]> 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
Clean-up in libman: make all calls to man_*msg not check return value.
[mandoc.git] / mdoc_macro.c
1 /* $Id: mdoc_macro.c,v 1.103 2011/03/17 11:30:23 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 != mandoc_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 mdoc_pmsg(m, line, ppos, MANDOCERR_ARGSLOST);
714
715 if ( ! rew_sub(MDOC_BODY, m, tok, line, ppos))
716 return(0);
717 return(rew_sub(MDOC_BLOCK, m, tok, line, ppos));
718 }
719
720 if ( ! rew_sub(MDOC_BODY, m, tok, line, ppos))
721 return(0);
722
723 if (NULL == later && maxargs > 0)
724 if ( ! mdoc_tail_alloc(m, line, ppos, rew_alt(tok)))
725 return(0);
726
727 for (flushed = j = 0; ; j++) {
728 lastarg = *pos;
729
730 if (j == maxargs && ! flushed) {
731 if ( ! rew_sub(MDOC_BLOCK, m, tok, line, ppos))
732 return(0);
733 flushed = 1;
734 }
735
736 ac = mdoc_args(m, line, pos, buf, tok, &p);
737
738 if (ARGS_ERROR == ac)
739 return(0);
740 if (ARGS_PUNCT == ac)
741 break;
742 if (ARGS_EOLN == ac)
743 break;
744
745 ntok = ARGS_QWORD == ac ? MDOC_MAX : lookup(tok, p);
746
747 if (MDOC_MAX == ntok) {
748 if ( ! mdoc_word_alloc(m, line, lastarg, p))
749 return(0);
750 continue;
751 }
752
753 if ( ! flushed) {
754 if ( ! rew_sub(MDOC_BLOCK, m, tok, line, ppos))
755 return(0);
756 flushed = 1;
757 }
758 if ( ! mdoc_macro(m, ntok, line, lastarg, pos, buf))
759 return(0);
760 break;
761 }
762
763 if ( ! flushed && ! rew_sub(MDOC_BLOCK, m, tok, line, ppos))
764 return(0);
765
766 if ( ! nl)
767 return(1);
768 return(append_delims(m, line, pos, buf));
769 }
770
771
772 static int
773 in_line(MACRO_PROT_ARGS)
774 {
775 int la, scope, cnt, nc, nl;
776 enum margverr av;
777 enum mdoct ntok;
778 enum margserr ac;
779 enum mdelim d;
780 struct mdoc_arg *arg;
781 char *p;
782
783 nl = MDOC_NEWLINE & m->flags;
784
785 /*
786 * Whether we allow ignored elements (those without content,
787 * usually because of reserved words) to squeak by.
788 */
789
790 switch (tok) {
791 case (MDOC_An):
792 /* FALLTHROUGH */
793 case (MDOC_Ar):
794 /* FALLTHROUGH */
795 case (MDOC_Fl):
796 /* FALLTHROUGH */
797 case (MDOC_Mt):
798 /* FALLTHROUGH */
799 case (MDOC_Nm):
800 /* FALLTHROUGH */
801 case (MDOC_Pa):
802 nc = 1;
803 break;
804 default:
805 nc = 0;
806 break;
807 }
808
809 for (arg = NULL;; ) {
810 la = *pos;
811 av = mdoc_argv(m, line, tok, &arg, pos, buf);
812
813 if (ARGV_WORD == av) {
814 *pos = la;
815 break;
816 }
817 if (ARGV_EOLN == av)
818 break;
819 if (ARGV_ARG == av)
820 continue;
821
822 mdoc_argv_free(arg);
823 return(0);
824 }
825
826 for (cnt = scope = 0;; ) {
827 la = *pos;
828 ac = mdoc_args(m, line, pos, buf, tok, &p);
829
830 if (ARGS_ERROR == ac)
831 return(0);
832 if (ARGS_EOLN == ac)
833 break;
834 if (ARGS_PUNCT == ac)
835 break;
836
837 ntok = ARGS_QWORD == ac ? MDOC_MAX : lookup(tok, p);
838
839 /*
840 * In this case, we've located a submacro and must
841 * execute it. Close out scope, if open. If no
842 * elements have been generated, either create one (nc)
843 * or raise a warning.
