]> 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 some tight spots in mandoc's default mode: pessimistically
[mandoc.git] / mdoc_macro.c
1 /* $Id: mdoc_macro.c,v 1.93 2010/07/18 17:00:26 schwarze Exp $ */
2 /*
3 * Copyright (c) 2008, 2009, 2010 Kristaps Dzonsons <kristaps@bsd.lv>
4 * Copyright (c) 2010 Ingo Schwarze <schwarze@openbsd.org>
5 *
6 * Permission to use, copy, modify, and distribute this software for any
7 * purpose with or without fee is hereby granted, provided that the above
8 * copyright notice and this permission notice appear in all copies.
9 *
10 * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
11 * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
12 * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
13 * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
14 * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
15 * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
16 * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
17 */
18 #ifdef HAVE_CONFIG_H
19 #include "config.h"
20 #endif
21
22 #include <assert.h>
23 #include <ctype.h>
24 #include <stdlib.h>
25 #include <stdio.h>
26 #include <string.h>
27 #include <time.h>
28
29 #include "mandoc.h"
30 #include "libmdoc.h"
31 #include "libmandoc.h"
32
33 enum rew { /* see rew_dohalt() */
34 REWIND_NONE,
35 REWIND_THIS,
36 REWIND_MORE,
37 REWIND_LATER,
38 REWIND_ERROR
39 };
40
41 static int blk_full(MACRO_PROT_ARGS);
42 static int blk_exp_close(MACRO_PROT_ARGS);
43 static int blk_part_exp(MACRO_PROT_ARGS);
44 static int blk_part_imp(MACRO_PROT_ARGS);
45 static int ctx_synopsis(MACRO_PROT_ARGS);
46 static int in_line_eoln(MACRO_PROT_ARGS);
47 static int in_line_argn(MACRO_PROT_ARGS);
48 static int in_line(MACRO_PROT_ARGS);
49 static int obsolete(MACRO_PROT_ARGS);
50 static int phrase_ta(MACRO_PROT_ARGS);
51
52 static int append_delims(struct mdoc *,
53 int, int *, char *);
54 static enum mdoct lookup(enum mdoct, const char *);
55 static enum mdoct lookup_raw(const char *);
56 static int make_pending(struct mdoc_node *, enum mdoct,
57 struct mdoc *, int, int);
58 static int phrase(struct mdoc *, int, int, char *);
59 static enum mdoct rew_alt(enum mdoct);
60 static enum rew rew_dohalt(enum mdoct, enum mdoc_type,
61 const struct mdoc_node *);
62 static int rew_elem(struct mdoc *, enum mdoct);
63 static int rew_last(struct mdoc *,
64 const struct mdoc_node *);
65 static int rew_sub(enum mdoc_type, struct mdoc *,
66 enum mdoct, int, int);
67
68 const struct mdoc_macro __mdoc_macros[MDOC_MAX] = {
69 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* Ap */
70 { in_line_eoln, MDOC_PROLOGUE }, /* Dd */
71 { in_line_eoln, MDOC_PROLOGUE }, /* Dt */
72 { in_line_eoln, MDOC_PROLOGUE }, /* Os */
73 { blk_full, 0 }, /* Sh */
74 { blk_full, 0 }, /* Ss */
75 { in_line_eoln, 0 }, /* Pp */
76 { blk_part_imp, MDOC_PARSED }, /* D1 */
77 { blk_part_imp, MDOC_PARSED }, /* Dl */
78 { blk_full, MDOC_EXPLICIT }, /* Bd */
79 { blk_exp_close, MDOC_EXPLICIT }, /* Ed */
80 { blk_full, MDOC_EXPLICIT }, /* Bl */
81 { blk_exp_close, MDOC_EXPLICIT }, /* El */
82 { blk_full, MDOC_PARSED }, /* It */
83 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Ad */
84 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* An */
85 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Ar */
86 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Cd */
87 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Cm */
88 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Dv */
89 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Er */
90 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Ev */
91 { in_line_eoln, 0 }, /* Ex */
92 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Fa */
93 { in_line_eoln, 0 }, /* Fd */
94 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Fl */
95 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Fn */
96 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Ft */
97 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Ic */
98 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* In */
99 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Li */
100 { blk_full, 0 }, /* Nd */
101 { ctx_synopsis, MDOC_CALLABLE | MDOC_PARSED }, /* Nm */
102 { blk_part_imp, MDOC_CALLABLE | MDOC_PARSED }, /* Op */
103 { obsolete, 0 }, /* Ot */
104 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Pa */
105 { in_line_eoln, 0 }, /* Rv */
106 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* St */
107 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Va */
108 { ctx_synopsis, MDOC_CALLABLE | MDOC_PARSED }, /* Vt */
109 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* Xr */
110 { in_line_eoln, 0 }, /* %A */
111 { in_line_eoln, 0 }, /* %B */
