1 /* $Id: mdoc_macro.c,v 1.234 2020/01/19 18:02:00 schwarze Exp $ */
2 /*
3 * Copyright (c) 2008-2012 Kristaps Dzonsons <kristaps@bsd.lv>
4 * Copyright (c) 2010, 2012-2020 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 AUTHORS DISCLAIM ALL WARRANTIES
11 * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
12 * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHORS 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 "mandoc.h"
30 #include "roff.h"
31 #include "mdoc.h"
32 #include "libmandoc.h"
33 #include "roff_int.h"
34 #include "libmdoc.h"
35
36 static void blk_full(MACRO_PROT_ARGS);
37 static void blk_exp_close(MACRO_PROT_ARGS);
38 static void blk_part_exp(MACRO_PROT_ARGS);
39 static void blk_part_imp(MACRO_PROT_ARGS);
40 static void ctx_synopsis(MACRO_PROT_ARGS);
41 static void in_line_eoln(MACRO_PROT_ARGS);
42 static void in_line_argn(MACRO_PROT_ARGS);
43 static void in_line(MACRO_PROT_ARGS);
44 static void phrase_ta(MACRO_PROT_ARGS);
45
46 static void append_delims(struct roff_man *, int, int *, char *);
47 static void dword(struct roff_man *, int, int, const char *,
48 enum mdelim, int);
49 static int find_pending(struct roff_man *, enum roff_tok,
50 int, int, struct roff_node *);
51 static int lookup(struct roff_man *, int, int, int, const char *);
52 static int macro_or_word(MACRO_PROT_ARGS, char *, int);
53 static void break_intermediate(struct roff_node *,
54 struct roff_node *);
55 static int parse_rest(struct roff_man *, enum roff_tok,
56 int, int *, char *);
57 static enum roff_tok rew_alt(enum roff_tok);
58 static void rew_elem(struct roff_man *, enum roff_tok);
59 static void rew_last(struct roff_man *, const struct roff_node *);
60 static void rew_pending(struct roff_man *,
61 const struct roff_node *);
62
63 static const struct mdoc_macro mdoc_macros[MDOC_MAX - MDOC_Dd] = {
64 { in_line_eoln, MDOC_PROLOGUE | MDOC_JOIN }, /* Dd */
65 { in_line_eoln, MDOC_PROLOGUE }, /* Dt */
66 { in_line_eoln, MDOC_PROLOGUE }, /* Os */
67 { blk_full, MDOC_PARSED | MDOC_JOIN }, /* Sh */
68 { blk_full, MDOC_PARSED | MDOC_JOIN }, /* Ss */
69 { in_line_eoln, 0 }, /* Pp */
70 { blk_part_imp, MDOC_PARSED | MDOC_JOIN }, /* D1 */
71 { blk_part_imp, MDOC_PARSED | MDOC_JOIN }, /* Dl */
72 { blk_full, MDOC_EXPLICIT }, /* Bd */
73 { blk_exp_close, MDOC_EXPLICIT | MDOC_JOIN }, /* Ed */
74 { blk_full, MDOC_EXPLICIT }, /* Bl */
75 { blk_exp_close, MDOC_EXPLICIT | MDOC_JOIN }, /* El */
76 { blk_full, MDOC_PARSED | MDOC_JOIN }, /* It */
77 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Ad */
78 { in_line, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* An */
79 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED |
80 MDOC_IGNDELIM | MDOC_JOIN }, /* Ap */
81 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Ar */
82 { in_line, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* Cd */
83 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Cm */
84 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Dv */
85 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Er */
86 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Ev */
87 { in_line_eoln, 0 }, /* Ex */
88 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Fa */
89 { in_line_eoln, 0 }, /* Fd */
90 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Fl */
91 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Fn */
92 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Ft */
