]> 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
Have `Bx' accept two arguments, not just one, and join these arguments
[mandoc.git] / mdoc_macro.c
1 /* $Id: mdoc_macro.c,v 1.101 2011/01/25 15:17:18 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 struct mdoc_node *n, *np;
256
257 assert(to);
258 mdoc->next = MDOC_NEXT_SIBLING;
259
260 /* LINTED */
261 while (mdoc->last != to) {
262 /*
263 * Save the parent here, because we may delete the
264 * m->last node in the post-validation phase and reset
265 * it to m->last->parent, causing a step in the closing
266 * out to be lost.
267 */
268 np = mdoc->last->parent;
269 if ( ! mdoc_valid_post(mdoc))
270 return(0);
271 n = mdoc->last;
272 mdoc->last = np;
273 assert(mdoc->last);
274 mdoc->last->last = n;
275 }
276
277 return(mdoc_valid_post(mdoc));
278 }
279
280
281 /*
282 * For a block closing macro, return the corresponding opening one.
283 * Otherwise, return the macro itself.
284 */
285 static enum mdoct
286 rew_alt(enum mdoct tok)
287 {
288 switch (tok) {
289 case (MDOC_Ac):
290 return(MDOC_Ao);
291 case (MDOC_Bc):
292 return(MDOC_Bo);
293 case (MDOC_Brc):
294 return(MDOC_Bro);
295 case (MDOC_Dc):
296 return(MDOC_Do);
297 case (MDOC_Ec):
298 return(MDOC_Eo);
299 case (MDOC_Ed):
300 return(MDOC_Bd);
301 case (MDOC_Ef):
302 return(MDOC_Bf);
303 case (MDOC_Ek):
304 return(MDOC_Bk);
305 case (MDOC_El):
306 return(MDOC_Bl);
307 case (MDOC_Fc):
308 return(MDOC_Fo);
309 case (MDOC_Oc):
310 return(MDOC_Oo);
311 case (MDOC_Pc):
312 return(MDOC_Po);
313 case (MDOC_Qc):
314 return(MDOC_Qo);
315 case (MDOC_Re):
316 return(MDOC_Rs);
317 case (MDOC_Sc):
318 return(MDOC_So);
319 case (MDOC_Xc):
320 return(MDOC_Xo);
321 default:
322 return(tok);
323 }
324 /* NOTREACHED */
325 }
326
327
328 /*
329 * Rewinding to tok, how do we have to handle *p?
330 * REWIND_NONE: *p would delimit tok, but no tok scope is open
331 * inside *p, so there is no need to rewind anything at all.
332 * REWIND_THIS: *p matches tok, so rewind *p and nothing else.
333 * REWIND_MORE: *p is implicit, rewind it and keep searching for tok.
334 * REWIND_FORCE: *p is explicit, but tok is full, force rewinding *p.
335 * REWIND_LATER: *p is explicit and still open, postpone rewinding.
336 * REWIND_ERROR: No tok block is open at all.
337 */
338 static enum rew
339 rew_dohalt(enum mdoct tok, enum mdoc_type type,
340 const struct mdoc_node *p)
341 {
342
343 /*
344 * No matching token, no delimiting block, no broken block.
345 * This can happen when full implicit macros are called for
346 * the first time but try to rewind their previous
347 * instance anyway.
348 */
349 if (MDOC_ROOT == p->type)
350 return(MDOC_BLOCK == type &&
351 MDOC_EXPLICIT & mdoc_macros[tok].flags ?
352 REWIND_ERROR : REWIND_NONE);
353
354 /*
355 * When starting to rewind, skip plain text
356 * and nodes that have already been rewound.
357 */
358 if (MDOC_TEXT == p->type || MDOC_VALID & p->flags)
359 return(REWIND_MORE);
360
361 /*
362 * The easiest case: Found a matching token.
363 * This applies to both blocks and elements.
364 */
365 tok = rew_alt(tok);
366 if (tok == p->tok)
367 return(p->end ? REWIND_NONE :
368 type == p->type ? REWIND_THIS : REWIND_MORE);
369
370 /*
371 * While elements do require rewinding for themselves,
372 * they never affect rewinding of other nodes.
373 */
374 if (MDOC_ELEM == p->type)
375 return(REWIND_MORE);
376
377 /*
378 * Blocks delimited by our target token get REWIND_MORE.
379 * Blocks delimiting our target token get REWIND_NONE.
380 */
381 switch (tok) {
382 case (MDOC_Bl):
383 if (MDOC_It == p->tok)
384 return(REWIND_MORE);
385 break;
386 case (MDOC_It):
387 if (MDOC_BODY == p->type && MDOC_Bl == p->tok)
388 return(REWIND_NONE);
389 break;
390 /*
391 * XXX Badly nested block handling still fails badly
392 * when one block is breaking two blocks of the same type.
393 * This is an incomplete and extremely ugly workaround,
394 * required to let the OpenBSD tree build.
395 */
396 case (MDOC_Oo):
397 if (MDOC_Op == p->tok)
398 return(REWIND_MORE);
399 break;
400 case (MDOC_Nm):
401 return(REWIND_NONE);
402 case (MDOC_Nd):
403 /* FALLTHROUGH */
404 case (MDOC_Ss):
405 if (MDOC_BODY == p->type && MDOC_Sh == p->tok)
406 return(REWIND_NONE);
407 /* FALLTHROUGH */
408 case (MDOC_Sh):
409 if (MDOC_Nd == p->tok || MDOC_Ss == p->tok ||
410 MDOC_Sh == p->tok)
411 return(REWIND_MORE);
412 break;
413 default:
414 break;
415 }
416
417 /*
418 * Default block rewinding rules.
419 * In particular, always skip block end markers,
420 * and let all blocks rewind Nm children.
421 */
422 if (ENDBODY_NOT != p->end || MDOC_Nm == p->tok ||
423 (MDOC_BLOCK == p->type &&
424 ! (MDOC_EXPLICIT & mdoc_macros[tok].flags)))
425 return(REWIND_MORE);
426
427 /*
428 * By default, closing out full blocks
429 * forces closing of broken explicit blocks,
430 * while closing out partial blocks
431 * allows delayed rewinding by default.