112 { in_line_eoln, 0 }, /* %D */
113 { in_line_eoln, 0 }, /* %I */
114 { in_line_eoln, 0 }, /* %J */
115 { in_line_eoln, 0 }, /* %N */
116 { in_line_eoln, 0 }, /* %O */
117 { in_line_eoln, 0 }, /* %P */
118 { in_line_eoln, 0 }, /* %R */
119 { in_line_eoln, 0 }, /* %T */
120 { in_line_eoln, 0 }, /* %V */
121 { blk_exp_close, MDOC_EXPLICIT | MDOC_CALLABLE | MDOC_PARSED }, /* Ac */
122 { blk_part_exp, MDOC_CALLABLE | MDOC_PARSED | MDOC_EXPLICIT }, /* Ao */
123 { blk_part_imp, MDOC_CALLABLE | MDOC_PARSED }, /* Aq */
124 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* At */
125 { blk_exp_close, MDOC_EXPLICIT | MDOC_CALLABLE | MDOC_PARSED }, /* Bc */
126 { blk_full, MDOC_EXPLICIT }, /* Bf */
127 { blk_part_exp, MDOC_CALLABLE | MDOC_PARSED | MDOC_EXPLICIT }, /* Bo */
128 { blk_part_imp, MDOC_CALLABLE | MDOC_PARSED }, /* Bq */
129 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* Bsx */
130 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* Bx */
131 { in_line_eoln, 0 }, /* Db */
132 { blk_exp_close, MDOC_EXPLICIT | MDOC_CALLABLE | MDOC_PARSED }, /* Dc */
133 { blk_part_exp, MDOC_CALLABLE | MDOC_PARSED | MDOC_EXPLICIT }, /* Do */
134 { blk_part_imp, MDOC_CALLABLE | MDOC_PARSED }, /* Dq */
135 { blk_exp_close, MDOC_EXPLICIT | MDOC_CALLABLE | MDOC_PARSED }, /* Ec */
136 { blk_exp_close, MDOC_EXPLICIT }, /* Ef */
137 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Em */
138 { blk_part_exp, MDOC_CALLABLE | MDOC_PARSED | MDOC_EXPLICIT }, /* Eo */
139 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* Fx */
140 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Ms */
141 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* No */
142 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* Ns */
143 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* Nx */
144 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* Ox */
145 { blk_exp_close, MDOC_EXPLICIT | MDOC_CALLABLE | MDOC_PARSED }, /* Pc */
146 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED | MDOC_IGNDELIM }, /* Pf */
147 { blk_part_exp, MDOC_CALLABLE | MDOC_PARSED | MDOC_EXPLICIT }, /* Po */
148 { blk_part_imp, MDOC_CALLABLE | MDOC_PARSED }, /* Pq */
149 { blk_exp_close, MDOC_EXPLICIT | MDOC_CALLABLE | MDOC_PARSED }, /* Qc */
150 { blk_part_imp, MDOC_CALLABLE | MDOC_PARSED }, /* Ql */
151 { blk_part_exp, MDOC_CALLABLE | MDOC_PARSED | MDOC_EXPLICIT }, /* Qo */
152 { blk_part_imp, MDOC_CALLABLE | MDOC_PARSED }, /* Qq */
153 { blk_exp_close, MDOC_EXPLICIT }, /* Re */
154 { blk_full, MDOC_EXPLICIT }, /* Rs */
155 { blk_exp_close, MDOC_EXPLICIT | MDOC_CALLABLE | MDOC_PARSED }, /* Sc */
156 { blk_part_exp, MDOC_CALLABLE | MDOC_PARSED | MDOC_EXPLICIT }, /* So */
157 { blk_part_imp, MDOC_CALLABLE | MDOC_PARSED }, /* Sq */
158 { in_line_eoln, 0 }, /* Sm */
159 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Sx */
160 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Sy */
161 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Tn */
162 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* Ux */
163 { blk_exp_close, MDOC_EXPLICIT | MDOC_CALLABLE | MDOC_PARSED }, /* Xc */
164 { blk_part_exp, MDOC_CALLABLE | MDOC_PARSED | MDOC_EXPLICIT }, /* Xo */
165 { blk_full, MDOC_EXPLICIT | MDOC_CALLABLE }, /* Fo */
166 { blk_exp_close, MDOC_EXPLICIT | MDOC_CALLABLE | MDOC_PARSED }, /* Fc */
167 { blk_part_exp, MDOC_CALLABLE | MDOC_PARSED | MDOC_EXPLICIT }, /* Oo */
168 { blk_exp_close, MDOC_EXPLICIT | MDOC_CALLABLE | MDOC_PARSED }, /* Oc */
169 { blk_full, MDOC_EXPLICIT }, /* Bk */
170 { blk_exp_close, MDOC_EXPLICIT }, /* Ek */
171 { in_line_eoln, 0 }, /* Bt */
172 { in_line_eoln, 0 }, /* Hf */
173 { obsolete, 0 }, /* Fr */
174 { in_line_eoln, 0 }, /* Ud */
175 { in_line, 0 }, /* Lb */
176 { in_line_eoln, 0 }, /* Lp */
177 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Lk */
178 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Mt */
179 { blk_part_imp, MDOC_CALLABLE | MDOC_PARSED }, /* Brq */
180 { blk_part_exp, MDOC_CALLABLE | MDOC_PARSED | MDOC_EXPLICIT }, /* Bro */
181 { blk_exp_close, MDOC_EXPLICIT | MDOC_CALLABLE | MDOC_PARSED }, /* Brc */
182 { in_line_eoln, 0 }, /* %C */
183 { obsolete, 0 }, /* Es */
184 { obsolete, 0 }, /* En */
185 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* Dx */
186 { in_line_eoln, 0 }, /* %Q */
187 { in_line_eoln, 0 }, /* br */
188 { in_line_eoln, 0 }, /* sp */
189 { in_line_eoln, 0 }, /* %U */
190 { phrase_ta, MDOC_CALLABLE | MDOC_PARSED }, /* Ta */
191 };
192
193 const struct mdoc_macro * const mdoc_macros = __mdoc_macros;
194
195
196 /*
197 * This is called at the end of parsing. It must traverse up the tree,
198 * closing out open [implicit] scopes. Obviously, open explicit scopes
199 * are errors.
