1 /* $Id: mdoc_macro.c,v 1.183 2015/02/12 12:24:33 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 int parse_rest(struct mdoc *, enum mdoct,
51 int, int *, char *);
52 static enum mdoct rew_alt(enum mdoct);
53 static void rew_elem(struct mdoc *, enum mdoct);
54 static void rew_last(struct mdoc *, const struct mdoc_node *);
55 static void rew_pending(struct mdoc *, const struct mdoc_node *);
56
57 const struct mdoc_macro __mdoc_macros[MDOC_MAX] = {
58 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* Ap */
59 { in_line_eoln, MDOC_PROLOGUE }, /* Dd */
60 { in_line_eoln, MDOC_PROLOGUE }, /* Dt */
61 { in_line_eoln, MDOC_PROLOGUE }, /* Os */
62 { blk_full, MDOC_PARSED | MDOC_JOIN }, /* Sh */
63 { blk_full, MDOC_PARSED | MDOC_JOIN }, /* Ss */
64 { in_line_eoln, 0 }, /* Pp */
65 { blk_part_imp, MDOC_PARSED | MDOC_JOIN }, /* D1 */
66 { blk_part_imp, MDOC_PARSED | MDOC_JOIN }, /* Dl */
67 { blk_full, MDOC_EXPLICIT }, /* Bd */
68 { blk_exp_close, MDOC_EXPLICIT | MDOC_JOIN }, /* Ed */
69 { blk_full, MDOC_EXPLICIT }, /* Bl */
70 { blk_exp_close, MDOC_EXPLICIT | MDOC_JOIN }, /* El */
71 { blk_full, MDOC_PARSED | MDOC_JOIN }, /* It */
72 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Ad */
73 { in_line, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* An */
74 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Ar */
75 { in_line, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* Cd */
76 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Cm */
77 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Dv */
78 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Er */
79 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Ev */
80 { in_line_eoln, 0 }, /* Ex */
81 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Fa */
82 { in_line_eoln, 0 }, /* Fd */
83 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Fl */
84 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Fn */
85 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Ft */
86 { in_line, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* Ic */
87 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* In */
88 { in_line, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* Li */
89 { blk_full, MDOC_JOIN }, /* Nd */
90 { ctx_synopsis, MDOC_CALLABLE | MDOC_PARSED }, /* Nm */
91 { blk_part_imp, MDOC_CALLABLE | MDOC_PARSED }, /* Op */
92 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Ot */
93 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Pa */
94 { in_line_eoln, 0 }, /* Rv */
95 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* St */
96 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Va */
97 { ctx_synopsis, MDOC_CALLABLE | MDOC_PARSED }, /* Vt */
98 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* Xr */
99 { in_line_eoln, MDOC_JOIN }, /* %A */
100 { in_line_eoln, MDOC_JOIN }, /* %B */
101 { in_line_eoln, MDOC_JOIN }, /* %D */
102 { in_line_eoln, MDOC_JOIN }, /* %I */
103 { in_line_eoln, MDOC_JOIN }, /* %J */
104 { in_line_eoln, 0 }, /* %N */
105 { in_line_eoln, MDOC_JOIN }, /* %O */
106 { in_line_eoln, 0 }, /* %P */
107 { in_line_eoln, MDOC_JOIN }, /* %R */
108 { in_line_eoln, MDOC_JOIN }, /* %T */
109 { in_line_eoln, 0 }, /* %V */
110 { blk_exp_close, MDOC_CALLABLE | MDOC_PARSED |
111 MDOC_EXPLICIT | MDOC_JOIN }, /* Ac */
112 { blk_part_exp, MDOC_CALLABLE | MDOC_PARSED |
113 MDOC_EXPLICIT | MDOC_JOIN }, /* Ao */
114 { blk_part_imp, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* Aq */
115 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* At */
116 { blk_exp_close, MDOC_CALLABLE | MDOC_PARSED |
117 MDOC_EXPLICIT | MDOC_JOIN }, /* Bc */
118 { blk_full, MDOC_EXPLICIT }, /* Bf */
119 { blk_part_exp, MDOC_CALLABLE | MDOC_PARSED |
120 MDOC_EXPLICIT | MDOC_JOIN }, /* Bo */
121 { blk_part_imp, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* Bq */
122 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* Bsx */
123 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* Bx */
124 { in_line_eoln, 0 }, /* Db */
125 { blk_exp_close, MDOC_CALLABLE | MDOC_PARSED |
126 MDOC_EXPLICIT | MDOC_JOIN }, /* Dc */
127 { blk_part_exp, MDOC_CALLABLE | MDOC_PARSED |
128 MDOC_EXPLICIT | MDOC_JOIN }, /* Do */
129 { blk_part_imp, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* Dq */
130 { blk_exp_close, MDOC_CALLABLE | MDOC_PARSED | MDOC_EXPLICIT }, /* Ec */
131 { blk_exp_close, MDOC_EXPLICIT | MDOC_JOIN }, /* Ef */
132 { in_line, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* Em */
133 { blk_part_exp, MDOC_CALLABLE | MDOC_PARSED | MDOC_EXPLICIT }, /* Eo */
134 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* Fx */
135 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Ms */
136 { in_line, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* No */
137 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED |
138 MDOC_IGNDELIM | MDOC_JOIN }, /* Ns */
139 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* Nx */
140 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* Ox */
141 { blk_exp_close, MDOC_CALLABLE | MDOC_PARSED |
142 MDOC_EXPLICIT | MDOC_JOIN }, /* Pc */
143 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED | MDOC_IGNDELIM }, /* Pf */
144 { blk_part_exp, MDOC_CALLABLE | MDOC_PARSED |
145 MDOC_EXPLICIT | MDOC_JOIN }, /* Po */
146 { blk_part_imp, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* Pq */
147 { blk_exp_close, MDOC_CALLABLE | MDOC_PARSED |
148 MDOC_EXPLICIT | MDOC_JOIN }, /* Qc */
149 { blk_part_imp, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* Ql */
150 { blk_part_exp, MDOC_CALLABLE | MDOC_PARSED |