844 */
845
846 if (MDOC_MAX != ntok) {
847 if (scope && ! rew_elem(m, tok))
848 return(0);
849 if (nc && 0 == cnt) {
850 if ( ! mdoc_elem_alloc(m, line, ppos, tok, arg))
851 return(0);
852 if ( ! rew_last(m, m->last))
853 return(0);
854 } else if ( ! nc && 0 == cnt) {
855 mdoc_argv_free(arg);
856 mdoc_pmsg(m, line, ppos, MANDOCERR_MACROEMPTY);
857 }
858
859 if ( ! mdoc_macro(m, ntok, line, la, pos, buf))
860 return(0);
861 if ( ! nl)
862 return(1);
863 return(append_delims(m, line, pos, buf));
864 }
865
866 /*
867 * Non-quote-enclosed punctuation. Set up our scope, if
868 * a word; rewind the scope, if a delimiter; then append
869 * the word.
870 */
871
872 d = ARGS_QWORD == ac ? DELIM_NONE : mandoc_isdelim(p);
873
874 if (DELIM_NONE != d) {
875 /*
876 * If we encounter closing punctuation, no word
877 * has been omitted, no scope is open, and we're
878 * allowed to have an empty element, then start
879 * a new scope. `Ar', `Fl', and `Li', only do
880 * this once per invocation. There may be more
881 * of these (all of them?).
882 */
883 if (0 == cnt && (nc || MDOC_Li == tok) &&
884 DELIM_CLOSE == d && ! scope) {
885 if ( ! mdoc_elem_alloc(m, line, ppos, tok, arg))
886 return(0);
887 if (MDOC_Ar == tok || MDOC_Li == tok ||
888 MDOC_Fl == tok)
889 cnt++;
890 scope = 1;
891 }
892 /*
893 * Close out our scope, if one is open, before
894 * any punctuation.
895 */
896 if (scope && ! rew_elem(m, tok))
897 return(0);
898 scope = 0;
899 } else if ( ! scope) {
900 if ( ! mdoc_elem_alloc(m, line, ppos, tok, arg))
901 return(0);
902 scope = 1;
903 }
904
905 if (DELIM_NONE == d)
906 cnt++;
907 if ( ! mdoc_word_alloc(m, line, la, p))
908 return(0);
909
910 /*
911 * `Fl' macros have their scope re-opened with each new
912 * word so that the `-' can be added to each one without
913 * having to parse out spaces.
914 */
915 if (scope && MDOC_Fl == tok) {
916 if ( ! rew_elem(m, tok))
917 return(0);
918 scope = 0;
919 }
920 }
921
922 if (scope && ! rew_elem(m, tok))
923 return(0);
924
925 /*
926 * If no elements have been collected and we're allowed to have
927 * empties (nc), open a scope and close it out. Otherwise,
928 * raise a warning.
929 */
930
931 if (nc && 0 == cnt) {
932 if ( ! mdoc_elem_alloc(m, line, ppos, tok, arg))
933 return(0);
934 if ( ! rew_last(m, m->last))
935 return(0);
936 } else if ( ! nc && 0 == cnt) {
937 mdoc_argv_free(arg);
938 mdoc_pmsg(m, line, ppos, MANDOCERR_MACROEMPTY);
939 }
940
941 if ( ! nl)
942 return(1);
943 return(append_delims(m, line, pos, buf));
944 }
945
946
947 static int
948 blk_full(MACRO_PROT_ARGS)
949 {
950 int la, nl;
951 struct mdoc_arg *arg;
952 struct mdoc_node *head; /* save of head macro */
953 struct mdoc_node *body; /* save of body macro */
954 struct mdoc_node *n;
955 enum mdoc_type mtt;
956 enum mdoct ntok;
957 enum margserr ac, lac;
958 enum margverr av;
959 char *p;
960
961 nl = MDOC_NEWLINE & m->flags;
962
963 /* Close out prior implicit scope. */
964
965 if ( ! (MDOC_EXPLICIT & mdoc_macros[tok].flags)) {
966 if ( ! rew_sub(MDOC_BODY, m, tok, line, ppos))
967 return(0);
968 if ( ! rew_sub(MDOC_BLOCK, m, tok, line, ppos))
969 return(0);
970 }
971
972 /*
973 * This routine accomodates implicitly- and explicitly-scoped
974 * macro openings. Implicit ones first close out prior scope
975 * (seen above). Delay opening the head until necessary to
976 * allow leading punctuation to print. Special consideration
977 * for `It -column', which has phrase-part syntax instead of
978 * regular child nodes.