200 */
201 int
202 mdoc_macroend(struct mdoc *m)
203 {
204 struct mdoc_node *n;
205
206 /* Scan for open explicit scopes. */
207
208 n = MDOC_VALID & m->last->flags ? m->last->parent : m->last;
209
210 for ( ; n; n = n->parent) {
211 if (MDOC_BLOCK != n->type)
212 continue;
213 if ( ! (MDOC_EXPLICIT & mdoc_macros[n->tok].flags))
214 continue;
215 mdoc_nmsg(m, n, MANDOCERR_SYNTSCOPE);
216 return(0);
217 }
218
219 /* Rewind to the first. */
220
221 return(rew_last(m, m->first));
222 }
223
224
225 /*
226 * Look up a macro from within a subsequent context.
227 */
228 static enum mdoct
229 lookup(enum mdoct from, const char *p)
230 {
231 /* FIXME: make -diag lists be un-PARSED. */
232
233 if ( ! (MDOC_PARSED & mdoc_macros[from].flags))
234 return(MDOC_MAX);
235 return(lookup_raw(p));
236 }
237
238
239 /*
240 * Lookup a macro following the initial line macro.
241 */
242 static enum mdoct
243 lookup_raw(const char *p)
244 {
245 enum mdoct res;
246
247 if (MDOC_MAX == (res = mdoc_hash_find(p)))
248 return(MDOC_MAX);
249 if (MDOC_CALLABLE & mdoc_macros[res].flags)
250 return(res);
251 return(MDOC_MAX);
252 }
253
254
255 static int
256 rew_last(struct mdoc *mdoc, const struct mdoc_node *to)
257 {
258
259 assert(to);
260 mdoc->next = MDOC_NEXT_SIBLING;
261
262 /* LINTED */
263 while (mdoc->last != to) {
264 if ( ! mdoc_valid_post(mdoc))
265 return(0);
266 if ( ! mdoc_action_post(mdoc))
267 return(0);
268 mdoc->last = mdoc->last->parent;
269 assert(mdoc->last);
270 }
271
272 if ( ! mdoc_valid_post(mdoc))
273 return(0);
274 return(mdoc_action_post(mdoc));
275 }
276
277
278 /*
279 * For a block closing macro, return the corresponding opening one.
280 * Otherwise, return the macro itself.
281 */
282 static enum mdoct
283 rew_alt(enum mdoct tok)
284 {
285 switch (tok) {
286 case (MDOC_Ac):
287 return(MDOC_Ao);
288 case (MDOC_Bc):
289 return(MDOC_Bo);
290 case (MDOC_Brc):
291 return(MDOC_Bro);
292 case (MDOC_Dc):
293 return(MDOC_Do);
294 case (MDOC_Ec):
295 return(MDOC_Eo);
296 case (MDOC_Ed):
297 return(MDOC_Bd);
298 case (MDOC_Ef):
299 return(MDOC_Bf);
300 case (MDOC_Ek):
301 return(MDOC_Bk);
302 case (MDOC_El):
303 return(MDOC_Bl);
304 case (MDOC_Fc):
305 return(MDOC_Fo);
306 case (MDOC_Oc):
307 return(MDOC_Oo);
308 case (MDOC_Pc):
309 return(MDOC_Po);
310 case (MDOC_Qc):
311 return(MDOC_Qo);
312 case (MDOC_Re):
313 return(MDOC_Rs);
314 case (MDOC_Sc):
315 return(MDOC_So);
316 case (MDOC_Xc):
317 return(MDOC_Xo);
318 default:
319 return(tok);
320 }
321 /* NOTREACHED */
322 }
323
324
325 /*
326 * Rewinding to tok, how do we have to handle *p?
327 * REWIND_NONE: *p would delimit tok, but no tok scope is open
328 * inside *p, so there is no need to rewind anything at all.
329 * REWIND_THIS: *p matches tok, so rewind *p and nothing else.
330 * REWIND_MORE: *p is implicit, rewind it and keep searching for tok.
331 * REWIND_LATER: *p is explicit and still open, postpone rewinding.
