]> git.cameronkatri.com Git - mandoc.git/blob - mdoc_macro.c
discard .Rs head arguments and improve .Rs diagnostics
[mandoc.git] / mdoc_macro.c
1 /* $Id: mdoc_macro.c,v 1.173 2015/02/04 18:03:47 schwarze Exp $ */
2 /*
3 * Copyright (c) 2008-2012 Kristaps Dzonsons <kristaps@bsd.lv>
4 * Copyright (c) 2010, 2012-2015 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 #include "config.h"
19
20 #include <sys/types.h>
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 "mdoc.h"
30 #include "mandoc.h"
31 #include "libmdoc.h"
32 #include "libmandoc.h"
33
34 static void blk_full(MACRO_PROT_ARGS);
35 static void blk_exp_close(MACRO_PROT_ARGS);
36 static void blk_part_exp(MACRO_PROT_ARGS);
37 static void blk_part_imp(MACRO_PROT_ARGS);
38 static void ctx_synopsis(MACRO_PROT_ARGS);
39 static void in_line_eoln(MACRO_PROT_ARGS);
40 static void in_line_argn(MACRO_PROT_ARGS);
41 static void in_line(MACRO_PROT_ARGS);
42 static void phrase_ta(MACRO_PROT_ARGS);
43
44 static void dword(struct mdoc *, int, int, const char *,
45 enum mdelim, int);
46 static void append_delims(struct mdoc *, int, int *, char *);
47 static enum mdoct lookup(struct mdoc *, enum mdoct,
48 int, int, const char *);
49 static int macro_or_word(MACRO_PROT_ARGS, int);
50 static void make_pending(struct mdoc *, struct mdoc_node *,
51 struct mdoc_node *, int, int);
52 static int parse_rest(struct mdoc *, enum mdoct,
53 int, int *, char *);
54 static enum mdoct rew_alt(enum mdoct);
55 static void rew_elem(struct mdoc *, enum mdoct);
56 static void rew_last(struct mdoc *, const struct mdoc_node *);
57 static void rew_pending(struct mdoc *, const struct mdoc_node *);
58
59 const struct mdoc_macro __mdoc_macros[MDOC_MAX] = {
60 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* Ap */
61 { in_line_eoln, MDOC_PROLOGUE }, /* Dd */
62 { in_line_eoln, MDOC_PROLOGUE }, /* Dt */
63 { in_line_eoln, MDOC_PROLOGUE }, /* Os */
64 { blk_full, MDOC_PARSED | MDOC_JOIN }, /* Sh */
65 { blk_full, MDOC_PARSED | MDOC_JOIN }, /* Ss */
66 { in_line_eoln, 0 }, /* Pp */
67 { blk_part_imp, MDOC_PARSED | MDOC_JOIN }, /* D1 */
68 { blk_part_imp, MDOC_PARSED | MDOC_JOIN }, /* Dl */
69 { blk_full, MDOC_EXPLICIT }, /* Bd */
70 { blk_exp_close, MDOC_EXPLICIT | MDOC_JOIN }, /* Ed */
71 { blk_full, MDOC_EXPLICIT }, /* Bl */
72 { blk_exp_close, MDOC_EXPLICIT | MDOC_JOIN }, /* El */
73 { blk_full, MDOC_PARSED | MDOC_JOIN }, /* It */
74 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Ad */
75 { in_line, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* An */
76 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Ar */
77 { in_line, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* Cd */
78 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Cm */
79 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Dv */
80 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Er */
81 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Ev */
82 { in_line_eoln, 0 }, /* Ex */
83 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Fa */
84 { in_line_eoln, 0 }, /* Fd */
85 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Fl */
86 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Fn */
87 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Ft */
88 { in_line, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* Ic */
89 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* In */
90 { in_line, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* Li */
91 { blk_full, MDOC_JOIN }, /* Nd */
92 { ctx_synopsis, MDOC_CALLABLE | MDOC_PARSED }, /* Nm */
93 { blk_part_imp, MDOC_CALLABLE | MDOC_PARSED }, /* Op */
94 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Ot */
95 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Pa */
96 { in_line_eoln, 0 }, /* Rv */
97 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* St */
98 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Va */
99 { ctx_synopsis, MDOC_CALLABLE | MDOC_PARSED }, /* Vt */
100 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* Xr */
101 { in_line_eoln, MDOC_JOIN }, /* %A */
102 { in_line_eoln, MDOC_JOIN }, /* %B */
103 { in_line_eoln, MDOC_JOIN }, /* %D */
104 { in_line_eoln, MDOC_JOIN }, /* %I */
105 { in_line_eoln, MDOC_JOIN }, /* %J */
106 { in_line_eoln, 0 }, /* %N */
107 { in_line_eoln, MDOC_JOIN }, /* %O */
108 { in_line_eoln, 0 }, /* %P */
109 { in_line_eoln, MDOC_JOIN }, /* %R */
110 { in_line_eoln, MDOC_JOIN }, /* %T */
111 { in_line_eoln, 0 }, /* %V */
112 { blk_exp_close, MDOC_CALLABLE | MDOC_PARSED |
113 MDOC_EXPLICIT | MDOC_JOIN }, /* Ac */
114 { blk_part_exp, MDOC_CALLABLE | MDOC_PARSED |
115 MDOC_EXPLICIT | MDOC_JOIN }, /* Ao */
116 { blk_part_imp, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* Aq */
117 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* At */
118 { blk_exp_close, MDOC_CALLABLE | MDOC_PARSED |
119 MDOC_EXPLICIT | MDOC_JOIN }, /* Bc */
120 { blk_full, MDOC_EXPLICIT }, /* Bf */
121 { blk_part_exp, MDOC_CALLABLE | MDOC_PARSED |
122 MDOC_EXPLICIT | MDOC_JOIN }, /* Bo */
123 { blk_part_imp, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* Bq */
124 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* Bsx */
125 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* Bx */
126 { in_line_eoln, 0 }, /* Db */
127 { blk_exp_close, MDOC_CALLABLE | MDOC_PARSED |
128 MDOC_EXPLICIT | MDOC_JOIN }, /* Dc */
129 { blk_part_exp, MDOC_CALLABLE | MDOC_PARSED |
130 MDOC_EXPLICIT | MDOC_JOIN }, /* Do */
131 { blk_part_imp, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* Dq */
132 { blk_exp_close, MDOC_CALLABLE | MDOC_PARSED | MDOC_EXPLICIT }, /* Ec */
133 { blk_exp_close, MDOC_EXPLICIT | MDOC_JOIN }, /* Ef */
134 { in_line, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* Em */
135 { blk_part_exp, MDOC_CALLABLE | MDOC_PARSED | MDOC_EXPLICIT }, /* Eo */