432 */
433 return (&blk_full == mdoc_macros[tok].fp ?
434 REWIND_FORCE : REWIND_LATER);
435 }
436
437
438 static int
439 rew_elem(struct mdoc *mdoc, enum mdoct tok)
440 {
441 struct mdoc_node *n;
442
443 n = mdoc->last;
444 if (MDOC_ELEM != n->type)
445 n = n->parent;
446 assert(MDOC_ELEM == n->type);
447 assert(tok == n->tok);
448
449 return(rew_last(mdoc, n));
450 }
451
452
453 /*
454 * We are trying to close a block identified by tok,
455 * but the child block *broken is still open.
456 * Thus, postpone closing the tok block
457 * until the rew_sub call closing *broken.
458 */
459 static int
460 make_pending(struct mdoc_node *broken, enum mdoct tok,
461 struct mdoc *m, int line, int ppos)
462 {
463 struct mdoc_node *breaker;
464
465 /*
466 * Iterate backwards, searching for the block matching tok,
467 * that is, the block breaking the *broken block.
468 */
469 for (breaker = broken->parent; breaker; breaker = breaker->parent) {
470
471 /*
472 * If the *broken block had already been broken before
473 * and we encounter its breaker, make the tok block
474 * pending on the inner breaker.
475 * Graphically, "[A breaker=[B broken=[C->B B] tok=A] C]"
476 * becomes "[A broken=[B [C->B B] tok=A] C]"
477 * and finally "[A [B->A [C->B B] A] C]".
478 */
479 if (breaker == broken->pending) {
480 broken = breaker;
481 continue;
482 }
483
484 if (REWIND_THIS != rew_dohalt(tok, MDOC_BLOCK, breaker))
485 continue;
486 if (MDOC_BODY == broken->type)
487 broken = broken->parent;
488
489 /*
490 * Found the breaker.
491 * If another, outer breaker is already pending on
492 * the *broken block, we must not clobber the link
493 * to the outer breaker, but make it pending on the
494 * new, now inner breaker.
495 * Graphically, "[A breaker=[B broken=[C->A A] tok=B] C]"
496 * becomes "[A breaker=[B->A broken=[C A] tok=B] C]"
497 * and finally "[A [B->A [C->B A] B] C]".
498 */
499 if (broken->pending) {
500 struct mdoc_node *taker;
501
502 /*
503 * If the breaker had also been broken before,
504 * it cannot take on the outer breaker itself,
505 * but must hand it on to its own breakers.
506 * Graphically, this is the following situation:
507 * "[A [B breaker=[C->B B] broken=[D->A A] tok=C] D]"
508 * "[A taker=[B->A breaker=[C->B B] [D->C A] C] D]"
509 */
510 taker = breaker;
511 while (taker->pending)
512 taker = taker->pending;
513 taker->pending = broken->pending;
514 }
515 broken->pending = breaker;
516 mdoc_vmsg(m, MANDOCERR_SCOPENEST, line, ppos,
517 "%s breaks %s", mdoc_macronames[tok],
518 mdoc_macronames[broken->tok]);
519 return(1);
520 }
521
522 /*
523 * Found no matching block for tok.
524 * Are you trying to close a block that is not open?
525 */
526 return(0);
527 }
528
529
530 static int
531 rew_sub(enum mdoc_type t, struct mdoc *m,
532 enum mdoct tok, int line, int ppos)
533 {
534 struct mdoc_node *n;
535
536 n = m->last;
537 while (n) {
538 switch (rew_dohalt(tok, t, n)) {
539 case (REWIND_NONE):
540 return(1);
541 case (REWIND_THIS):
542 break;
543 case (REWIND_FORCE):
544 mdoc_vmsg(m, MANDOCERR_SCOPEBROKEN, line, ppos,
545 "%s breaks %s", mdoc_macronames[tok],
546 mdoc_macronames[n->tok]);
547 /* FALLTHROUGH */
548 case (REWIND_MORE):
549 n = n->parent;
550 continue;
551 case (REWIND_LATER):
552 if (make_pending(n, tok, m, line, ppos) ||
553 MDOC_BLOCK != t)
554 return(1);
555 /* FALLTHROUGH */
556 case (REWIND_ERROR):
557 mdoc_pmsg(m, line, ppos, MANDOCERR_NOSCOPE);
558 return(1);
559 }
560 break;
561 }
562
563 assert(n);
564 if ( ! rew_last(m, n))
565 return(0);
566
567 /*
568 * The current block extends an enclosing block.
569 * Now that the current block ends, close the enclosing block, too.
570 */
571 while (NULL != (n = n->pending)) {
572 if ( ! rew_last(m, n))
573 return(0);
574 if (MDOC_HEAD == n->type &&
575 ! mdoc_body_alloc(m, n->line, n->pos, n->tok))
576 return(0);
577 }
578
579 return(1);
580 }
581
582
583 static int
584 append_delims(struct mdoc *m, int line, int *pos, char *buf)
585 {
586 int la;
587 enum margserr ac;
588 char *p;
589
590 if ('\0' == buf[*pos])
591 return(1);
592
593 for (;;) {
594 la = *pos;
595 ac = mdoc_zargs(m, line, pos, buf, ARGS_NOWARN, &p);
596
597 if (ARGS_ERROR == ac)
598 return(0);
599 else if (ARGS_EOLN == ac)
600 break;
601
602 assert(DELIM_NONE != mdoc_isdelim(p));
603 if ( ! mdoc_word_alloc(m, line, la, p))
604 return(0);
605
606 /*
607 * If we encounter end-of-sentence symbols, then trigger
608 * the double-space.
609 *
610 * XXX: it's easy to allow this to propogate outward to
611 * the last symbol, such that `. )' will cause the
612 * correct double-spacing. However, (1) groff isn't
613 * smart enough to do this and (2) it would require
614 * knowing which symbols break this behaviour, for
615 * example, `. ;' shouldn't propogate the double-space.
616 */
617 if (mandoc_eos(p, strlen(p), 0))
618 m->last->flags |= MDOC_EOS;
619 }
620
621 return(1);
622 }
623
624
625 /*
626 * Close out block partial/full explicit.