93 { in_line, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* Ic */
94 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* In */
95 { in_line, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* Li */
96 { blk_full, MDOC_JOIN }, /* Nd */
97 { ctx_synopsis, MDOC_CALLABLE | MDOC_PARSED }, /* Nm */
98 { blk_part_imp, MDOC_CALLABLE | MDOC_PARSED }, /* Op */
99 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Ot */
100 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Pa */
101 { in_line_eoln, 0 }, /* Rv */
102 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* St */
103 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Va */
104 { ctx_synopsis, MDOC_CALLABLE | MDOC_PARSED }, /* Vt */
105 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* Xr */
106 { in_line_eoln, MDOC_JOIN }, /* %A */
107 { in_line_eoln, MDOC_JOIN }, /* %B */
108 { in_line_eoln, MDOC_JOIN }, /* %D */
109 { in_line_eoln, MDOC_JOIN }, /* %I */
110 { in_line_eoln, MDOC_JOIN }, /* %J */
111 { in_line_eoln, 0 }, /* %N */
112 { in_line_eoln, MDOC_JOIN }, /* %O */
113 { in_line_eoln, 0 }, /* %P */
114 { in_line_eoln, MDOC_JOIN }, /* %R */
115 { in_line_eoln, MDOC_JOIN }, /* %T */
116 { in_line_eoln, 0 }, /* %V */
117 { blk_exp_close, MDOC_CALLABLE | MDOC_PARSED |
118 MDOC_EXPLICIT | MDOC_JOIN }, /* Ac */
119 { blk_part_exp, MDOC_CALLABLE | MDOC_PARSED |
120 MDOC_EXPLICIT | MDOC_JOIN }, /* Ao */
121 { blk_part_imp, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* Aq */
122 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* At */
123 { blk_exp_close, MDOC_CALLABLE | MDOC_PARSED |
124 MDOC_EXPLICIT | MDOC_JOIN }, /* Bc */
125 { blk_full, MDOC_EXPLICIT }, /* Bf */
126 { blk_part_exp, MDOC_CALLABLE | MDOC_PARSED |
127 MDOC_EXPLICIT | MDOC_JOIN }, /* Bo */
128 { blk_part_imp, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* 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_CALLABLE | MDOC_PARSED |
133 MDOC_EXPLICIT | MDOC_JOIN }, /* Dc */
134 { blk_part_exp, MDOC_CALLABLE | MDOC_PARSED |
135 MDOC_EXPLICIT | MDOC_JOIN }, /* Do */
136 { blk_part_imp, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* Dq */
137 { blk_exp_close, MDOC_CALLABLE | MDOC_PARSED | MDOC_EXPLICIT }, /* Ec */
138 { blk_exp_close, MDOC_EXPLICIT | MDOC_JOIN }, /* Ef */
139 { in_line, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* Em */
140 { blk_part_exp, MDOC_CALLABLE | MDOC_PARSED | MDOC_EXPLICIT }, /* Eo */
141 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* Fx */
142 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Ms */
143 { in_line, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* No */
144 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED |
145 MDOC_IGNDELIM | MDOC_JOIN }, /* Ns */
146 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* Nx */
147 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* Ox */
148 { blk_exp_close, MDOC_CALLABLE | MDOC_PARSED |
149 MDOC_EXPLICIT | MDOC_JOIN }, /* Pc */
150 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED | MDOC_IGNDELIM }, /* Pf */
151 { blk_part_exp, MDOC_CALLABLE | MDOC_PARSED |
152 MDOC_EXPLICIT | MDOC_JOIN }, /* Po */
153 { blk_part_imp, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* Pq */
154 { blk_exp_close, MDOC_CALLABLE | MDOC_PARSED |
155 MDOC_EXPLICIT | MDOC_JOIN }, /* Qc */
156 { blk_part_imp, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* Ql */
157 { blk_part_exp, MDOC_CALLABLE | MDOC_PARSED |
158 MDOC_EXPLICIT | MDOC_JOIN }, /* Qo */