136 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* Fx */
137 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Ms */
138 { in_line, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* No */
139 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED |
140 MDOC_IGNDELIM | MDOC_JOIN }, /* Ns */
141 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* Nx */
142 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* Ox */
143 { blk_exp_close, MDOC_CALLABLE | MDOC_PARSED |
144 MDOC_EXPLICIT | MDOC_JOIN }, /* Pc */
145 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED | MDOC_IGNDELIM }, /* Pf */
146 { blk_part_exp, MDOC_CALLABLE | MDOC_PARSED |
147 MDOC_EXPLICIT | MDOC_JOIN }, /* Po */
148 { blk_part_imp, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* Pq */
149 { blk_exp_close, MDOC_CALLABLE | MDOC_PARSED |
150 MDOC_EXPLICIT | MDOC_JOIN }, /* Qc */
151 { blk_part_imp, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* Ql */
152 { blk_part_exp, MDOC_CALLABLE | MDOC_PARSED |
153 MDOC_EXPLICIT | MDOC_JOIN }, /* Qo */
154 { blk_part_imp, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* Qq */
155 { blk_exp_close, MDOC_EXPLICIT | MDOC_JOIN }, /* Re */
156 { blk_full, MDOC_EXPLICIT }, /* Rs */
157 { blk_exp_close, MDOC_CALLABLE | MDOC_PARSED |
158 MDOC_EXPLICIT | MDOC_JOIN }, /* Sc */
159 { blk_part_exp, MDOC_CALLABLE | MDOC_PARSED |
160 MDOC_EXPLICIT | MDOC_JOIN }, /* So */
161 { blk_part_imp, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* Sq */
162 { in_line_argn, 0 }, /* Sm */
163 { in_line, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* Sx */
164 { in_line, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* Sy */
165 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Tn */
166 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* Ux */
167 { blk_exp_close, MDOC_EXPLICIT | MDOC_CALLABLE | MDOC_PARSED }, /* Xc */
168 { blk_part_exp, MDOC_CALLABLE | MDOC_PARSED | MDOC_EXPLICIT }, /* Xo */
169 { blk_full, MDOC_EXPLICIT | MDOC_CALLABLE }, /* Fo */
170 { blk_exp_close, MDOC_CALLABLE | MDOC_PARSED |
171 MDOC_EXPLICIT | MDOC_JOIN }, /* Fc */
172 { blk_part_exp, MDOC_CALLABLE | MDOC_PARSED |
173 MDOC_EXPLICIT | MDOC_JOIN }, /* Oo */
174 { blk_exp_close, MDOC_CALLABLE | MDOC_PARSED |
175 MDOC_EXPLICIT | MDOC_JOIN }, /* Oc */
176 { blk_full, MDOC_EXPLICIT }, /* Bk */
177 { blk_exp_close, MDOC_EXPLICIT | MDOC_JOIN }, /* Ek */
178 { in_line_eoln, 0 }, /* Bt */
179 { in_line_eoln, 0 }, /* Hf */
180 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Fr */
181 { in_line_eoln, 0 }, /* Ud */
182 { in_line, 0 }, /* Lb */
183 { in_line_eoln, 0 }, /* Lp */
184 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Lk */
185 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Mt */
186 { blk_part_imp, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* Brq */
187 { blk_part_exp, MDOC_CALLABLE | MDOC_PARSED |
188 MDOC_EXPLICIT | MDOC_JOIN }, /* Bro */
189 { blk_exp_close, MDOC_CALLABLE | MDOC_PARSED |
190 MDOC_EXPLICIT | MDOC_JOIN }, /* Brc */
191 { in_line_eoln, MDOC_JOIN }, /* %C */
192 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* Es */
193 { blk_part_imp, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* En */
194 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* Dx */
195 { in_line_eoln, MDOC_JOIN }, /* %Q */
196 { in_line_eoln, 0 }, /* br */
197 { in_line_eoln, 0 }, /* sp */
198 { in_line_eoln, 0 }, /* %U */
199 { phrase_ta, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* Ta */
200 { in_line_eoln, MDOC_PROLOGUE }, /* ll */
201 };
202
203 const struct mdoc_macro * const mdoc_macros = __mdoc_macros;
204
205
206 /*
207 * This is called at the end of parsing. It must traverse up the tree,
208 * closing out open [implicit] scopes. Obviously, open explicit scopes
209 * are errors.
210 */
211 void
212 mdoc_macroend(struct mdoc *mdoc)
213 {
214 struct mdoc_node *n;
215
216 /* Scan for open explicit scopes. */
217
218 n = mdoc->last->flags & MDOC_VALID ?
219 mdoc->last->parent : mdoc->last;
220
221 for ( ; n; n = n->parent)
222 if (n->type == MDOC_BLOCK &&
223 mdoc_macros[n->tok].flags & MDOC_EXPLICIT)
224 mandoc_msg(MANDOCERR_BLK_NOEND, mdoc->parse,
225 n->line, n->pos, mdoc_macronames[n->tok]);
226
227 /* Rewind to the first. */
228
229 rew_last(mdoc, mdoc->first);
230 }
231
232 /*
233 * Look up the macro at *p called by "from",
234 * or as a line macro if from == MDOC_MAX.
235 */
236 static enum mdoct
237 lookup(struct mdoc *mdoc, enum mdoct from, int line, int ppos, const char *p)
238 {
239 enum mdoct res;
240
241 if (from == MDOC_MAX || mdoc_macros[from].flags & MDOC_PARSED) {
242 res = mdoc_hash_find(p);
243 if (res != MDOC_MAX) {
244 if (mdoc_macros[res].flags & MDOC_CALLABLE)
245 return(res);
246 if (res != MDOC_br && res != MDOC_sp && res != MDOC_ll)
247 mandoc_msg(MANDOCERR_MACRO_CALL,
248 mdoc->parse, line, ppos, p);
249 }
250 }
251 return(MDOC_MAX);
252 }
253
254 /*
255 * Rewind up to and including a specific node.
256 */
257 static void
258 rew_last(struct mdoc *mdoc, const struct mdoc_node *to)
259 {
260 struct mdoc_node *n, *np;
261
262 assert(to);
263 mdoc->next = MDOC_NEXT_SIBLING;
264 while (mdoc->last != to) {
265 if ( ! (mdoc->last->flags & MDOC_VALID))
266 mdoc->last->lastline = to->lastline -
267 (mdoc->flags & MDOC_NEWLINE ? 1 : 0);
268 /*
269 * Save the parent here, because we may delete the
270 * mdoc->last node in the post-validation phase and reset
271 * it to mdoc->last->parent, causing a step in the closing
272 * out to be lost.
273 */
274 np = mdoc->last->parent;
275 mdoc_valid_post(mdoc);
276 n = mdoc->last;
277 mdoc->last = np;
278 assert(mdoc->last);
279 mdoc->last->last = n;
280 }
281 mdoc_valid_post(mdoc);
282 }
283
284 /*
285 * Rewind up to a specific block, including all blocks that broke it.