979 */
980
981 for (arg = NULL;; ) {
982 la = *pos;
983 av = mdoc_argv(m, line, tok, &arg, pos, buf);
984
985 if (ARGV_WORD == av) {
986 *pos = la;
987 break;
988 }
989
990 if (ARGV_EOLN == av)
991 break;
992 if (ARGV_ARG == av)
993 continue;
994
995 mdoc_argv_free(arg);
996 return(0);
997 }
998
999 if ( ! mdoc_block_alloc(m, line, ppos, tok, arg))
1000 return(0);
1001
1002 head = body = NULL;
1003
1004 /*
1005 * The `Nd' macro has all arguments in its body: it's a hybrid
1006 * of block partial-explicit and full-implicit. Stupid.
1007 */
1008
1009 if (MDOC_Nd == tok) {
1010 if ( ! mdoc_head_alloc(m, line, ppos, tok))
1011 return(0);
1012 head = m->last;
1013 if ( ! rew_sub(MDOC_HEAD, m, tok, line, ppos))
1014 return(0);
1015 if ( ! mdoc_body_alloc(m, line, ppos, tok))
1016 return(0);
1017 body = m->last;
1018 }
1019
1020 ac = ARGS_ERROR;
1021
1022 for ( ; ; ) {
1023 la = *pos;
1024 /* Initialise last-phrase-type with ARGS_PEND. */
1025 lac = ARGS_ERROR == ac ? ARGS_PEND : ac;
1026 ac = mdoc_args(m, line, pos, buf, tok, &p);
1027
1028 if (ARGS_PUNCT == ac)
1029 break;
1030
1031 if (ARGS_ERROR == ac)
1032 return(0);
1033
1034 if (ARGS_EOLN == ac) {
1035 if (ARGS_PPHRASE != lac && ARGS_PHRASE != lac)
1036 break;
1037 /*
1038 * This is necessary: if the last token on a
1039 * line is a `Ta' or tab, then we'll get
1040 * ARGS_EOLN, so we must be smart enough to
1041 * reopen our scope if the last parse was a
1042 * phrase or partial phrase.
1043 */
1044 if ( ! rew_sub(MDOC_BODY, m, tok, line, ppos))
1045 return(0);
1046 if ( ! mdoc_body_alloc(m, line, ppos, tok))
1047 return(0);
1048 body = m->last;
1049 break;
1050 }
1051
1052 /*
1053 * Emit leading punctuation (i.e., punctuation before
1054 * the MDOC_HEAD) for non-phrase types.
1055 */
1056
1057 if (NULL == head &&
1058 ARGS_PEND != ac &&
1059 ARGS_PHRASE != ac &&
1060 ARGS_PPHRASE != ac &&
1061 ARGS_QWORD != ac &&
1062 DELIM_OPEN == mandoc_isdelim(p)) {
1063 if ( ! mdoc_word_alloc(m, line, la, p))
1064 return(0);
1065 continue;
1066 }
1067
1068 /* Open a head if one hasn't been opened. */
1069
1070 if (NULL == head) {
1071 if ( ! mdoc_head_alloc(m, line, ppos, tok))
1072 return(0);
1073 head = m->last;
1074 }
1075
1076 if (ARGS_PHRASE == ac ||
1077 ARGS_PEND == ac ||
1078 ARGS_PPHRASE == ac) {
1079 /*
1080 * If we haven't opened a body yet, rewind the
1081 * head; if we have, rewind that instead.
1082 */
1083
1084 mtt = body ? MDOC_BODY : MDOC_HEAD;
1085 if ( ! rew_sub(mtt, m, tok, line, ppos))
1086 return(0);
1087
1088 /* Then allocate our body context. */
1089
1090 if ( ! mdoc_body_alloc(m, line, ppos, tok))
1091 return(0);
1092 body = m->last;
1093
1094 /*
1095 * Process phrases: set whether we're in a
1096 * partial-phrase (this effects line handling)
1097 * then call down into the phrase parser.