151 MDOC_EXPLICIT | MDOC_JOIN }, /* Qo */
152 { blk_part_imp, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* Qq */
153 { blk_exp_close, MDOC_EXPLICIT | MDOC_JOIN }, /* Re */
154 { blk_full, MDOC_EXPLICIT }, /* Rs */
155 { blk_exp_close, MDOC_CALLABLE | MDOC_PARSED |
156 MDOC_EXPLICIT | MDOC_JOIN }, /* Sc */
157 { blk_part_exp, MDOC_CALLABLE | MDOC_PARSED |
158 MDOC_EXPLICIT | MDOC_JOIN }, /* So */
159 { blk_part_imp, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* Sq */
160 { in_line_argn, 0 }, /* Sm */
161 { in_line, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* Sx */
162 { in_line, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* Sy */
163 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Tn */
164 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* Ux */
165 { blk_exp_close, MDOC_EXPLICIT | MDOC_CALLABLE | MDOC_PARSED }, /* Xc */
166 { blk_part_exp, MDOC_CALLABLE | MDOC_PARSED | MDOC_EXPLICIT }, /* Xo */
167 { blk_full, MDOC_EXPLICIT | MDOC_CALLABLE }, /* Fo */
168 { blk_exp_close, MDOC_CALLABLE | MDOC_PARSED |
169 MDOC_EXPLICIT | MDOC_JOIN }, /* Fc */
170 { blk_part_exp, MDOC_CALLABLE | MDOC_PARSED |
171 MDOC_EXPLICIT | MDOC_JOIN }, /* Oo */
172 { blk_exp_close, MDOC_CALLABLE | MDOC_PARSED |
173 MDOC_EXPLICIT | MDOC_JOIN }, /* Oc */
174 { blk_full, MDOC_EXPLICIT }, /* Bk */
175 { blk_exp_close, MDOC_EXPLICIT | MDOC_JOIN }, /* Ek */
176 { in_line_eoln, 0 }, /* Bt */
177 { in_line_eoln, 0 }, /* Hf */
178 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Fr */
179 { in_line_eoln, 0 }, /* Ud */
180 { in_line, 0 }, /* Lb */
181 { in_line_eoln, 0 }, /* Lp */
182 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Lk */
183 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Mt */
184 { blk_part_imp, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* Brq */
185 { blk_part_exp, MDOC_CALLABLE | MDOC_PARSED |
186 MDOC_EXPLICIT | MDOC_JOIN }, /* Bro */
187 { blk_exp_close, MDOC_CALLABLE | MDOC_PARSED |
188 MDOC_EXPLICIT | MDOC_JOIN }, /* Brc */
189 { in_line_eoln, MDOC_JOIN }, /* %C */
190 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* Es */
191 { blk_part_imp, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* En */
192 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* Dx */
193 { in_line_eoln, MDOC_JOIN }, /* %Q */
194 { in_line_eoln, 0 }, /* br */
195 { in_line_eoln, 0 }, /* sp */
196 { in_line_eoln, 0 }, /* %U */
197 { phrase_ta, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* Ta */
198 { in_line_eoln, MDOC_PROLOGUE }, /* ll */
199 };
200
201 const struct mdoc_macro * const mdoc_macros = __mdoc_macros;
202
203
204 /*
205 * This is called at the end of parsing. It must traverse up the tree,
206 * closing out open [implicit] scopes. Obviously, open explicit scopes
207 * are errors.
208 */
209 void
mdoc_macroend(struct mdoc * mdoc)210 mdoc_macroend(struct mdoc *mdoc)
211 {
212 struct mdoc_node *n;
213
214 /* Scan for open explicit scopes. */
215
216 n = mdoc->last->flags & MDOC_VALID ?
217 mdoc->last->parent : mdoc->last;
218
219 for ( ; n; n = n->parent)
220 if (n->type == MDOC_BLOCK &&
221 mdoc_macros[n->tok].flags & MDOC_EXPLICIT)
222 mandoc_msg(MANDOCERR_BLK_NOEND, mdoc->parse,
223 n->line, n->pos, mdoc_macronames[n->tok]);
224
225 /* Rewind to the first. */
226
227 rew_last(mdoc, mdoc->first);
228 }
229
230 /*
231 * Look up the macro at *p called by "from",
232 * or as a line macro if from == MDOC_MAX.
233 */
234 static enum mdoct
lookup(struct mdoc * mdoc,enum mdoct from,int line,int ppos,const char * p)235 lookup(struct mdoc *mdoc, enum mdoct from, int line, int ppos, const char *p)
236 {
237 enum mdoct res;
238
239 if (from == MDOC_MAX || mdoc_macros[from].flags & MDOC_PARSED) {
240 res = mdoc_hash_find(p);
241 if (res != MDOC_MAX) {
242 if (mdoc_macros[res].flags & MDOC_CALLABLE)
243 return(res);
244 if (res != MDOC_br && res != MDOC_sp && res != MDOC_ll)
245 mandoc_msg(MANDOCERR_MACRO_CALL,
246 mdoc->parse, line, ppos, p);
247 }
248 }
249 return(MDOC_MAX);
250 }
251
252 /*
253 * Rewind up to and including a specific node.
254 */
255 static void
rew_last(struct mdoc * mdoc,const struct mdoc_node * to)256 rew_last(struct mdoc *mdoc, const struct mdoc_node *to)
257 {
258 struct mdoc_node *n, *np;
259
260 assert(to);
261 mdoc->next = MDOC_NEXT_SIBLING;
262 while (mdoc->last != to) {
263 /*
264 * Save the parent here, because we may delete the
265 * mdoc->last node in the post-validation phase and reset
266 * it to mdoc->last->parent, causing a step in the closing
267 * out to be lost.
268 */
269 np = mdoc->last->parent;
270 mdoc_valid_post(mdoc);
271 n = mdoc->last;
272 mdoc->last = np;
273 assert(mdoc->last);
274 mdoc->last->last = n;
275 }
276 mdoc_valid_post(mdoc);
277 }
278
279 /*
280 * Rewind up to a specific block, including all blocks that broke it.
281 */
282 static void
rew_pending(struct mdoc * mdoc,const struct mdoc_node * n)283 rew_pending(struct mdoc *mdoc, const struct mdoc_node *n)
284 {
285
286 for (;;) {
287 rew_last(mdoc, n);
288
289 switch (n->type) {
290 case MDOC_HEAD:
291 mdoc_body_alloc(mdoc, n->line, n->pos, n->tok);
292 return;
293 case MDOC_BLOCK:
294 break;
295 default:
296 return;
297 }
298
299 if ( ! (n->flags & MDOC_BROKEN))
300 return;
301
302 for (;;) {
303 if ((n = n->parent) == NULL)
304 return;
305
306 if (n->type == MDOC_BLOCK ||
307 n->type == MDOC_HEAD) {
308 if (n->flags & MDOC_ENDED)
309 break;
310 else
311 return;
312 }
313 }
314 }
315 }
316
317 /*
318 * For a block closing macro, return the corresponding opening one.
319 * Otherwise, return the macro itself.
