 for ( ; ; ) {
1020 la = *pos;
1021 /* Initialise last-phrase-type with ARGS_PEND. */
1022 lac = ARGS_ERROR == ac ? ARGS_PEND : ac;
1023 ac = mdoc_args(m, line, pos, buf, tok, &p);
1024
1025 if (ARGS_PUNCT == ac)
1026 break;
1027
1028 if (ARGS_ERROR == ac)
1029 return(0);
1030
1031 if (ARGS_EOLN == ac) {
1032 if (ARGS_PPHRASE != lac && ARGS_PHRASE != lac)
1033 break;
1034 /*
1035 * This is necessary: if the last token on a
1036 * line is a `Ta' or tab, then we'll get
1037 * ARGS_EOLN, so we must be smart enough to
1038 * reopen our scope if the last parse was a
1039 * phrase or partial phrase.
1040 */
1041 if ( ! rew_sub(MDOC_BODY, m, tok, line, ppos))
1042 return(0);
1043 if ( ! mdoc_body_alloc(m, line, ppos, tok))
1044 return(0);
1045 body = m->last;
1046 break;
1047 }
1048
1049 /*
1050 * Emit leading punctuation (i.e., punctuation before
1051 * the MDOC_HEAD) for non-phrase types.
1052 */
1053
1054 if (NULL == head &&
1055 ARGS_PEND != ac &&
1056 ARGS_PHRASE != ac &&
1057 ARGS_PPHRASE != ac &&
1058 ARGS_QWORD != ac &&
1059 DELIM_OPEN == mdoc_isdelim(p)) {
1060 if ( ! mdoc_word_alloc(m, line, la, p))
1061 return(0);
1062 continue;
1063 }
1064
1065 /* Open a head if one hasn't been opened. */
1066
1067 if (NULL == head) {
1068 if ( ! mdoc_head_alloc(m, line, ppos, tok))
1069 return(0);
1070 head = m->last;
1071 }
1072
1073 if (ARGS_PHRASE == ac ||
1074 ARGS_PEND == ac ||
1075 ARGS_PPHRASE == ac) {
1076 /*
1077 * If we haven't opened a body yet, rewind the
1078 * head; if we have, rewind that instead.
1079 */
1080
1081 mtt = body ? MDOC_BODY : MDOC_HEAD;
1082 if ( ! rew_sub(mtt, m, tok, line, ppos))
1083 return(0);
1084
1085 /* Then allocate our body context. */
1086
1087 if ( ! mdoc_body_alloc(m, line, ppos, tok))
1088 return(0);
1089 body = m->last;
1090
1091 /*
1092 * Process phrases: set whether we're in a
1093 * partial-phrase (this effects line handling)
1094 * then call down into the phrase parser.
1095 */
1096
1097 if (ARGS_PPHRASE == ac)
1098 m->flags |= MDOC_PPHRASE;
1099 if (ARGS_PEND == ac && ARGS_PPHRASE == lac)
1100 m->flags |= MDOC_PPHRASE;
1101
1102 if ( ! phrase(m, line, la, buf))
1103 return(0);
1104
1105 m->flags &= ~MDOC_PPHRASE;
1106 continue;
1107 }
1108
1109 ntok = ARGS_QWORD == ac ? MDOC_MAX : lookup(tok, p);
1110
1111 if (MDOC_MAX == ntok) {
1112 if ( ! mdoc_word_alloc(m, line, la, p))
1113 return(0);
1114 continue;
1115 }
1116
1117 if ( ! mdoc_macro(m, ntok, line, la, pos, buf))
1118 return(0);
1119 break;
1120 }
1121
1122 if (NULL == head) {
1123 if ( ! mdoc_head_alloc(m, line, ppos, tok))
1124 return(0);
1125 head = m->last;
1126 }
1127
1128 if (nl && ! append_delims(m, line, pos, buf))
1129 return(0);
1130
1131 /* If we've already opened our body, exit now. */
1132
1133 if (NULL != body)
1134 goto out;
1135
1136 /*
1137 * If there is an open (i.e., unvalidated) sub-block requiring
1138 * explicit close-out, postpone switching the current block from
1139 * head to body until the rew_sub() call closing out that
1140 * sub-block.