627 */
628 static int
629 blk_exp_close(MACRO_PROT_ARGS)
630 {
631 struct mdoc_node *body; /* Our own body. */
632 struct mdoc_node *later; /* A sub-block starting later. */
633 struct mdoc_node *n; /* For searching backwards. */
634
635 int j, lastarg, maxargs, flushed, nl;
636 enum margserr ac;
637 enum mdoct atok, ntok;
638 char *p;
639
640 nl = MDOC_NEWLINE & m->flags;
641
642 switch (tok) {
643 case (MDOC_Ec):
644 maxargs = 1;
645 break;
646 default:
647 maxargs = 0;
648 break;
649 }
650
651 /*
652 * Search backwards for beginnings of blocks,
653 * both of our own and of pending sub-blocks.
654 */
655 atok = rew_alt(tok);
656 body = later = NULL;
657 for (n = m->last; n; n = n->parent) {
658 if (MDOC_VALID & n->flags)
659 continue;
660
661 /* Remember the start of our own body. */
662 if (MDOC_BODY == n->type && atok == n->tok) {
663 if (ENDBODY_NOT == n->end)
664 body = n;
665 continue;
666 }
667
668 if (MDOC_BLOCK != n->type || MDOC_Nm == n->tok)
669 continue;
670 if (atok == n->tok) {
671 assert(body);
672
673 /*
674 * Found the start of our own block.
675 * When there is no pending sub block,
676 * just proceed to closing out.
677 */
678 if (NULL == later)
679 break;
680
681 /*
682 * When there is a pending sub block,
683 * postpone closing out the current block
684 * until the rew_sub() closing out the sub-block.
685 */
686 make_pending(later, tok, m, line, ppos);
687
688 /*
689 * Mark the place where the formatting - but not
690 * the scope - of the current block ends.
691 */
692 if ( ! mdoc_endbody_alloc(m, line, ppos,
693 atok, body, ENDBODY_SPACE))
694 return(0);
695 break;
696 }
697
698 /*
699 * When finding an open sub block, remember the last
700 * open explicit block, or, in case there are only
701 * implicit ones, the first open implicit block.
702 */
703 if (later &&
704 MDOC_EXPLICIT & mdoc_macros[later->tok].flags)
705 continue;
706 if (MDOC_CALLABLE & mdoc_macros[n->tok].flags)
707 later = n;
708 }
709
710 if ( ! (MDOC_CALLABLE & mdoc_macros[tok].flags)) {
711 /* FIXME: do this in validate */
712 if (buf[*pos])
713 if ( ! mdoc_pmsg(m, line, ppos, MANDOCERR_ARGSLOST))
714 return(0);
715
716 if ( ! rew_sub(MDOC_BODY, m, tok, line, ppos))
717 return(0);
718 return(rew_sub(MDOC_BLOCK, m, tok, line, ppos));
719 }
720
721 if ( ! rew_sub(MDOC_BODY, m, tok, line, ppos))
722 return(0);
723
724 if (NULL == later && maxargs > 0)
725 if ( ! mdoc_tail_alloc(m, line, ppos, rew_alt(tok)))
726 return(0);
727
728 for (flushed = j = 0; ; j++) {
729 lastarg = *pos;
730
731 if (j == maxargs && ! flushed) {
732 if ( ! rew_sub(MDOC_BLOCK, m, tok, line, ppos))
733 return(0);
734 flushed = 1;
735 }
736
737 ac = mdoc_args(m, line, pos, buf, tok, &p);
738
739 if (ARGS_ERROR == ac)
740 return(0);
741 if (ARGS_PUNCT == ac)
742 break;
743 if (ARGS_EOLN == ac)
744 break;
745
746 ntok = ARGS_QWORD == ac ? MDOC_MAX : lookup(tok, p);
747
748 if (MDOC_MAX == ntok) {
749 if ( ! mdoc_word_alloc(m, line, lastarg, p))
750 return(0);
751 continue;
752 }
753
754 if ( ! flushed) {
755 if ( ! rew_sub(MDOC_BLOCK, m, tok, line, ppos))
756 return(0);
757 flushed = 1;
758 }
759 if ( ! mdoc_macro(m, ntok, line, lastarg, pos, buf))
760 return(0);
761 break;
762 }
763
764 if ( ! flushed && ! rew_sub(MDOC_BLOCK, m, tok, line, ppos))
765 return(0);
766
767 if ( ! nl)
768 return(1);
769 return(append_delims(m, line, pos, buf));
770 }
771
772
773 static int
774 in_line(MACRO_PROT_ARGS)
775 {
776 int la, scope, cnt, nc, nl;
777 enum margverr av;
778 enum mdoct ntok;
779 enum margserr ac;
780 enum mdelim d;
781 struct mdoc_arg *arg;
782 char *p;
783
784 nl = MDOC_NEWLINE & m->flags;
785
786 /*
787 * Whether we allow ignored elements (those without content,
788 * usually because of reserved words) to squeak by.
789 */
790
791 switch (tok) {
792 case (MDOC_An):
793 /* FALLTHROUGH */
794 case (MDOC_Ar):
795 /* FALLTHROUGH */
796 case (MDOC_Fl):
797 /* FALLTHROUGH */
798 case (MDOC_Mt):
799 /* FALLTHROUGH */
800 case (MDOC_Nm):
801 /* FALLTHROUGH */
802 case (MDOC_Pa):
803 nc = 1;
804 break;
805 default:
806 nc = 0;
807 break;
808 }
809
810 for (arg = NULL;; ) {
811 la = *pos;
812 av = mdoc_argv(m, line, tok, &arg, pos, buf);
813
814 if (ARGV_WORD == av) {
815 *pos = la;
816 break;
817 }
818 if (ARGV_EOLN == av)
819 break;
820 if (ARGV_ARG == av)
821 continue;
822
823 mdoc_argv_free(arg);
824 return(0);
825 }
826
827 for (cnt = scope = 0;; ) {
828 la = *pos;
829 ac = mdoc_args(m, line, pos, buf, tok, &p);
830
831 if (ARGS_ERROR == ac)
832 return(0);
833 if (ARGS_EOLN == ac)
834 break;
835 if (ARGS_PUNCT == ac)
836 break;
837
838 ntok = ARGS_QWORD == ac ? MDOC_MAX : lookup(tok, p);
839
840 /*
841 * In this case, we've located a submacro and must
842 * execute it. Close out scope, if open. If no
843 * elements have been generated, either create one (nc)
844 * or raise a warning.