320 */
321 static enum mdoct
rew_alt(enum mdoct tok)322 rew_alt(enum mdoct tok)
323 {
324 switch (tok) {
325 case MDOC_Ac:
326 return(MDOC_Ao);
327 case MDOC_Bc:
328 return(MDOC_Bo);
329 case MDOC_Brc:
330 return(MDOC_Bro);
331 case MDOC_Dc:
332 return(MDOC_Do);
333 case MDOC_Ec:
334 return(MDOC_Eo);
335 case MDOC_Ed:
336 return(MDOC_Bd);
337 case MDOC_Ef:
338 return(MDOC_Bf);
339 case MDOC_Ek:
340 return(MDOC_Bk);
341 case MDOC_El:
342 return(MDOC_Bl);
343 case MDOC_Fc:
344 return(MDOC_Fo);
345 case MDOC_Oc:
346 return(MDOC_Oo);
347 case MDOC_Pc:
348 return(MDOC_Po);
349 case MDOC_Qc:
350 return(MDOC_Qo);
351 case MDOC_Re:
352 return(MDOC_Rs);
353 case MDOC_Sc:
354 return(MDOC_So);
355 case MDOC_Xc:
356 return(MDOC_Xo);
357 default:
358 return(tok);
359 }
360 /* NOTREACHED */
361 }
362
363 static void
rew_elem(struct mdoc * mdoc,enum mdoct tok)364 rew_elem(struct mdoc *mdoc, enum mdoct tok)
365 {
366 struct mdoc_node *n;
367
368 n = mdoc->last;
369 if (MDOC_ELEM != n->type)
370 n = n->parent;
371 assert(MDOC_ELEM == n->type);
372 assert(tok == n->tok);
373 rew_last(mdoc, n);
374 }
375
376 /*
377 * Allocate a word and check whether it's punctuation or not.
378 * Punctuation consists of those tokens found in mdoc_isdelim().
379 */
380 static void
dword(struct mdoc * mdoc,int line,int col,const char * p,enum mdelim d,int may_append)381 dword(struct mdoc *mdoc, int line, int col, const char *p,
382 enum mdelim d, int may_append)
383 {
384
385 if (d == DELIM_MAX)
386 d = mdoc_isdelim(p);
387
388 if (may_append &&
389 ! (mdoc->flags & (MDOC_SYNOPSIS | MDOC_KEEP | MDOC_SMOFF)) &&
390 d == DELIM_NONE && mdoc->last->type == MDOC_TEXT &&
391 mdoc_isdelim(mdoc->last->string) == DELIM_NONE) {
392 mdoc_word_append(mdoc, p);
393 return;
394 }
395
396 mdoc_word_alloc(mdoc, line, col, p);
397
398 /*
399 * If the word consists of a bare delimiter,
400 * flag the new node accordingly,
401 * unless doing so was vetoed by the invoking macro.
402 * Always clear the veto, it is only valid for one word.
403 */
404
405 if (d == DELIM_OPEN)
406 mdoc->last->flags |= MDOC_DELIMO;
407 else if (d == DELIM_CLOSE &&
408 ! (mdoc->flags & MDOC_NODELIMC) &&
409 mdoc->last->parent->tok != MDOC_Fd)
410 mdoc->last->flags |= MDOC_DELIMC;
411 mdoc->flags &= ~MDOC_NODELIMC;
412 }
413
414 static void
append_delims(struct mdoc * mdoc,int line,int * pos,char * buf)415 append_delims(struct mdoc *mdoc, int line, int *pos, char *buf)
416 {
417 char *p;
418 int la;
419
420 if (buf[*pos] == '\0')
421 return;
422
423 for (;;) {
424 la = *pos;
425 if (mdoc_args(mdoc, line, pos, buf, MDOC_MAX, &p) == ARGS_EOLN)
426 break;
427 dword(mdoc, line, la, p, DELIM_MAX, 1);
428
429 /*
430 * If we encounter end-of-sentence symbols, then trigger
431 * the double-space.
432 *
433 * XXX: it's easy to allow this to propagate outward to
434 * the last symbol, such that `. )' will cause the
435 * correct double-spacing. However, (1) groff isn't
436 * smart enough to do this and (2) it would require
437 * knowing which symbols break this behaviour, for
438 * example, `. ;' shouldn't propagate the double-space.
439 */
440
441 if (mandoc_eos(p, strlen(p)))
442 mdoc->last->flags |= MDOC_EOS;
443 }
444 }
445
446 /*
447 * Parse one word.
448 * If it is a macro, call it and return 1.
449 * Otherwise, allocate it and return 0.
450 */
451 static int
macro_or_word(MACRO_PROT_ARGS,int parsed)452 macro_or_word(MACRO_PROT_ARGS, int parsed)
453 {
454 char *p;
455 enum mdoct ntok;
456
457 p = buf + ppos;
458 ntok = MDOC_MAX;
459 if (*p == '"')
460 p++;
461 else if (parsed && ! (mdoc->flags & MDOC_PHRASELIT))
462 ntok = lookup(mdoc, tok, line, ppos, p);
463
464 if (ntok == MDOC_MAX) {
465 dword(mdoc, line, ppos, p, DELIM_MAX, tok == MDOC_MAX ||
466 mdoc_macros[tok].flags & MDOC_JOIN);
467 return(0);
468 } else {
469 if (mdoc_macros[tok].fp == in_line_eoln)
470 rew_elem(mdoc, tok);
471 mdoc_macro(mdoc, ntok, line, ppos, pos, buf);
472 if (tok == MDOC_MAX)
473 append_delims(mdoc, line, pos, buf);
474 return(1);
475 }
476 }
477
478 /*
479 * Close out block partial/full explicit.
480 */
481 static void
blk_exp_close(MACRO_PROT_ARGS)482 blk_exp_close(MACRO_PROT_ARGS)
483 {
484 struct mdoc_node *body; /* Our own body. */
485 struct mdoc_node *endbody; /* Our own end marker. */
486 struct mdoc_node *itblk; /* An It block starting later. */
487 struct mdoc_node *later; /* A sub-block starting later. */
488 struct mdoc_node *n; /* Search back to our block. */
489
490 int j, lastarg, maxargs, nl;
491 enum margserr ac;
492 enum mdoct atok, ntok;
493 char *p;
494
495 nl = MDOC_NEWLINE & mdoc->flags;
496
497 switch (tok) {
498 case MDOC_Ec:
499 maxargs = 1;
500 break;
501 case MDOC_Ek:
502 mdoc->flags &= ~MDOC_KEEP;
503 /* FALLTHROUGH */
504 default:
505 maxargs = 0;
506 break;
507 }
508
509 /*
510 * Search backwards for beginnings of blocks,
511 * both of our own and of pending sub-blocks.
512 */
513
514 atok = rew_alt(tok);
515 body = endbody = itblk = later = NULL;
516 for (n = mdoc->last; n; n = n->parent) {
517 if (n->flags & MDOC_ENDED) {
518 if ( ! (n->flags & MDOC_VALID))
519 n->flags |= MDOC_BROKEN;
520 continue;
521 }
522
523 /* Remember the start of our own body. */
524
525 if (n->type == MDOC_BODY && atok == n->tok) {
526 if (n->end == ENDBODY_NOT)
527 body = n;
528 continue;
529 }
530
531 if (n->type != MDOC_BLOCK || n->tok == MDOC_Nm)
532 continue;
533
534 if (n->tok == MDOC_It) {
535 itblk = n;
536 continue;
537 }
538
539 if (atok == n->tok) {
540 assert(body);
541
542 /*
543 * Found the start of our own block.