1141 */
1142 for (n = m->last; n && n != head; n = n->parent) {
1143 if (MDOC_BLOCK == n->type &&
1144 MDOC_EXPLICIT & mdoc_macros[n->tok].flags &&
1145 ! (MDOC_VALID & n->flags)) {
1146 assert( ! (MDOC_ACTED & n->flags));
1147 n->pending = head;
1148 return(1);
1149 }
1150 }
1151
1152 /* Close out scopes to remain in a consistent state. */
1153
1154 if ( ! rew_sub(MDOC_HEAD, m, tok, line, ppos))
1155 return(0);
1156 if ( ! mdoc_body_alloc(m, line, ppos, tok))
1157 return(0);
1158
1159 out:
1160 if ( ! (MDOC_FREECOL & m->flags))
1161 return(1);
1162
1163 if ( ! rew_sub(MDOC_BODY, m, tok, line, ppos))
1164 return(0);
1165 if ( ! rew_sub(MDOC_BLOCK, m, tok, line, ppos))
1166 return(0);
1167
1168 m->flags &= ~MDOC_FREECOL;
1169 return(1);
1170 }
1171
1172
1173 static int
1174 blk_part_imp(MACRO_PROT_ARGS)
1175 {
1176 int la, nl;
1177 enum mdoct ntok;
1178 enum margserr ac;
1179 char *p;
1180 struct mdoc_node *blk; /* saved block context */
1181 struct mdoc_node *body; /* saved body context */
1182 struct mdoc_node *n;
1183
1184 nl = MDOC_NEWLINE & m->flags;
1185
1186 /*
1187 * A macro that spans to the end of the line. This is generally
1188 * (but not necessarily) called as the first macro. The block
1189 * has a head as the immediate child, which is always empty,
1190 * followed by zero or more opening punctuation nodes, then the
1191 * body (which may be empty, depending on the macro), then zero
1192 * or more closing punctuation nodes.
1193 */
1194
1195 if ( ! mdoc_block_alloc(m, line, ppos, tok, NULL))
1196 return(0);
1197
1198 blk = m->last;
1199
1200 if ( ! mdoc_head_alloc(m, line, ppos, tok))
1201 return(0);
1202 if ( ! rew_sub(MDOC_HEAD, m, tok, line, ppos))
1203 return(0);
1204
1205 /*
1206 * Open the body scope "on-demand", that is, after we've
1207 * processed all our the leading delimiters (open parenthesis,
1208 * etc.).
1209 */
1210
1211 for (body = NULL; ; ) {
1212 la = *pos;
1213 ac = mdoc_args(m, line, pos, buf, tok, &p);
1214
1215 if (ARGS_ERROR == ac)
1216 return(0);
1217 if (ARGS_EOLN == ac)
1218 break;
1219 if (ARGS_PUNCT == ac)
1220 break;
1221
1222 if (NULL == body && ARGS_QWORD != ac &&
1223 DELIM_OPEN == mdoc_isdelim(p)) {
1224 if ( ! mdoc_word_alloc(m, line, la, p))
1225 return(0);
1226 continue;
1227 }
1228
1229 if (NULL == body) {
1230 if ( ! mdoc_body_alloc(m, line, ppos, tok))
1231 return(0);
1232 body = m->last;
1233 }
1234
1235 ntok = ARGS_QWORD == ac ? MDOC_MAX : lookup(tok, p);
1236
1237 if (MDOC_MAX == ntok) {
1238 if ( ! mdoc_word_alloc(m, line, la, p))
1239 return(0);
1240 continue;
1241 }
1242
1243 if ( ! mdoc_macro(m, ntok, line, la, pos, buf))
1244 return(0);
1245 break;
1246 }
1247
1248 /* Clean-ups to leave in a consistent state. */
1249
1250 if (NULL == body) {
1251 if ( ! mdoc_body_alloc(m, line, ppos, tok))
1252 return(0);
1253 body = m->last;
1254 }
1255
1256 for (n = body->child; n && n->next; n = n->next)
1257 /* Do nothing. */ ;
1258
1259 /*
1260 * End of sentence spacing: if the last node is a text node and
1261 * has a trailing period, then mark it as being end-of-sentence.