845 */
846
847 if (MDOC_MAX != ntok) {
848 if (scope && ! rew_elem(m, tok))
849 return(0);
850 if (nc && 0 == cnt) {
851 if ( ! mdoc_elem_alloc(m, line, ppos, tok, arg))
852 return(0);
853 if ( ! rew_last(m, m->last))
854 return(0);
855 } else if ( ! nc && 0 == cnt) {
856 mdoc_argv_free(arg);
857 if ( ! mdoc_pmsg(m, line, ppos, MANDOCERR_MACROEMPTY))
858 return(0);
859 }
860 if ( ! mdoc_macro(m, ntok, line, la, pos, buf))
861 return(0);
862 if ( ! nl)
863 return(1);
864 return(append_delims(m, line, pos, buf));
865 }
866
867 /*
868 * Non-quote-enclosed punctuation. Set up our scope, if
869 * a word; rewind the scope, if a delimiter; then append
870 * the word.
871 */
872
873 d = ARGS_QWORD == ac ? DELIM_NONE : mdoc_isdelim(p);
874
875 if (DELIM_NONE != d) {
876 /*
877 * If we encounter closing punctuation, no word
878 * has been omitted, no scope is open, and we're
879 * allowed to have an empty element, then start
880 * a new scope. `Ar', `Fl', and `Li', only do
881 * this once per invocation. There may be more
882 * of these (all of them?).
883 */
884 if (0 == cnt && (nc || MDOC_Li == tok) &&
885 DELIM_CLOSE == d && ! scope) {
886 if ( ! mdoc_elem_alloc(m, line, ppos, tok, arg))
887 return(0);
888 if (MDOC_Ar == tok || MDOC_Li == tok ||
889 MDOC_Fl == tok)
890 cnt++;
891 scope = 1;
892 }
893 /*
894 * Close out our scope, if one is open, before
895 * any punctuation.
896 */
897 if (scope && ! rew_elem(m, tok))
898 return(0);
899 scope = 0;
900 } else if ( ! scope) {
901 if ( ! mdoc_elem_alloc(m, line, ppos, tok, arg))
902 return(0);
903 scope = 1;
904 }
905
906 if (DELIM_NONE == d)
907 cnt++;
908 if ( ! mdoc_word_alloc(m, line, la, p))
909 return(0);
910
911 /*
912 * `Fl' macros have their scope re-opened with each new
913 * word so that the `-' can be added to each one without
914 * having to parse out spaces.
915 */
916 if (scope && MDOC_Fl == tok) {
917 if ( ! rew_elem(m, tok))
918 return(0);
919 scope = 0;
920 }
921 }
922
923 if (scope && ! rew_elem(m, tok))
924 return(0);
925
926 /*
927 * If no elements have been collected and we're allowed to have
928 * empties (nc), open a scope and close it out. Otherwise,
929 * raise a warning.
930 */
931
932 if (nc && 0 == cnt) {
933 if ( ! mdoc_elem_alloc(m, line, ppos, tok, arg))
934 return(0);
935 if ( ! rew_last(m, m->last))
936 return(0);
937 } else if ( ! nc && 0 == cnt) {
938 mdoc_argv_free(arg);
939 if ( ! mdoc_pmsg(m, line, ppos, MANDOCERR_MACROEMPTY))
940 return(0);
941 }
942
943 if ( ! nl)
944 return(1);
945 return(append_delims(m, line, pos, buf));
946 }
947
948
949 static int
950 blk_full(MACRO_PROT_ARGS)
951 {
952 int la, nl;
953 struct mdoc_arg *arg;
954 struct mdoc_node *head; /* save of head macro */
955 struct mdoc_node *body; /* save of body macro */
956 struct mdoc_node *n;
957 enum mdoc_type mtt;
958 enum mdoct ntok;
959 enum margserr ac, lac;
960 enum margverr av;
961 char *p;
962
963 nl = MDOC_NEWLINE & m->flags;
964
965 /* Close out prior implicit scope. */
966
967 if ( ! (MDOC_EXPLICIT & mdoc_macros[tok].flags)) {
968 if ( ! rew_sub(MDOC_BODY, m, tok, line, ppos))
969 return(0);
970 if ( ! rew_sub(MDOC_BLOCK, m, tok, line, ppos))
971 return(0);
972 }
973
974 /*
975 * This routine accomodates implicitly- and explicitly-scoped
976 * macro openings. Implicit ones first close out prior scope
977 * (seen above). Delay opening the head until necessary to
978 * allow leading punctuation to print. Special consideration
979 * for `It -column', which has phrase-part syntax instead of
980 * regular child nodes.
981 */
982
983 for (arg = NULL;; ) {
984 la = *pos;
985 av = mdoc_argv(m, line, tok, &arg, pos, buf);
986
987 if (ARGV_WORD == av) {
988 *pos = la;
989 break;
990 }
991
992 if (ARGV_EOLN == av)
993 break;
994 if (ARGV_ARG == av)
995 continue;
996
997 mdoc_argv_free(arg);
998 return(0);
999 }
1000
1001 if ( ! mdoc_block_alloc(m, line, ppos, tok, arg))
1002 return(0);
1003
1004 head = body = NULL;
1005
1006 /*
1007 * The `Nd' macro has all arguments in its body: it's a hybrid
1008 * of block partial-explicit and full-implicit. Stupid.