544 * When there is no pending sub block,
545 * just proceed to closing out.
546 */
547
548 if (later == NULL ||
549 (tok == MDOC_El && itblk == NULL))
550 break;
551
552 /*
553 * When there is a pending sub block, postpone
554 * closing out the current block until the
555 * rew_pending() closing out the sub-block.
556 * Mark the place where the formatting - but not
557 * the scope - of the current block ends.
558 */
559
560 mandoc_vmsg(MANDOCERR_BLK_NEST, mdoc->parse,
561 line, ppos, "%s breaks %s",
562 mdoc_macronames[atok],
563 mdoc_macronames[later->tok]);
564
565 endbody = mdoc_endbody_alloc(mdoc, line, ppos,
566 atok, body, ENDBODY_SPACE);
567
568 if (tok == MDOC_El)
569 itblk->flags |= MDOC_ENDED | MDOC_BROKEN;
570
571 /*
572 * If a block closing macro taking arguments
573 * breaks another block, put the arguments
574 * into the end marker.
575 */
576
577 if (maxargs)
578 mdoc->next = MDOC_NEXT_CHILD;
579 break;
580 }
581
582 /* Explicit blocks close out description lines. */
583
584 if (n->tok == MDOC_Nd) {
585 rew_last(mdoc, n);
586 continue;
587 }
588
589 /* Breaking an open sub block. */
590
591 n->flags |= MDOC_BROKEN;
592 if (later == NULL)
593 later = n;
594 }
595
596 if (body == NULL) {
597 mandoc_msg(MANDOCERR_BLK_NOTOPEN, mdoc->parse,
598 line, ppos, mdoc_macronames[tok]);
599 if (maxargs && endbody == NULL) {
600 /*
601 * Stray .Ec without previous .Eo:
602 * Break the output line, keep the arguments.
603 */
604 mdoc_elem_alloc(mdoc, line, ppos, MDOC_br, NULL);
605 rew_elem(mdoc, MDOC_br);
606 }
607 } else if (endbody == NULL) {
608 rew_last(mdoc, body);
609 if (maxargs)
610 mdoc_tail_alloc(mdoc, line, ppos, atok);
611 }
612
613 if ( ! (mdoc_macros[tok].flags & MDOC_PARSED)) {
614 if (buf[*pos] != '\0')
615 mandoc_vmsg(MANDOCERR_ARG_SKIP,
616 mdoc->parse, line, ppos,
617 "%s %s", mdoc_macronames[tok],
618 buf + *pos);
619 if (endbody == NULL && n != NULL)
620 rew_pending(mdoc, n);
621 return;
622 }
623
624 if (endbody != NULL)
625 n = endbody;
626 for (j = 0; ; j++) {
627 lastarg = *pos;
628
629 if (j == maxargs && n != NULL) {
630 rew_pending(mdoc, n);
631 n = NULL;
632 }
633
634 ac = mdoc_args(mdoc, line, pos, buf, tok, &p);
635 if (ac == ARGS_PUNCT || ac == ARGS_EOLN)
636 break;
637
638 ntok = ac == ARGS_QWORD ? MDOC_MAX :
639 lookup(mdoc, tok, line, lastarg, p);
640
641 if (ntok == MDOC_MAX) {
642 dword(mdoc, line, lastarg, p, DELIM_MAX,
643 MDOC_JOIN & mdoc_macros[tok].flags);
644 continue;
645 }
646
647 if (n != NULL) {
648 rew_pending(mdoc, n);
649 n = NULL;
650 }
651 mdoc->flags &= ~MDOC_NEWLINE;
652 mdoc_macro(mdoc, ntok, line, lastarg, pos, buf);
653 break;
654 }
655
656 if (n != NULL)
657 rew_pending(mdoc, n);
658 if (nl)
659 append_delims(mdoc, line, pos, buf);
660 }
661
662 static void
in_line(MACRO_PROT_ARGS)663 in_line(MACRO_PROT_ARGS)
664 {
665 int la, scope, cnt, firstarg, mayopen, nc, nl;
666 enum mdoct ntok;
667 enum margserr ac;
668 enum mdelim d;
669 struct mdoc_arg *arg;
670 char *p;
671
672 nl = MDOC_NEWLINE & mdoc->flags;
673
674 /*
675 * Whether we allow ignored elements (those without content,
676 * usually because of reserved words) to squeak by.
677 */
678
679 switch (tok) {
680 case MDOC_An:
681 /* FALLTHROUGH */
682 case MDOC_Ar:
683 /* FALLTHROUGH */
684 case MDOC_Fl:
685 /* FALLTHROUGH */
686 case MDOC_Mt:
687 /* FALLTHROUGH */
688 case MDOC_Nm:
689 /* FALLTHROUGH */
690 case MDOC_Pa:
691 nc = 1;
692 break;
693 default:
694 nc = 0;
695 break;
696 }
697
698 mdoc_argv(mdoc, line, tok, &arg, pos, buf);
699
700 d = DELIM_NONE;
701 firstarg = 1;
702 mayopen = 1;
703 for (cnt = scope = 0;; ) {
704 la = *pos;
705 ac = mdoc_args(mdoc, line, pos, buf, tok, &p);
706
707 /*
708 * At the end of a macro line,
709 * opening delimiters do not suppress spacing.
710 */
711
712 if (ac == ARGS_EOLN) {
713 if (d == DELIM_OPEN)
714 mdoc->last->flags &= ~MDOC_DELIMO;
715 break;
716 }
717
718 /*
719 * The rest of the macro line is only punctuation,
720 * to be handled by append_delims().
721 * If there were no other arguments,
722 * do not allow the first one to suppress spacing,
723 * even if it turns out to be a closing one.
724 */
725
726 if (ac == ARGS_PUNCT) {
727 if (cnt == 0 && (nc == 0 || tok == MDOC_An))
728 mdoc->flags |= MDOC_NODELIMC;
729 break;
730 }
731
732 ntok = (ac == ARGS_QWORD || (tok == MDOC_Fn && !cnt)) ?
733 MDOC_MAX : lookup(mdoc, tok, line, la, p);
734
735 /*
736 * In this case, we've located a submacro and must
737 * execute it. Close out scope, if open. If no
738 * elements have been generated, either create one (nc)
739 * or raise a warning.