1262 */
1263
1264 if (n && MDOC_TEXT == n->type && n->string)
1265 if (mandoc_eos(n->string, strlen(n->string), 1))
1266 n->flags |= MDOC_EOS;
1267
1268 /* Up-propogate the end-of-space flag. */
1269
1270 if (n && (MDOC_EOS & n->flags)) {
1271 body->flags |= MDOC_EOS;
1272 body->parent->flags |= MDOC_EOS;
1273 }
1274
1275 /*
1276 * If there is an open sub-block requiring explicit close-out,
1277 * postpone closing out the current block
1278 * until the rew_sub() call closing out the sub-block.
1279 */
1280 for (n = m->last; n && n != body && n != blk->parent; n = n->parent) {
1281 if (MDOC_BLOCK == n->type &&
1282 MDOC_EXPLICIT & mdoc_macros[n->tok].flags &&
1283 ! (MDOC_VALID & n->flags)) {
1284 assert( ! (MDOC_ACTED & n->flags));
1285 if ( ! make_pending(n, tok, m, line, ppos))
1286 return(0);
1287 if ( ! mdoc_endbody_alloc(m, line, ppos,
1288 tok, body, ENDBODY_NOSPACE))
1289 return(0);
1290 return(1);
1291 }
1292 }
1293
1294 /*
1295 * If we can't rewind to our body, then our scope has already
1296 * been closed by another macro (like `Oc' closing `Op'). This
1297 * is ugly behaviour nodding its head to OpenBSD's overwhelming
1298 * crufty use of `Op' breakage.
1299 */
1300 if (n != body && ! mdoc_vmsg(m, MANDOCERR_SCOPENEST,
1301 line, ppos, "%s broken", mdoc_macronames[tok]))
1302 return(0);
1303
1304 if (n && ! rew_sub(MDOC_BODY, m, tok, line, ppos))
1305 return(0);
1306
1307 /* Standard appending of delimiters. */
1308
1309 if (nl && ! append_delims(m, line, pos, buf))
1310 return(0);
1311
1312 /* Rewind scope, if applicable. */
1313
1314 if (n && ! rew_sub(MDOC_BLOCK, m, tok, line, ppos))
1315 return(0);
1316
1317 return(1);
1318 }
1319
1320
1321 static int
1322 blk_part_exp(MACRO_PROT_ARGS)
1323 {
1324 int la, nl;
1325 enum margserr ac;
1326 struct mdoc_node *head; /* keep track of head */
1327 struct mdoc_node *body; /* keep track of body */
1328 char *p;
1329 enum mdoct ntok;
1330
1331 nl = MDOC_NEWLINE & m->flags;
1332
1333 /*
1334 * The opening of an explicit macro having zero or more leading
1335 * punctuation nodes; a head with optional single element (the
1336 * case of `Eo'); and a body that may be empty.
1337 */
1338
1339 if ( ! mdoc_block_alloc(m, line, ppos, tok, NULL))
1340 return(0);
1341
1342 for (head = body = NULL; ; ) {
1343 la = *pos;
1344 ac = mdoc_args(m, line, pos, buf, tok, &p);
1345
1346 if (ARGS_ERROR == ac)
1347 return(0);
1348 if (ARGS_PUNCT == ac)
1349 break;
1350 if (ARGS_EOLN == ac)
1351 break;
1352
1353 /* Flush out leading punctuation. */
1354
1355 if (NULL == head && ARGS_QWORD != ac &&
1356 DELIM_OPEN == mdoc_isdelim(p)) {
1357 assert(NULL == body);
1358 if ( ! mdoc_word_alloc(m, line, la, p))
1359 return(0);
1360 continue;
1361 }
1362
1363 if (NULL == head) {
1364 assert(NULL == body);
1365 if ( ! mdoc_head_alloc(m, line, ppos, tok))
1366 return(0);
1367 head = m->last;
1368 }
1369
1370 /*
1371 * `Eo' gobbles any data into the head, but most other
1372 * macros just immediately close out and begin the body.