332 * REWIND_ERROR: No tok block is open at all.
333 */
334 static enum rew
335 rew_dohalt(enum mdoct tok, enum mdoc_type type,
336 const struct mdoc_node *p)
337 {
338
339 /*
340 * No matching token, no delimiting block, no broken block.
341 * This can happen when full implicit macros are called for
342 * the first time but try to rewind their previous
343 * instance anyway.
344 */
345 if (MDOC_ROOT == p->type)
346 return(MDOC_BLOCK == type &&
347 MDOC_EXPLICIT & mdoc_macros[tok].flags ?
348 REWIND_ERROR : REWIND_NONE);
349
350 /*
351 * When starting to rewind, skip plain text
352 * and nodes that have already been rewound.
353 */
354 if (MDOC_TEXT == p->type || MDOC_VALID & p->flags)
355 return(REWIND_MORE);
356
357 /*
358 * The easiest case: Found a matching token.
359 * This applies to both blocks and elements.
360 */
361 tok = rew_alt(tok);
362 if (tok == p->tok)
363 return(p->end ? REWIND_NONE :
364 type == p->type ? REWIND_THIS : REWIND_MORE);
365
366 /*
367 * While elements do require rewinding for themselves,
368 * they never affect rewinding of other nodes.
369 */
370 if (MDOC_ELEM == p->type)
371 return(REWIND_MORE);
372
373 /*
374 * Blocks delimited by our target token get REWIND_MORE.
375 * Blocks delimiting our target token get REWIND_NONE.
376 */
377 switch (tok) {
378 case (MDOC_Bl):
379 if (MDOC_It == p->tok)
380 return(REWIND_MORE);
381 break;
382 case (MDOC_It):
383 if (MDOC_BODY == p->type && MDOC_Bl == p->tok)
384 return(REWIND_NONE);
385 break;
386 /*
387 * XXX Badly nested block handling still fails badly
388 * when one block is breaking two blocks of the same type.
389 * This is an incomplete and extremely ugly workaround,
390 * required to let the OpenBSD tree build.
391 */
392 case (MDOC_Oo):
393 if (MDOC_Op == p->tok)
394 return(REWIND_MORE);
395 break;
396 case (MDOC_Nm):
397 return(REWIND_NONE);
398 case (MDOC_Nd):
399 /* FALLTHROUGH */
400 case (MDOC_Ss):
401 if (MDOC_BODY == p->type && MDOC_Sh == p->tok)
402 return(REWIND_NONE);
403 /* FALLTHROUGH */
404 case (MDOC_Sh):
405 if (MDOC_Nd == p->tok || MDOC_Ss == p->tok ||
406 MDOC_Sh == p->tok)
407 return(REWIND_MORE);
408 break;
409 default:
410 break;
411 }
412
413 /*
414 * Default block rewinding rules.
415 * In particular, always skip block end markers,
416 * and let all blocks rewind Nm children.
417 */
418 if (ENDBODY_NOT != p->end || MDOC_Nm == p->tok ||
419 (MDOC_BLOCK == p->type &&
420 ! (MDOC_EXPLICIT & mdoc_macros[tok].flags)))
421 return(REWIND_MORE);
422
423 /*
424 * Partial blocks allow delayed rewinding by default.
425 */
426 if (&blk_full != mdoc_macros[tok].fp)
427 return (REWIND_LATER);
428
429 /*
430 * Full blocks can only be rewound when matching
431 * or when there is an explicit rule.
432 */
433 return(REWIND_ERROR);
434 }
435
436
437 static int
438 rew_elem(struct mdoc *mdoc, enum mdoct tok)
439 {
440 struct mdoc_node *n;
441
442 n = mdoc->last;
443 if (MDOC_ELEM != n->type)
444 n = n->parent;
445 assert(MDOC_ELEM == n->type);
446 assert(tok == n->tok);
447
448 return(rew_last(mdoc, n));
449 }
450
451
452 /*
453 * We are trying to close a block identified by tok,
454 * but the child block *broken is still open.
455 * Thus, postpone closing the tok block
456 * until the rew_sub call closing *broken.
457 */
458 static int
459 make_pending(struct mdoc_node *broken, enum mdoct tok,
460 struct mdoc *m, int line, int ppos)
461 {
462 struct mdoc_node *breaker;
463
464 /*
465 * Iterate backwards, searching for the block matching tok,
466 * that is, the block breaking the *broken block.
467 */
468 for (breaker = broken->parent; breaker; breaker = breaker->parent) {
469
470 /*
471 * If the *broken block had already been broken before
472 * and we encounter its breaker, make the tok block
473 * pending on the inner breaker.
474 * Graphically, "[A breaker=[B broken=[C->B B] tok=A] C]"
475 * becomes "[A broken=[B [C->B B] tok=A] C]"
476 * and finally "[A [B->A [C->B B] A] C]".
477 */
478 if (breaker == broken->pending) {
479 broken = breaker;
480 continue;
481 }
482
483 if (REWIND_THIS != rew_dohalt(tok, MDOC_BLOCK, breaker))
484 continue;
485 if (MDOC_BODY == broken->type)
486 broken = broken->parent;
487
488 /*
489 * Found the breaker.