286 */
287 static void
288 rew_pending(struct mdoc *mdoc, const struct mdoc_node *n)
289 {
290
291 rew_last(mdoc, n);
292
293 if (n->type != MDOC_BLOCK)
294 return;
295
296 while ((n = n->pending) != NULL) {
297 rew_last(mdoc, n);
298 if (n->type == MDOC_HEAD)
299 mdoc_body_alloc(mdoc, n->line, n->pos, n->tok);
300 }
301 }
302
303 /*
304 * For a block closing macro, return the corresponding opening one.
305 * Otherwise, return the macro itself.
306 */
307 static enum mdoct
308 rew_alt(enum mdoct tok)
309 {
310 switch (tok) {
311 case MDOC_Ac:
312 return(MDOC_Ao);
313 case MDOC_Bc:
314 return(MDOC_Bo);
315 case MDOC_Brc:
316 return(MDOC_Bro);
317 case MDOC_Dc:
318 return(MDOC_Do);
319 case MDOC_Ec:
320 return(MDOC_Eo);
321 case MDOC_Ed:
322 return(MDOC_Bd);
323 case MDOC_Ef:
324 return(MDOC_Bf);
325 case MDOC_Ek:
326 return(MDOC_Bk);
327 case MDOC_El:
328 return(MDOC_Bl);
329 case MDOC_Fc:
330 return(MDOC_Fo);
331 case MDOC_Oc:
332 return(MDOC_Oo);
333 case MDOC_Pc:
334 return(MDOC_Po);
335 case MDOC_Qc:
336 return(MDOC_Qo);
337 case MDOC_Re:
338 return(MDOC_Rs);
339 case MDOC_Sc:
340 return(MDOC_So);
341 case MDOC_Xc:
342 return(MDOC_Xo);
343 default:
344 return(tok);
345 }
346 /* NOTREACHED */
347 }
348
349 static void
350 rew_elem(struct mdoc *mdoc, enum mdoct tok)
351 {
352 struct mdoc_node *n;
353
354 n = mdoc->last;
355 if (MDOC_ELEM != n->type)
356 n = n->parent;
357 assert(MDOC_ELEM == n->type);
358 assert(tok == n->tok);
359 rew_last(mdoc, n);
360 }
361
362 /*
363 * We are trying to close the block *breaker,
364 * but the child block *broken is still open.
365 * Thus, postpone closing the *breaker
366 * until the rew_pending() call closing *broken.
367 */
368 static void
369 make_pending(struct mdoc *mdoc, struct mdoc_node *breaker,
370 struct mdoc_node *broken, int line, int ppos)
371 {
372 struct mdoc_node *n;
373
374 mandoc_vmsg(MANDOCERR_BLK_NEST, mdoc->parse, line, ppos,
375 "%s breaks %s", mdoc_macronames[breaker->tok],
376 mdoc_macronames[broken->tok]);
377
378 /*
379 * If the *broken block (Z) is already broken by a block (B)
380 * contained in the breaker (A), make the breaker pending
381 * on that inner breaker (B). Graphically,
382 *
383 * breaker=[A! broken=n=[B!->A (old broken=)[Z->B B] A] Z]
384 *
385 * In these graphics, "->" indicates the "pending" pointer and
386 * "!" indicates the MDOC_BREAK flag. Each of the cases gets
387 * one additional pointer (B->A) and one additional flag (A!).
388 */
389
390 for (n = broken->parent; ; n = n->parent)
391 if (n == broken->pending)
392 broken = n;
393 else if (n == breaker)
394 break;
395
396 /*
397 * Found the breaker.
398 *
399 * If another, outer breaker (X) is already pending on
400 * the *broken block (B), we must not clobber the link
401 * to the outer breaker, but make it pending on the new,
402 * now inner breaker (A). Graphically,
403 *
404 * [X! n=breaker=[A!->X broken=[B(->X)->A X] A] B].
405 */
406
407 if (broken->pending != NULL) {
408 n = breaker;
409
410 /*
411 * If the inner breaker (A) is already broken, too,
412 * it cannot take on the outer breaker (X) but must
413 * hand it on to its own breakers (Y). Graphically,
414 *
415 * [X! n=[Y!->X breaker=[A!->Y Y] broken=[B(->X)->A X] A] B]
416 */
417
418 while (n->pending)
419 n = n->pending;
420 n->pending = broken->pending;
421 }
422
423 /*
424 * Now we have reduced the situation to the simplest case:
425 * breaker=[A! broken=[B->A A] B].
426 */
427
428 broken->pending = breaker;
429 breaker->flags |= MDOC_BREAK;
430 if (breaker->body != NULL)
431 breaker->body->flags |= MDOC_BREAK;
432 }
433
434 /*
435 * Allocate a word and check whether it's punctuation or not.
436 * Punctuation consists of those tokens found in mdoc_isdelim().
437 */
438 static void
439 dword(struct mdoc *mdoc, int line, int col, const char *p,
440 enum mdelim d, int may_append)
441 {
442
443 if (d == DELIM_MAX)
444 d = mdoc_isdelim(p);
445
446 if (may_append &&
447 ! (mdoc->flags & (MDOC_SYNOPSIS | MDOC_KEEP | MDOC_SMOFF)) &&
448 d == DELIM_NONE && mdoc->last->type == MDOC_TEXT &&
449 mdoc_isdelim(mdoc->last->string) == DELIM_NONE) {
450 mdoc_word_append(mdoc, p);
451 return;
452 }
453
454 mdoc_word_alloc(mdoc, line, col, p);
455
456 /*
457 * If the word consists of a bare delimiter,
458 * flag the new node accordingly,
459 * unless doing so was vetoed by the invoking macro.
460 * Always clear the veto, it is only valid for one word.
461 */
462
463 if (d == DELIM_OPEN)
464 mdoc->last->flags |= MDOC_DELIMO;
465 else if (d == DELIM_CLOSE &&
466 ! (mdoc->flags & MDOC_NODELIMC) &&
467 mdoc->last->parent->tok != MDOC_Fd)
468 mdoc->last->flags |= MDOC_DELIMC;
469 mdoc->flags &= ~MDOC_NODELIMC;
470 }
471
472 static void
473 append_delims(struct mdoc *mdoc, int line, int *pos, char *buf)
474 {
475 char *p;
476 int la;
477
478 if (buf[*pos] == '\0')
479 return;
480
481 for (;;) {
482 la = *pos;
483 if (mdoc_args(mdoc, line, pos, buf, MDOC_MAX, &p) == ARGS_EOLN)
484 break;
485 dword(mdoc, line, la, p, DELIM_MAX, 1);
486
487 /*
488 * If we encounter end-of-sentence symbols, then trigger
489 * the double-space.