1098 */
1099
1100 if (ARGS_PPHRASE == ac)
1101 m->flags |= MDOC_PPHRASE;
1102 if (ARGS_PEND == ac && ARGS_PPHRASE == lac)
1103 m->flags |= MDOC_PPHRASE;
1104
1105 if ( ! phrase(m, line, la, buf))
1106 return(0);
1107
1108 m->flags &= ~MDOC_PPHRASE;
1109 continue;
1110 }
1111
1112 ntok = ARGS_QWORD == ac ? MDOC_MAX : lookup(tok, p);
1113
1114 if (MDOC_MAX == ntok) {
1115 if ( ! mdoc_word_alloc(m, line, la, p))
1116 return(0);
1117 continue;
1118 }
1119
1120 if ( ! mdoc_macro(m, ntok, line, la, pos, buf))
1121 return(0);
1122 break;
1123 }
1124
1125 if (NULL == head) {
1126 if ( ! mdoc_head_alloc(m, line, ppos, tok))
1127 return(0);
1128 head = m->last;
1129 }
1130
1131 if (nl && ! append_delims(m, line, pos, buf))
1132 return(0);
1133
1134 /* If we've already opened our body, exit now. */
1135
1136 if (NULL != body)
1137 goto out;
1138
1139 /*
1140 * If there is an open (i.e., unvalidated) sub-block requiring
1141 * explicit close-out, postpone switching the current block from
1142 * head to body until the rew_sub() call closing out that
1143 * sub-block.
1144 */
1145 for (n = m->last; n && n != head; n = n->parent) {
1146 if (MDOC_BLOCK == n->type &&
1147 MDOC_EXPLICIT & mdoc_macros[n->tok].flags &&
1148 ! (MDOC_VALID & n->flags)) {
1149 n->pending = head;
1150 return(1);
1151 }
1152 }
1153
1154 /* Close out scopes to remain in a consistent state. */
1155
1156 if ( ! rew_sub(MDOC_HEAD, m, tok, line, ppos))
1157 return(0);
1158 if ( ! mdoc_body_alloc(m, line, ppos, tok))
1159 return(0);
1160
1161 out:
1162 if ( ! (MDOC_FREECOL & m->flags))
1163 return(1);
1164
1165 if ( ! rew_sub(MDOC_BODY, m, tok, line, ppos))
1166 return(0);
1167 if ( ! rew_sub(MDOC_BLOCK, m, tok, line, ppos))
1168 return(0);
1169
1170 m->flags &= ~MDOC_FREECOL;
1171 return(1);
1172 }
1173
1174
1175 static int
1176 blk_part_imp(MACRO_PROT_ARGS)
1177 {
1178 int la, nl;
1179 enum mdoct ntok;
1180 enum margserr ac;
1181 char *p;
1182 struct mdoc_node *blk; /* saved block context */
1183 struct mdoc_node *body; /* saved body context */
1184 struct mdoc_node *n;
1185
1186 nl = MDOC_NEWLINE & m->flags;
1187
1188 /*
1189 * A macro that spans to the end of the line. This is generally
1190 * (but not necessarily) called as the first macro. The block
1191 * has a head as the immediate child, which is always empty,
1192 * followed by zero or more opening punctuation nodes, then the
1193 * body (which may be empty, depending on the macro), then zero
1194 * or more closing punctuation nodes.
1195 */
1196
1197 if ( ! mdoc_block_alloc(m, line, ppos, tok, NULL))
1198 return(0);
1199
1200 blk = m->last;
1201
1202 if ( ! mdoc_head_alloc(m, line, ppos, tok))
1203 return(0);
1204 if ( ! rew_sub(MDOC_HEAD, m, tok, line, ppos))
1205 return(0);
1206
1207 /*
1208 * Open the body scope "on-demand", that is, after we've
1209 * processed all our the leading delimiters (open parenthesis,
1210 * etc.).