159 { blk_part_imp, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* Qq */
160 { blk_exp_close, MDOC_EXPLICIT | MDOC_JOIN }, /* Re */
161 { blk_full, MDOC_EXPLICIT }, /* Rs */
162 { blk_exp_close, MDOC_CALLABLE | MDOC_PARSED |
163 MDOC_EXPLICIT | MDOC_JOIN }, /* Sc */
164 { blk_part_exp, MDOC_CALLABLE | MDOC_PARSED |
165 MDOC_EXPLICIT | MDOC_JOIN }, /* So */
166 { blk_part_imp, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* Sq */
167 { in_line_argn, 0 }, /* Sm */
168 { in_line, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* Sx */
169 { in_line, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* Sy */
170 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Tn */
171 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* Ux */
172 { blk_exp_close, MDOC_EXPLICIT | MDOC_CALLABLE | MDOC_PARSED }, /* Xc */
173 { blk_part_exp, MDOC_CALLABLE | MDOC_PARSED | MDOC_EXPLICIT }, /* Xo */
174 { blk_full, MDOC_EXPLICIT | MDOC_CALLABLE }, /* Fo */
175 { blk_exp_close, MDOC_CALLABLE | MDOC_PARSED |
176 MDOC_EXPLICIT | MDOC_JOIN }, /* Fc */
177 { blk_part_exp, MDOC_CALLABLE | MDOC_PARSED |
178 MDOC_EXPLICIT | MDOC_JOIN }, /* Oo */
179 { blk_exp_close, MDOC_CALLABLE | MDOC_PARSED |
180 MDOC_EXPLICIT | MDOC_JOIN }, /* Oc */
181 { blk_full, MDOC_EXPLICIT }, /* Bk */
182 { blk_exp_close, MDOC_EXPLICIT | MDOC_JOIN }, /* Ek */
183 { in_line_eoln, 0 }, /* Bt */
184 { in_line_eoln, 0 }, /* Hf */
185 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Fr */
186 { in_line_eoln, 0 }, /* Ud */
187 { in_line, 0 }, /* Lb */
188 { in_line_eoln, 0 }, /* Lp */
189 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Lk */
190 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Mt */
191 { blk_part_imp, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* Brq */
192 { blk_part_exp, MDOC_CALLABLE | MDOC_PARSED |
193 MDOC_EXPLICIT | MDOC_JOIN }, /* Bro */
194 { blk_exp_close, MDOC_CALLABLE | MDOC_PARSED |
195 MDOC_EXPLICIT | MDOC_JOIN }, /* Brc */
196 { in_line_eoln, MDOC_JOIN }, /* %C */
197 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* Es */
198 { blk_part_imp, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* En */
199 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* Dx */
200 { in_line_eoln, MDOC_JOIN }, /* %Q */
201 { in_line_eoln, 0 }, /* %U */
202 { phrase_ta, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* Ta */
203 { in_line_eoln, 0 }, /* Tg */
204 };
205
206
207 const struct mdoc_macro *
mdoc_macro(enum roff_tok tok)208 mdoc_macro(enum roff_tok tok)
209 {
210 assert(tok >= MDOC_Dd && tok < MDOC_MAX);
211 return mdoc_macros + (tok - MDOC_Dd);
212 }
213
214 /*
215 * This is called at the end of parsing. It must traverse up the tree,
216 * closing out open [implicit] scopes. Obviously, open explicit scopes
217 * are errors.
218 */
219 void
mdoc_endparse(struct roff_man * mdoc)220 mdoc_endparse(struct roff_man *mdoc)
221 {
222 struct roff_node *n;
223
224 /* Scan for open explicit scopes. */
225
226 n = mdoc->last->flags & NODE_VALID ?
227 mdoc->last->parent : mdoc->last;
228
229 for ( ; n; n = n->parent)
230 if (n->type == ROFFT_BLOCK &&
231 mdoc_macro(n->tok)->flags & MDOC_EXPLICIT)
232 mandoc_msg(MANDOCERR_BLK_NOEND,
233 n->line, n->pos, "%s", roff_name[n->tok]);
234
235 /* Rewind to the first. */
236
237 rew_last(mdoc, mdoc->meta.first);
238 }
239
240 /*
241 * Look up the macro at *p called by "from",
242 * or as a line macro if from == TOKEN_NONE.