490 * If another, outer breaker is already pending on
491 * the *broken block, we must not clobber the link
492 * to the outer breaker, but make it pending on the
493 * new, now inner breaker.
494 * Graphically, "[A breaker=[B broken=[C->A A] tok=B] C]"
495 * becomes "[A breaker=[B->A broken=[C A] tok=B] C]"
496 * and finally "[A [B->A [C->B A] B] C]".
497 */
498 if (broken->pending) {
499 struct mdoc_node *taker;
500
501 /*
502 * If the breaker had also been broken before,
503 * it cannot take on the outer breaker itself,
504 * but must hand it on to its own breakers.
505 * Graphically, this is the following situation:
506 * "[A [B breaker=[C->B B] broken=[D->A A] tok=C] D]"
507 * "[A taker=[B->A breaker=[C->B B] [D->C A] C] D]"
508 */
509 taker = breaker;
510 while (taker->pending)
511 taker = taker->pending;
512 taker->pending = broken->pending;
513 }
514 broken->pending = breaker;
515 mdoc_vmsg(m, MANDOCERR_SCOPENEST, line, ppos,
516 "%s breaks %s", mdoc_macronames[tok],
517 mdoc_macronames[broken->tok]);
518 return(1);
519 }
520
521 /*
522 * Found no matching block for tok.
523 * Are you trying to close a block that is not open?
524 * XXX Make this non-fatal.
525 */
526 mdoc_pmsg(m, line, ppos, MANDOCERR_SYNTNOSCOPE);
527 return(0);
528 }
529
530
531 static int
532 rew_sub(enum mdoc_type t, struct mdoc *m,
533 enum mdoct tok, int line, int ppos)
534 {
535 struct mdoc_node *n;
536
537 n = m->last;
538 while (n) {
539 switch (rew_dohalt(tok, t, n)) {
540 case (REWIND_NONE):
541 return(1);
542 case (REWIND_THIS):
543 break;
544 case (REWIND_MORE):
545 n = n->parent;
546 continue;
547 case (REWIND_LATER):
548 return(make_pending(n, tok, m, line, ppos));
549 case (REWIND_ERROR):
550 /* XXX Make this non-fatal. */
551 mdoc_vmsg(m, MANDOCERR_SCOPEFATAL, line, ppos,
552 "%s cannot break %s", mdoc_macronames[tok],
553 mdoc_macronames[n->tok]);
554 return 0;
555 }
556 break;
557 }
558
559 assert(n);
560 if ( ! rew_last(m, n))
561 return(0);
562
563 /*
564 * The current block extends an enclosing block.
565 * Now that the current block ends, close the enclosing block, too.
566 */
567 while (NULL != (n = n->pending)) {
568 if ( ! rew_last(m, n))
569 return(0);
570 if (MDOC_HEAD == n->type &&
571 ! mdoc_body_alloc(m, n->line, n->pos, n->tok))
572 return(0);
573 }
574
575 return(1);
576 }
577
578
579 static int
580 append_delims(struct mdoc *m, int line, int *pos, char *buf)
581 {
582 int la;
583 enum margserr ac;
584 char *p;
585
586 if ('\0' == buf[*pos])
587 return(1);
588
589 for (;;) {
590 la = *pos;
591 ac = mdoc_zargs(m, line, pos, buf, ARGS_NOWARN, &p);
592
593 if (ARGS_ERROR == ac)
594 return(0);
595 else if (ARGS_EOLN == ac)
596 break;
597
598 assert(DELIM_NONE != mdoc_isdelim(p));
599 if ( ! mdoc_word_alloc(m, line, la, p))
600 return(0);
601
602 /*
603 * If we encounter end-of-sentence symbols, then trigger
604 * the double-space.
605 *
606 * XXX: it's easy to allow this to propogate outward to
607 * the last symbol, such that `. )' will cause the
608 * correct double-spacing. However, (1) groff isn't
609 * smart enough to do this and (2) it would require
610 * knowing which symbols break this behaviour, for
611 * example, `. ;' shouldn't propogate the double-space.
612 */
613 if (mandoc_eos(p, strlen(p), 0))
614 m->last->flags |= MDOC_EOS;
615 }
616
617 return(1);
618 }
619
620
621 /*
622 * Close out block partial/full explicit.
623 */
624 static int
625 blk_exp_close(MACRO_PROT_ARGS)
626 {
627 struct mdoc_node *body; /* Our own body. */
628 struct mdoc_node *later; /* A sub-block starting later. */
629 struct mdoc_node *n; /* For searching backwards. */
630
631 int j, lastarg, maxargs, flushed, nl;
632 enum margserr ac;
633 enum mdoct atok, ntok;
634 char *p;
635
636 nl = MDOC_NEWLINE & m->flags;
637
638 switch (tok) {
639 case (MDOC_Ec):
640 maxargs = 1;
641 break;
642 default:
643 maxargs = 0;
644 break;
645 }
646
647 /*
648 * Search backwards for beginnings of blocks,
649 * both of our own and of pending sub-blocks.
