]> 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
Merge schwarze@'s relaxation of scope-breaking rules: allow implicit
[mandoc.git] / mdoc_macro.c
1 /* $Id: mdoc_macro.c,v 1.98 2010/12/06 11:01:19 kristaps 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_FORCE,
38 REWIND_LATER,
39 REWIND_ERROR
40 };
41
42 static int blk_full(MACRO_PROT_ARGS);
43 static int blk_exp_close(MACRO_PROT_ARGS);
44 static int blk_part_exp(MACRO_PROT_ARGS);
45 static int blk_part_imp(MACRO_PROT_ARGS);
46 static int ctx_synopsis(MACRO_PROT_ARGS);
47 static int in_line_eoln(MACRO_PROT_ARGS);
48 static int in_line_argn(MACRO_PROT_ARGS);
49 static int in_line(MACRO_PROT_ARGS);
50 static int obsolete(MACRO_PROT_ARGS);
51 static int phrase_ta(MACRO_PROT_ARGS);
52
53 static int append_delims(struct mdoc *,
54 int, int *, char *);
55 static enum mdoct lookup(enum mdoct, const char *);
56 static enum mdoct lookup_raw(const char *);
57 static int make_pending(struct mdoc_node *, enum mdoct,
58 struct mdoc *, int, int);
59 static int phrase(struct mdoc *, int, int, char *);
60 static enum mdoct rew_alt(enum mdoct);
61 static enum rew rew_dohalt(enum mdoct, enum mdoc_type,
62 const struct mdoc_node *);
63 static int rew_elem(struct mdoc *, enum mdoct);
64 static int rew_last(struct mdoc *,
65 const struct mdoc_node *);
66 static int rew_sub(enum mdoc_type, struct mdoc *,
67 enum mdoct, int, int);
68
69 const struct mdoc_macro __mdoc_macros[MDOC_MAX] = {
70 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* Ap */
71 { in_line_eoln, MDOC_PROLOGUE }, /* Dd */
72 { in_line_eoln, MDOC_PROLOGUE }, /* Dt */
73 { in_line_eoln, MDOC_PROLOGUE }, /* Os */
74 { blk_full, 0 }, /* Sh */
75 { blk_full, 0 }, /* Ss */
76 { in_line_eoln, 0 }, /* Pp */
77 { blk_part_imp, MDOC_PARSED }, /* D1 */
78 { blk_part_imp, MDOC_PARSED }, /* Dl */
79 { blk_full, MDOC_EXPLICIT }, /* Bd */
80 { blk_exp_close, MDOC_EXPLICIT }, /* Ed */
81 { blk_full, MDOC_EXPLICIT }, /* Bl */
82 { blk_exp_close, MDOC_EXPLICIT }, /* El */
83 { blk_full, MDOC_PARSED }, /* It */
84 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Ad */
85 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* An */
86 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Ar */
87 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Cd */
88 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Cm */
89 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Dv */
90 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Er */
91 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Ev */
92 { in_line_eoln, 0 }, /* Ex */
93 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Fa */
94 { in_line_eoln, 0 }, /* Fd */
95 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Fl */
96 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Fn */
97 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Ft */
98 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Ic */
99 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* In */
100 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Li */
101 { blk_full, 0 }, /* Nd */
102 { ctx_synopsis, MDOC_CALLABLE | MDOC_PARSED }, /* Nm */
103 { blk_part_imp, MDOC_CALLABLE | MDOC_PARSED }, /* Op */
104 { obsolete, 0 }, /* Ot */
105 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Pa */
106 { in_line_eoln, 0 }, /* Rv */
107 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* St */
108 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Va */
109 { ctx_synopsis, MDOC_CALLABLE | MDOC_PARSED }, /* Vt */
110 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* Xr */
111 { in_line_eoln, 0 }, /* %A */
112 { in_line_eoln, 0 }, /* %B */
113 { in_line_eoln, 0 }, /* %D */
114 { in_line_eoln, 0 }, /* %I */
115 { in_line_eoln, 0 }, /* %J */
116 { in_line_eoln, 0 }, /* %N */
117 { in_line_eoln, 0 }, /* %O */
118 { in_line_eoln, 0 }, /* %P */
119 { in_line_eoln, 0 }, /* %R */
120 { in_line_eoln, 0 }, /* %T */
121 { in_line_eoln, 0 }, /* %V */
122 { blk_exp_close, MDOC_EXPLICIT | MDOC_CALLABLE | MDOC_PARSED }, /* Ac */
123 { blk_part_exp, MDOC_CALLABLE | MDOC_PARSED | MDOC_EXPLICIT }, /* Ao */
124 { blk_part_imp, MDOC_CALLABLE | MDOC_PARSED }, /* Aq */
125 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* At */
126 { blk_exp_close, MDOC_EXPLICIT | MDOC_CALLABLE | MDOC_PARSED }, /* Bc */
127 { blk_full, MDOC_EXPLICIT }, /* Bf */
128 { blk_part_exp, MDOC_CALLABLE | MDOC_PARSED | MDOC_EXPLICIT }, /* Bo */
129 { blk_part_imp, MDOC_CALLABLE | MDOC_PARSED }, /* Bq */
130 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* Bsx */
131 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* Bx */
132 { in_line_eoln, 0 }, /* Db */
133 { blk_exp_close, MDOC_EXPLICIT | MDOC_CALLABLE | MDOC_PARSED }, /* Dc */
134 { blk_part_exp, MDOC_CALLABLE | MDOC_PARSED | MDOC_EXPLICIT }, /* Do */
135 { blk_part_imp, MDOC_CALLABLE | MDOC_PARSED }, /* Dq */
136 { blk_exp_close, MDOC_EXPLICIT | MDOC_CALLABLE | MDOC_PARSED }, /* Ec */
137 { blk_exp_close, MDOC_EXPLICIT }, /* Ef */
138 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Em */
139 { blk_part_exp, MDOC_CALLABLE | MDOC_PARSED | MDOC_EXPLICIT }, /* Eo */
140 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* Fx */
141 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Ms */
142 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED | MDOC_IGNDELIM }, /* No */
143 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED | MDOC_IGNDELIM }, /* Ns */
144 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* Nx */
145 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* Ox */
146 { blk_exp_close, MDOC_EXPLICIT | MDOC_CALLABLE | MDOC_PARSED }, /* Pc */
147 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED | MDOC_IGNDELIM }, /* Pf */
148 { blk_part_exp, MDOC_CALLABLE | MDOC_PARSED | MDOC_EXPLICIT }, /* Po */
149 { blk_part_imp, MDOC_CALLABLE | MDOC_PARSED }, /* Pq */
150 { blk_exp_close, MDOC_EXPLICIT | MDOC_CALLABLE | MDOC_PARSED }, /* Qc */
151 { blk_part_imp, MDOC_CALLABLE | MDOC_PARSED }, /* Ql */