740 */
741
742 if (ntok != MDOC_MAX) {
743 if (scope)
744 rew_elem(mdoc, tok);
745 if (nc && ! cnt) {
746 mdoc_elem_alloc(mdoc, line, ppos, tok, arg);
747 rew_last(mdoc, mdoc->last);
748 } else if ( ! nc && ! cnt) {
749 mdoc_argv_free(arg);
750 mandoc_msg(MANDOCERR_MACRO_EMPTY,
751 mdoc->parse, line, ppos,
752 mdoc_macronames[tok]);
753 }
754 mdoc_macro(mdoc, ntok, line, la, pos, buf);
755 if (nl)
756 append_delims(mdoc, line, pos, buf);
757 return;
758 }
759
760 /*
761 * Non-quote-enclosed punctuation. Set up our scope, if
762 * a word; rewind the scope, if a delimiter; then append
763 * the word.
764 */
765
766 d = ac == ARGS_QWORD ? DELIM_NONE : mdoc_isdelim(p);
767
768 if (DELIM_NONE != d) {
769 /*
770 * If we encounter closing punctuation, no word
771 * has been emitted, no scope is open, and we're
772 * allowed to have an empty element, then start
773 * a new scope.
774 */
775 if ((d == DELIM_CLOSE ||
776 (d == DELIM_MIDDLE && tok == MDOC_Fl)) &&
777 !cnt && !scope && nc && mayopen) {
778 mdoc_elem_alloc(mdoc, line, ppos, tok, arg);
779 scope = 1;
780 cnt++;
781 if (tok == MDOC_Nm)
782 mayopen = 0;
783 }
784 /*
785 * Close out our scope, if one is open, before
786 * any punctuation.
787 */
788 if (scope)
789 rew_elem(mdoc, tok);
790 scope = 0;
791 if (tok == MDOC_Fn)
792 mayopen = 0;
793 } else if (mayopen && !scope) {
794 mdoc_elem_alloc(mdoc, line, ppos, tok, arg);
795 scope = 1;
796 cnt++;
797 }
798
799 dword(mdoc, line, la, p, d,
800 MDOC_JOIN & mdoc_macros[tok].flags);
801
802 /*
803 * If the first argument is a closing delimiter,
804 * do not suppress spacing before it.
805 */
806
807 if (firstarg && d == DELIM_CLOSE && !nc)
808 mdoc->last->flags &= ~MDOC_DELIMC;
809 firstarg = 0;
810
811 /*
812 * `Fl' macros have their scope re-opened with each new
813 * word so that the `-' can be added to each one without
814 * having to parse out spaces.
815 */
816 if (scope && tok == MDOC_Fl) {
817 rew_elem(mdoc, tok);
818 scope = 0;
819 }
820 }
821
822 if (scope)
823 rew_elem(mdoc, tok);
824
825 /*
826 * If no elements have been collected and we're allowed to have
827 * empties (nc), open a scope and close it out. Otherwise,
828 * raise a warning.
829 */
830
831 if ( ! cnt) {
832 if (nc) {
833 mdoc_elem_alloc(mdoc, line, ppos, tok, arg);
834 rew_last(mdoc, mdoc->last);
835 } else {
836 mdoc_argv_free(arg);
837 mandoc_msg(MANDOCERR_MACRO_EMPTY, mdoc->parse,
838 line, ppos, mdoc_macronames[tok]);
839 }
840 }
841 if (nl)
842 append_delims(mdoc, line, pos, buf);
843 }
844
845 static void
blk_full(MACRO_PROT_ARGS)846 blk_full(MACRO_PROT_ARGS)
847 {
848 int la, nl, parsed;
849 struct mdoc_arg *arg;
850 struct mdoc_node *blk; /* Our own or a broken block. */
851 struct mdoc_node *head; /* Our own head. */
852 struct mdoc_node *body; /* Our own body. */
853 struct mdoc_node *n;
854 enum margserr ac, lac;
855 char *p;
856
857 nl = MDOC_NEWLINE & mdoc->flags;
858
859 if (buf[*pos] == '\0' && (tok == MDOC_Sh || tok == MDOC_Ss)) {
860 mandoc_msg(MANDOCERR_MACRO_EMPTY, mdoc->parse,
861 line, ppos, mdoc_macronames[tok]);
862 return;
863 }
864
865 if ( ! (mdoc_macros[tok].flags & MDOC_EXPLICIT)) {
866
867 /* Here, tok is one of Sh Ss Nm Nd It. */
868
869 blk = NULL;
870 for (n = mdoc->last; n != NULL; n = n->parent) {
871 if (n->flags & MDOC_ENDED) {
872 if ( ! (n->flags & MDOC_VALID))
873 n->flags |= MDOC_BROKEN;
874 continue;
875 }
876 if (n->type != MDOC_BLOCK)
877 continue;
878
879 if (tok == MDOC_It && n->tok == MDOC_Bl) {
880 if (blk != NULL) {
881 mandoc_vmsg(MANDOCERR_BLK_BROKEN,
882 mdoc->parse, line, ppos,
883 "It breaks %s",
884 mdoc_macronames[blk->tok]);
885 rew_pending(mdoc, blk);
886 }
887 break;
888 }
889
890 if (mdoc_macros[n->tok].flags & MDOC_EXPLICIT) {
891 switch (tok) {
892 case MDOC_Sh:
893 /* FALLTHROUGH */
894 case MDOC_Ss:
895 mandoc_vmsg(MANDOCERR_BLK_BROKEN,
896 mdoc->parse, line, ppos,
897 "%s breaks %s",
898 mdoc_macronames[tok],
899 mdoc_macronames[n->tok]);
900 rew_pending(mdoc, n);
901 n = mdoc->last;
902 continue;
903 case MDOC_It:
904 /* Delay in case it's astray. */
905 blk = n;
906 continue;
907 default:
908 break;
909 }
910 break;
911 }
912
913 /* Here, n is one of Sh Ss Nm Nd It. */
914
915 if (tok != MDOC_Sh && (n->tok == MDOC_Sh ||
916 (tok != MDOC_Ss && (n->tok == MDOC_Ss ||
917 (tok != MDOC_It && n->tok == MDOC_It)))))
918 break;
919
920 /* Item breaking an explicit block. */
921
922 if (blk != NULL) {
923 mandoc_vmsg(MANDOCERR_BLK_BROKEN,
924 mdoc->parse, line, ppos,
925 "It breaks %s",
926 mdoc_macronames[blk->tok]);
927 rew_pending(mdoc, blk);
928 blk = NULL;
929 }
930
931 /* Close out prior implicit scopes. */
932
933 rew_last(mdoc, n);
934 }
935
936 /* Skip items outside lists. */
937
938 if (tok == MDOC_It && (n == NULL || n->tok != MDOC_Bl)) {
939 mandoc_vmsg(MANDOCERR_IT_STRAY, mdoc->parse,
940 line, ppos, "It %s", buf + *pos);
941 mdoc_elem_alloc(mdoc, line, ppos, MDOC_br, NULL);
942 rew_elem(mdoc, MDOC_br);
943 return;
944 }
945 }
946
947 /*
948 * This routine accommodates implicitly- and explicitly-scoped
949 * macro openings. Implicit ones first close out prior scope
950 * (seen above). Delay opening the head until necessary to
951 * allow leading punctuation to print. Special consideration
952 * for `It -column', which has phrase-part syntax instead of
953 * regular child nodes.