650 */
651 atok = rew_alt(tok);
652 body = later = NULL;
653 for (n = m->last; n; n = n->parent) {
654 if (MDOC_VALID & n->flags)
655 continue;
656
657 /* Remember the start of our own body. */
658 if (MDOC_BODY == n->type && atok == n->tok) {
659 if (ENDBODY_NOT == n->end)
660 body = n;
661 continue;
662 }
663
664 if (MDOC_BLOCK != n->type || MDOC_Nm == n->tok)
665 continue;
666 if (atok == n->tok) {
667 assert(body);
668
669 /*
670 * Found the start of our own block.
671 * When there is no pending sub block,
672 * just proceed to closing out.
673 */
674 if (NULL == later)
675 break;
676
677 /*
678 * When there is a pending sub block,
679 * postpone closing out the current block
680 * until the rew_sub() closing out the sub-block.
681 */
682 if ( ! make_pending(later, tok, m, line, ppos))
683 return(0);
684
685 /*
686 * Mark the place where the formatting - but not
687 * the scope - of the current block ends.
688 */
689 if ( ! mdoc_endbody_alloc(m, line, ppos,
690 atok, body, ENDBODY_SPACE))
691 return(0);
692 break;
693 }
694
695 /*
696 * When finding an open sub block, remember the last
697 * open explicit block, or, in case there are only
698 * implicit ones, the first open implicit block.
699 */
700 if (later &&
701 MDOC_EXPLICIT & mdoc_macros[later->tok].flags)
702 continue;
703 if (MDOC_CALLABLE & mdoc_macros[n->tok].flags) {
704 assert( ! (MDOC_ACTED & n->flags));
705 later = n;
706 }
707 }
708
709 if ( ! (MDOC_CALLABLE & mdoc_macros[tok].flags)) {
710 /* FIXME: do this in validate */
711 if (buf[*pos])
712 if ( ! mdoc_pmsg(m, line, ppos, MANDOCERR_ARGSLOST))
713 return(0);
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_Lk):
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 if ( ! mdoc_pmsg(m, line, ppos, MANDOCERR_MACROEMPTY))
857 return(0);
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 : mdoc_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 if ( ! mdoc_pmsg(m, line, ppos, MANDOCERR_MACROEMPTY))
939 return(0);
940 }
941
942 if ( ! nl)
943 return(1);
944 return(append_delims(m, line, pos, buf));
945 }
946
947
948 static int
949 blk_full(MACRO_PROT_ARGS)
950 {
951 int la, nl;
952 struct mdoc_arg *arg;
953 struct mdoc_node *head; /* save of head macro */
954 struct mdoc_node *body; /* save of body macro */
955 struct mdoc_node *n;
956 enum mdoc_type mtt;
957 enum mdoct ntok;
958 enum margserr ac, lac;
959 enum margverr av;
960 char *p;
961
962 nl = MDOC_NEWLINE & m->flags;
963
964 /* Close out prior implicit scope. */
965
966 if ( ! (MDOC_EXPLICIT & mdoc_macros[tok].flags)) {
967 if ( ! rew_sub(MDOC_BODY, m, tok, line, ppos))
968 return(0);
969 if ( ! rew_sub(MDOC_BLOCK, m, tok, line, ppos))
970 return(0);
971 }
972
973 /*
974 * This routine accomodates implicitly- and explicitly-scoped
975 * macro openings. Implicit ones first close out prior scope
976 * (seen above). Delay opening the head until necessary to
977 * allow leading punctuation to print. Special consideration
978 * for `It -column', which has phrase-part syntax instead of
979 * regular child nodes.
980 */
981
982 for (arg = NULL;; ) {
983 la = *pos;
984 av = mdoc_argv(m, line, tok, &arg, pos, buf);
985
986 if (ARGV_WORD == av) {
987 *pos = la;
988 break;
989 }
990
991 if (ARGV_EOLN == av)
992 break;
993 if (ARGV_ARG == av)
994 continue;
995
996 mdoc_argv_free(arg);
997 return(0);
998 }
999
1000 if ( ! mdoc_block_alloc(m, line, ppos, tok, arg))
1001 return(0);
1002
1003 head = body = NULL;
1004
1005 /*
1006 * The `Nd' macro has all arguments in its body: it's a hybrid
1007 * of block partial-explicit and full-implicit. Stupid.
1008 */
1009
1010 if (MDOC_Nd == tok) {
1011 if ( ! mdoc_head_alloc(m, line, ppos, tok))
1012 return(0);
1013 head = m->last;
1014 if ( ! rew_sub(MDOC_HEAD, m, tok, line, ppos))
1015 return(0);
1016 if ( ! mdoc_body_alloc(m, line, ppos, tok))
1017 return(0);
1018 body = m->last;
1019 }
1020
1021 ac = ARGS_ERROR;
1022
1023 for ( ; ; ) {
1024 la = *pos;
1025 /* Initialise last-phrase-type with ARGS_PEND. */
1026 lac = ARGS_ERROR == ac ? ARGS_PEND : ac;
1027 ac = mdoc_args(m, line, pos, buf, tok, &p);
1028
1029 if (ARGS_PUNCT == ac)
1030 break;
1031
1032 if (ARGS_ERROR == ac)
1033 return(0);
1034
1035 if (ARGS_EOLN == ac) {
1036 if (ARGS_PPHRASE != lac && ARGS_PHRASE != lac)
1037 break;
1038 /*
1039 * This is necessary: if the last token on a
1040 * line is a `Ta' or tab, then we'll get
1041 * ARGS_EOLN, so we must be smart enough to
1042 * reopen our scope if the last parse was a
1043 * phrase or partial phrase.