490 *
491 * XXX: it's easy to allow this to propagate outward to
492 * the last symbol, such that `. )' will cause the
493 * correct double-spacing. However, (1) groff isn't
494 * smart enough to do this and (2) it would require
495 * knowing which symbols break this behaviour, for
496 * example, `. ;' shouldn't propagate the double-space.
497 */
498
499 if (mandoc_eos(p, strlen(p)))
500 mdoc->last->flags |= MDOC_EOS;
501 }
502 }
503
504 /*
505 * Parse one word.
506 * If it is a macro, call it and return 1.
507 * Otherwise, allocate it and return 0.
508 */
509 static int
510 macro_or_word(MACRO_PROT_ARGS, int parsed)
511 {
512 char *p;
513 enum mdoct ntok;
514
515 p = buf + ppos;
516 ntok = MDOC_MAX;
517 if (*p == '"')
518 p++;
519 else if (parsed && ! (mdoc->flags & MDOC_PHRASELIT))
520 ntok = lookup(mdoc, tok, line, ppos, p);
521
522 if (ntok == MDOC_MAX) {
523 dword(mdoc, line, ppos, p, DELIM_MAX, tok == MDOC_MAX ||
524 mdoc_macros[tok].flags & MDOC_JOIN);
525 return(0);
526 } else {
527 if (mdoc_macros[tok].fp == in_line_eoln)
528 rew_elem(mdoc, tok);
529 mdoc_macro(mdoc, ntok, line, ppos, pos, buf);
530 if (tok == MDOC_MAX)
531 append_delims(mdoc, line, pos, buf);
532 return(1);
533 }
534 }
535
536 /*
537 * Close out block partial/full explicit.
538 */
539 static void
540 blk_exp_close(MACRO_PROT_ARGS)
541 {
542 struct mdoc_node *body; /* Our own body. */
543 struct mdoc_node *endbody; /* Our own end marker. */
544 struct mdoc_node *itblk; /* An It block starting later. */
545 struct mdoc_node *later; /* A sub-block starting later. */
546 struct mdoc_node *n; /* Search back to our block. */
547
548 int j, lastarg, maxargs, nl;
549 enum margserr ac;
550 enum mdoct atok, ntok;
551 char *p;
552
553 nl = MDOC_NEWLINE & mdoc->flags;
554
555 switch (tok) {
556 case MDOC_Ec:
557 maxargs = 1;
558 break;
559 case MDOC_Ek:
560 mdoc->flags &= ~MDOC_KEEP;
561 /* FALLTHROUGH */
562 default:
563 maxargs = 0;
564 break;
565 }
566
567 /*
568 * Search backwards for beginnings of blocks,
569 * both of our own and of pending sub-blocks.
570 */
571
572 atok = rew_alt(tok);
573 body = endbody = itblk = later = NULL;
574 for (n = mdoc->last; n; n = n->parent) {
575 if (n->flags & (MDOC_VALID | MDOC_BREAK))
576 continue;
577
578 /* Remember the start of our own body. */
579
580 if (n->type == MDOC_BODY && atok == n->tok) {
581 if (n->end == ENDBODY_NOT) {
582 body = n;
583 n->lastline = line;
584 }
585 continue;
586 }
587
588 if (n->type != MDOC_BLOCK || n->tok == MDOC_Nm)
589 continue;
590
591 if (n->tok == MDOC_It) {
592 itblk = n;
593 continue;
594 }
595
596 if (atok == n->tok) {
597 n->lastline = line;
598 assert(body);
599
600 /*
601 * Found the start of our own block.
602 * When there is no pending sub block,
603 * just proceed to closing out.
604 */
605
606 if (later == NULL ||
607 (tok == MDOC_El && itblk == NULL))
608 break;
609
610 /*
611 * When there is a pending sub block, postpone
612 * closing out the current block until the
613 * rew_pending() closing out the sub-block.
614 */
615
616 make_pending(mdoc, n, later, line, ppos);
617 if (tok == MDOC_El)
618 itblk->flags |= MDOC_BREAK;
619
620 /*
621 * Mark the place where the formatting - but not
622 * the scope - of the current block ends.
623 */
624
625 endbody = mdoc_endbody_alloc(mdoc, line, ppos,
626 atok, body, ENDBODY_SPACE);
627
628 /*
629 * If a block closing macro taking arguments
630 * breaks another block, put the arguments
631 * into the end marker.
632 */
633
634 if (maxargs)
635 mdoc->next = MDOC_NEXT_CHILD;
636 break;
637 }
638
639 /*
640 * When finding an open sub block, remember the last
641 * open explicit block, or, in case there are only
642 * implicit ones, the first open implicit block.
643 */
644
645 if (later == NULL ||
646 ! (mdoc_macros[later->tok].flags & MDOC_EXPLICIT))
647 later = n;
648 }
649
650 if (body == NULL) {
651 mandoc_msg(MANDOCERR_BLK_NOTOPEN, mdoc->parse,
652 line, ppos, mdoc_macronames[tok]);
653 if (maxargs && endbody == NULL) {
654 /*
655 * Stray .Ec without previous .Eo:
656 * Break the output line, keep the arguments.
657 */
658 mdoc_elem_alloc(mdoc, line, ppos, MDOC_br, NULL);
659 rew_elem(mdoc, MDOC_br);
660 }
661 } else if (endbody == NULL) {
662 rew_last(mdoc, body);
663 if (maxargs)
664 mdoc_tail_alloc(mdoc, line, ppos, atok);
665 }
666
667 if ( ! (mdoc_macros[tok].flags & MDOC_PARSED)) {
668 if (buf[*pos] != '\0')
669 mandoc_vmsg(MANDOCERR_ARG_SKIP,
670 mdoc->parse, line, ppos,
671 "%s %s", mdoc_macronames[tok],
672 buf + *pos);
673 if (endbody == NULL && n != NULL)
674 rew_pending(mdoc, n);
675 return;
676 }
677
678 if (endbody != NULL)
679 n = endbody;
680 for (j = 0; ; j++) {
681 lastarg = *pos;
682
683 if (j == maxargs && n != NULL) {
684 rew_pending(mdoc, n);
685 n = NULL;
686 }
687
688 ac = mdoc_args(mdoc, line, pos, buf, tok, &p);
689 if (ac == ARGS_PUNCT || ac == ARGS_EOLN)
690 break;
691
692 ntok = ac == ARGS_QWORD ? MDOC_MAX :
693 lookup(mdoc, tok, line, lastarg, p);
694
695 if (ntok == MDOC_MAX) {
696 dword(mdoc, line, lastarg, p, DELIM_MAX,
697 MDOC_JOIN & mdoc_macros[tok].flags);
698 continue;
699 }
700
701 if (n != NULL) {
702 rew_pending(mdoc, n);
703 n = NULL;
704 }
705 mdoc->flags &= ~MDOC_NEWLINE;
706 mdoc_macro(mdoc, ntok, line, lastarg, pos, buf);
707 break;
708 }
709
710 if (n != NULL)
711 rew_pending(mdoc, n);
712 if (nl)
713 append_delims(mdoc, line, pos, buf);
714 }
715
716 static void
717 in_line(MACRO_PROT_ARGS)
718 {
719 int la, scope, cnt, firstarg, mayopen, nc, nl;
720 enum mdoct ntok;
721 enum margserr ac;
722 enum mdelim d;
723 struct mdoc_arg *arg;
724 char *p;
725
726 nl = MDOC_NEWLINE & mdoc->flags;
727
728 /*
729 * Whether we allow ignored elements (those without content,
730 * usually because of reserved words) to squeak by.