954 */
955
956 mdoc_argv(mdoc, line, tok, &arg, pos, buf);
957 blk = mdoc_block_alloc(mdoc, line, ppos, tok, arg);
958 head = body = NULL;
959
960 /*
961 * Exception: Heads of `It' macros in `-diag' lists are not
962 * parsed, even though `It' macros in general are parsed.
963 */
964
965 parsed = tok != MDOC_It ||
966 mdoc->last->parent->tok != MDOC_Bl ||
967 mdoc->last->parent->norm->Bl.type != LIST_diag;
968
969 /*
970 * The `Nd' macro has all arguments in its body: it's a hybrid
971 * of block partial-explicit and full-implicit. Stupid.
972 */
973
974 if (tok == MDOC_Nd) {
975 head = mdoc_head_alloc(mdoc, line, ppos, tok);
976 rew_last(mdoc, head);
977 body = mdoc_body_alloc(mdoc, line, ppos, tok);
978 }
979
980 if (tok == MDOC_Bk)
981 mdoc->flags |= MDOC_KEEP;
982
983 ac = ARGS_PEND;
984 for (;;) {
985 la = *pos;
986 lac = ac;
987 ac = mdoc_args(mdoc, line, pos, buf, tok, &p);
988 if (ac == ARGS_EOLN) {
989 if (lac != ARGS_PPHRASE && lac != ARGS_PHRASE)
990 break;
991 /*
992 * This is necessary: if the last token on a
993 * line is a `Ta' or tab, then we'll get
994 * ARGS_EOLN, so we must be smart enough to
995 * reopen our scope if the last parse was a
996 * phrase or partial phrase.
997 */
998 if (body != NULL)
999 rew_last(mdoc, body);
1000 body = mdoc_body_alloc(mdoc, line, ppos, tok);
1001 break;
1002 }
1003 if (tok == MDOC_Bd || tok == MDOC_Bk) {
1004 mandoc_vmsg(MANDOCERR_ARG_EXCESS,
1005 mdoc->parse, line, la, "%s ... %s",
1006 mdoc_macronames[tok], buf + la);
1007 break;
1008 }
1009 if (tok == MDOC_Rs) {
1010 mandoc_vmsg(MANDOCERR_ARG_SKIP, mdoc->parse,
1011 line, la, "Rs %s", buf + la);
1012 break;
1013 }
1014 if (ac == ARGS_PUNCT)
1015 break;
1016
1017 /*
1018 * Emit leading punctuation (i.e., punctuation before
1019 * the MDOC_HEAD) for non-phrase types.
1020 */
1021
1022 if (head == NULL &&
1023 ac != ARGS_PEND &&
1024 ac != ARGS_PHRASE &&
1025 ac != ARGS_PPHRASE &&
1026 ac != ARGS_QWORD &&
1027 mdoc_isdelim(p) == DELIM_OPEN) {
1028 dword(mdoc, line, la, p, DELIM_OPEN, 0);
1029 continue;
1030 }
1031
1032 /* Open a head if one hasn't been opened. */
1033
1034 if (head == NULL)
1035 head = mdoc_head_alloc(mdoc, line, ppos, tok);
1036
1037 if (ac == ARGS_PHRASE ||
1038 ac == ARGS_PEND ||
1039 ac == ARGS_PPHRASE) {
1040
1041 /*
1042 * If we haven't opened a body yet, rewind the
1043 * head; if we have, rewind that instead.
1044 */
1045
1046 rew_last(mdoc, body == NULL ? head : body);
1047 body = mdoc_body_alloc(mdoc, line, ppos, tok);
1048
1049 /*
1050 * Process phrases: set whether we're in a
1051 * partial-phrase (this effects line handling)
1052 * then call down into the phrase parser.
1053 */
1054
1055 if (ac == ARGS_PPHRASE)
1056 mdoc->flags |= MDOC_PPHRASE;
1057 if (ac == ARGS_PEND && lac == ARGS_PPHRASE)
1058 mdoc->flags |= MDOC_PPHRASE;
1059 parse_rest(mdoc, MDOC_MAX, line, &la, buf);
1060 mdoc->flags &= ~MDOC_PPHRASE;
1061 continue;
1062 }
1063
1064 if (macro_or_word(mdoc, tok, line, la, pos, buf, parsed))
1065 break;
1066 }
1067
1068 if (blk->flags & MDOC_VALID)
1069 return;
1070 if (head == NULL)
1071 head = mdoc_head_alloc(mdoc, line, ppos, tok);
1072 if (nl && tok != MDOC_Bd && tok != MDOC_Bl && tok != MDOC_Rs)
1073 append_delims(mdoc, line, pos, buf);
1074 if (body != NULL)
1075 goto out;
1076
1077 /*
1078 * If there is an open (i.e., unvalidated) sub-block requiring
1079 * explicit close-out, postpone switching the current block from
1080 * head to body until the rew_pending() call closing out that
1081 * sub-block.
1082 */
1083 for (n = mdoc->last; n && n != head; n = n->parent) {
1084 if (n->flags & MDOC_ENDED) {
1085 if ( ! (n->flags & MDOC_VALID))
1086 n->flags |= MDOC_BROKEN;
1087 continue;
1088 }
1089 if (n->type == MDOC_BLOCK &&
1090 mdoc_macros[n->tok].flags & MDOC_EXPLICIT) {
1091 n->flags = MDOC_BROKEN;
1092 head->flags = MDOC_ENDED;
1093 }
1094 }
1095 if (head->flags & MDOC_ENDED)
1096 return;
1097
1098 /* Close out scopes to remain in a consistent state. */
1099
1100 rew_last(mdoc, head);
1101 body = mdoc_body_alloc(mdoc, line, ppos, tok);
1102 out:
1103 if (mdoc->flags & MDOC_FREECOL) {
1104 rew_last(mdoc, body);
1105 rew_last(mdoc, blk);
1106 mdoc->flags &= ~MDOC_FREECOL;
1107 }
1108 }
1109
1110 static void
blk_part_imp(MACRO_PROT_ARGS)1111 blk_part_imp(MACRO_PROT_ARGS)
1112 {
1113 int la, nl;
1114 enum margserr ac;
1115 char *p;
1116 struct mdoc_node *blk; /* saved block context */
1117 struct mdoc_node *body; /* saved body context */
1118 struct mdoc_node *n;
1119
1120 nl = MDOC_NEWLINE & mdoc->flags;
1121
1122 /*
1123 * A macro that spans to the end of the line. This is generally
1124 * (but not necessarily) called as the first macro. The block
1125 * has a head as the immediate child, which is always empty,
1126 * followed by zero or more opening punctuation nodes, then the
1127 * body (which may be empty, depending on the macro), then zero
1128 * or more closing punctuation nodes.