1044 */
1045 if ( ! rew_sub(MDOC_BODY, m, tok, line, ppos))
1046 return(0);
1047 if ( ! mdoc_body_alloc(m, line, ppos, tok))
1048 return(0);
1049 body = m->last;
1050 break;
1051 }
1052
1053 /*
1054 * Emit leading punctuation (i.e., punctuation before
1055 * the MDOC_HEAD) for non-phrase types.
1056 */
1057
1058 if (NULL == head &&
1059 ARGS_PEND != ac &&
1060 ARGS_PHRASE != ac &&
1061 ARGS_PPHRASE != ac &&
1062 ARGS_QWORD != ac &&
1063 DELIM_OPEN == mdoc_isdelim(p)) {
1064 if ( ! mdoc_word_alloc(m, line, la, p))
1065 return(0);
1066 continue;
1067 }
1068
1069 /* Open a head if one hasn't been opened. */
1070
1071 if (NULL == head) {
1072 if ( ! mdoc_head_alloc(m, line, ppos, tok))
1073 return(0);
1074 head = m->last;
1075 }
1076
1077 if (ARGS_PHRASE == ac ||
1078 ARGS_PEND == ac ||
1079 ARGS_PPHRASE == ac) {
1080 /*
1081 * If we haven't opened a body yet, rewind the
1082 * head; if we have, rewind that instead.
1083 */
1084
1085 mtt = body ? MDOC_BODY : MDOC_HEAD;
1086 if ( ! rew_sub(mtt, m, tok, line, ppos))
1087 return(0);
1088
1089 /* Then allocate our body context. */
1090
1091 if ( ! mdoc_body_alloc(m, line, ppos, tok))
1092 return(0);
1093 body = m->last;
1094
1095 /*
1096 * Process phrases: set whether we're in a
1097 * partial-phrase (this effects line handling)
1098 * then call down into the phrase parser.
1099 */
1100
1101 if (ARGS_PPHRASE == ac)
1102 m->flags |= MDOC_PPHRASE;
1103 if (ARGS_PEND == ac && ARGS_PPHRASE == lac)
1104 m->flags |= MDOC_PPHRASE;
1105
1106 if ( ! phrase(m, line, la, buf))
1107 return(0);
1108
1109 m->flags &= ~MDOC_PPHRASE;
1110 continue;
1111 }
1112
1113 ntok = ARGS_QWORD == ac ? MDOC_MAX : lookup(tok, p);
1114
1115 if (MDOC_MAX == ntok) {
1116 if ( ! mdoc_word_alloc(m, line, la, p))
1117 return(0);
1118 continue;
1119 }
1120
1121 if ( ! mdoc_macro(m, ntok, line, la, pos, buf))
1122 return(0);
1123 break;
1124 }
1125
1126 if (NULL == head) {
1127 if ( ! mdoc_head_alloc(m, line, ppos, tok))
1128 return(0);
1129 head = m->last;
1130 }
1131
1132 if (nl && ! append_delims(m, line, pos, buf))
1133 return(0);
1134
1135 /* If we've already opened our body, exit now. */
1136
1137 if (NULL != body)
1138 goto out;
1139
1140 /*
1141 * If there is an open (i.e., unvalidated) sub-block requiring
1142 * explicit close-out, postpone switching the current block from
1143 * head to body until the rew_sub() call closing out that
1144 * sub-block.
1145 */
1146 for (n = m->last; n && n != head; n = n->parent) {
1147 if (MDOC_BLOCK == n->type &&
1148 MDOC_EXPLICIT & mdoc_macros[n->tok].flags &&
1149 ! (MDOC_VALID & n->flags)) {
1150 assert( ! (MDOC_ACTED & 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 assert( ! (MDOC_ACTED & n->flags));
1289 if ( ! make_pending(n, tok, m, line, ppos))
1290 return(0);
1291 if ( ! mdoc_endbody_alloc(m, line, ppos,
1292 tok, body, ENDBODY_NOSPACE))
1293 return(0);
1294 return(1);
1295 }
1296 }
1297
1298 /*
1299 * If we can't rewind to our body, then our scope has already
1300 * been closed by another macro (like `Oc' closing `Op'). This
1301 * is ugly behaviour nodding its head to OpenBSD's overwhelming
1302 * crufty use of `Op' breakage.