731 */
732
733 switch (tok) {
734 case MDOC_An:
735 /* FALLTHROUGH */
736 case MDOC_Ar:
737 /* FALLTHROUGH */
738 case MDOC_Fl:
739 /* FALLTHROUGH */
740 case MDOC_Mt:
741 /* FALLTHROUGH */
742 case MDOC_Nm:
743 /* FALLTHROUGH */
744 case MDOC_Pa:
745 nc = 1;
746 break;
747 default:
748 nc = 0;
749 break;
750 }
751
752 mdoc_argv(mdoc, line, tok, &arg, pos, buf);
753
754 d = DELIM_NONE;
755 firstarg = 1;
756 mayopen = 1;
757 for (cnt = scope = 0;; ) {
758 la = *pos;
759 ac = mdoc_args(mdoc, line, pos, buf, tok, &p);
760
761 /*
762 * At the end of a macro line,
763 * opening delimiters do not suppress spacing.
764 */
765
766 if (ac == ARGS_EOLN) {
767 if (d == DELIM_OPEN)
768 mdoc->last->flags &= ~MDOC_DELIMO;
769 break;
770 }
771
772 /*
773 * The rest of the macro line is only punctuation,
774 * to be handled by append_delims().
775 * If there were no other arguments,
776 * do not allow the first one to suppress spacing,
777 * even if it turns out to be a closing one.
778 */
779
780 if (ac == ARGS_PUNCT) {
781 if (cnt == 0 && nc == 0)
782 mdoc->flags |= MDOC_NODELIMC;
783 break;
784 }
785
786 ntok = (ac == ARGS_QWORD || (tok == MDOC_Fn && !cnt)) ?
787 MDOC_MAX : lookup(mdoc, tok, line, la, p);
788
789 /*
790 * In this case, we've located a submacro and must
791 * execute it. Close out scope, if open. If no
792 * elements have been generated, either create one (nc)
793 * or raise a warning.
794 */
795
796 if (ntok != MDOC_MAX) {
797 if (scope)
798 rew_elem(mdoc, tok);
799 if (nc && ! cnt) {
800 mdoc_elem_alloc(mdoc, line, ppos, tok, arg);
801 rew_last(mdoc, mdoc->last);
802 } else if ( ! nc && ! cnt) {
803 mdoc_argv_free(arg);
804 mandoc_msg(MANDOCERR_MACRO_EMPTY,
805 mdoc->parse, line, ppos,
806 mdoc_macronames[tok]);
807 }
808 mdoc_macro(mdoc, ntok, line, la, pos, buf);
809 if (nl)
810 append_delims(mdoc, line, pos, buf);
811 return;
812 }
813
814 /*
815 * Non-quote-enclosed punctuation. Set up our scope, if
816 * a word; rewind the scope, if a delimiter; then append
817 * the word.
818 */
819
820 d = ac == ARGS_QWORD ? DELIM_NONE : mdoc_isdelim(p);
821
822 if (DELIM_NONE != d) {
823 /*
824 * If we encounter closing punctuation, no word
825 * has been emitted, no scope is open, and we're
826 * allowed to have an empty element, then start
827 * a new scope.
828 */
829 if ((d == DELIM_CLOSE ||
830 (d == DELIM_MIDDLE && tok == MDOC_Fl)) &&
831 !cnt && !scope && nc && mayopen) {
832 mdoc_elem_alloc(mdoc, line, ppos, tok, arg);
833 scope = 1;
834 cnt++;
835 if (tok == MDOC_Nm)
836 mayopen = 0;
837 }
838 /*
839 * Close out our scope, if one is open, before
840 * any punctuation.
841 */
842 if (scope)
843 rew_elem(mdoc, tok);
844 scope = 0;
845 if (tok == MDOC_Fn)
846 mayopen = 0;
847 } else if (mayopen && !scope) {
848 mdoc_elem_alloc(mdoc, line, ppos, tok, arg);
849 scope = 1;
850 cnt++;
851 }
852
853 dword(mdoc, line, la, p, d,
854 MDOC_JOIN & mdoc_macros[tok].flags);
855
856 /*
857 * If the first argument is a closing delimiter,
858 * do not suppress spacing before it.
859 */
860
861 if (firstarg && d == DELIM_CLOSE && !nc)
862 mdoc->last->flags &= ~MDOC_DELIMC;
863 firstarg = 0;
864
865 /*
866 * `Fl' macros have their scope re-opened with each new
867 * word so that the `-' can be added to each one without
868 * having to parse out spaces.
869 */
870 if (scope && tok == MDOC_Fl) {
871 rew_elem(mdoc, tok);
872 scope = 0;
873 }
874 }
875
876 if (scope)
877 rew_elem(mdoc, tok);
878
879 /*
880 * If no elements have been collected and we're allowed to have
881 * empties (nc), open a scope and close it out. Otherwise,
882 * raise a warning.