152 { blk_part_exp, MDOC_CALLABLE | MDOC_PARSED | MDOC_EXPLICIT }, /* Qo */
153 { blk_part_imp, MDOC_CALLABLE | MDOC_PARSED }, /* Qq */
154 { blk_exp_close, MDOC_EXPLICIT }, /* Re */
155 { blk_full, MDOC_EXPLICIT }, /* Rs */
156 { blk_exp_close, MDOC_EXPLICIT | MDOC_CALLABLE | MDOC_PARSED }, /* Sc */
157 { blk_part_exp, MDOC_CALLABLE | MDOC_PARSED | MDOC_EXPLICIT }, /* So */
158 { blk_part_imp, MDOC_CALLABLE | MDOC_PARSED }, /* Sq */
159 { in_line_eoln, 0 }, /* Sm */
160 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Sx */
161 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Sy */
162 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Tn */
163 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* Ux */
164 { blk_exp_close, MDOC_EXPLICIT | MDOC_CALLABLE | MDOC_PARSED }, /* Xc */
165 { blk_part_exp, MDOC_CALLABLE | MDOC_PARSED | MDOC_EXPLICIT }, /* Xo */
166 { blk_full, MDOC_EXPLICIT | MDOC_CALLABLE }, /* Fo */
167 { blk_exp_close, MDOC_EXPLICIT | MDOC_CALLABLE | MDOC_PARSED }, /* Fc */
168 { blk_part_exp, MDOC_CALLABLE | MDOC_PARSED | MDOC_EXPLICIT }, /* Oo */
169 { blk_exp_close, MDOC_EXPLICIT | MDOC_CALLABLE | MDOC_PARSED }, /* Oc */
170 { blk_full, MDOC_EXPLICIT }, /* Bk */
171 { blk_exp_close, MDOC_EXPLICIT }, /* Ek */
172 { in_line_eoln, 0 }, /* Bt */
173 { in_line_eoln, 0 }, /* Hf */
174 { obsolete, 0 }, /* Fr */
175 { in_line_eoln, 0 }, /* Ud */
176 { in_line, 0 }, /* Lb */
177 { in_line_eoln, 0 }, /* Lp */
178 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Lk */
179 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Mt */
180 { blk_part_imp, MDOC_CALLABLE | MDOC_PARSED }, /* Brq */
181 { blk_part_exp, MDOC_CALLABLE | MDOC_PARSED | MDOC_EXPLICIT }, /* Bro */
182 { blk_exp_close, MDOC_EXPLICIT | MDOC_CALLABLE | MDOC_PARSED }, /* Brc */
183 { in_line_eoln, 0 }, /* %C */
184 { obsolete, 0 }, /* Es */
185 { obsolete, 0 }, /* En */
186 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* Dx */
187 { in_line_eoln, 0 }, /* %Q */
188 { in_line_eoln, 0 }, /* br */
189 { in_line_eoln, 0 }, /* sp */
190 { in_line_eoln, 0 }, /* %U */
191 { phrase_ta, MDOC_CALLABLE | MDOC_PARSED }, /* Ta */
192 };
193
194 const struct mdoc_macro * const mdoc_macros = __mdoc_macros;
195
196
197 /*
198 * This is called at the end of parsing. It must traverse up the tree,
199 * closing out open [implicit] scopes. Obviously, open explicit scopes
200 * are errors.
201 */
202 int
203 mdoc_macroend(struct mdoc *m)
204 {
205 struct mdoc_node *n;
206
207 /* Scan for open explicit scopes. */
208
209 n = MDOC_VALID & m->last->flags ? m->last->parent : m->last;
210
211 for ( ; n; n = n->parent)
212 if (MDOC_BLOCK == n->type &&
213 MDOC_EXPLICIT & mdoc_macros[n->tok].flags)
214 mdoc_nmsg(m, n, MANDOCERR_SCOPEEXIT);
215
216 /* Rewind to the first. */
217
218 return(rew_last(m, m->first));
219 }
220
221
222 /*
223 * Look up a macro from within a subsequent context.
224 */
225 static enum mdoct
226 lookup(enum mdoct from, const char *p)
227 {
228 /* FIXME: make -diag lists be un-PARSED. */
229
230 if ( ! (MDOC_PARSED & mdoc_macros[from].flags))
231 return(MDOC_MAX);
232 return(lookup_raw(p));
233 }
234
235
236 /*
237 * Lookup a macro following the initial line macro.
238 */
239 static enum mdoct
240 lookup_raw(const char *p)
241 {
242 enum mdoct res;
243
244 if (MDOC_MAX == (res = mdoc_hash_find(p)))
245 return(MDOC_MAX);
246 if (MDOC_CALLABLE & mdoc_macros[res].flags)
247 return(res);
248 return(MDOC_MAX);
249 }
250
251
252 static int
253 rew_last(struct mdoc *mdoc, const struct mdoc_node *to)
254 {
255
256 assert(to);
257 mdoc->next = MDOC_NEXT_SIBLING;
258
259 /* LINTED */
260 while (mdoc->last != to) {
261 if ( ! mdoc_valid_post(mdoc))
262 return(0);
263 mdoc->last = mdoc->last->parent;
264 assert(mdoc->last);
265 }
266
267 return(mdoc_valid_post(mdoc));
268 }
269
270
271 /*
272 * For a block closing macro, return the corresponding opening one.
273 * Otherwise, return the macro itself.
274 */
275 static enum mdoct
276 rew_alt(enum mdoct tok)
277 {
278 switch (tok) {
279 case (MDOC_Ac):
280 return(MDOC_Ao);
281 case (MDOC_Bc):
282 return(MDOC_Bo);
283 case (MDOC_Brc):
284 return(MDOC_Bro);
285 case (MDOC_Dc):
286 return(MDOC_Do);
287 case (MDOC_Ec):
288 return(MDOC_Eo);
289 case (MDOC_Ed):
290 return(MDOC_Bd);
291 case (MDOC_Ef):
292 return(MDOC_Bf);
293 case (MDOC_Ek):
294 return(MDOC_Bk);
295 case (MDOC_El):
296 return(MDOC_Bl);
297 case (MDOC_Fc):
298 return(MDOC_Fo);
299 case (MDOC_Oc):
300 return(MDOC_Oo);
301 case (MDOC_Pc):
302 return(MDOC_Po);
303 case (MDOC_Qc):
304 return(MDOC_Qo);
305 case (MDOC_Re):
306 return(MDOC_Rs);
307 case (MDOC_Sc):
308 return(MDOC_So);
309 case (MDOC_Xc):
310 return(MDOC_Xo);
311 default:
312 return(tok);
313 }
314 /* NOTREACHED */
315 }
316
317
318 /*
319 * Rewinding to tok, how do we have to handle *p?
320 * REWIND_NONE: *p would delimit tok, but no tok scope is open
321 * inside *p, so there is no need to rewind anything at all.
322 * REWIND_THIS: *p matches tok, so rewind *p and nothing else.
323 * REWIND_MORE: *p is implicit, rewind it and keep searching for tok.
324 * REWIND_FORCE: *p is explicit, but tok is full, force rewinding *p.
325 * REWIND_LATER: *p is explicit and still open, postpone rewinding.
326 * REWIND_ERROR: No tok block is open at all.
327 */
328 static enum rew
329 rew_dohalt(enum mdoct tok, enum mdoc_type type,
330 const struct mdoc_node *p)
331 {
332
333 /*
334 * No matching token, no delimiting block, no broken block.
335 * This can happen when full implicit macros are called for
336 * the first time but try to rewind their previous
337 * instance anyway.
338 */
339 if (MDOC_ROOT == p->type)
340 return(MDOC_BLOCK == type &&
341 MDOC_EXPLICIT & mdoc_macros[tok].flags ?
342 REWIND_ERROR : REWIND_NONE);
343
344 /*
345 * When starting to rewind, skip plain text
346 * and nodes that have already been rewound.
347 */
348 if (MDOC_TEXT == p->type || MDOC_VALID & p->flags)
349 return(REWIND_MORE);
350
351 /*
352 * The easiest case: Found a matching token.
353 * This applies to both blocks and elements.