1129 */
1130
1131 blk = mdoc_block_alloc(mdoc, line, ppos, tok, NULL);
1132 rew_last(mdoc, mdoc_head_alloc(mdoc, line, ppos, tok));
1133
1134 /*
1135 * Open the body scope "on-demand", that is, after we've
1136 * processed all our the leading delimiters (open parenthesis,
1137 * etc.).
1138 */
1139
1140 for (body = NULL; ; ) {
1141 la = *pos;
1142 ac = mdoc_args(mdoc, line, pos, buf, tok, &p);
1143 if (ac == ARGS_EOLN || ac == ARGS_PUNCT)
1144 break;
1145
1146 if (body == NULL && ac != ARGS_QWORD &&
1147 mdoc_isdelim(p) == DELIM_OPEN) {
1148 dword(mdoc, line, la, p, DELIM_OPEN, 0);
1149 continue;
1150 }
1151
1152 if (body == NULL)
1153 body = mdoc_body_alloc(mdoc, line, ppos, tok);
1154
1155 if (macro_or_word(mdoc, tok, line, la, pos, buf, 1))
1156 break;
1157 }
1158 if (body == NULL)
1159 body = mdoc_body_alloc(mdoc, line, ppos, tok);
1160
1161 /*
1162 * If there is an open sub-block requiring explicit close-out,
1163 * postpone closing out the current block until the
1164 * rew_pending() call closing out the sub-block.
1165 */
1166
1167 for (n = mdoc->last; n && n != body && n != blk->parent;
1168 n = n->parent) {
1169 if (n->flags & MDOC_ENDED) {
1170 if ( ! (n->flags & MDOC_VALID))
1171 n->flags |= MDOC_BROKEN;
1172 continue;
1173 }
1174 if (n->type == MDOC_BLOCK &&
1175 mdoc_macros[n->tok].flags & MDOC_EXPLICIT) {
1176 n->flags |= MDOC_BROKEN;
1177 if ( ! (body->flags & MDOC_ENDED)) {
1178 mandoc_vmsg(MANDOCERR_BLK_NEST,
1179 mdoc->parse, line, ppos,
1180 "%s breaks %s", mdoc_macronames[tok],
1181 mdoc_macronames[n->tok]);
1182 mdoc_endbody_alloc(mdoc, line, ppos,
1183 tok, body, ENDBODY_NOSPACE);
1184 }
1185 }
1186 }
1187 assert(n == body);
1188 if (body->flags & MDOC_ENDED)
1189 return;
1190
1191 rew_last(mdoc, body);
1192 if (nl)
1193 append_delims(mdoc, line, pos, buf);
1194 rew_pending(mdoc, blk);
1195
1196 /* Move trailing .Ns out of scope. */
1197
1198 for (n = body->child; n && n->next; n = n->next)
1199 /* Do nothing. */ ;
1200 if (n && n->tok == MDOC_Ns)
1201 mdoc_node_relink(mdoc, n);
1202 }
1203
1204 static void
blk_part_exp(MACRO_PROT_ARGS)1205 blk_part_exp(MACRO_PROT_ARGS)
1206 {
1207 int la, nl;
1208 enum margserr ac;
1209 struct mdoc_node *head; /* keep track of head */
1210 char *p;
1211
1212 nl = MDOC_NEWLINE & mdoc->flags;
1213
1214 /*
1215 * The opening of an explicit macro having zero or more leading
1216 * punctuation nodes; a head with optional single element (the
1217 * case of `Eo'); and a body that may be empty.
1218 */
1219
1220 mdoc_block_alloc(mdoc, line, ppos, tok, NULL);
1221 head = NULL;
1222 for (;;) {
1223 la = *pos;
1224 ac = mdoc_args(mdoc, line, pos, buf, tok, &p);
1225 if (ac == ARGS_PUNCT || ac == ARGS_EOLN)
1226 break;
1227
1228 /* Flush out leading punctuation. */
1229
1230 if (head == NULL && ac != ARGS_QWORD &&
1231 mdoc_isdelim(p) == DELIM_OPEN) {
1232 dword(mdoc, line, la, p, DELIM_OPEN, 0);
1233 continue;
1234 }
1235
1236 if (head == NULL) {
1237 head = mdoc_head_alloc(mdoc, line, ppos, tok);
1238 if (tok == MDOC_Eo) /* Not parsed. */
1239 dword(mdoc, line, la, p, DELIM_MAX, 0);
1240 rew_last(mdoc, head);
1241 mdoc_body_alloc(mdoc, line, ppos, tok);
1242 if (tok == MDOC_Eo)
1243 continue;
1244 }
1245
1246 if (macro_or_word(mdoc, tok, line, la, pos, buf, 1))
1247 break;
1248 }
1249
1250 /* Clean-up to leave in a consistent state. */
1251
1252 if (head == NULL) {
1253 rew_last(mdoc, mdoc_head_alloc(mdoc, line, ppos, tok));
1254 mdoc_body_alloc(mdoc, line, ppos, tok);
1255 }
1256 if (nl)
1257 append_delims(mdoc, line, pos, buf);
1258 }
1259
1260 static void
in_line_argn(MACRO_PROT_ARGS)1261 in_line_argn(MACRO_PROT_ARGS)
1262 {
1263 struct mdoc_arg *arg;
1264 char *p;
1265 enum margserr ac;
1266 enum mdoct ntok;
1267 int state; /* arg#; -1: not yet open; -2: closed */
1268 int la, maxargs, nl;
1269
1270 nl = mdoc->flags & MDOC_NEWLINE;
1271
1272 /*
1273 * A line macro that has a fixed number of arguments (maxargs).
1274 * Only open the scope once the first non-leading-punctuation is
1275 * found (unless MDOC_IGNDELIM is noted, like in `Pf'), then
1276 * keep it open until the maximum number of arguments are
1277 * exhausted.