883 */
884
885 if ( ! cnt) {
886 if (nc) {
887 mdoc_elem_alloc(mdoc, line, ppos, tok, arg);
888 rew_last(mdoc, mdoc->last);
889 } else {
890 mdoc_argv_free(arg);
891 mandoc_msg(MANDOCERR_MACRO_EMPTY, mdoc->parse,
892 line, ppos, mdoc_macronames[tok]);
893 }
894 }
895 if (nl)
896 append_delims(mdoc, line, pos, buf);
897 }
898
899 static void
900 blk_full(MACRO_PROT_ARGS)
901 {
902 int la, nl, parsed;
903 struct mdoc_arg *arg;
904 struct mdoc_node *blk; /* Our own or a broken block. */
905 struct mdoc_node *head; /* Our own head. */
906 struct mdoc_node *body; /* Our own body. */
907 struct mdoc_node *n;
908 enum margserr ac, lac;
909 char *p;
910
911 nl = MDOC_NEWLINE & mdoc->flags;
912
913 if ( ! (mdoc_macros[tok].flags & MDOC_EXPLICIT)) {
914
915 /* Here, tok is one of Sh Ss Nm Nd It. */
916
917 blk = NULL;
918 for (n = mdoc->last; n != NULL; n = n->parent) {
919 if (n->flags & (MDOC_VALID | MDOC_BREAK) ||
920 n->type != MDOC_BLOCK)
921 continue;
922 if (tok == MDOC_It && n->tok == MDOC_Bl) {
923 if (blk != NULL) {
924 mandoc_vmsg(MANDOCERR_BLK_BROKEN,
925 mdoc->parse, line, ppos,
926 "It breaks %s",
927 mdoc_macronames[blk->tok]);
928 rew_pending(mdoc, blk);
929 }
930 break;
931 }
932
933 if (mdoc_macros[n->tok].flags & MDOC_EXPLICIT) {
934 switch (tok) {
935 case MDOC_Sh:
936 /* FALLTHROUGH */
937 case MDOC_Ss:
938 mandoc_vmsg(MANDOCERR_BLK_BROKEN,
939 mdoc->parse, line, ppos,
940 "%s breaks %s",
941 mdoc_macronames[tok],
942 mdoc_macronames[n->tok]);
943 rew_pending(mdoc, n);
944 continue;
945 case MDOC_It:
946 /* Delay in case it's astray. */
947 blk = n;
948 continue;
949 default:
950 break;
951 }
952 break;
953 }
954
955 /* Here, n is one of Sh Ss Nm Nd It. */
956
957 if (tok != MDOC_Sh && (n->tok == MDOC_Sh ||
958 (tok != MDOC_Ss && (n->tok == MDOC_Ss ||
959 (tok != MDOC_It && n->tok == MDOC_It)))))
960 break;
961
962 /* Item breaking an explicit block. */
963
964 if (blk != NULL) {
965 mandoc_vmsg(MANDOCERR_BLK_BROKEN,
966 mdoc->parse, line, ppos,
967 "It breaks %s",
968 mdoc_macronames[blk->tok]);
969 rew_pending(mdoc, blk);
970 blk = NULL;
971 }
972
973 /* Close out prior implicit scopes. */
974
975 rew_last(mdoc, n);
976 }
977
978 /* Skip items outside lists. */
979
980 if (tok == MDOC_It && (n == NULL || n->tok != MDOC_Bl)) {
981 mandoc_vmsg(MANDOCERR_IT_STRAY, mdoc->parse,
982 line, ppos, "It %s", buf + *pos);
983 mdoc_elem_alloc(mdoc, line, ppos, MDOC_br, NULL);
984 rew_elem(mdoc, MDOC_br);
985 return;
986 }
987 }
988
989 /*
990 * This routine accommodates implicitly- and explicitly-scoped
991 * macro openings. Implicit ones first close out prior scope
992 * (seen above). Delay opening the head until necessary to
993 * allow leading punctuation to print. Special consideration
994 * for `It -column', which has phrase-part syntax instead of
995 * regular child nodes.
996 */
997
998 mdoc_argv(mdoc, line, tok, &arg, pos, buf);
999 blk = mdoc_block_alloc(mdoc, line, ppos, tok, arg);
1000 head = body = NULL;
1001
1002 /*
1003 * Exception: Heads of `It' macros in `-diag' lists are not
1004 * parsed, even though `It' macros in general are parsed.
1005 */
1006
1007 parsed = tok != MDOC_It ||
1008 mdoc->last->parent->tok != MDOC_Bl ||
1009 mdoc->last->parent->norm->Bl.type != LIST_diag;
1010
1011 /*
1012 * The `Nd' macro has all arguments in its body: it's a hybrid
1013 * of block partial-explicit and full-implicit. Stupid.
1014 */
1015
1016 if (tok == MDOC_Nd) {
1017 head = mdoc_head_alloc(mdoc, line, ppos, tok);
1018 rew_last(mdoc, head);
1019 body = mdoc_body_alloc(mdoc, line, ppos, tok);
1020 }
1021
1022 if (tok == MDOC_Bk)
1023 mdoc->flags |= MDOC_KEEP;
1024
1025 ac = ARGS_PEND;
1026 for (;;) {
1027 la = *pos;
1028 lac = ac;
1029 ac = mdoc_args(mdoc, line, pos, buf, tok, &p);
1030 if (ac == ARGS_EOLN) {
1031 if (lac != ARGS_PPHRASE && lac != ARGS_PHRASE)
1032 break;
1033 /*
1034 * This is necessary: if the last token on a
1035 * line is a `Ta' or tab, then we'll get
1036 * ARGS_EOLN, so we must be smart enough to
1037 * reopen our scope if the last parse was a
1038 * phrase or partial phrase.
1039 */
1040 if (body != NULL)
1041 rew_last(mdoc, body);
1042 body = mdoc_body_alloc(mdoc, line, ppos, tok);
1043 break;
1044 }
1045 if (tok == MDOC_Rs) {
1046 mandoc_vmsg(MANDOCERR_ARG_SKIP, mdoc->parse,
1047 line, la, "Rs %s", buf + la);
1048 break;
1049 }
1050 if (ac == ARGS_PUNCT)
1051 break;
1052
1053 /*
1054 * Emit leading punctuation (i.e., punctuation before
1055 * the MDOC_HEAD) for non-phrase types.
1056 */
1057
1058 if (head == NULL &&
1059 ac != ARGS_PEND &&
1060 ac != ARGS_PHRASE &&
1061 ac != ARGS_PPHRASE &&
1062 ac != ARGS_QWORD &&
1063 mdoc_isdelim(p) == DELIM_OPEN) {
1064 dword(mdoc, line, la, p, DELIM_OPEN, 0);
1065 continue;
1066 }
1067
1068 /* Open a head if one hasn't been opened. */
1069
1070 if (head == NULL)
1071 head = mdoc_head_alloc(mdoc, line, ppos, tok);
1072
1073 if (ac == ARGS_PHRASE ||
1074 ac == ARGS_PEND ||
1075 ac == ARGS_PPHRASE) {
1076
1077 /*
1078 * If we haven't opened a body yet, rewind the
1079 * head; if we have, rewind that instead.
1080 */
1081
1082 rew_last(mdoc, body == NULL ? head : body);
1083 body = mdoc_body_alloc(mdoc, line, ppos, tok);
1084
1085 /*
1086 * Process phrases: set whether we're in a
1087 * partial-phrase (this effects line handling)
1088 * then call down into the phrase parser.
1089 */
1090
1091 if (ac == ARGS_PPHRASE)
1092 mdoc->flags |= MDOC_PPHRASE;
1093 if (ac == ARGS_PEND && lac == ARGS_PPHRASE)
1094 mdoc->flags |= MDOC_PPHRASE;
1095 parse_rest(mdoc, MDOC_MAX, line, &la, buf);
1096 mdoc->flags &= ~MDOC_PPHRASE;
1097 continue;
1098 }
1099
1100 if (macro_or_word(mdoc, tok, line, la, pos, buf, parsed))
1101 break;
1102 }
1103
1104 if (blk->flags & MDOC_VALID)
1105 return;
1106 if (head == NULL)
1107 head = mdoc_head_alloc(mdoc, line, ppos, tok);
1108 if (nl && tok != MDOC_Rs)
1109 append_delims(mdoc, line, pos, buf);
1110 if (body != NULL)
1111 goto out;
1112
1113 /*
1114 * If there is an open (i.e., unvalidated) sub-block requiring
1115 * explicit close-out, postpone switching the current block from
1116 * head to body until the rew_pending() call closing out that
1117 * sub-block.