354 */
355 tok = rew_alt(tok);
356 if (tok == p->tok)
357 return(p->end ? REWIND_NONE :
358 type == p->type ? REWIND_THIS : REWIND_MORE);
359
360 /*
361 * While elements do require rewinding for themselves,
362 * they never affect rewinding of other nodes.
363 */
364 if (MDOC_ELEM == p->type)
365 return(REWIND_MORE);
366
367 /*
368 * Blocks delimited by our target token get REWIND_MORE.
369 * Blocks delimiting our target token get REWIND_NONE.
370 */
371 switch (tok) {
372 case (MDOC_Bl):
373 if (MDOC_It == p->tok)
374 return(REWIND_MORE);
375 break;
376 case (MDOC_It):
377 if (MDOC_BODY == p->type && MDOC_Bl == p->tok)
378 return(REWIND_NONE);
379 break;
380 /*
381 * XXX Badly nested block handling still fails badly
382 * when one block is breaking two blocks of the same type.
383 * This is an incomplete and extremely ugly workaround,
384 * required to let the OpenBSD tree build.
385 */
386 case (MDOC_Oo):
387 if (MDOC_Op == p->tok)
388 return(REWIND_MORE);
389 break;
390 case (MDOC_Nm):
391 return(REWIND_NONE);
392 case (MDOC_Nd):
393 /* FALLTHROUGH */
394 case (MDOC_Ss):
395 if (MDOC_BODY == p->type && MDOC_Sh == p->tok)
396 return(REWIND_NONE);
397 /* FALLTHROUGH */
398 case (MDOC_Sh):
399 if (MDOC_Nd == p->tok || MDOC_Ss == p->tok ||
400 MDOC_Sh == p->tok)
401 return(REWIND_MORE);
402 break;
403 default:
404 break;
405 }
406
407 /*
408 * Default block rewinding rules.
409 * In particular, always skip block end markers,
410 * and let all blocks rewind Nm children.
411 */
412 if (ENDBODY_NOT != p->end || MDOC_Nm == p->tok ||
413 (MDOC_BLOCK == p->type &&
414 ! (MDOC_EXPLICIT & mdoc_macros[tok].flags)))
415 return(REWIND_MORE);
416
417 /*
418 * By default, closing out full blocks
419 * forces closing of broken explicit blocks,
420 * while closing out partial blocks
421 * allows delayed rewinding by default.
422 */
423 return (&blk_full == mdoc_macros[tok].fp ?
424 REWIND_FORCE : REWIND_LATER);
425 }
426
427
428 static int
429 rew_elem(struct mdoc *mdoc, enum mdoct tok)
430 {
431 struct mdoc_node *n;
432
433 n = mdoc->last;
434 if (MDOC_ELEM != n->type)
435 n = n->parent;
436 assert(MDOC_ELEM == n->type);
437 assert(tok == n->tok);
438
439 return(rew_last(mdoc, n));
440 }
441
442
443 /*
444 * We are trying to close a block identified by tok,
445 * but the child block *broken is still open.
446 * Thus, postpone closing the tok block
447 * until the rew_sub call closing *broken.
448 */
449 static int
450 make_pending(struct mdoc_node *broken, enum mdoct tok,
451 struct mdoc *m, int line, int ppos)
452 {
453 struct mdoc_node *breaker;
454
455 /*
456 * Iterate backwards, searching for the block matching tok,
457 * that is, the block breaking the *broken block.
458 */
459 for (breaker = broken->parent; breaker; breaker = breaker->parent) {
460
461 /*
462 * If the *broken block had already been broken before
463 * and we encounter its breaker, make the tok block
464 * pending on the inner breaker.
465 * Graphically, "[A breaker=[B broken=[C->B B] tok=A] C]"
466 * becomes "[A broken=[B [C->B B] tok=A] C]"
467 * and finally "[A [B->A [C->B B] A] C]".
468 */
469 if (breaker == broken->pending) {
470 broken = breaker;
471 continue;
472 }
473
474 if (REWIND_THIS != rew_dohalt(tok, MDOC_BLOCK, breaker))
475 continue;
476 if (MDOC_BODY == broken->type)
477 broken = broken->parent;
478
479 /*
480 * Found the breaker.
481 * If another, outer breaker is already pending on
482 * the *broken block, we must not clobber the link
483 * to the outer breaker, but make it pending on the
484 * new, now inner breaker.
485 * Graphically, "[A breaker=[B broken=[C->A A] tok=B] C]"
486 * becomes "[A breaker=[B->A broken=[C A] tok=B] C]"
487 * and finally "[A [B->A [C->B A] B] C]".
488 */
489 if (broken->pending) {
490 struct mdoc_node *taker;
491
492 /*
493 * If the breaker had also been broken before,
494 * it cannot take on the outer breaker itself,
495 * but must hand it on to its own breakers.
496 * Graphically, this is the following situation:
497 * "[A [B breaker=[C->B B] broken=[D->A A] tok=C] D]"
498 * "[A taker=[B->A breaker=[C->B B] [D->C A] C] D]"
499 */
500 taker = breaker;
501 while (taker->pending)
502 taker = taker->pending;
503 taker->pending = broken->pending;
504 }
505 broken->pending = breaker;
506 mdoc_vmsg(m, MANDOCERR_SCOPENEST, line, ppos,
507 "%s breaks %s", mdoc_macronames[tok],
508 mdoc_macronames[broken->tok]);
509 return(1);
510 }
511
512 /*
513 * Found no matching block for tok.
514 * Are you trying to close a block that is not open?
515 */
516 return(0);
517 }
518
519
520 static int
521 rew_sub(enum mdoc_type t, struct mdoc *m,
522 enum mdoct tok, int line, int ppos)
523 {
524 struct mdoc_node *n;
525
526 n = m->last;
527 while (n) {
528 switch (rew_dohalt(tok, t, n)) {
529 case (REWIND_NONE):
530 return(1);
531 case (REWIND_THIS):
532 break;
533 case (REWIND_FORCE):
534 mdoc_vmsg(m, MANDOCERR_SCOPEBROKEN, line, ppos,
535 "%s breaks %s", mdoc_macronames[tok],
536 mdoc_macronames[n->tok]);
537 /* FALLTHROUGH */
538 case (REWIND_MORE):
539 n = n->parent;
540 continue;
541 case (REWIND_LATER):
542 if (make_pending(n, tok, m, line, ppos) ||
543 MDOC_BLOCK != t)
544 return(1);
545 /* FALLTHROUGH */
546 case (REWIND_ERROR):
547 mdoc_pmsg(m, line, ppos, MANDOCERR_NOSCOPE);
548 return(1);
549 }
550 break;
551 }
552
553 assert(n);
554 if ( ! rew_last(m, n))
555 return(0);
556
557 /*
558 * The current block extends an enclosing block.
559 * Now that the current block ends, close the enclosing block, too.
560 */
561 while (NULL != (n = n->pending)) {
562 if ( ! rew_last(m, n))
563 return(0);
564 if (MDOC_HEAD == n->type &&
565 ! mdoc_body_alloc(m, n->line, n->pos, n->tok))
566 return(0);
567 }
568
569 return(1);
570 }
571
572
573 static int
574 append_delims(struct mdoc *m, int line, int *pos, char *buf)
575 {
576 int la;
577 enum margserr ac;
578 char *p;
579
580 if ('\0' == buf[*pos])
581 return(1);
582
583 for (;;) {
584 la = *pos;
585 ac = mdoc_zargs(m, line, pos, buf, ARGS_NOWARN, &p);
586
587 if (ARGS_ERROR == ac)
588 return(0);
589 else if (ARGS_EOLN == ac)
590 break;
591
592 assert(DELIM_NONE != mdoc_isdelim(p));
593 if ( ! mdoc_word_alloc(m, line, la, p))
594 return(0);
595
596 /*
597 * If we encounter end-of-sentence symbols, then trigger
598 * the double-space.