1009 */
1010
1011 if (MDOC_Nd == tok) {
1012 if ( ! mdoc_head_alloc(m, line, ppos, tok))
1013 return(0);
1014 head = m->last;
1015 if ( ! rew_sub(MDOC_HEAD, m, tok, line, ppos))
1016 return(0);
1017 if ( ! mdoc_body_alloc(m, line, ppos, tok))
1018 return(0);
1019 body = m->last;
1020 }
1021
1022 ac = ARGS_ERROR;
1023
1024 for ( ; ; ) {
1025 la = *pos;
1026 /* Initialise last-phrase-type with ARGS_PEND. */
1027 lac = ARGS_ERROR == ac ? ARGS_PEND : ac;
1028 ac = mdoc_args(m, line, pos, buf, tok, &p);
1029
1030 if (ARGS_PUNCT == ac)
1031 break;
1032
1033 if (ARGS_ERROR == ac)
1034 return(0);
1035
1036 if (ARGS_EOLN == ac) {
1037 if (ARGS_PPHRASE != lac && ARGS_PHRASE != lac)
1038 break;
1039 /*
1040 * This is necessary: if the last token on a
1041 * line is a `Ta' or tab, then we'll get
1042 * ARGS_EOLN, so we must be smart enough to
1043 * reopen our scope if the last parse was a
1044 * phrase or partial phrase.
1045 */
1046 if ( ! rew_sub(MDOC_BODY, m, tok, line, ppos))
1047 return(0);
1048 if ( ! mdoc_body_alloc(m, line, ppos, tok))
1049 return(0);
1050 body = m->last;
1051 break;
1052 }
1053
1054 /*
1055 * Emit leading punctuation (i.e., punctuation before
1056 * the MDOC_HEAD) for non-phrase types.
1057 */
1058
1059 if (NULL == head &&
1060 ARGS_PEND != ac &&
1061 ARGS_PHRASE != ac &&
1062 ARGS_PPHRASE != ac &&
1063 ARGS_QWORD != ac &&
1064 DELIM_OPEN == mdoc_isdelim(p)) {
1065 if ( ! mdoc_word_alloc(m, line, la, p))
1066 return(0);
1067 continue;
1068 }
1069
1070 /* Open a head if one hasn't been opened. */
1071
1072 if (NULL == head) {
1073 if ( ! mdoc_head_alloc(m, line, ppos, tok))
1074 return(0);
1075 head = m->last;
1076 }
1077
1078 if (ARGS_PHRASE == ac ||
1079 ARGS_PEND == ac ||
1080 ARGS_PPHRASE == ac) {
1081 /*
1082 * If we haven't opened a body yet, rewind the
1083 * head; if we have, rewind that instead.
1084 */
1085
1086 mtt = body ? MDOC_BODY : MDOC_HEAD;
1087 if ( ! rew_sub(mtt, m, tok, line, ppos))
1088 return(0);
1089
1090 /* Then allocate our body context. */
1091
1092 if ( ! mdoc_body_alloc(m, line, ppos, tok))
1093 return(0);
1094 body = m->last;
1095
1096 /*
1097 * Process phrases: set whether we're in a
1098 * partial-phrase (this effects line handling)
1099 * then call down into the phrase parser.
1100 */
1101
1102 if (ARGS_PPHRASE == ac)
1103 m->flags |= MDOC_PPHRASE;
1104 if (ARGS_PEND == ac && ARGS_PPHRASE == lac)
1105 m->flags |= MDOC_PPHRASE;
1106
1107 if ( ! phrase(m, line, la, buf))
1108 return(0);
1109
1110 m->flags &= ~MDOC_PPHRASE;
1111 continue;
1112 }
1113
1114 ntok = ARGS_QWORD == ac ? MDOC_MAX : lookup(tok, p);
1115
1116 if (MDOC_MAX == ntok) {
1117 if ( ! mdoc_word_alloc(m, line, la, p))
1118 return(0);
1119 continue;
1120 }
1121
1122 if ( ! mdoc_macro(m, ntok, line, la, pos, buf))
1123 return(0);
1124 break;
1125 }
1126
1127 if (NULL == head) {
1128 if ( ! mdoc_head_alloc(m, line, ppos, tok))
1129 return(0);
1130 head = m->last;
1131 }
1132
1133 if (nl && ! append_delims(m, line, pos, buf))
1134 return(0);
1135
1136 /* If we've already opened our body, exit now. */
1137
1138 if (NULL != body)
1139 goto out;
1140
1141 /*
1142 * If there is an open (i.e., unvalidated) sub-block requiring
1143 * explicit close-out, postpone switching the current block from
1144 * head to body until the rew_sub() call closing out that
1145 * sub-block.
1146 */
1147 for (n = m->last; n && n != head; n = n->parent) {
1148 if (MDOC_BLOCK == n->type &&
1149 MDOC_EXPLICIT & mdoc_macros[n->tok].flags &&
1150 ! (MDOC_VALID & n->flags)) {
1151 n->pending = head;
1152 return(1);
1153 }
1154 }
1155
1156 /* Close out scopes to remain in a consistent state. */
1157
1158 if ( ! rew_sub(MDOC_HEAD, m, tok, line, ppos))
1159 return(0);
1160 if ( ! mdoc_body_alloc(m, line, ppos, tok))
1161 return(0);
1162
1163 out:
1164 if ( ! (MDOC_FREECOL & m->flags))
1165 return(1);
1166
1167 if ( ! rew_sub(MDOC_BODY, m, tok, line, ppos))
1168 return(0);
1169 if ( ! rew_sub(MDOC_BLOCK, m, tok, line, ppos))
1170 return(0);
1171
1172 m->flags &= ~MDOC_FREECOL;
1173 return(1);
1174 }
1175
1176
1177 static int
1178 blk_part_imp(MACRO_PROT_ARGS)
1179 {
1180 int la, nl;
1181 enum mdoct ntok;
1182 enum margserr ac;
1183 char *p;
1184 struct mdoc_node *blk; /* saved block context */
1185 struct mdoc_node *body; /* saved body context */
1186 struct mdoc_node *n;
1187
1188 nl = MDOC_NEWLINE & m->flags;
1189
1190 /*
1191 * A macro that spans to the end of the line. This is generally
1192 * (but not necessarily) called as the first macro. The block
1193 * has a head as the immediate child, which is always empty,
1194 * followed by zero or more opening punctuation nodes, then the
1195 * body (which may be empty, depending on the macro), then zero
1196 * or more closing punctuation nodes.