1118 */
1119 for (n = mdoc->last; n && n != head; n = n->parent) {
1120 if (n->type == MDOC_BLOCK &&
1121 mdoc_macros[n->tok].flags & MDOC_EXPLICIT &&
1122 ! (n->flags & MDOC_VALID)) {
1123 n->pending = head;
1124 return;
1125 }
1126 }
1127
1128 /* Close out scopes to remain in a consistent state. */
1129
1130 rew_last(mdoc, head);
1131 body = mdoc_body_alloc(mdoc, line, ppos, tok);
1132 out:
1133 if (mdoc->flags & MDOC_FREECOL) {
1134 rew_last(mdoc, body);
1135 rew_last(mdoc, blk);
1136 mdoc->flags &= ~MDOC_FREECOL;
1137 }
1138 }
1139
1140 static void
1141 blk_part_imp(MACRO_PROT_ARGS)
1142 {
1143 int la, nl;
1144 enum margserr ac;
1145 char *p;
1146 struct mdoc_node *blk; /* saved block context */
1147 struct mdoc_node *body; /* saved body context */
1148 struct mdoc_node *n;
1149
1150 nl = MDOC_NEWLINE & mdoc->flags;
1151
1152 /*
1153 * A macro that spans to the end of the line. This is generally
1154 * (but not necessarily) called as the first macro. The block
1155 * has a head as the immediate child, which is always empty,
1156 * followed by zero or more opening punctuation nodes, then the
1157 * body (which may be empty, depending on the macro), then zero
1158 * or more closing punctuation nodes.
1159 */
1160
1161 blk = mdoc_block_alloc(mdoc, line, ppos, tok, NULL);
1162 rew_last(mdoc, mdoc_head_alloc(mdoc, line, ppos, tok));
1163
1164 /*
1165 * Open the body scope "on-demand", that is, after we've
1166 * processed all our the leading delimiters (open parenthesis,
1167 * etc.).
1168 */
1169
1170 for (body = NULL; ; ) {
1171 la = *pos;
1172 ac = mdoc_args(mdoc, line, pos, buf, tok, &p);
1173 if (ac == ARGS_EOLN || ac == ARGS_PUNCT)
1174 break;
1175
1176 if (body == NULL && ac != ARGS_QWORD &&
1177 mdoc_isdelim(p) == DELIM_OPEN) {
1178 dword(mdoc, line, la, p, DELIM_OPEN, 0);
1179 continue;
1180 }
1181
1182 if (body == NULL)
1183 body = mdoc_body_alloc(mdoc, line, ppos, tok);
1184
1185 if (macro_or_word(mdoc, tok, line, la, pos, buf, 1))
1186 break;
1187 }
1188 if (body == NULL)
1189 body = mdoc_body_alloc(mdoc, line, ppos, tok);
1190
1191 /*
1192 * If there is an open sub-block requiring explicit close-out,
1193 * postpone closing out the current block until the
1194 * rew_pending() call closing out the sub-block.
1195 */
1196
1197 for (n = mdoc->last; n && n != body && n != blk->parent;
1198 n = n->parent) {
1199 if (n->type == MDOC_BLOCK &&
1200 mdoc_macros[n->tok].flags & MDOC_EXPLICIT &&
1201 ! (n->flags & MDOC_VALID)) {
1202 make_pending(mdoc, blk, n, line, ppos);
1203 mdoc_endbody_alloc(mdoc, line, ppos,
1204 tok, body, ENDBODY_NOSPACE);
1205 return;
1206 }
1207 }
1208 assert(n == body);
1209 rew_last(mdoc, body);
1210 if (nl)
1211 append_delims(mdoc, line, pos, buf);
1212 rew_pending(mdoc, blk);
1213
1214 /* Move trailing .Ns out of scope. */
1215
1216 for (n = body->child; n && n->next; n = n->next)
1217 /* Do nothing. */ ;
1218 if (n && n->tok == MDOC_Ns)
1219 mdoc_node_relink(mdoc, n);
1220 }
1221
1222 static void
1223 blk_part_exp(MACRO_PROT_ARGS)
1224 {
1225 int la, nl;
1226 enum margserr ac;
1227 struct mdoc_node *head; /* keep track of head */
1228 char *p;
1229
1230 nl = MDOC_NEWLINE & mdoc->flags;
1231
1232 /*
1233 * The opening of an explicit macro having zero or more leading
1234 * punctuation nodes; a head with optional single element (the
1235 * case of `Eo'); and a body that may be empty.
1236 */
1237
1238 mdoc_block_alloc(mdoc, line, ppos, tok, NULL);
1239 head = NULL;
1240 for (;;) {
1241 la = *pos;
1242 ac = mdoc_args(mdoc, line, pos, buf, tok, &p);
1243 if (ac == ARGS_PUNCT || ac == ARGS_EOLN)
1244 break;
1245
1246 /* Flush out leading punctuation. */
1247
1248 if (head == NULL && ac != ARGS_QWORD &&
1249 mdoc_isdelim(p) == DELIM_OPEN) {
1250 dword(mdoc, line, la, p, DELIM_OPEN, 0);
1251 continue;
1252 }
1253
1254 if (head == NULL) {
1255 head = mdoc_head_alloc(mdoc, line, ppos, tok);
1256 if (tok == MDOC_Eo) /* Not parsed. */
1257 dword(mdoc, line, la, p, DELIM_MAX, 0);
1258 rew_last(mdoc, head);
1259 mdoc_body_alloc(mdoc, line, ppos, tok);
1260 if (tok == MDOC_Eo)
1261 continue;
1262 }
1263
1264 if (macro_or_word(mdoc, tok, line, la, pos, buf, 1))
1265 break;
1266 }
1267
1268 /* Clean-up to leave in a consistent state. */
1269
1270 if (head == NULL) {
1271 rew_last(mdoc, mdoc_head_alloc(mdoc, line, ppos, tok));
1272 mdoc_body_alloc(mdoc, line, ppos, tok);
1273 }
1274 if (nl)
1275 append_delims(mdoc, line, pos, buf);
1276 }
1277
1278 static void
1279 in_line_argn(MACRO_PROT_ARGS)
1280 {
1281 int la, flushed, j, maxargs, nl;
1282 enum margserr ac;
1283 struct mdoc_arg *arg;
1284 char *p;
1285 enum mdoct ntok;
1286
1287 nl = mdoc->flags & MDOC_NEWLINE;
1288
1289 /*
1290 * A line macro that has a fixed number of arguments (maxargs).