599 *
600 * XXX: it's easy to allow this to propogate outward to
601 * the last symbol, such that `. )' will cause the
602 * correct double-spacing. However, (1) groff isn't
603 * smart enough to do this and (2) it would require
604 * knowing which symbols break this behaviour, for
605 * example, `. ;' shouldn't propogate the double-space.
606 */
607 if (mandoc_eos(p, strlen(p), 0))
608 m->last->flags |= MDOC_EOS;
609 }
610
611 return(1);
612 }
613
614
615 /*
616 * Close out block partial/full explicit.
617 */
618 static int
619 blk_exp_close(MACRO_PROT_ARGS)
620 {
621 struct mdoc_node *body; /* Our own body. */
622 struct mdoc_node *later; /* A sub-block starting later. */
623 struct mdoc_node *n; /* For searching backwards. */
624
625 int j, lastarg, maxargs, flushed, nl;
626 enum margserr ac;
627 enum mdoct atok, ntok;
628 char *p;
629
630 nl = MDOC_NEWLINE & m->flags;
631
632 switch (tok) {
633 case (MDOC_Ec):
634 maxargs = 1;
635 break;
636 default:
637 maxargs = 0;
638 break;
639 }
640
641 /*
642 * Search backwards for beginnings of blocks,
643 * both of our own and of pending sub-blocks.
644 */
645 atok = rew_alt(tok);
646 body = later = NULL;
647 for (n = m->last; n; n = n->parent) {
648 if (MDOC_VALID & n->flags)
649 continue;
650
651 /* Remember the start of our own body. */
652 if (MDOC_BODY == n->type && atok == n->tok) {
653 if (ENDBODY_NOT == n->end)
654 body = n;
655 continue;
656 }
657
658 if (MDOC_BLOCK != n->type || MDOC_Nm == n->tok)
659 continue;
660 if (atok == n->tok) {
661 assert(body);
662
663 /*
664 * Found the start of our own block.
665 * When there is no pending sub block,
666 * just proceed to closing out.
667 */
668 if (NULL == later)
669 break;
670
671 /*
672 * When there is a pending sub block,
673 * postpone closing out the current block
674 * until the rew_sub() closing out the sub-block.
675 */
676 make_pending(later, tok, m, line, ppos);
677
678 /*
679 * Mark the place where the formatting - but not
680 * the scope - of the current block ends.
681 */
682 if ( ! mdoc_endbody_alloc(m, line, ppos,
683 atok, body, ENDBODY_SPACE))
684 return(0);
685 break;
686 }
687
688 /*
689 * When finding an open sub block, remember the last
690 * open explicit block, or, in case there are only
691 * implicit ones, the first open implicit block.
692 */
693 if (later &&
694 MDOC_EXPLICIT & mdoc_macros[later->tok].flags)
695 continue;
696 if (MDOC_CALLABLE & mdoc_macros[n->tok].flags)
697 later = n;
698 }
699
700 if ( ! (MDOC_CALLABLE & mdoc_macros[tok].flags)) {
701 /* FIXME: do this in validate */
702 if (buf[*pos])
703 if ( ! mdoc_pmsg(m, line, ppos, MANDOCERR_ARGSLOST))
704 return(0);
705
706 if ( ! rew_sub(MDOC_BODY, m, tok, line, ppos))
707 return(0);
708 return(rew_sub(MDOC_BLOCK, m, tok, line, ppos));
709 }
710
711 if ( ! rew_sub(MDOC_BODY, m, tok, line, ppos))
712 return(0);
713
714 if (NULL == later && maxargs > 0)
715 if ( ! mdoc_tail_alloc(m, line, ppos, rew_alt(tok)))
716 return(0);
717
718 for (flushed = j = 0; ; j++) {
719 lastarg = *pos;
720
721 if (j == maxargs && ! flushed) {
722 if ( ! rew_sub(MDOC_BLOCK, m, tok, line, ppos))
723 return(0);
724 flushed = 1;
725 }
726
727 ac = mdoc_args(m, line, pos, buf, tok, &p);
728
729 if (ARGS_ERROR == ac)
730 return(0);
731 if (ARGS_PUNCT == ac)
732 break;
733 if (ARGS_EOLN == ac)
734 break;
735
736 ntok = ARGS_QWORD == ac ? MDOC_MAX : lookup(tok, p);
737
738 if (MDOC_MAX == ntok) {
739 if ( ! mdoc_word_alloc(m, line, lastarg, p))
740 return(0);
741 continue;
742 }
743
744 if ( ! flushed) {
745 if ( ! rew_sub(MDOC_BLOCK, m, tok, line, ppos))
746 return(0);
747 flushed = 1;
748 }
749 if ( ! mdoc_macro(m, ntok, line, lastarg, pos, buf))
750 return(0);
751 break;
752 }
753
754 if ( ! flushed && ! rew_sub(MDOC_BLOCK, m, tok, line, ppos))
755 return(0);
756
757 if ( ! nl)
758 return(1);
759 return(append_delims(m, line, pos, buf));
760 }
761
762
763 static int
764 in_line(MACRO_PROT_ARGS)
765 {
766 int la, scope, cnt, nc, nl;
767 enum margverr av;
768 enum mdoct ntok;
769 enum margserr ac;
770 enum mdelim d;
771 struct mdoc_arg *arg;
772 char *p;
773
774 nl = MDOC_NEWLINE & m->flags;
775
776 /*
777 * Whether we allow ignored elements (those without content,
778 * usually because of reserved words) to squeak by.
779 */
780
781 switch (tok) {
782 case (MDOC_An):
783 /* FALLTHROUGH */
784 case (MDOC_Ar):
785 /* FALLTHROUGH */
786 case (MDOC_Fl):
787 /* FALLTHROUGH */
788 case (MDOC_Mt):
789 /* FALLTHROUGH */
790 case (MDOC_Nm):
791 /* FALLTHROUGH */
792 case (MDOC_Pa):
793 nc = 1;
794 break;
795 default:
796 nc = 0;
797 break;
798 }
799
800 for (arg = NULL;; ) {
801 la = *pos;
802 av = mdoc_argv(m, line, tok, &arg, pos, buf);
803
804 if (ARGV_WORD == av) {
805 *pos = la;
806 break;
807 }
808 if (ARGV_EOLN == av)
809 break;
810 if (ARGV_ARG == av)
811 continue;
812
813 mdoc_argv_free(arg);
814 return(0);
815 }
816
817 for (cnt = scope = 0;; ) {
818 la = *pos;
819 ac = mdoc_args(m, line, pos, buf, tok, &p);
820
821 if (ARGS_ERROR == ac)
822 return(0);
823 if (ARGS_EOLN == ac)
824 break;
825 if (ARGS_PUNCT == ac)
826 break;
827
828 ntok = ARGS_QWORD == ac ? MDOC_MAX : lookup(tok, p);
829
830 /*
831 * In this case, we've located a submacro and must
832 * execute it. Close out scope, if open. If no
833 * elements have been generated, either create one (nc)
834 * or raise a warning.