243 */
244 static int
lookup(struct roff_man * mdoc,int from,int line,int ppos,const char * p)245 lookup(struct roff_man *mdoc, int from, int line, int ppos, const char *p)
246 {
247 enum roff_tok res;
248
249 if (mdoc->flags & MDOC_PHRASEQF) {
250 mdoc->flags &= ~MDOC_PHRASEQF;
251 return TOKEN_NONE;
252 }
253 if (from == TOKEN_NONE || mdoc_macro(from)->flags & MDOC_PARSED) {
254 res = roffhash_find(mdoc->mdocmac, p, 0);
255 if (res != TOKEN_NONE) {
256 if (mdoc_macro(res)->flags & MDOC_CALLABLE)
257 return res;
258 mandoc_msg(MANDOCERR_MACRO_CALL, line, ppos, "%s", p);
259 }
260 }
261 return TOKEN_NONE;
262 }
263
264 /*
265 * Rewind up to and including a specific node.
266 */
267 static void
rew_last(struct roff_man * mdoc,const struct roff_node * to)268 rew_last(struct roff_man *mdoc, const struct roff_node *to)
269 {
270
271 if (to->flags & NODE_VALID)
272 return;
273
274 while (mdoc->last != to) {
275 mdoc_state(mdoc, mdoc->last);
276 mdoc->last->flags |= NODE_VALID | NODE_ENDED;
277 mdoc->last = mdoc->last->parent;
278 }
279 mdoc_state(mdoc, mdoc->last);
280 mdoc->last->flags |= NODE_VALID | NODE_ENDED;
281 mdoc->next = ROFF_NEXT_SIBLING;
282 }
283
284 /*
285 * Rewind up to a specific block, including all blocks that broke it.
286 */
287 static void
rew_pending(struct roff_man * mdoc,const struct roff_node * n)288 rew_pending(struct roff_man *mdoc, const struct roff_node *n)
289 {
290
291 for (;;) {
292 rew_last(mdoc, n);
293
294 if (mdoc->last == n) {
295 switch (n->type) {
296 case ROFFT_HEAD:
297 roff_body_alloc(mdoc, n->line, n->pos,
298 n->tok);
299 if (n->tok == MDOC_Ss)
300 mdoc->flags &= ~ROFF_NONOFILL;
301 break;
302 case ROFFT_BLOCK:
303 break;
304 default:
305 return;
306 }
307 if ( ! (n->flags & NODE_BROKEN))
308 return;
309 } else
310 n = mdoc->last;
311
312 for (;;) {
313 if ((n = n->parent) == NULL)
314 return;
315
316 if (n->type == ROFFT_BLOCK ||
317 n->type == ROFFT_HEAD) {
318 if (n->flags & NODE_ENDED)
319 break;
320 else
321 return;
322 }
323 }
324 }
325 }
326
327 /*
328 * For a block closing macro, return the corresponding opening one.
329 * Otherwise, return the macro itself.
330 */
331 static enum roff_tok
rew_alt(enum roff_tok tok)332 rew_alt(enum roff_tok tok)
333 {
334 switch (tok) {
335 case MDOC_Ac:
336 return MDOC_Ao;
337 case MDOC_Bc:
338 return MDOC_Bo;
339 case MDOC_Brc:
340 return MDOC_Bro;
341 case MDOC_Dc:
342 return MDOC_Do;
343 case MDOC_Ec:
344 return MDOC_Eo;
345 case MDOC_Ed:
346 return MDOC_Bd;
347 case MDOC_Ef:
348 return MDOC_Bf;
349 case MDOC_Ek:
350 return MDOC_Bk;
351 case MDOC_El:
352 return MDOC_Bl;
353 case MDOC_Fc:
354 return MDOC_Fo;
355 case MDOC_Oc:
356 return MDOC_Oo;
357 case MDOC_Pc:
358 return MDOC_Po;
359 case MDOC_Qc:
360 return MDOC_Qo;
361 case MDOC_Re:
362 return MDOC_Rs;
363 case MDOC_Sc:
364 return MDOC_So;
365 case MDOC_Xc:
366 return MDOC_Xo;
367 default:
368 return tok;
369 }
370 }
371
372 static void
rew_elem(struct roff_man * mdoc,enum roff_tok tok)373 rew_elem(struct roff_man *mdoc, enum roff_tok tok)
374 {
375 struct roff_node *n;
376
377 n = mdoc->last;
378 if (n->type != ROFFT_ELEM)
379 n = n->parent;
380 assert(n->type == ROFFT_ELEM);
381 assert(tok == n->tok);
382 rew_last(mdoc, n);
383 }
384
385 static void
break_intermediate(struct roff_node * n,struct roff_node * breaker)386 break_intermediate(struct roff_node *n, struct roff_node *breaker)
387 {
388 if (n != breaker &&
389 n->type != ROFFT_BLOCK && n->type != ROFFT_HEAD &&
390 (n->type != ROFFT_BODY || n->end != ENDBODY_NOT))
391 n = n->parent;
392 while (n != breaker) {
393 if ( ! (n->flags & NODE_VALID))
394 n->flags |= NODE_BROKEN;
395 n = n->parent;
396 }
397 }
398
399 /*
400 * If there is an open sub-block of the target requiring
401 * explicit close-out, postpone closing out the target until
402 * the rew_pending() call closing out the sub-block.