1291 * Only open the scope once the first non-leading-punctuation is
1292 * found (unless MDOC_IGNDELIM is noted, like in `Pf'), then
1293 * keep it open until the maximum number of arguments are
1294 * exhausted.
1295 */
1296
1297 switch (tok) {
1298 case MDOC_Ap:
1299 /* FALLTHROUGH */
1300 case MDOC_Ns:
1301 /* FALLTHROUGH */
1302 case MDOC_Ux:
1303 maxargs = 0;
1304 break;
1305 case MDOC_Bx:
1306 /* FALLTHROUGH */
1307 case MDOC_Es:
1308 /* FALLTHROUGH */
1309 case MDOC_Xr:
1310 maxargs = 2;
1311 break;
1312 default:
1313 maxargs = 1;
1314 break;
1315 }
1316
1317 mdoc_argv(mdoc, line, tok, &arg, pos, buf);
1318
1319 p = NULL;
1320 flushed = j = 0;
1321 for (;;) {
1322 la = *pos;
1323 ac = mdoc_args(mdoc, line, pos, buf, tok, &p);
1324 if (ac == ARGS_PUNCT || ac == ARGS_EOLN) {
1325 if (j < 2 && tok == MDOC_Pf)
1326 mandoc_vmsg(MANDOCERR_PF_SKIP,
1327 mdoc->parse, line, ppos, "Pf %s",
1328 p == NULL ? "at eol" : p);
1329 break;
1330 }
1331
1332 if ( ! (mdoc_macros[tok].flags & MDOC_IGNDELIM) &&
1333 ac != ARGS_QWORD && j == 0 &&
1334 mdoc_isdelim(p) == DELIM_OPEN) {
1335 dword(mdoc, line, la, p, DELIM_OPEN, 0);
1336 continue;
1337 } else if (j == 0)
1338 mdoc_elem_alloc(mdoc, line, ppos, tok, arg);
1339
1340 if (j == maxargs && ! flushed) {
1341 rew_elem(mdoc, tok);
1342 flushed = 1;
1343 }
1344
1345 ntok = (ac == ARGS_QWORD || (tok == MDOC_Pf && j == 0)) ?
1346 MDOC_MAX : lookup(mdoc, tok, line, la, p);
1347
1348 if (ntok != MDOC_MAX) {
1349 if ( ! flushed)
1350 rew_elem(mdoc, tok);
1351 flushed = 1;
1352 mdoc_macro(mdoc, ntok, line, la, pos, buf);
1353 j++;
1354 break;
1355 }
1356
1357 if ( ! (mdoc_macros[tok].flags & MDOC_IGNDELIM) &&
1358 ac != ARGS_QWORD && ! flushed &&
1359 mdoc_isdelim(p) != DELIM_NONE) {
1360 rew_elem(mdoc, tok);
1361 flushed = 1;
1362 }
1363
1364 dword(mdoc, line, la, p, DELIM_MAX,
1365 MDOC_JOIN & mdoc_macros[tok].flags);
1366 j++;
1367 }
1368
1369 if (j == 0) {
1370 mdoc_elem_alloc(mdoc, line, ppos, tok, arg);
1371 if (ac == ARGS_PUNCT && tok == MDOC_Pf)
1372 append_delims(mdoc, line, pos, buf);
1373 }
1374 if ( ! flushed)
1375 rew_elem(mdoc, tok);
1376 if (nl)
1377 append_delims(mdoc, line, pos, buf);
1378 }
1379
1380 static void
1381 in_line_eoln(MACRO_PROT_ARGS)
1382 {
1383 struct mdoc_node *n;
1384 struct mdoc_arg *arg;
1385
1386 if ((tok == MDOC_Pp || tok == MDOC_Lp) &&
1387 ! (mdoc->flags & MDOC_SYNOPSIS)) {
1388 n = mdoc->last;
1389 if (mdoc->next == MDOC_NEXT_SIBLING)
1390 n = n->parent;
1391 if (n->tok == MDOC_Nm)
1392 rew_last(mdoc, mdoc->last->parent);
1393 }
1394
1395 mdoc_argv(mdoc, line, tok, &arg, pos, buf);
1396 mdoc_elem_alloc(mdoc, line, ppos, tok, arg);
1397 if (parse_rest(mdoc, tok, line, pos, buf))
1398 return;
1399 rew_elem(mdoc, tok);
1400 }
1401
1402 /*
1403 * The simplest argument parser available: Parse the remaining
1404 * words until the end of the phrase or line and return 0
1405 * or until the next macro, call that macro, and return 1.
1406 */
1407 static int
1408 parse_rest(struct mdoc *mdoc, enum mdoct tok, int line, int *pos, char *buf)
1409 {
1410 int la;
1411
1412 for (;;) {
1413 la = *pos;
1414 if (mdoc_args(mdoc, line, pos, buf, tok, NULL) == ARGS_EOLN)
1415 return(0);
1416 if (macro_or_word(mdoc, tok, line, la, pos, buf, 1))
1417 return(1);
1418 }
1419 }
1420
1421 static void
1422 ctx_synopsis(MACRO_PROT_ARGS)
1423 {
1424
1425 if (~mdoc->flags & (MDOC_SYNOPSIS | MDOC_NEWLINE))
1426 in_line(mdoc, tok, line, ppos, pos, buf);
1427 else if (tok == MDOC_Nm)
1428 blk_full(mdoc, tok, line, ppos, pos, buf);
1429 else {
1430 assert(tok == MDOC_Vt);
1431 blk_part_imp(mdoc, tok, line, ppos, pos, buf);
1432 }
1433 }
1434
1435 /*
1436 * Phrases occur within `Bl -column' entries, separated by `Ta' or tabs.
1437 * They're unusual because they're basically free-form text until a
1438 * macro is encountered.
1439 */
1440 static void
1441 phrase_ta(MACRO_PROT_ARGS)
1442 {
1443 struct mdoc_node *body, *n;
1444
1445 /* Make sure we are in a column list or ignore this macro. */
1446
1447 body = NULL;
1448 for (n = mdoc->last; n != NULL; n = n->parent) {
1449 if (n->flags & (MDOC_VALID | MDOC_BREAK))
1450 continue;
1451 if (n->tok == MDOC_It && n->type == MDOC_BODY)
1452 body = n;
1453 if (n->tok == MDOC_Bl)
1454 break;
1455 }
1456
1457 if (n == NULL || n->norm->Bl.type != LIST_column) {
1458 mandoc_msg(MANDOCERR_TA_STRAY, mdoc->parse,
1459 line, ppos, "Ta");
1460 return;
1461 }
1462
1463 /* Advance to the next column. */
1464
1465 rew_last(mdoc, body);
1466 mdoc_body_alloc(mdoc, line, ppos, MDOC_It);
1467 parse_rest(mdoc, MDOC_MAX, line, pos, buf);
1468 }