835 */
836
837 if (MDOC_MAX != ntok) {
838 if (scope && ! rew_elem(m, tok))
839 return(0);
840 if (nc && 0 == cnt) {
841 if ( ! mdoc_elem_alloc(m, line, ppos, tok, arg))
842 return(0);
843 if ( ! rew_last(m, m->last))
844 return(0);
845 } else if ( ! nc && 0 == cnt) {
846 mdoc_argv_free(arg);
847 if ( ! mdoc_pmsg(m, line, ppos, MANDOCERR_MACROEMPTY))
848 return(0);
849 }
850 if ( ! mdoc_macro(m, ntok, line, la, pos, buf))
851 return(0);
852 if ( ! nl)
853 return(1);
854 return(append_delims(m, line, pos, buf));
855 }
856
857 /*
858 * Non-quote-enclosed punctuation. Set up our scope, if
859 * a word; rewind the scope, if a delimiter; then append
860 * the word.
861 */
862
863 d = ARGS_QWORD == ac ? DELIM_NONE : mdoc_isdelim(p);
864
865 if (DELIM_NONE != d) {
866 /*
867 * If we encounter closing punctuation, no word
868 * has been omitted, no scope is open, and we're
869 * allowed to have an empty element, then start
870 * a new scope. `Ar', `Fl', and `Li', only do
871 * this once per invocation. There may be more
872 * of these (all of them?).
873 */
874 if (0 == cnt && (nc || MDOC_Li == tok) &&
875 DELIM_CLOSE == d && ! scope) {
876 if ( ! mdoc_elem_alloc(m, line, ppos, tok, arg))
877 return(0);
878 if (MDOC_Ar == tok || MDOC_Li == tok ||
879 MDOC_Fl == tok)
880 cnt++;
881 scope = 1;
882 }
883 /*
884 * Close out our scope, if one is open, before
885 * any punctuation.
886 */
887 if (scope && ! rew_elem(m, tok))
888 return(0);
889 scope = 0;
890 } else if ( ! scope) {
891 if ( ! mdoc_elem_alloc(m, line, ppos, tok, arg))
892 return(0);
893 scope = 1;
894 }
895
896 if (DELIM_NONE == d)
897 cnt++;
898 if ( ! mdoc_word_alloc(m, line, la, p))
899 return(0);
900
901 /*
902 * `Fl' macros have their scope re-opened with each new
903 * word so that the `-' can be added to each one without
904 * having to parse out spaces.
905 */
906 if (scope && MDOC_Fl == tok) {
907 if ( ! rew_elem(m, tok))
908 return(0);
909 scope = 0;
910 }
911 }
912
913 if (scope && ! rew_elem(m, tok))
914 return(0);
915
916 /*
917 * If no elements have been collected and we're allowed to have
918 * empties (nc), open a scope and close it out. Otherwise,
919 * raise a warning.
920 */
921
922 if (nc && 0 == cnt) {
923 if ( ! mdoc_elem_alloc(m, line, ppos, tok, arg))
924 return(0);
925 if ( ! rew_last(m, m->last))
926 return(0);
927 } else if ( ! nc && 0 == cnt) {
928 mdoc_argv_free(arg);
929 if ( ! mdoc_pmsg(m, line, ppos, MANDOCERR_MACROEMPTY))
930 return(0);
931 }
932
933 if ( ! nl)
934 return(1);
935 return(append_delims(m, line, pos, buf));
936 }
937
938
939 static int
940 blk_full(MACRO_PROT_ARGS)
941 {
942 int la, nl;
943 struct mdoc_arg *arg;
944 struct mdoc_node *head; /* save of head macro */
945 struct mdoc_node *body; /* save of body macro */
946 struct mdoc_node *n;
947 enum mdoc_type mtt;
948 enum mdoct ntok;
949 enum margserr ac, lac;
950 enum margverr av;
951 char *p;
952
953 nl = MDOC_NEWLINE & m->flags;
954
955 /* Close out prior implicit scope. */
956
957 if ( ! (MDOC_EXPLICIT & mdoc_macros[tok].flags)) {
958 if ( ! rew_sub(MDOC_BODY, m, tok, line, ppos))
959 return(0);
960 if ( ! rew_sub(MDOC_BLOCK, m, tok, line, ppos))
961 return(0);
962 }
963
964 /*
965 * This routine accomodates implicitly- and explicitly-scoped
966 * macro openings. Implicit ones first close out prior scope
967 * (seen above). Delay opening the head until necessary to
968 * allow leading punctuation to print. Special consideration
969 * for `It -column', which has phrase-part syntax instead of
970 * regular child nodes.
971 */
972
973 for (arg = NULL;; ) {
974 la = *pos;
975 av = mdoc_argv(m, line, tok, &arg, pos, buf);
976
977 if (ARGV_WORD == av) {
978 *pos = la;
979 break;
980 }
981
982 if (ARGV_EOLN == av)
983 break;
984 if (ARGV_ARG == av)
985 continue;
986
987 mdoc_argv_free(arg);
988 return(0);
989 }
990
991 if ( ! mdoc_block_alloc(m, line, ppos, tok, arg))
992 return(0);
993
994 head = body = NULL;
995
996 /*
997 * The `Nd' macro has all arguments in its body: it's a hybrid
998 * of block partial-explicit and full-implicit. Stupid.
999 */
1000
1001 if (MDOC_Nd == tok) {
1002 if ( ! mdoc_head_alloc(m, line, ppos, tok))
1003 return(0);
1004 head = m->last;
1005 if ( ! rew_sub(MDOC_HEAD, m, tok, line, ppos))
1006 return(0);
1007 if ( ! mdoc_body_alloc(m, line, ppos, tok))
1008 return(0);
1009 body = m->last;
1010 }
1011
1012 ac = ARGS_ERROR;
1013
1014 for ( ; ; ) {
1015 la = *pos;
1016 /* Initialise last-phrase-type with ARGS_PEND. */
1017 lac = ARGS_ERROR == ac ? ARGS_PEND : ac;
1018 ac = mdoc_args(m, line, pos, buf, tok, &p);
1019
1020 if (ARGS_PUNCT == ac)
1021 break;
1022
1023 if (ARGS_ERROR == ac)
1024 return(0);
1025
1026 if (ARGS_EOLN == ac) {
1027 if (ARGS_PPHRASE != lac && ARGS_PHRASE != lac)
1028 break;
1029 /*
1030 * This is necessary: if the last token on a
1031 * line is a `Ta' or tab, then we'll get
1032 * ARGS_EOLN, so we must be smart enough to
1033 * reopen our scope if the last parse was a
1034 * phrase or partial phrase.
1035 */
1036 if ( ! rew_sub(MDOC_BODY, m, tok, line, ppos))
1037 return(0);
1038 if ( ! mdoc_body_alloc(m, line, ppos, tok))
1039 return(0);
1040 body = m->last;
1041 break;
1042 }
1043
1044 /*
1045 * Emit leading punctuation (i.e., punctuation before
1046 * the MDOC_HEAD) for non-phrase types.
1047 */
1048
1049 if (NULL == head &&
1050 ARGS_PEND != ac &&
1051 ARGS_PHRASE != ac &&
1052 ARGS_PPHRASE != ac &&
1053 ARGS_QWORD != ac &&
1054 DELIM_OPEN == mdoc_isdelim(p)) {
1055 if ( ! mdoc_word_alloc(m, line, la, p))
1056 return(0);
1057 continue;
1058 }
1059
1060 /* Open a head if one hasn't been opened. */
1061
1062 if (NULL == head) {
1063 if ( ! mdoc_head_alloc(m, line, ppos, tok))
1064 return(0);
1065 head = m->last;
1066 }
1067
1068 if (ARGS_PHRASE == ac ||
1069 ARGS_PEND == ac ||
1070 ARGS_PPHRASE == ac) {
1071 /*
1072 * If we haven't opened a body yet, rewind the
1073 * head; if we have, rewind that instead.