1197 */
1198
1199 if ( ! mdoc_block_alloc(m, line, ppos, tok, NULL))
1200 return(0);
1201
1202 blk = m->last;
1203
1204 if ( ! mdoc_head_alloc(m, line, ppos, tok))
1205 return(0);
1206 if ( ! rew_sub(MDOC_HEAD, m, tok, line, ppos))
1207 return(0);
1208
1209 /*
1210 * Open the body scope "on-demand", that is, after we've
1211 * processed all our the leading delimiters (open parenthesis,
1212 * etc.).
1213 */
1214
1215 for (body = NULL; ; ) {
1216 la = *pos;
1217 ac = mdoc_args(m, line, pos, buf, tok, &p);
1218
1219 if (ARGS_ERROR == ac)
1220 return(0);
1221 if (ARGS_EOLN == ac)
1222 break;
1223 if (ARGS_PUNCT == ac)
1224 break;
1225
1226 if (NULL == body && ARGS_QWORD != ac &&
1227 DELIM_OPEN == mdoc_isdelim(p)) {
1228 if ( ! mdoc_word_alloc(m, line, la, p))
1229 return(0);
1230 continue;
1231 }
1232
1233 if (NULL == body) {
1234 if ( ! mdoc_body_alloc(m, line, ppos, tok))
1235 return(0);
1236 body = m->last;
1237 }
1238
1239 ntok = ARGS_QWORD == ac ? MDOC_MAX : lookup(tok, p);
1240
1241 if (MDOC_MAX == ntok) {
1242 if ( ! mdoc_word_alloc(m, line, la, p))
1243 return(0);
1244 continue;
1245 }
1246
1247 if ( ! mdoc_macro(m, ntok, line, la, pos, buf))
1248 return(0);
1249 break;
1250 }
1251
1252 /* Clean-ups to leave in a consistent state. */
1253
1254 if (NULL == body) {
1255 if ( ! mdoc_body_alloc(m, line, ppos, tok))
1256 return(0);
1257 body = m->last;
1258 }
1259
1260 for (n = body->child; n && n->next; n = n->next)
1261 /* Do nothing. */ ;
1262
1263 /*
1264 * End of sentence spacing: if the last node is a text node and
1265 * has a trailing period, then mark it as being end-of-sentence.
1266 */
1267
1268 if (n && MDOC_TEXT == n->type && n->string)
1269 if (mandoc_eos(n->string, strlen(n->string), 1))
1270 n->flags |= MDOC_EOS;
1271
1272 /* Up-propogate the end-of-space flag. */
1273
1274 if (n && (MDOC_EOS & n->flags)) {
1275 body->flags |= MDOC_EOS;
1276 body->parent->flags |= MDOC_EOS;
1277 }
1278
1279 /*
1280 * If there is an open sub-block requiring explicit close-out,
1281 * postpone closing out the current block
1282 * until the rew_sub() call closing out the sub-block.
1283 */
1284 for (n = m->last; n && n != body && n != blk->parent; n = n->parent) {
1285 if (MDOC_BLOCK == n->type &&
1286 MDOC_EXPLICIT & mdoc_macros[n->tok].flags &&
1287 ! (MDOC_VALID & n->flags)) {
1288 make_pending(n, tok, m, line, ppos);
1289 if ( ! mdoc_endbody_alloc(m, line, ppos,
1290 tok, body, ENDBODY_NOSPACE))
1291 return(0);
1292 return(1);
1293 }
1294 }
1295
1296 /*
1297 * If we can't rewind to our body, then our scope has already
1298 * been closed by another macro (like `Oc' closing `Op'). This
1299 * is ugly behaviour nodding its head to OpenBSD's overwhelming
1300 * crufty use of `Op' breakage.
1301 */
1302 if (n != body && ! mdoc_vmsg(m, MANDOCERR_SCOPENEST,
1303 line, ppos, "%s broken", mdoc_macronames[tok]))
1304 return(0);
1305
1306 if (n && ! rew_sub(MDOC_BODY, m, tok, line, ppos))
1307 return(0);
1308
1309 /* Standard appending of delimiters. */
1310
1311 if (nl && ! append_delims(m, line, pos, buf))
1312 return(0);
1313
1314 /* Rewind scope, if applicable. */
1315
1316 if (n && ! rew_sub(MDOC_BLOCK, m, tok, line, ppos))
1317 return(0);
1318
1319 return(1);
1320 }
1321
1322
1323 static int
1324 blk_part_exp(MACRO_PROT_ARGS)
1325 {
1326 int la, nl;
1327 enum margserr ac;
1328 struct mdoc_node *head; /* keep track of head */
1329 struct mdoc_node *body; /* keep track of body */
1330 char *p;
1331 enum mdoct ntok;
1332
1333 nl = MDOC_NEWLINE & m->flags;
1334
1335 /*
1336 * The opening of an explicit macro having zero or more leading
1337 * punctuation nodes; a head with optional single element (the
1338 * case of `Eo'); and a body that may be empty.
1339 */
1340
1341 if ( ! mdoc_block_alloc(m, line, ppos, tok, NULL))
1342 return(0);
1343
1344 for (head = body = NULL; ; ) {
1345 la = *pos;
1346 ac = mdoc_args(m, line, pos, buf, tok, &p);
1347
1348 if (ARGS_ERROR == ac)
1349 return(0);
1350 if (ARGS_PUNCT == ac)
1351 break;
1352 if (ARGS_EOLN == ac)
1353 break;
1354
1355 /* Flush out leading punctuation. */
1356
1357 if (NULL == head && ARGS_QWORD != ac &&
1358 DELIM_OPEN == mdoc_isdelim(p)) {
1359 assert(NULL == body);
1360 if ( ! mdoc_word_alloc(m, line, la, p))
1361 return(0);
1362 continue;
1363 }
1364
1365 if (NULL == head) {
1366 assert(NULL == body);
1367 if ( ! mdoc_head_alloc(m, line, ppos, tok))
1368 return(0);
1369 head = m->last;
1370 }
1371
1372 /*
1373 * `Eo' gobbles any data into the head, but most other
1374 * macros just immediately close out and begin the body.