403 */
404 static int
find_pending(struct roff_man * mdoc,enum roff_tok tok,int line,int ppos,struct roff_node * target)405 find_pending(struct roff_man *mdoc, enum roff_tok tok, int line, int ppos,
406 struct roff_node *target)
407 {
408 struct roff_node *n;
409 int irc;
410
411 if (target->flags & NODE_VALID)
412 return 0;
413
414 irc = 0;
415 for (n = mdoc->last; n != NULL && n != target; n = n->parent) {
416 if (n->flags & NODE_ENDED)
417 continue;
418 if (n->type == ROFFT_BLOCK &&
419 mdoc_macro(n->tok)->flags & MDOC_EXPLICIT) {
420 irc = 1;
421 break_intermediate(mdoc->last, target);
422 if (target->type == ROFFT_HEAD)
423 target->flags |= NODE_ENDED;
424 else if ( ! (target->flags & NODE_ENDED)) {
425 mandoc_msg(MANDOCERR_BLK_NEST,
426 line, ppos, "%s breaks %s",
427 roff_name[tok], roff_name[n->tok]);
428 mdoc_endbody_alloc(mdoc, line, ppos,
429 tok, target);
430 }
431 }
432 }
433 return irc;
434 }
435
436 /*
437 * Allocate a word and check whether it's punctuation or not.
438 * Punctuation consists of those tokens found in mdoc_isdelim().
439 */
440 static void
dword(struct roff_man * mdoc,int line,int col,const char * p,enum mdelim d,int may_append)441 dword(struct roff_man *mdoc, int line, int col, const char *p,
442 enum mdelim d, int may_append)
443 {
444
445 if (d == DELIM_MAX)
446 d = mdoc_isdelim(p);
447
448 if (may_append &&
449 ! (mdoc->flags & (MDOC_SYNOPSIS | MDOC_KEEP | MDOC_SMOFF)) &&
450 d == DELIM_NONE && mdoc->last->type == ROFFT_TEXT &&
451 mdoc_isdelim(mdoc->last->string) == DELIM_NONE) {
452 roff_word_append(mdoc, p);
453 return;
454 }
455
456 roff_word_alloc(mdoc, line, col, p);
457
458 /*
459 * If the word consists of a bare delimiter,
460 * flag the new node accordingly,
461 * unless doing so was vetoed by the invoking macro.
462 * Always clear the veto, it is only valid for one word.
463 */
464
465 if (d == DELIM_OPEN)
466 mdoc->last->flags |= NODE_DELIMO;
467 else if (d == DELIM_CLOSE &&
468 ! (mdoc->flags & MDOC_NODELIMC) &&
469 mdoc->last->parent->tok != MDOC_Fd)
470 mdoc->last->flags |= NODE_DELIMC;
471 mdoc->flags &= ~MDOC_NODELIMC;
472 }
473
474 static void
append_delims(struct roff_man * mdoc,int line,int * pos,char * buf)475 append_delims(struct roff_man *mdoc, int line, int *pos, char *buf)
476 {
477 char *p;
478 int la;
479 enum margserr ac;
480
481 if (buf[*pos] == '\0')
482 return;
483
484 for (;;) {
485 la = *pos;
486 ac = mdoc_args(mdoc, line, pos, buf, TOKEN_NONE, &p);
487 if (ac == ARGS_EOLN)
488 break;
489 dword(mdoc, line, la, p, DELIM_MAX, 1);
490
491 /*
492 * If we encounter end-of-sentence symbols, then trigger
493 * the double-space.