1074 */
1075
1076 mtt = body ? MDOC_BODY : MDOC_HEAD;
1077 if ( ! rew_sub(mtt, m, tok, line, ppos))
1078 return(0);
1079
1080 /* Then allocate our body context. */
1081
1082 if ( ! mdoc_body_alloc(m, line, ppos, tok))
1083 return(0);
1084 body = m->last;
1085
1086 /*
1087 * Process phrases: set whether we're in a
1088 * partial-phrase (this effects line handling)
1089 * then call down into the phrase parser.
1090 */
1091
1092 if (ARGS_PPHRASE == ac)
1093 m->flags |= MDOC_PPHRASE;
1094 if (ARGS_PEND == ac && ARGS_PPHRASE == lac)
1095 m->flags |= MDOC_PPHRASE;
1096
1097 if ( ! phrase(m, line, la, buf))
1098 return(0);
1099
1100 m->flags &= ~MDOC_PPHRASE;
1101 continue;
1102 }
1103
1104 ntok = ARGS_QWORD == ac ? MDOC_MAX : lookup(tok, p);
1105
1106 if (MDOC_MAX == ntok) {
1107 if ( ! mdoc_word_alloc(m, line, la, p))
1108 return(0);
1109 continue;
1110 }
1111
1112 if ( ! mdoc_macro(m, ntok, line, la, pos, buf))
1113 return(0);
1114 break;
1115 }
1116
1117 if (NULL == head) {
1118 if ( ! mdoc_head_alloc(m, line, ppos, tok))
1119 return(0);
1120 head = m->last;
1121 }
1122
1123 if (nl && ! append_delims(m, line, pos, buf))
1124 return(0);
1125
1126 /* If we've already opened our body, exit now. */
1127
1128 if (NULL != body)
1129 goto out;
1130
1131 /*
1132 * If there is an open (i.e., unvalidated) sub-block requiring
1133 * explicit close-out, postpone switching the current block from
1134 * head to body until the rew_sub() call closing out that
1135 * sub-block.
1136 */
1137 for (n = m->last; n && n != head; n = n->parent) {
1138 if (MDOC_BLOCK == n->type &&
1139 MDOC_EXPLICIT & mdoc_macros[n->tok].flags &&
1140 ! (MDOC_VALID & n->flags)) {
1141 n->pending = head;
1142 return(1);
1143 }
1144 }
1145
1146 /* Close out scopes to remain in a consistent state. */
1147
1148 if ( ! rew_sub(MDOC_HEAD, m, tok, line, ppos))
1149 return(0);
1150 if ( ! mdoc_body_alloc(m, line, ppos, tok))
1151 return(0);
1152
1153 out:
1154 if ( ! (MDOC_FREECOL & m->flags))
1155 return(1);
1156
1157 if ( ! rew_sub(MDOC_BODY, m, tok, line, ppos))
1158 return(0);
1159 if ( ! rew_sub(MDOC_BLOCK, m, tok, line, ppos))
1160 return(0);
1161
1162 m->flags &= ~MDOC_FREECOL;
1163 return(1);
1164 }
1165
1166
1167 static int
1168 blk_part_imp(MACRO_PROT_ARGS)
1169 {
1170 int la, nl;
1171 enum mdoct ntok;
1172 enum margserr ac;
1173 char *p;
1174 struct mdoc_node *blk; /* saved block context */
1175 struct mdoc_node *body; /* saved body context */
1176 struct mdoc_node *n;
1177
1178 nl = MDOC_NEWLINE & m->flags;
1179
1180 /*
1181 * A macro that spans to the end of the line. This is generally
1182 * (but not necessarily) called as the first macro. The block
1183 * has a head as the immediate child, which is always empty,
1184 * followed by zero or more opening punctuation nodes, then the
1185 * body (which may be empty, depending on the macro), then zero
1186 * or more closing punctuation nodes.
1187 */
1188
1189 if ( ! mdoc_block_alloc(m, line, ppos, tok, NULL))
1190 return(0);
1191
1192 blk = m->last;
1193
1194 if ( ! mdoc_head_alloc(m, line, ppos, tok))
1195 return(0);
1196 if ( ! rew_sub(MDOC_HEAD, m, tok, line, ppos))
1197 return(0);
1198
1199 /*
1200 * Open the body scope "on-demand", that is, after we've
1201 * processed all our the leading delimiters (open parenthesis,
1202 * etc.).
1203 */
1204
1205 for (body = NULL; ; ) {
1206 la = *pos;
1207 ac = mdoc_args(m, line, pos, buf, tok, &p);
1208
1209 if (ARGS_ERROR == ac)
1210 return(0);
1211 if (ARGS_EOLN == ac)
1212 break;
1213 if (ARGS_PUNCT == ac)
1214 break;
1215
1216 if (NULL == body && ARGS_QWORD != ac &&
1217 DELIM_OPEN == mdoc_isdelim(p)) {
1218 if ( ! mdoc_word_alloc(m, line, la, p))
1219 return(0);
1220 continue;
1221 }
1222
1223 if (NULL == body) {
1224 if ( ! mdoc_body_alloc(m, line, ppos, tok))
1225 return(0);
1226 body = m->last;
1227 }
1228
1229 ntok = ARGS_QWORD == ac ? MDOC_MAX : lookup(tok, p);
1230
1231 if (MDOC_MAX == ntok) {
1232 if ( ! mdoc_word_alloc(m, line, la, p))
1233 return(0);
1234 continue;
1235 }
1236
1237 if ( ! mdoc_macro(m, ntok, line, la, pos, buf))
1238 return(0);
1239 break;
1240 }
1241
1242 /* Clean-ups to leave in a consistent state. */
1243
1244 if (NULL == body) {
1245 if ( ! mdoc_body_alloc(m, line, ppos, tok))
1246 return(0);
1247 body = m->last;
1248 }
1249
1250 for (n = body->child; n && n->next; n = n->next)
1251 /* Do nothing. */ ;
1252
1253 /*
1254 * End of sentence spacing: if the last node is a text node and
1255 * has a trailing period, then mark it as being end-of-sentence.
1256 */
1257
1258 if (n && MDOC_TEXT == n->type && n->string)
1259 if (mandoc_eos(n->string, strlen(n->string), 1))
1260 n->flags |= MDOC_EOS;
1261
1262 /* Up-propogate the end-of-space flag. */
1263
1264 if (n && (MDOC_EOS & n->flags)) {
1265 body->flags |= MDOC_EOS;
1266 body->parent->flags |= MDOC_EOS;
1267 }
1268
1269 /*
1270 * If there is an open sub-block requiring explicit close-out,
1271 * postpone closing out the current block
1272 * until the rew_sub() call closing out the sub-block.
1273 */
1274 for (n = m->last; n && n != body && n != blk->parent; n = n->parent) {
1275 if (MDOC_BLOCK == n->type &&
1276 MDOC_EXPLICIT & mdoc_macros[n->tok].flags &&
1277 ! (MDOC_VALID & n->flags)) {
1278 make_pending(n, tok, m, line, ppos);
1279 if ( ! mdoc_endbody_alloc(m, line, ppos,
1280 tok, body, ENDBODY_NOSPACE))
1281 return(0);
1282 return(1);
1283 }
1284 }
1285
1286 /*
1287 * If we can't rewind to our body, then our scope has already
1288 * been closed by another macro (like `Oc' closing `Op'). This
1289 * is ugly behaviour nodding its head to OpenBSD's overwhelming
1290 * crufty use of `Op' breakage.