1211 */
1212
1213 for (body = NULL; ; ) {
1214 la = *pos;
1215 ac = mdoc_args(m, line, pos, buf, tok, &p);
1216
1217 if (ARGS_ERROR == ac)
1218 return(0);
1219 if (ARGS_EOLN == ac)
1220 break;
1221 if (ARGS_PUNCT == ac)
1222 break;
1223
1224 if (NULL == body && ARGS_QWORD != ac &&
1225 DELIM_OPEN == mandoc_isdelim(p)) {
1226 if ( ! mdoc_word_alloc(m, line, la, p))
1227 return(0);
1228 continue;
1229 }
1230
1231 if (NULL == body) {
1232 if ( ! mdoc_body_alloc(m, line, ppos, tok))
1233 return(0);
1234 body = m->last;
1235 }
1236
1237 ntok = ARGS_QWORD == ac ? MDOC_MAX : lookup(tok, p);
1238
1239 if (MDOC_MAX == ntok) {
1240 if ( ! mdoc_word_alloc(m, line, la, p))
1241 return(0);
1242 continue;
1243 }
1244
1245 if ( ! mdoc_macro(m, ntok, line, la, pos, buf))
1246 return(0);
1247 break;
1248 }
1249
1250 /* Clean-ups to leave in a consistent state. */
1251
1252 if (NULL == body) {
1253 if ( ! mdoc_body_alloc(m, line, ppos, tok))
1254 return(0);
1255 body = m->last;
1256 }
1257
1258 for (n = body->child; n && n->next; n = n->next)
1259 /* Do nothing. */ ;
1260
1261 /*
1262 * End of sentence spacing: if the last node is a text node and
1263 * has a trailing period, then mark it as being end-of-sentence.
1264 */
1265
1266 if (n && MDOC_TEXT == n->type && n->string)
1267 if (mandoc_eos(n->string, strlen(n->string), 1))
1268 n->flags |= MDOC_EOS;
1269
1270 /* Up-propogate the end-of-space flag. */
1271
1272 if (n && (MDOC_EOS & n->flags)) {
1273 body->flags |= MDOC_EOS;
1274 body->parent->flags |= MDOC_EOS;
1275 }
1276
1277 /*
1278 * If there is an open sub-block requiring explicit close-out,
1279 * postpone closing out the current block
1280 * until the rew_sub() call closing out the sub-block.
1281 */
1282 for (n = m->last; n && n != body && n != blk->parent; n = n->parent) {
1283 if (MDOC_BLOCK == n->type &&
1284 MDOC_EXPLICIT & mdoc_macros[n->tok].flags &&
1285 ! (MDOC_VALID & n->flags)) {
1286 make_pending(n, tok, m, line, ppos);
1287 if ( ! mdoc_endbody_alloc(m, line, ppos,
1288 tok, body, ENDBODY_NOSPACE))
1289 return(0);
1290 return(1);
1291 }
1292 }
1293
1294 /*
1295 * If we can't rewind to our body, then our scope has already
1296 * been closed by another macro (like `Oc' closing `Op'). This
1297 * is ugly behaviour nodding its head to OpenBSD's overwhelming
1298 * crufty use of `Op' breakage.
1299 */
1300 if (n != body)
1301 mdoc_vmsg(m, MANDOCERR_SCOPENEST, line, ppos,
1302 "%s broken", mdoc_macronames[tok]);
1303
1304 if (n && ! rew_sub(MDOC_BODY, m, tok, line, ppos))
1305 return(0);
1306
1307 /* Standard appending of delimiters. */
1308
1309 if (nl && ! append_delims(m, line, pos, buf))
1310 return(0);
1311
1312 /* Rewind scope, if applicable. */
1313
1314 if (n && ! rew_sub(MDOC_BLOCK, m, tok, line, ppos))
1315 return(0);
1316
1317 return(1);
1318 }
1319
1320
1321 static int
1322 blk_part_exp(MACRO_PROT_ARGS)
1323 {
1324 int la, nl;
1325 enum margserr ac;
1326 struct mdoc_node *head; /* keep track of head */
1327 struct mdoc_node *body; /* keep track of body */
1328 char *p;
1329 enum mdoct ntok;
1330
1331 nl = MDOC_NEWLINE & m->flags;
1332
1333 /*
1334 * The opening of an explicit macro having zero or more leading
1335 * punctuation nodes; a head with optional single element (the
1336 * case of `Eo'); and a body that may be empty.
1337 */
1338
1339 if ( ! mdoc_block_alloc(m, line, ppos, tok, NULL))
1340 return(0);
1341
1342 for (head = body = NULL; ; ) {
1343 la = *pos;
1344 ac = mdoc_args(m, line, pos, buf, tok, &p);
1345
1346 if (ARGS_ERROR == ac)
1347 return(0);
1348 if (ARGS_PUNCT == ac)
1349 break;
1350 if (ARGS_EOLN == ac)
1351 break;
1352
1353 /* Flush out leading punctuation. */
1354
1355 if (NULL == head && ARGS_QWORD != ac &&
1356 DELIM_OPEN == mandoc_isdelim(p)) {
1357 assert(NULL == body);
1358 if ( ! mdoc_word_alloc(m, line, la, p))
1359 return(0);
1360 continue;
1361 }
1362
1363 if (NULL == head) {
1364 assert(NULL == body);
1365 if ( ! mdoc_head_alloc(m, line, ppos, tok))
1366 return(0);
1367 head = m->last;
1368 }
1369
1370 /*
1371 * `Eo' gobbles any data into the head, but most other
1372 * macros just immediately close out and begin the body.