1375 */
1376
1377 if (NULL == body) {
1378 assert(head);
1379 /* No check whether it's a macro! */
1380 if (MDOC_Eo == tok)
1381 if ( ! mdoc_word_alloc(m, line, la, p))
1382 return(0);
1383
1384 if ( ! rew_sub(MDOC_HEAD, m, tok, line, ppos))
1385 return(0);
1386 if ( ! mdoc_body_alloc(m, line, ppos, tok))
1387 return(0);
1388 body = m->last;
1389
1390 if (MDOC_Eo == tok)
1391 continue;
1392 }
1393
1394 assert(NULL != head && NULL != body);
1395
1396 ntok = ARGS_QWORD == ac ? MDOC_MAX : lookup(tok, p);
1397
1398 if (MDOC_MAX == ntok) {
1399 if ( ! mdoc_word_alloc(m, line, la, p))
1400 return(0);
1401 continue;
1402 }
1403
1404 if ( ! mdoc_macro(m, ntok, line, la, pos, buf))
1405 return(0);
1406 break;
1407 }
1408
1409 /* Clean-up to leave in a consistent state. */
1410
1411 if (NULL == head) {
1412 if ( ! mdoc_head_alloc(m, line, ppos, tok))
1413 return(0);
1414 head = m->last;
1415 }
1416
1417 if (NULL == body) {
1418 if ( ! rew_sub(MDOC_HEAD, m, tok, line, ppos))
1419 return(0);
1420 if ( ! mdoc_body_alloc(m, line, ppos, tok))
1421 return(0);
1422 body = m->last;
1423 }
1424
1425 /* Standard appending of delimiters. */
1426
1427 if ( ! nl)
1428 return(1);
1429 return(append_delims(m, line, pos, buf));
1430 }
1431
1432
1433 /* ARGSUSED */
1434 static int
1435 in_line_argn(MACRO_PROT_ARGS)
1436 {
1437 int la, flushed, j, maxargs, nl;
1438 enum margserr ac;
1439 enum margverr av;
1440 struct mdoc_arg *arg;
1441 char *p;
1442 enum mdoct ntok;
1443
1444 nl = MDOC_NEWLINE & m->flags;
1445
1446 /*
1447 * A line macro that has a fixed number of arguments (maxargs).
1448 * Only open the scope once the first non-leading-punctuation is
1449 * found (unless MDOC_IGNDELIM is noted, like in `Pf'), then
1450 * keep it open until the maximum number of arguments are
1451 * exhausted.
1452 */
1453
1454 switch (tok) {
1455 case (MDOC_Ap):
1456 /* FALLTHROUGH */
1457 case (MDOC_No):
1458 /* FALLTHROUGH */
1459 case (MDOC_Ns):
1460 /* FALLTHROUGH */
1461 case (MDOC_Ux):
1462 maxargs = 0;
1463 break;
1464 case (MDOC_Bx):
1465 /* FALLTHROUGH */
1466 case (MDOC_Xr):
1467 maxargs = 2;
1468 break;
1469 default:
1470 maxargs = 1;
1471 break;
1472 }
1473
1474 for (arg = NULL; ; ) {
1475 la = *pos;
1476 av = mdoc_argv(m, line, tok, &arg, pos, buf);
1477
1478 if (ARGV_WORD == av) {
1479 *pos = la;
1480 break;
1481 }
1482
1483 if (ARGV_EOLN == av)
1484 break;
1485 if (ARGV_ARG == av)
1486 continue;
1487
1488 mdoc_argv_free(arg);
1489 return(0);
1490 }
1491
1492 for (flushed = j = 0; ; ) {
1493 la = *pos;
1494 ac = mdoc_args(m, line, pos, buf, tok, &p);
1495
1496 if (ARGS_ERROR == ac)
1497 return(0);
1498 if (ARGS_PUNCT == ac)
1499 break;
1500 if (ARGS_EOLN == ac)
1501 break;
1502
1503 if ( ! (MDOC_IGNDELIM & mdoc_macros[tok].flags) &&
1504 ARGS_QWORD != ac &&
1505 0 == j && DELIM_OPEN == mdoc_isdelim(p)) {
1506 if ( ! mdoc_word_alloc(m, line, la, p))
1507 return(0);
1508 continue;
1509 } else if (0 == j)
1510 if ( ! mdoc_elem_alloc(m, line, la, tok, arg))
1511 return(0);
1512
1513 if (j == maxargs && ! flushed) {
1514 if ( ! rew_elem(m, tok))
1515 return(0);
1516 flushed = 1;
1517 }
1518
1519 ntok = ARGS_QWORD == ac ? MDOC_MAX : lookup(tok, p);
1520
1521 if (MDOC_MAX != ntok) {
1522 if ( ! flushed && ! rew_elem(m, tok))
1523 return(0);
1524 flushed = 1;
1525 if ( ! mdoc_macro(m, ntok, line, la, pos, buf))
1526 return(0);
1527 j++;
1528 break;
1529 }
1530
1531 if ( ! (MDOC_IGNDELIM & mdoc_macros[tok].flags) &&
1532 ARGS_QWORD != ac &&
1533 ! flushed &&
1534 DELIM_NONE != mdoc_isdelim(p)) {
1535 if ( ! rew_elem(m, tok))
1536 return(0);
1537 flushed = 1;
1538 }
1539
1540 /*
1541 * XXX: this is a hack to work around groff's ugliness
1542 * as regards `Xr' and extraneous arguments. It should
1543 * ideally be deprecated behaviour, but because this is
1544 * code is no here, it's unlikely to be removed.