1373 */
1374
1375 if (NULL == body) {
1376 assert(head);
1377 /* No check whether it's a macro! */
1378 if (MDOC_Eo == tok)
1379 if ( ! mdoc_word_alloc(m, line, la, p))
1380 return(0);
1381
1382 if ( ! rew_sub(MDOC_HEAD, m, tok, line, ppos))
1383 return(0);
1384 if ( ! mdoc_body_alloc(m, line, ppos, tok))
1385 return(0);
1386 body = m->last;
1387
1388 if (MDOC_Eo == tok)
1389 continue;
1390 }
1391
1392 assert(NULL != head && NULL != body);
1393
1394 ntok = ARGS_QWORD == ac ? MDOC_MAX : lookup(tok, p);
1395
1396 if (MDOC_MAX == ntok) {
1397 if ( ! mdoc_word_alloc(m, line, la, p))
1398 return(0);
1399 continue;
1400 }
1401
1402 if ( ! mdoc_macro(m, ntok, line, la, pos, buf))
1403 return(0);
1404 break;
1405 }
1406
1407 /* Clean-up to leave in a consistent state. */
1408
1409 if (NULL == head) {
1410 if ( ! mdoc_head_alloc(m, line, ppos, tok))
1411 return(0);
1412 head = m->last;
1413 }
1414
1415 if (NULL == body) {
1416 if ( ! rew_sub(MDOC_HEAD, m, tok, line, ppos))
1417 return(0);
1418 if ( ! mdoc_body_alloc(m, line, ppos, tok))
1419 return(0);
1420 body = m->last;
1421 }
1422
1423 /* Standard appending of delimiters. */
1424
1425 if ( ! nl)
1426 return(1);
1427 return(append_delims(m, line, pos, buf));
1428 }
1429
1430
1431 /* ARGSUSED */
1432 static int
1433 in_line_argn(MACRO_PROT_ARGS)
1434 {
1435 int la, flushed, j, maxargs, nl;
1436 enum margserr ac;
1437 enum margverr av;
1438 struct mdoc_arg *arg;
1439 char *p;
1440 enum mdoct ntok;
1441
1442 nl = MDOC_NEWLINE & m->flags;
1443
1444 /*
1445 * A line macro that has a fixed number of arguments (maxargs).
1446 * Only open the scope once the first non-leading-punctuation is
1447 * found (unless MDOC_IGNDELIM is noted, like in `Pf'), then
1448 * keep it open until the maximum number of arguments are
1449 * exhausted.
1450 */
1451
1452 switch (tok) {
1453 case (MDOC_Ap):
1454 /* FALLTHROUGH */
1455 case (MDOC_No):
1456 /* FALLTHROUGH */
1457 case (MDOC_Ns):
1458 /* FALLTHROUGH */
1459 case (MDOC_Ux):
1460 maxargs = 0;
1461 break;
1462 case (MDOC_Xr):
1463 maxargs = 2;
1464 break;
1465 default:
1466 maxargs = 1;
1467 break;
1468 }
1469
1470 for (arg = NULL; ; ) {
1471 la = *pos;
1472 av = mdoc_argv(m, line, tok, &arg, pos, buf);
1473
1474 if (ARGV_WORD == av) {
1475 *pos = la;
1476 break;
1477 }
1478
1479 if (ARGV_EOLN == av)
1480 break;
1481 if (ARGV_ARG == av)
1482 continue;
1483
1484 mdoc_argv_free(arg);
1485 return(0);
1486 }
1487
1488 for (flushed = j = 0; ; ) {
1489 la = *pos;
1490 ac = mdoc_args(m, line, pos, buf, tok, &p);
1491
1492 if (ARGS_ERROR == ac)
1493 return(0);
1494 if (ARGS_PUNCT == ac)
1495 break;
1496 if (ARGS_EOLN == ac)
1497 break;
1498
1499 if ( ! (MDOC_IGNDELIM & mdoc_macros[tok].flags) &&
1500 ARGS_QWORD != ac &&
1501 0 == j && DELIM_OPEN == mdoc_isdelim(p)) {
1502 if ( ! mdoc_word_alloc(m, line, la, p))
1503 return(0);
1504 continue;
1505 } else if (0 == j)
1506 if ( ! mdoc_elem_alloc(m, line, la, tok, arg))
1507 return(0);
1508
1509 if (j == maxargs && ! flushed) {
1510 if ( ! rew_elem(m, tok))
1511 return(0);
1512 flushed = 1;
1513 }
1514
1515 ntok = ARGS_QWORD == ac ? MDOC_MAX : lookup(tok, p);
1516
1517 if (MDOC_MAX != ntok) {