1373 */
1374
1375 if (NULL == body) {
1376 assert(head);
1377 /* No check whether it's a macro! */
1378 if (MDOC_Eo == tok)
1379 if ( ! mdoc_word_alloc(m, line, la, p))
1380 return(0);
1381
1382 if ( ! rew_sub(MDOC_HEAD, m, tok, line, ppos))
1383 return(0);
1384 if ( ! mdoc_body_alloc(m, line, ppos, tok))
1385 return(0);
1386 body = m->last;
1387
1388 if (MDOC_Eo == tok)
1389 continue;
1390 }
1391
1392 assert(NULL != head && NULL != body);
1393
1394 ntok = ARGS_QWORD == ac ? MDOC_MAX : lookup(tok, p);
1395
1396 if (MDOC_MAX == ntok) {
1397 if ( ! mdoc_word_alloc(m, line, la, p))
1398 return(0);
1399 continue;
1400 }
1401
1402 if ( ! mdoc_macro(m, ntok, line, la, pos, buf))
1403 return(0);
1404 break;
1405 }
1406
1407 /* Clean-up to leave in a consistent state. */
1408
1409 if (NULL == head) {
1410 if ( ! mdoc_head_alloc(m, line, ppos, tok))
1411 return(0);
1412 head = m->last;
1413 }
1414
1415 if (NULL == body) {
1416 if ( ! rew_sub(MDOC_HEAD, m, tok, line, ppos))
1417 return(0);
1418 if ( ! mdoc_body_alloc(m, line, ppos, tok))
1419 return(0);
1420 body = m->last;
1421 }
1422
1423 /* Standard appending of delimiters. */
1424
1425 if ( ! nl)
1426 return(1);
1427 return(append_delims(m, line, pos, buf));
1428 }
1429
1430
1431 /* ARGSUSED */
1432 static int
1433 in_line_argn(MACRO_PROT_ARGS)
1434 {
1435 int la, flushed, j, maxargs, nl;
1436 enum margserr ac;
1437 enum margverr av;
1438 struct mdoc_arg *arg;
1439 char *p;
1440 enum mdoct ntok;
1441
1442 nl = MDOC_NEWLINE & m->flags;
1443
1444 /*
1445 * A line macro that has a fixed number of arguments (maxargs).
1446 * Only open the scope once the first non-leading-punctuation is
1447 * found (unless MDOC_IGNDELIM is noted, like in `Pf'), then
1448 * keep it open until the maximum number of arguments are
1449 * exhausted.
1450 */
1451
1452 switch (tok) {
1453 case (MDOC_Ap):
1454 /* FALLTHROUGH */
1455 case (MDOC_No):
1456 /* FALLTHROUGH */
1457 case (MDOC_Ns):
1458 /* FALLTHROUGH */
1459 case (MDOC_Ux):
1460 maxargs = 0;
1461 break;
1462 case (MDOC_Bx):
1463 /* FALLTHROUGH */
1464 case (MDOC_Xr):
1465 maxargs = 2;
1466 break;
1467 default:
1468 maxargs = 1;
1469 break;
1470 }
1471
1472 for (arg = NULL; ; ) {
1473 la = *pos;
1474 av = mdoc_argv(m, line, tok, &arg, pos, buf);
1475
1476 if (ARGV_WORD == av) {
1477 *pos = la;
1478 break;
1479 }
1480
1481 if (ARGV_EOLN == av)
1482 break;
1483 if (ARGV_ARG == av)
1484 continue;
1485
1486 mdoc_argv_free(arg);
1487 return(0);
1488 }
1489
1490 for (flushed = j = 0; ; ) {
1491 la = *pos;
1492 ac = mdoc_args(m, line, pos, buf, tok, &p);
1493
1494 if (ARGS_ERROR == ac)
1495 return(0);
1496 if (ARGS_PUNCT == ac)
1497 break;
1498 if (ARGS_EOLN == ac)
1499 break;
1500
1501 if ( ! (MDOC_IGNDELIM & mdoc_macros[tok].flags) &&
1502 ARGS_QWORD != ac &&
1503 0 == j && DELIM_OPEN == mandoc_isdelim(p)) {
1504 if ( ! mdoc_word_alloc(m, line, la, p))
1505 return(0);
1506 continue;
1507 } else if (0 == j)
1508 if ( ! mdoc_elem_alloc(m, line, la, tok, arg))
1509 return(0);
1510
1511 if (j == maxargs && ! flushed) {
1512 if ( ! rew_elem(m, tok))
1513 return(0);
1514 flushed = 1;
1515 }
1516
1517 ntok = ARGS_QWORD == ac ? MDOC_MAX : lookup(tok, p);
1518
1519 if (MDOC_MAX != ntok) {