494 *
495 * XXX: it's easy to allow this to propagate outward to
496 * the last symbol, such that `. )' will cause the
497 * correct double-spacing. However, (1) groff isn't
498 * smart enough to do this and (2) it would require
499 * knowing which symbols break this behaviour, for
500 * example, `. ;' shouldn't propagate the double-space.
501 */
502
503 if (mandoc_eos(p, strlen(p)))
504 mdoc->last->flags |= NODE_EOS;
505 if (ac == ARGS_ALLOC)
506 free(p);
507 }
508 }
509
510 /*
511 * Parse one word.
512 * If it is a macro, call it and return 1.
513 * Otherwise, allocate it and return 0.
514 */
515 static int
macro_or_word(MACRO_PROT_ARGS,char * p,int parsed)516 macro_or_word(MACRO_PROT_ARGS, char *p, int parsed)
517 {
518 int ntok;
519
520 ntok = buf[ppos] == '"' || parsed == 0 ||
521 mdoc->flags & MDOC_PHRASELIT ? TOKEN_NONE :
522 lookup(mdoc, tok, line, ppos, p);
523
524 if (ntok == TOKEN_NONE) {
525 dword(mdoc, line, ppos, p, DELIM_MAX, tok == TOKEN_NONE ||
526 mdoc_macro(tok)->flags & MDOC_JOIN);
527 return 0;
528 } else {
529 if (tok != TOKEN_NONE &&
530 mdoc_macro(tok)->fp == in_line_eoln)
531 rew_elem(mdoc, tok);
532 (*mdoc_macro(ntok)->fp)(mdoc, ntok, line, ppos, pos, buf);
533 if (tok == TOKEN_NONE)
534 append_delims(mdoc, line, pos, buf);
535 return 1;
536 }
537 }
538
539 /*
540 * Close out block partial/full explicit.
541 */
542 static void
blk_exp_close(MACRO_PROT_ARGS)543 blk_exp_close(MACRO_PROT_ARGS)
544 {
545 struct roff_node *body; /* Our own body. */
546 struct roff_node *endbody; /* Our own end marker. */
547 struct roff_node *itblk; /* An It block starting later. */
548 struct roff_node *later; /* A sub-block starting later. */
549 struct roff_node *n; /* Search back to our block. */
550 struct roff_node *target; /* For find_pending(). */
551
552 int j, lastarg, maxargs, nl, pending;
553 enum margserr ac;
554 enum roff_tok atok, ntok;
555 char *p;
556
557 nl = MDOC_NEWLINE & mdoc->flags;
558
559 switch (tok) {
560 case MDOC_Ec:
561 maxargs = 1;
562 break;
563 case MDOC_Ek:
564 mdoc->flags &= ~MDOC_KEEP;
565 /* FALLTHROUGH */
566 default:
567 maxargs = 0;
568 break;
569 }
570
571 /* Search backwards for the beginning of our own body. */
572
573 atok = rew_alt(tok);
574 body = NULL;
575 for (n = mdoc->last; n; n = n->parent) {
576 if (n->flags & NODE_ENDED || n->tok != atok ||
577 n->type != ROFFT_BODY || n->end != ENDBODY_NOT)
578 continue;
579 body = n;
580 break;
581 }
582
583 /*
584 * Search backwards for beginnings of blocks,
585 * both of our own and of pending sub-blocks.
586 */
587
588 endbody = itblk = later = NULL;
589 for (n = mdoc->last; n; n = n->parent) {
590 if (n->flags & NODE_ENDED)
591 continue;
592
593 /*
594 * Mismatching end macros can never break anything
595 * and we only care about the breaking of BLOCKs.
596 */
597
598 if (body == NULL || n->type != ROFFT_BLOCK)
599 continue;
600
601 /*
602 * SYNOPSIS name blocks can not be broken themselves,
603 * but they do get broken together with a broken child.
604 */
605
606 if (n->tok == MDOC_Nm) {
607 if (later != NULL)
608 n->flags |= NODE_BROKEN | NODE_ENDED;
609 continue;
610 }
611
612 if (n->tok == MDOC_It) {
613 itblk = n;
614 continue;
615 }
616
617 if (atok == n->tok) {
618
619 /*
620 * Found the start of our own block.