1518 if ( ! flushed && ! rew_elem(m, tok))
1519 return(0);
1520 flushed = 1;
1521 if ( ! mdoc_macro(m, ntok, line, la, pos, buf))
1522 return(0);
1523 j++;
1524 break;
1525 }
1526
1527 if ( ! (MDOC_IGNDELIM & mdoc_macros[tok].flags) &&
1528 ARGS_QWORD != ac &&
1529 ! flushed &&
1530 DELIM_NONE != mdoc_isdelim(p)) {
1531 if ( ! rew_elem(m, tok))
1532 return(0);
1533 flushed = 1;
1534 }
1535
1536 /*
1537 * XXX: this is a hack to work around groff's ugliness
1538 * as regards `Xr' and extraneous arguments. It should
1539 * ideally be deprecated behaviour, but because this is
1540 * code is no here, it's unlikely to be removed.
1541 */
1542
1543 #ifdef __OpenBSD__
1544 if (MDOC_Xr == tok && j == maxargs) {
1545 if ( ! mdoc_elem_alloc(m, line, la, MDOC_Ns, NULL))
1546 return(0);
1547 if ( ! rew_elem(m, MDOC_Ns))
1548 return(0);
1549 }
1550 #endif
1551
1552 if ( ! mdoc_word_alloc(m, line, la, p))
1553 return(0);
1554 j++;
1555 }
1556
1557 if (0 == j && ! mdoc_elem_alloc(m, line, la, tok, arg))
1558 return(0);
1559
1560 /* Close out in a consistent state. */
1561
1562 if ( ! flushed && ! rew_elem(m, tok))
1563 return(0);
1564 if ( ! nl)
1565 return(1);
1566 return(append_delims(m, line, pos, buf));
1567 }
1568
1569
1570 static int
1571 in_line_eoln(MACRO_PROT_ARGS)
1572 {
1573 int la;
1574 enum margserr ac;
1575 enum margverr av;
1576 struct mdoc_arg *arg;
1577 char *p;
1578 enum mdoct ntok;
1579
1580 assert( ! (MDOC_PARSED & mdoc_macros[tok].flags));
1581
1582 if (tok == MDOC_Pp)
1583 rew_sub(MDOC_BLOCK, m, MDOC_Nm, line, ppos);
1584
1585 /* Parse macro arguments. */
1586
1587 for (arg = NULL; ; ) {
1588 la = *pos;
1589 av = mdoc_argv(m, line, tok, &arg, pos, buf);
1590
1591 if (ARGV_WORD == av) {
1592 *pos = la;
1593 break;
1594 }
1595 if (ARGV_EOLN == av)
1596 break;
1597 if (ARGV_ARG == av)
1598 continue;
1599
1600 mdoc_argv_free(arg);
1601 return(0);
1602 }
1603
1604 /* Open element scope. */
1605
1606 if ( ! mdoc_elem_alloc(m, line, ppos, tok, arg))
1607 return(0);
1608
1609 /* Parse argument terms. */
1610
1611 for (;;) {
1612 la = *pos;
1613 ac = mdoc_args(m, line, pos, buf, tok, &p);
1614
1615 if (ARGS_ERROR == ac)
1616 return(0);
1617 if (ARGS_EOLN == ac)
1618 break;
1619
1620 ntok = ARGS_QWORD == ac ? MDOC_MAX : lookup(tok, p);
1621
1622 if (MDOC_MAX == ntok) {
1623 if ( ! mdoc_word_alloc(m, line, la, p))
1624 return(0);
1625 continue;
1626 }
1627
1628 if ( ! rew_elem(m, tok))
1629 return(0);
1630 return(mdoc_macro(m, ntok, line, la, pos, buf));
1631 }
1632
1633 /* Close out (no delimiters). */
1634
1635 return(rew_elem(m, tok));
1636 }
1637
1638
1639 /* ARGSUSED */
1640 static int
1641 ctx_synopsis(MACRO_PROT_ARGS)
1642 {
1643 int nl;
1644
1645 nl = MDOC_NEWLINE & m->flags;
1646
1647 /* If we're not in the SYNOPSIS, go straight to in-line. */
1648 if ( ! (MDOC_SYNOPSIS & m->flags))
1649 return(in_line(m, tok, line, ppos, pos, buf));
1650
1651 /* If we're a nested call, same place. */
1652 if ( ! nl)
1653 return(in_line(m, tok, line, ppos, pos, buf));
1654
1655 /*
1656 * XXX: this will open a block scope; however, if later we end
1657 * up formatting the block scope, then child nodes will inherit
1658 * the formatting. Be careful.