1520 if ( ! flushed && ! rew_elem(m, tok))
1521 return(0);
1522 flushed = 1;
1523 if ( ! mdoc_macro(m, ntok, line, la, pos, buf))
1524 return(0);
1525 j++;
1526 break;
1527 }
1528
1529 if ( ! (MDOC_IGNDELIM & mdoc_macros[tok].flags) &&
1530 ARGS_QWORD != ac &&
1531 ! flushed &&
1532 DELIM_NONE != mandoc_isdelim(p)) {
1533 if ( ! rew_elem(m, tok))
1534 return(0);
1535 flushed = 1;
1536 }
1537
1538 /*
1539 * XXX: this is a hack to work around groff's ugliness
1540 * as regards `Xr' and extraneous arguments. It should
1541 * ideally be deprecated behaviour, but because this is
1542 * code is no here, it's unlikely to be removed.
1543 */
1544
1545 #ifdef __OpenBSD__
1546 if (MDOC_Xr == tok && j == maxargs) {
1547 if ( ! mdoc_elem_alloc(m, line, la, MDOC_Ns, NULL))
1548 return(0);
1549 if ( ! rew_elem(m, MDOC_Ns))
1550 return(0);
1551 }
1552 #endif
1553
1554 if ( ! mdoc_word_alloc(m, line, la, p))
1555 return(0);
1556 j++;
1557 }
1558
1559 if (0 == j && ! mdoc_elem_alloc(m, line, la, tok, arg))
1560 return(0);
1561
1562 /* Close out in a consistent state. */
1563
1564 if ( ! flushed && ! rew_elem(m, tok))
1565 return(0);
1566 if ( ! nl)
1567 return(1);
1568 return(append_delims(m, line, pos, buf));
1569 }
1570
1571
1572 static int
1573 in_line_eoln(MACRO_PROT_ARGS)
1574 {
1575 int la;
1576 enum margserr ac;
1577 enum margverr av;
1578 struct mdoc_arg *arg;
1579 char *p;
1580 enum mdoct ntok;
1581
1582 assert( ! (MDOC_PARSED & mdoc_macros[tok].flags));
1583
1584 if (tok == MDOC_Pp)
1585 rew_sub(MDOC_BLOCK, m, MDOC_Nm, line, ppos);
1586
1587 /* Parse macro arguments. */
1588
1589 for (arg = NULL; ; ) {
1590 la = *pos;
1591 av = mdoc_argv(m, line, tok, &arg, pos, buf);
1592
1593 if (ARGV_WORD == av) {
1594 *pos = la;
1595 break;
1596 }
1597 if (ARGV_EOLN == av)
1598 break;
1599 if (ARGV_ARG == av)
1600 continue;
1601
1602 mdoc_argv_free(arg);
1603 return(0);
1604 }
1605
1606 /* Open element scope. */
1607
1608 if ( ! mdoc_elem_alloc(m, line, ppos, tok, arg))
1609 return(0);
1610
1611 /* Parse argument terms. */
1612
1613 for (;;) {
1614 la = *pos;
1615 ac = mdoc_args(m, line, pos, buf, tok, &p);
1616
1617 if (ARGS_ERROR == ac)
1618 return(0);
1619 if (ARGS_EOLN == ac)
1620 break;
1621
1622 ntok = ARGS_QWORD == ac ? MDOC_MAX : lookup(tok, p);
1623
1624 if (MDOC_MAX == ntok) {
1625 if ( ! mdoc_word_alloc(m, line, la, p))
1626 return(0);
1627 continue;
1628 }
1629
1630 if ( ! rew_elem(m, tok))
1631 return(0);
1632 return(mdoc_macro(m, ntok, line, la, pos, buf));
1633 }
1634
1635 /* Close out (no delimiters). */
1636
1637 return(rew_elem(m, tok));
1638 }
1639
1640
1641 /* ARGSUSED */
1642 static int
1643 ctx_synopsis(MACRO_PROT_ARGS)
1644 {
1645 int nl;
1646
1647 nl = MDOC_NEWLINE & m->flags;
1648
1649 /* If we're not in the SYNOPSIS, go straight to in-line. */
1650 if ( ! (MDOC_SYNOPSIS & m->flags))
1651 return(in_line(m, tok, line, ppos, pos, buf));
1652
1653 /* If we're a nested call, same place. */
1654 if ( ! nl)
1655 return(in_line(m, tok, line, ppos, pos, buf));
1656
1657 /*
1658 * XXX: this will open a block scope; however, if later we end
1659 * up formatting the block scope, then child nodes will inherit
1660 * the formatting. Be careful.