621 * When there is no pending sub block,
622 * just proceed to closing out.
623 */
624
625 if (later == NULL ||
626 (tok == MDOC_El && itblk == NULL))
627 break;
628
629 /*
630 * When there is a pending sub block, postpone
631 * closing out the current block until the
632 * rew_pending() closing out the sub-block.
633 * Mark the place where the formatting - but not
634 * the scope - of the current block ends.
635 */
636
637 mandoc_msg(MANDOCERR_BLK_NEST,
638 line, ppos, "%s breaks %s",
639 roff_name[atok], roff_name[later->tok]);
640
641 endbody = mdoc_endbody_alloc(mdoc, line, ppos,
642 atok, body);
643
644 if (tok == MDOC_El)
645 itblk->flags |= NODE_ENDED | NODE_BROKEN;
646
647 /*
648 * If a block closing macro taking arguments
649 * breaks another block, put the arguments
650 * into the end marker.
651 */
652
653 if (maxargs)
654 mdoc->next = ROFF_NEXT_CHILD;
655 break;
656 }
657
658 /*
659 * Explicit blocks close out description lines, but
660 * even those can get broken together with a child.
661 */
662
663 if (n->tok == MDOC_Nd) {
664 if (later != NULL)
665 n->flags |= NODE_BROKEN | NODE_ENDED;
666 else
667 rew_last(mdoc, n);
668 continue;
669 }
670
671 /* Breaking an open sub block. */
672
673 break_intermediate(mdoc->last, body);
674 n->flags |= NODE_BROKEN;
675 if (later == NULL)
676 later = n;
677 }
678
679 if (body == NULL) {
680 mandoc_msg(MANDOCERR_BLK_NOTOPEN, line, ppos,
681 "%s", roff_name[tok]);
682 if (maxargs && endbody == NULL) {
683 /*
684 * Stray .Ec without previous .Eo:
685 * Break the output line, keep the arguments.
686 */
687 roff_elem_alloc(mdoc, line, ppos, ROFF_br);
688 rew_elem(mdoc, ROFF_br);
689 }
690 } else if (endbody == NULL) {
691 rew_last(mdoc, body);
692 if (maxargs)
693 mdoc_tail_alloc(mdoc, line, ppos, atok);
694 }
695
696 if ((mdoc_macro(tok)->flags & MDOC_PARSED) == 0) {
697 if (buf[*pos] != '\0')
698 mandoc_msg(MANDOCERR_ARG_SKIP, line, ppos,
699 "%s %s", roff_name[tok], buf + *pos);
700 if (endbody == NULL && n != NULL)
701 rew_pending(mdoc, n);
702
703 /*
704 * Restore the fill mode that was set before the display.
705 * This needs to be done here rather than during validation
706 * such that subsequent nodes get the right flags.
707 */
708
709 if (tok == MDOC_Ed && body != NULL) {
710 if (body->flags & NODE_NOFILL)
711 mdoc->flags |= ROFF_NOFILL;
712 else
713 mdoc->flags &= ~ROFF_NOFILL;
714 }
715 return;
716 }
717
718 if (endbody != NULL)
719 n = endbody;
720
721 ntok = TOKEN_NONE;
722 for (j = 0; ; j++) {
723 lastarg = *pos;
724
725 if (j == maxargs && n != NULL)
726 rew_last(mdoc, n);
727
728 ac = mdoc_args(mdoc, line, pos, buf, tok, &p);
729 if (ac == ARGS_PUNCT || ac == ARGS_EOLN)
730 break;
731
732 ntok = lookup(mdoc, tok, line, lastarg, p);
733
734 if (ntok == TOKEN_NONE) {
735 dword(mdoc, line, lastarg, p, DELIM_MAX,
736 mdoc_macro(tok)->flags & MDOC_JOIN);
737 if (ac == ARGS_ALLOC)
738 free(p);
739 continue;
740 }
741 if (ac == ARGS_ALLOC)
742 free(p);
743
744 if (n != NULL)
745 rew_last(mdoc, n);
746 mdoc->flags &= ~MDOC_NEWLINE;
747 (*mdoc_macro(ntok)->fp)(mdoc, ntok, line, lastarg, pos, buf);
748 break;
749 }
750
751 if (n != NULL) {
752 pending = 0;
753 if (ntok != TOKEN_NONE && n->flags & NODE_BROKEN) {
754 target = n;
755 do
756 target = target->parent;
757 while ( ! (target->flags & NODE_ENDED));
758 pending = find_pending(mdoc, ntok, line, ppos, target);
759 }
760 if ( ! pending)
761 rew_pending(mdoc, n);
762 }
763 if (nl)
764 append_delims(mdoc, line, pos, buf);
765 }
766
767 static void
in_line(MACRO_PROT_ARGS)768 in_line(MACRO_PROT_ARGS)
769 {
770 int la, scope, cnt, firstarg, mayopen, nc, nl;
771 enum roff_tok ntok;
772 enum margserr ac;
773 enum mdelim d;
774 struct mdoc_arg *arg;
775 char *p;
776
777 nl = MDOC_NEWLINE & mdoc->flags;
778
779 /*
780 * Whether we allow ignored elements (those without content,
781 * usually because of reserved words) to squeak by.