1291 */
1292 if (n != body && ! mdoc_vmsg(m, MANDOCERR_SCOPENEST,
1293 line, ppos, "%s broken", mdoc_macronames[tok]))
1294 return(0);
1295
1296 if (n && ! rew_sub(MDOC_BODY, m, tok, line, ppos))
1297 return(0);
1298
1299 /* Standard appending of delimiters. */
1300
1301 if (nl && ! append_delims(m, line, pos, buf))
1302 return(0);
1303
1304 /* Rewind scope, if applicable. */
1305
1306 if (n && ! rew_sub(MDOC_BLOCK, m, tok, line, ppos))
1307 return(0);
1308
1309 return(1);
1310 }
1311
1312
1313 static int
1314 blk_part_exp(MACRO_PROT_ARGS)
1315 {
1316 int la, nl;
1317 enum margserr ac;
1318 struct mdoc_node *head; /* keep track of head */
1319 struct mdoc_node *body; /* keep track of body */
1320 char *p;
1321 enum mdoct ntok;
1322
1323 nl = MDOC_NEWLINE & m->flags;
1324
1325 /*
1326 * The opening of an explicit macro having zero or more leading
1327 * punctuation nodes; a head with optional single element (the
1328 * case of `Eo'); and a body that may be empty.
1329 */
1330
1331 if ( ! mdoc_block_alloc(m, line, ppos, tok, NULL))
1332 return(0);
1333
1334 for (head = body = NULL; ; ) {
1335 la = *pos;
1336 ac = mdoc_args(m, line, pos, buf, tok, &p);
1337
1338 if (ARGS_ERROR == ac)
1339 return(0);
1340 if (ARGS_PUNCT == ac)
1341 break;
1342 if (ARGS_EOLN == ac)
1343 break;
1344
1345 /* Flush out leading punctuation. */
1346
1347 if (NULL == head && ARGS_QWORD != ac &&
1348 DELIM_OPEN == mdoc_isdelim(p)) {
1349 assert(NULL == body);
1350 if ( ! mdoc_word_alloc(m, line, la, p))
1351 return(0);
1352 continue;
1353 }
1354
1355 if (NULL == head) {
1356 assert(NULL == body);
1357 if ( ! mdoc_head_alloc(m, line, ppos, tok))
1358 return(0);
1359 head = m->last;
1360 }
1361
1362 /*
1363 * `Eo' gobbles any data into the head, but most other
1364 * macros just immediately close out and begin the body.
1365 */
1366
1367 if (NULL == body) {
1368 assert(head);
1369 /* No check whether it's a macro! */
1370 if (MDOC_Eo == tok)
1371 if ( ! mdoc_word_alloc(m, line, la, p))
1372 return(0);
1373
1374 if ( ! rew_sub(MDOC_HEAD, m, tok, line, ppos))
1375 return(0);
1376 if ( ! mdoc_body_alloc(m, line, ppos, tok))
1377 return(0);
1378 body = m->last;
1379
1380 if (MDOC_Eo == tok)
1381 continue;
1382 }
1383
1384 assert(NULL != head && NULL != body);
1385
1386 ntok = ARGS_QWORD == ac ? MDOC_MAX : lookup(tok, p);
1387
1388 if (MDOC_MAX == ntok) {
1389 if ( ! mdoc_word_alloc(m, line, la, p))
1390 return(0);
1391 continue;
1392 }
1393
1394 if ( ! mdoc_macro(m, ntok, line, la, pos, buf))
1395 return(0);
1396 break;
1397 }
1398
1399 /* Clean-up to leave in a consistent state. */
1400
1401 if (NULL == head) {
1402 if ( ! mdoc_head_alloc(m, line, ppos, tok))
1403 return(0);
1404 head = m->last;
1405 }
1406
1407 if (NULL == body) {
1408 if ( ! rew_sub(MDOC_HEAD, m, tok, line, ppos))
1409 return(0);
1410 if ( ! mdoc_body_alloc(m, line, ppos, tok))
1411 return(0);
1412 body = m->last;
1413 }
1414
1415 /* Standard appending of delimiters. */
1416
1417 if ( ! nl)
1418 return(1);
1419 return(append_delims(m, line, pos, buf));
1420 }
1421
1422
1423 /* ARGSUSED */
1424 static int
1425 in_line_argn(MACRO_PROT_ARGS)
1426 {
1427 int la, flushed, j, maxargs, nl;
1428 enum margserr ac;
1429 enum margverr av;
1430 struct mdoc_arg *arg;
1431 char *p;
1432 enum mdoct ntok;
1433
1434 nl = MDOC_NEWLINE & m->flags;
1435
1436 /*
1437 * A line macro that has a fixed number of arguments (maxargs).
1438 * Only open the scope once the first non-leading-punctuation is
1439 * found (unless MDOC_IGNDELIM is noted, like in `Pf'), then
1440 * keep it open until the maximum number of arguments are
1441 * exhausted.
1442 */
1443
1444 switch (tok) {
1445 case (MDOC_Ap):
1446 /* FALLTHROUGH */
1447 case (MDOC_No):
1448 /* FALLTHROUGH */
1449 case (MDOC_Ns):
1450 /* FALLTHROUGH */
1451 case (MDOC_Ux):
1452 maxargs = 0;
1453 break;
1454 case (MDOC_Xr):
1455 maxargs = 2;
1456 break;
1457 default:
1458 maxargs = 1;
1459 break;
1460 }
1461
1462 for (arg = NULL; ; ) {
1463 la = *pos;
1464 av = mdoc_argv(m, line, tok, &arg, pos, buf);
1465
1466 if (ARGV_WORD == av) {
1467 *pos = la;
1468 break;
1469 }
1470
1471 if (ARGV_EOLN == av)
1472 break;
1473 if (ARGV_ARG == av)
1474 continue;
1475
1476 mdoc_argv_free(arg);
1477 return(0);
1478 }
1479
1480 for (flushed = j = 0; ; ) {
1481 la = *pos;
1482 ac = mdoc_args(m, line, pos, buf, tok, &p);
1483
1484 if (ARGS_ERROR == ac)
1485 return(0);
1486 if (ARGS_PUNCT == ac)
1487 break;
1488 if (ARGS_EOLN == ac)
1489 break;
1490
1491 if ( ! (MDOC_IGNDELIM & mdoc_macros[tok].flags) &&
1492 ARGS_QWORD != ac &&
1493 0 == j && DELIM_OPEN == mdoc_isdelim(p)) {
1494 if ( ! mdoc_word_alloc(m, line, la, p))
1495 return(0);
1496 continue;
1497 } else if (0 == j)
1498 if ( ! mdoc_elem_alloc(m, line, la, tok, arg))
1499 return(0);
1500
1501 if (j == maxargs && ! flushed) {
1502 if ( ! rew_elem(m, tok))
1503 return(0);
1504 flushed = 1;
1505 }
1506
1507 ntok = ARGS_QWORD == ac ? MDOC_MAX : lookup(tok, p);
1508
1509 if (MDOC_MAX != ntok) {
1510 if ( ! flushed && ! rew_elem(m, tok))
1511 return(0);
1512 flushed = 1;
1513 if ( ! mdoc_macro(m, ntok, line, la, pos, buf))
1514 return(0);
1515 j++;
1516 break;
1517 }
1518
1519 if ( ! (MDOC_IGNDELIM & mdoc_macros[tok].flags) &&
1520 ARGS_QWORD != ac &&
1521 ! flushed &&
1522 DELIM_NONE != mdoc_isdelim(p)) {
1523 if ( ! rew_elem(m, tok))
1524 return(0);
1525 flushed = 1;
1526 }
1527
1528 /*
1529 * XXX: this is a hack to work around groff's ugliness
1530 * as regards `Xr' and extraneous arguments. It should
1531 * ideally be deprecated behaviour, but because this is
1532 * code is no here, it's unlikely to be removed.