1661 */
1662 if (MDOC_Nm == tok)
1663 return(blk_full(m, tok, line, ppos, pos, buf));
1664 assert(MDOC_Vt == tok);
1665 return(blk_part_imp(m, tok, line, ppos, pos, buf));
1666 }
1667
1668
1669 /* ARGSUSED */
1670 static int
1671 obsolete(MACRO_PROT_ARGS)
1672 {
1673
1674 mdoc_pmsg(m, line, ppos, MANDOCERR_MACROOBS);
1675 return(1);
1676 }
1677
1678
1679 /*
1680 * Phrases occur within `Bl -column' entries, separated by `Ta' or tabs.
1681 * They're unusual because they're basically free-form text until a
1682 * macro is encountered.
1683 */
1684 static int
1685 phrase(struct mdoc *m, int line, int ppos, char *buf)
1686 {
1687 int la, pos;
1688 enum margserr ac;
1689 enum mdoct ntok;
1690 char *p;
1691
1692 for (pos = ppos; ; ) {
1693 la = pos;
1694
1695 ac = mdoc_zargs(m, line, &pos, buf, 0, &p);
1696
1697 if (ARGS_ERROR == ac)
1698 return(0);
1699 if (ARGS_EOLN == ac)
1700 break;
1701
1702 ntok = ARGS_QWORD == ac ? MDOC_MAX : lookup_raw(p);
1703
1704 if (MDOC_MAX == ntok) {
1705 if ( ! mdoc_word_alloc(m, line, la, p))
1706 return(0);
1707 continue;
1708 }
1709
1710 if ( ! mdoc_macro(m, ntok, line, la, &pos, buf))
1711 return(0);
1712 return(append_delims(m, line, &pos, buf));
1713 }
1714
1715 return(1);
1716 }
1717
1718
1719 /* ARGSUSED */
1720 static int
1721 phrase_ta(MACRO_PROT_ARGS)
1722 {
1723 int la;
1724 enum mdoct ntok;
1725 enum margserr ac;
1726 char *p;
1727
1728 /*
1729 * FIXME: this is overly restrictive: if the `Ta' is unexpected,
1730 * it should simply error out with ARGSLOST.
1731 */
1732
1733 if ( ! rew_sub(MDOC_BODY, m, MDOC_It, line, ppos))
1734 return(0);
1735 if ( ! mdoc_body_alloc(m, line, ppos, MDOC_It))
1736 return(0);
1737
1738 for (;;) {
1739 la = *pos;
1740 ac = mdoc_zargs(m, line, pos, buf, 0, &p);
1741
1742 if (ARGS_ERROR == ac)
1743 return(0);
1744 if (ARGS_EOLN == ac)
1745 break;
1746
1747 ntok = ARGS_QWORD == ac ? MDOC_MAX : lookup_raw(p);
1748
1749 if (MDOC_MAX == ntok) {
1750 if ( ! mdoc_word_alloc(m, line, la, p))
1751 return(0);
1752 continue;
1753 }
1754
1755 if ( ! mdoc_macro(m, ntok, line, la, pos, buf))
1756 return(0);
1757 return(append_delims(m, line, pos, buf));
1758 }
1759
1760 return(1);
1761 }
A git conversion mirror of mandoc.bsd.lv