1545 */
1546
1547 #ifdef __OpenBSD__
1548 if (MDOC_Xr == tok && j == maxargs) {
1549 if ( ! mdoc_elem_alloc(m, line, la, MDOC_Ns, NULL))
1550 return(0);
1551 if ( ! rew_elem(m, MDOC_Ns))
1552 return(0);
1553 }
1554 #endif
1555
1556 if ( ! mdoc_word_alloc(m, line, la, p))
1557 return(0);
1558 j++;
1559 }
1560
1561 if (0 == j && ! mdoc_elem_alloc(m, line, la, tok, arg))
1562 return(0);
1563
1564 /* Close out in a consistent state. */
1565
1566 if ( ! flushed && ! rew_elem(m, tok))
1567 return(0);
1568 if ( ! nl)
1569 return(1);
1570 return(append_delims(m, line, pos, buf));
1571 }
1572
1573
1574 static int
1575 in_line_eoln(MACRO_PROT_ARGS)
1576 {
1577 int la;
1578 enum margserr ac;
1579 enum margverr av;
1580 struct mdoc_arg *arg;
1581 char *p;
1582 enum mdoct ntok;
1583
1584 assert( ! (MDOC_PARSED & mdoc_macros[tok].flags));
1585
1586 if (tok == MDOC_Pp)
1587 rew_sub(MDOC_BLOCK, m, MDOC_Nm, line, ppos);
1588
1589 /* Parse macro arguments. */
1590
1591 for (arg = NULL; ; ) {
1592 la = *pos;
1593 av = mdoc_argv(m, line, tok, &arg, pos, buf);
1594
1595 if (ARGV_WORD == av) {
1596 *pos = la;
1597 break;
1598 }
1599 if (ARGV_EOLN == av)
1600 break;
1601 if (ARGV_ARG == av)
1602 continue;
1603
1604 mdoc_argv_free(arg);
1605 return(0);
1606 }
1607
1608 /* Open element scope. */
1609
1610 if ( ! mdoc_elem_alloc(m, line, ppos, tok, arg))
1611 return(0);
1612
1613 /* Parse argument terms. */
1614
1615 for (;;) {
1616 la = *pos;
1617 ac = mdoc_args(m, line, pos, buf, tok, &p);
1618
1619 if (ARGS_ERROR == ac)
1620 return(0);
1621 if (ARGS_EOLN == ac)
1622 break;
1623
1624 ntok = ARGS_QWORD == ac ? MDOC_MAX : lookup(tok, p);
1625
1626 if (MDOC_MAX == ntok) {
1627 if ( ! mdoc_word_alloc(m, line, la, p))
1628 return(0);
1629 continue;
1630 }
1631
1632 if ( ! rew_elem(m, tok))
1633 return(0);
1634 return(mdoc_macro(m, ntok, line, la, pos, buf));
1635 }
1636
1637 /* Close out (no delimiters). */
1638
1639 return(rew_elem(m, tok));
1640 }
1641
1642
1643 /* ARGSUSED */
1644 static int
1645 ctx_synopsis(MACRO_PROT_ARGS)
1646 {
1647 int nl;
1648
1649 nl = MDOC_NEWLINE & m->flags;
1650
1651 /* If we're not in the SYNOPSIS, go straight to in-line. */
1652 if ( ! (MDOC_SYNOPSIS & m->flags))
1653 return(in_line(m, tok, line, ppos, pos, buf));
1654
1655 /* If we're a nested call, same place. */
1656 if ( ! nl)
1657 return(in_line(m, tok, line, ppos, pos, buf));
1658
1659 /*
1660 * XXX: this will open a block scope; however, if later we end
1661 * up formatting the block scope, then child nodes will inherit
1662 * the formatting. Be careful.
1663 */
1664 if (MDOC_Nm == tok)
1665 return(blk_full(m, tok, line, ppos, pos, buf));
1666 assert(MDOC_Vt == tok);
1667 return(blk_part_imp(m, tok, line, ppos, pos, buf));
1668 }
1669
1670
1671 /* ARGSUSED */
1672 static int
1673 obsolete(MACRO_PROT_ARGS)
1674 {
1675
1676 return(mdoc_pmsg(m, line, ppos, MANDOCERR_MACROOBS));
1677 }
1678
1679
1680 /*
1681 * Phrases occur within `Bl -column' entries, separated by `Ta' or tabs.
1682 * They're unusual because they're basically free-form text until a
1683 * macro is encountered.
1684 */
1685 static int
1686 phrase(struct mdoc *m, int line, int ppos, char *buf)
1687 {
1688 int la, pos;
1689 enum margserr ac;
1690 enum mdoct ntok;
1691 char *p;
1692
1693 for (pos = ppos; ; ) {
1694 la = pos;
1695
1696 ac = mdoc_zargs(m, line, &pos, buf, 0, &p);
1697
1698 if (ARGS_ERROR == ac)
1699 return(0);
1700 if (ARGS_EOLN == ac)
1701 break;
1702
1703 ntok = ARGS_QWORD == ac ? MDOC_MAX : lookup_raw(p);
1704
1705 if (MDOC_MAX == ntok) {
1706 if ( ! mdoc_word_alloc(m, line, la, p))
1707 return(0);
1708 continue;
1709 }
1710
1711 if ( ! mdoc_macro(m, ntok, line, la, &pos, buf))
1712 return(0);
1713 return(append_delims(m, line, &pos, buf));
1714 }
1715
1716 return(1);
1717 }
1718
1719
1720 /* ARGSUSED */
1721 static int
1722 phrase_ta(MACRO_PROT_ARGS)
1723 {
1724 int la;
1725 enum mdoct ntok;
1726 enum margserr ac;
1727 char *p;
1728
1729 /*
1730 * FIXME: this is overly restrictive: if the `Ta' is unexpected,
1731 * it should simply error out with ARGSLOST.
1732 */
1733
1734 if ( ! rew_sub(MDOC_BODY, m, MDOC_It, line, ppos))
1735 return(0);
1736 if ( ! mdoc_body_alloc(m, line, ppos, MDOC_It))
1737 return(0);
1738
1739 for (;;) {
1740 la = *pos;
1741 ac = mdoc_zargs(m, line, pos, buf, 0, &p);
1742
1743 if (ARGS_ERROR == ac)
1744 return(0);
1745 if (ARGS_EOLN == ac)
1746 break;
1747
1748 ntok = ARGS_QWORD == ac ? MDOC_MAX : lookup_raw(p);
1749
1750 if (MDOC_MAX == ntok) {
1751 if ( ! mdoc_word_alloc(m, line, la, p))
1752 return(0);
1753 continue;
1754 }
1755
1756 if ( ! mdoc_macro(m, ntok, line, la, pos, buf))
1757 return(0);
1758 return(append_delims(m, line, pos, buf));
1759 }
1760
1761 return(1);
1762 }
A git conversion mirror of mandoc.bsd.lv