1533 */
1534
1535 #ifdef __OpenBSD__
1536 if (MDOC_Xr == tok && j == maxargs) {
1537 if ( ! mdoc_elem_alloc(m, line, la, MDOC_Ns, NULL))
1538 return(0);
1539 if ( ! rew_elem(m, MDOC_Ns))
1540 return(0);
1541 }
1542 #endif
1543
1544 if ( ! mdoc_word_alloc(m, line, la, p))
1545 return(0);
1546 j++;
1547 }
1548
1549 if (0 == j && ! mdoc_elem_alloc(m, line, la, tok, arg))
1550 return(0);
1551
1552 /* Close out in a consistent state. */
1553
1554 if ( ! flushed && ! rew_elem(m, tok))
1555 return(0);
1556 if ( ! nl)
1557 return(1);
1558 return(append_delims(m, line, pos, buf));
1559 }
1560
1561
1562 static int
1563 in_line_eoln(MACRO_PROT_ARGS)
1564 {
1565 int la;
1566 enum margserr ac;
1567 enum margverr av;
1568 struct mdoc_arg *arg;
1569 char *p;
1570 enum mdoct ntok;
1571
1572 assert( ! (MDOC_PARSED & mdoc_macros[tok].flags));
1573
1574 if (tok == MDOC_Pp)
1575 rew_sub(MDOC_BLOCK, m, MDOC_Nm, line, ppos);
1576
1577 /* Parse macro arguments. */
1578
1579 for (arg = NULL; ; ) {
1580 la = *pos;
1581 av = mdoc_argv(m, line, tok, &arg, pos, buf);
1582
1583 if (ARGV_WORD == av) {
1584 *pos = la;
1585 break;
1586 }
1587 if (ARGV_EOLN == av)
1588 break;
1589 if (ARGV_ARG == av)
1590 continue;
1591
1592 mdoc_argv_free(arg);
1593 return(0);
1594 }
1595
1596 /* Open element scope. */
1597
1598 if ( ! mdoc_elem_alloc(m, line, ppos, tok, arg))
1599 return(0);
1600
1601 /* Parse argument terms. */
1602
1603 for (;;) {
1604 la = *pos;
1605 ac = mdoc_args(m, line, pos, buf, tok, &p);
1606
1607 if (ARGS_ERROR == ac)
1608 return(0);
1609 if (ARGS_EOLN == ac)
1610 break;
1611
1612 ntok = ARGS_QWORD == ac ? MDOC_MAX : lookup(tok, p);
1613
1614 if (MDOC_MAX == ntok) {
1615 if ( ! mdoc_word_alloc(m, line, la, p))
1616 return(0);
1617 continue;
1618 }
1619
1620 if ( ! rew_elem(m, tok))
1621 return(0);
1622 return(mdoc_macro(m, ntok, line, la, pos, buf));
1623 }
1624
1625 /* Close out (no delimiters). */
1626
1627 return(rew_elem(m, tok));
1628 }
1629
1630
1631 /* ARGSUSED */
1632 static int
1633 ctx_synopsis(MACRO_PROT_ARGS)
1634 {
1635 int nl;
1636
1637 nl = MDOC_NEWLINE & m->flags;
1638
1639 /* If we're not in the SYNOPSIS, go straight to in-line. */
1640 if ( ! (MDOC_SYNOPSIS & m->flags))
1641 return(in_line(m, tok, line, ppos, pos, buf));
1642
1643 /* If we're a nested call, same place. */
1644 if ( ! nl)
1645 return(in_line(m, tok, line, ppos, pos, buf));
1646
1647 /*
1648 * XXX: this will open a block scope; however, if later we end
1649 * up formatting the block scope, then child nodes will inherit
1650 * the formatting. Be careful.
1651 */
1652 if (MDOC_Nm == tok)
1653 return(blk_full(m, tok, line, ppos, pos, buf));
1654 assert(MDOC_Vt == tok);
1655 return(blk_part_imp(m, tok, line, ppos, pos, buf));
1656 }
1657
1658
1659 /* ARGSUSED */
1660 static int
1661 obsolete(MACRO_PROT_ARGS)
1662 {
1663
1664 return(mdoc_pmsg(m, line, ppos, MANDOCERR_MACROOBS));
1665 }
1666
1667
1668 /*
1669 * Phrases occur within `Bl -column' entries, separated by `Ta' or tabs.
1670 * They're unusual because they're basically free-form text until a
1671 * macro is encountered.
1672 */
1673 static int
1674 phrase(struct mdoc *m, int line, int ppos, char *buf)
1675 {
1676 int la, pos;
1677 enum margserr ac;
1678 enum mdoct ntok;
1679 char *p;
1680
1681 for (pos = ppos; ; ) {
1682 la = pos;
1683
1684 ac = mdoc_zargs(m, line, &pos, buf, 0, &p);
1685
1686 if (ARGS_ERROR == ac)
1687 return(0);
1688 if (ARGS_EOLN == ac)
1689 break;
1690
1691 ntok = ARGS_QWORD == ac ? MDOC_MAX : lookup_raw(p);
1692
1693 if (MDOC_MAX == ntok) {
1694 if ( ! mdoc_word_alloc(m, line, la, p))
1695 return(0);
1696 continue;
1697 }
1698
1699 if ( ! mdoc_macro(m, ntok, line, la, &pos, buf))
1700 return(0);
1701 return(append_delims(m, line, &pos, buf));
1702 }
1703
1704 return(1);
1705 }
1706
1707
1708 /* ARGSUSED */
1709 static int
1710 phrase_ta(MACRO_PROT_ARGS)
1711 {
1712 int la;
1713 enum mdoct ntok;
1714 enum margserr ac;
1715 char *p;
1716
1717 /*
1718 * FIXME: this is overly restrictive: if the `Ta' is unexpected,
1719 * it should simply error out with ARGSLOST.
1720 */
1721
1722 if ( ! rew_sub(MDOC_BODY, m, MDOC_It, line, ppos))
1723 return(0);
1724 if ( ! mdoc_body_alloc(m, line, ppos, MDOC_It))
1725 return(0);
1726
1727 for (;;) {
1728 la = *pos;
1729 ac = mdoc_zargs(m, line, pos, buf, 0, &p);
1730
1731 if (ARGS_ERROR == ac)
1732 return(0);
1733 if (ARGS_EOLN == ac)
1734 break;
1735
1736 ntok = ARGS_QWORD == ac ? MDOC_MAX : lookup_raw(p);
1737
1738 if (MDOC_MAX == ntok) {
1739 if ( ! mdoc_word_alloc(m, line, la, p))
1740 return(0);
1741 continue;
1742 }
1743
1744 if ( ! mdoc_macro(m, ntok, line, la, pos, buf))
1745 return(0);
1746 return(append_delims(m, line, pos, buf));
1747 }
1748
1749 return(1);
1750 }
A git conversion mirror of mandoc.bsd.lv