1303 */
1304 if (n != body && ! mdoc_vmsg(m, MANDOCERR_SCOPENEST,
1305 line, ppos, "%s broken", mdoc_macronames[tok]))
1306 return(0);
1307
1308 if (n && ! rew_sub(MDOC_BODY, m, tok, line, ppos))
1309 return(0);
1310
1311 /* Standard appending of delimiters. */
1312
1313 if (nl && ! append_delims(m, line, pos, buf))
1314 return(0);
1315
1316 /* Rewind scope, if applicable. */
1317
1318 if (n && ! rew_sub(MDOC_BLOCK, m, tok, line, ppos))
1319 return(0);
1320
1321 return(1);
1322 }
1323
1324
1325 static int
1326 blk_part_exp(MACRO_PROT_ARGS)
1327 {
1328 int la, nl;
1329 enum margserr ac;
1330 struct mdoc_node *head; /* keep track of head */
1331 struct mdoc_node *body; /* keep track of body */
1332 char *p;
1333 enum mdoct ntok;
1334
1335 nl = MDOC_NEWLINE & m->flags;
1336
1337 /*
1338 * The opening of an explicit macro having zero or more leading
1339 * punctuation nodes; a head with optional single element (the
1340 * case of `Eo'); and a body that may be empty.
1341 */
1342
1343 if ( ! mdoc_block_alloc(m, line, ppos, tok, NULL))
1344 return(0);
1345
1346 for (head = body = NULL; ; ) {
1347 la = *pos;
1348 ac = mdoc_args(m, line, pos, buf, tok, &p);
1349
1350 if (ARGS_ERROR == ac)
1351 return(0);
1352 if (ARGS_PUNCT == ac)
1353 break;
1354 if (ARGS_EOLN == ac)
1355 break;
1356
1357 /* Flush out leading punctuation. */
1358
1359 if (NULL == head && ARGS_QWORD != ac &&
1360 DELIM_OPEN == mdoc_isdelim(p)) {
1361 assert(NULL == body);
1362 if ( ! mdoc_word_alloc(m, line, la, p))
1363 return(0);
1364 continue;
1365 }
1366
1367 if (NULL == head) {
1368 assert(NULL == body);
1369 if ( ! mdoc_head_alloc(m, line, ppos, tok))
1370 return(0);
1371 head = m->last;
1372 }
1373
1374 /*
1375 * `Eo' gobbles any data into the head, but most other
1376 * macros just immediately close out and begin the body.
1377 */
1378
1379 if (NULL == body) {
1380 assert(head);
1381 /* No check whether it's a macro! */
1382 if (MDOC_Eo == tok)
1383 if ( ! mdoc_word_alloc(m, line, la, p))
1384 return(0);
1385
1386 if ( ! rew_sub(MDOC_HEAD, m, tok, line, ppos))
1387 return(0);
1388 if ( ! mdoc_body_alloc(m, line, ppos, tok))
1389 return(0);
1390 body = m->last;
1391
1392 if (MDOC_Eo == tok)
1393 continue;
1394 }
1395
1396 assert(NULL != head && NULL != body);
1397
1398 ntok = ARGS_QWORD == ac ? MDOC_MAX : lookup(tok, p);
1399
1400 if (MDOC_MAX == ntok) {
1401 if ( ! mdoc_word_alloc(m, line, la, p))
1402 return(0);
1403 continue;
1404 }
1405
1406 if ( ! mdoc_macro(m, ntok, line, la, pos, buf))
1407 return(0);
1408 break;
1409 }
1410
1411 /* Clean-up to leave in a consistent state. */
1412
1413 if (NULL == head) {
1414 if ( ! mdoc_head_alloc(m, line, ppos, tok))
1415 return(0);
1416 head = m->last;
1417 }
1418
1419 if (NULL == body) {
1420 if ( ! rew_sub(MDOC_HEAD, m, tok, line, ppos))
1421 return(0);
1422 if ( ! mdoc_body_alloc(m, line, ppos, tok))
1423 return(0);
1424 body = m->last;
1425 }
1426
1427 /* Standard appending of delimiters. */
1428
1429 if ( ! nl)
1430 return(1);
1431 return(append_delims(m, line, pos, buf));
1432 }
1433
1434
1435 /* ARGSUSED */
1436 static int
1437 in_line_argn(MACRO_PROT_ARGS)
1438 {
1439 int la, flushed, j, maxargs, nl;
1440 enum margserr ac;
1441 enum margverr av;
1442 struct mdoc_arg *arg;
1443 char *p;
1444 enum mdoct ntok;
1445
1446 nl = MDOC_NEWLINE & m->flags;
1447
1448 /*
1449 * A line macro that has a fixed number of arguments (maxargs).
1450 * Only open the scope once the first non-leading-punctuation is
1451 * found (unless MDOC_IGNDELIM is noted, like in `Pf'), then
1452 * keep it open until the maximum number of arguments are
1453 * exhausted.
1454 */
1455
1456 switch (tok) {
1457 case (MDOC_Ap):
1458 /* FALLTHROUGH */
1459 case (MDOC_No):
1460 /* FALLTHROUGH */
1461 case (MDOC_Ns):
1462 /* FALLTHROUGH */
1463 case (MDOC_Ux):
1464 maxargs = 0;
1465 break;
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