1659 */
1660 if (MDOC_Nm == tok)
1661 return(blk_full(m, tok, line, ppos, pos, buf));
1662 assert(MDOC_Vt == tok);
1663 return(blk_part_imp(m, tok, line, ppos, pos, buf));
1664 }
1665
1666
1667 /* ARGSUSED */
1668 static int
1669 obsolete(MACRO_PROT_ARGS)
1670 {
1671
1672 return(mdoc_pmsg(m, line, ppos, MANDOCERR_MACROOBS));
1673 }
1674
1675
1676 /*
1677 * Phrases occur within `Bl -column' entries, separated by `Ta' or tabs.
1678 * They're unusual because they're basically free-form text until a
1679 * macro is encountered.
1680 */
1681 static int
1682 phrase(struct mdoc *m, int line, int ppos, char *buf)
1683 {
1684 int la, pos;
1685 enum margserr ac;
1686 enum mdoct ntok;
1687 char *p;
1688
1689 for (pos = ppos; ; ) {
1690 la = pos;
1691
1692 ac = mdoc_zargs(m, line, &pos, buf, 0, &p);
1693
1694 if (ARGS_ERROR == ac)
1695 return(0);
1696 if (ARGS_EOLN == ac)
1697 break;
1698
1699 ntok = ARGS_QWORD == ac ? MDOC_MAX : lookup_raw(p);
1700
1701 if (MDOC_MAX == ntok) {
1702 if ( ! mdoc_word_alloc(m, line, la, p))
1703 return(0);
1704 continue;
1705 }
1706
1707 if ( ! mdoc_macro(m, ntok, line, la, &pos, buf))
1708 return(0);
1709 return(append_delims(m, line, &pos, buf));
1710 }
1711
1712 return(1);
1713 }
1714
1715
1716 /* ARGSUSED */
1717 static int
1718 phrase_ta(MACRO_PROT_ARGS)
1719 {
1720 int la;
1721 enum mdoct ntok;
1722 enum margserr ac;
1723 char *p;
1724
1725 /*
1726 * FIXME: this is overly restrictive: if the `Ta' is unexpected,
1727 * it should simply error out with ARGSLOST.
1728 */
1729
1730 if ( ! rew_sub(MDOC_BODY, m, MDOC_It, line, ppos))
1731 return(0);
1732 if ( ! mdoc_body_alloc(m, line, ppos, MDOC_It))
1733 return(0);
1734
1735 for (;;) {
1736 la = *pos;
1737 ac = mdoc_zargs(m, line, pos, buf, 0, &p);
1738
1739 if (ARGS_ERROR == ac)
1740 return(0);
1741 if (ARGS_EOLN == ac)
1742 break;
1743
1744 ntok = ARGS_QWORD == ac ? MDOC_MAX : lookup_raw(p);
1745
1746 if (MDOC_MAX == ntok) {
1747 if ( ! mdoc_word_alloc(m, line, la, p))
1748 return(0);
1749 continue;
1750 }
1751
1752 if ( ! mdoc_macro(m, ntok, line, la, pos, buf))
1753 return(0);
1754 return(append_delims(m, line, pos, buf));
1755 }
1756
1757 return(1);
1758 }
A git conversion mirror of mandoc.bsd.lv