1278 */
1279
1280 switch (tok) {
1281 case MDOC_Ap:
1282 /* FALLTHROUGH */
1283 case MDOC_Ns:
1284 /* FALLTHROUGH */
1285 case MDOC_Ux:
1286 maxargs = 0;
1287 break;
1288 case MDOC_Bx:
1289 /* FALLTHROUGH */
1290 case MDOC_Es:
1291 /* FALLTHROUGH */
1292 case MDOC_Xr:
1293 maxargs = 2;
1294 break;
1295 default:
1296 maxargs = 1;
1297 break;
1298 }
1299
1300 mdoc_argv(mdoc, line, tok, &arg, pos, buf);
1301
1302 state = -1;
1303 p = NULL;
1304 for (;;) {
1305 la = *pos;
1306 ac = mdoc_args(mdoc, line, pos, buf, tok, &p);
1307
1308 if (ac == ARGS_WORD && state == -1 &&
1309 ! (mdoc_macros[tok].flags & MDOC_IGNDELIM) &&
1310 mdoc_isdelim(p) == DELIM_OPEN) {
1311 dword(mdoc, line, la, p, DELIM_OPEN, 0);
1312 continue;
1313 }
1314
1315 if (state == -1 && tok != MDOC_In &&
1316 tok != MDOC_St && tok != MDOC_Xr) {
1317 mdoc_elem_alloc(mdoc, line, ppos, tok, arg);
1318 state = 0;
1319 }
1320
1321 if (ac == ARGS_PUNCT || ac == ARGS_EOLN) {
1322 if (abs(state) < 2 && tok == MDOC_Pf)
1323 mandoc_vmsg(MANDOCERR_PF_SKIP,
1324 mdoc->parse, line, ppos, "Pf %s",
1325 p == NULL ? "at eol" : p);
1326 break;
1327 }
1328
1329 if (state == maxargs) {
1330 rew_elem(mdoc, tok);
1331 state = -2;
1332 }
1333
1334 ntok = (ac == ARGS_QWORD || (tok == MDOC_Pf && state == 0)) ?
1335 MDOC_MAX : lookup(mdoc, tok, line, la, p);
1336
1337 if (ntok != MDOC_MAX) {
1338 if (state >= 0) {
1339 rew_elem(mdoc, tok);
1340 state = -2;
1341 }
1342 mdoc_macro(mdoc, ntok, line, la, pos, buf);
1343 break;
1344 }
1345
1346 if (ac == ARGS_QWORD ||
1347 mdoc_macros[tok].flags & MDOC_IGNDELIM ||
1348 mdoc_isdelim(p) == DELIM_NONE) {
1349 if (state == -1) {
1350 mdoc_elem_alloc(mdoc, line, ppos, tok, arg);
1351 state = 1;
1352 } else if (state >= 0)
1353 state++;
1354 } else if (state >= 0) {
1355 rew_elem(mdoc, tok);
1356 state = -2;
1357 }
1358
1359 dword(mdoc, line, la, p, DELIM_MAX,
1360 MDOC_JOIN & mdoc_macros[tok].flags);
1361 }
1362
1363 if (state == -1) {
1364 mandoc_msg(MANDOCERR_MACRO_EMPTY, mdoc->parse,
1365 line, ppos, mdoc_macronames[tok]);
1366 return;
1367 }
1368
1369 if (state == 0 && tok == MDOC_Pf)
1370 append_delims(mdoc, line, pos, buf);
1371 if (state >= 0)
1372 rew_elem(mdoc, tok);
1373 if (nl)
1374 append_delims(mdoc, line, pos, buf);
1375 }
1376
1377 static void
in_line_eoln(MACRO_PROT_ARGS)1378 in_line_eoln(MACRO_PROT_ARGS)
1379 {
1380 struct mdoc_node *n;
1381 struct mdoc_arg *arg;
1382
1383 if ((tok == MDOC_Pp || tok == MDOC_Lp) &&
1384 ! (mdoc->flags & MDOC_SYNOPSIS)) {
1385 n = mdoc->last;
1386 if (mdoc->next == MDOC_NEXT_SIBLING)
1387 n = n->parent;
1388 if (n->tok == MDOC_Nm)
1389 rew_last(mdoc, mdoc->last->parent);
1390 }
1391
1392 if (buf[*pos] == '\0' &&
1393 (tok == MDOC_Fd || mdoc_macronames[tok][0] == '%')) {
1394 mandoc_msg(MANDOCERR_MACRO_EMPTY, mdoc->parse,
1395 line, ppos, mdoc_macronames[tok]);
1396 return;
1397 }
1398
1399 mdoc_argv(mdoc, line, tok, &arg, pos, buf);
1400 mdoc_elem_alloc(mdoc, line, ppos, tok, arg);
1401 if (parse_rest(mdoc, tok, line, pos, buf))
1402 return;
1403 rew_elem(mdoc, tok);
1404 }
1405
1406 /*
1407 * The simplest argument parser available: Parse the remaining
1408 * words until the end of the phrase or line and return 0
1409 * or until the next macro, call that macro, and return 1.
1410 */
1411 static int
parse_rest(struct mdoc * mdoc,enum mdoct tok,int line,int * pos,char * buf)1412 parse_rest(struct mdoc *mdoc, enum mdoct tok, int line, int *pos, char *buf)
1413 {
1414 int la;
1415
1416 for (;;) {
1417 la = *pos;
1418 if (mdoc_args(mdoc, line, pos, buf, tok, NULL) == ARGS_EOLN)
1419 return(0);
1420 if (macro_or_word(mdoc, tok, line, la, pos, buf, 1))
1421 return(1);
1422 }
1423 }
1424
1425 static void
ctx_synopsis(MACRO_PROT_ARGS)1426 ctx_synopsis(MACRO_PROT_ARGS)
1427 {
1428
1429 if (~mdoc->flags & (MDOC_SYNOPSIS | MDOC_NEWLINE))
1430 in_line(mdoc, tok, line, ppos, pos, buf);
1431 else if (tok == MDOC_Nm)
1432 blk_full(mdoc, tok, line, ppos, pos, buf);
1433 else {
1434 assert(tok == MDOC_Vt);
1435 blk_part_imp(mdoc, tok, line, ppos, pos, buf);
1436 }
1437 }
1438
1439 /*
1440 * Phrases occur within `Bl -column' entries, separated by `Ta' or tabs.
1441 * They're unusual because they're basically free-form text until a
1442 * macro is encountered.
1443 */
1444 static void
phrase_ta(MACRO_PROT_ARGS)1445 phrase_ta(MACRO_PROT_ARGS)
1446 {
1447 struct mdoc_node *body, *n;
1448
1449 /* Make sure we are in a column list or ignore this macro. */
1450
1451 body = NULL;
1452 for (n = mdoc->last; n != NULL; n = n->parent) {
1453 if (n->flags & MDOC_ENDED)
1454 continue;
1455 if (n->tok == MDOC_It && n->type == MDOC_BODY)
1456 body = n;
1457 if (n->tok == MDOC_Bl)
1458 break;
1459 }
1460
1461 if (n == NULL || n->norm->Bl.type != LIST_column) {
1462 mandoc_msg(MANDOCERR_TA_STRAY, mdoc->parse,
1463 line, ppos, "Ta");
1464 return;
1465 }
1466
1467 /* Advance to the next column. */
1468
1469 rew_last(mdoc, body);
1470 mdoc_body_alloc(mdoc, line, ppos, MDOC_It);
1471 parse_rest(mdoc, MDOC_MAX, line, pos, buf);
1472 }
1473