782 */
783
784 switch (tok) {
785 case MDOC_An:
786 case MDOC_Ar:
787 case MDOC_Fl:
788 case MDOC_Mt:
789 case MDOC_Nm:
790 case MDOC_Pa:
791 nc = 1;
792 break;
793 default:
794 nc = 0;
795 break;
796 }
797
798 mdoc_argv(mdoc, line, tok, &arg, pos, buf);
799
800 d = DELIM_NONE;
801 firstarg = 1;
802 mayopen = 1;
803 for (cnt = scope = 0;; ) {
804 la = *pos;
805 ac = mdoc_args(mdoc, line, pos, buf, tok, &p);
806
807 /*
808 * At the end of a macro line,
809 * opening delimiters do not suppress spacing.
810 */
811
812 if (ac == ARGS_EOLN) {
813 if (d == DELIM_OPEN)
814 mdoc->last->flags &= ~NODE_DELIMO;
815 break;
816 }
817
818 /*
819 * The rest of the macro line is only punctuation,
820 * to be handled by append_delims().
821 * If there were no other arguments,
822 * do not allow the first one to suppress spacing,
823 * even if it turns out to be a closing one.
824 */
825
826 if (ac == ARGS_PUNCT) {
827 if (cnt == 0 && (nc == 0 || tok == MDOC_An))
828 mdoc->flags |= MDOC_NODELIMC;
829 break;
830 }
831
832 ntok = (tok == MDOC_Fn && !cnt) ?
833 TOKEN_NONE : lookup(mdoc, tok, line, la, p);
834
835 /*
836 * In this case, we've located a submacro and must
837 * execute it. Close out scope, if open. If no
838 * elements have been generated, either create one (nc)
839 * or raise a warning.
840 */
841
842 if (ntok != TOKEN_NONE) {
843 if (scope)
844 rew_elem(mdoc, tok);
845 if (nc && ! cnt) {
846 mdoc_elem_alloc(mdoc, line, ppos, tok, arg);
847 rew_last(mdoc, mdoc->last);
848 } else if ( ! nc && ! cnt) {
849 mdoc_argv_free(arg);
850 mandoc_msg(MANDOCERR_MACRO_EMPTY,
851 line, ppos, "%s", roff_name[tok]);
852 }
853 (*mdoc_macro(ntok)->fp)(mdoc, ntok,
854 line, la, pos, buf);
855 if (nl)
856 append_delims(mdoc, line, pos, buf);
857 if (ac == ARGS_ALLOC)
858 free(p);
859 return;
860 }
861
862 /*
863 * Handle punctuation. Set up our scope, if a word;
864 * rewind the scope, if a delimiter; then append the word.
865 */
866
867 if ((d = mdoc_isdelim(p)) != DELIM_NONE) {
868 /*
869 * If we encounter closing punctuation, no word
870 * has been emitted, no scope is open, and we're
871 * allowed to have an empty element, then start
872 * a new scope.
873 */
874 if ((d == DELIM_CLOSE ||
875 (d == DELIM_MIDDLE && tok == MDOC_Fl)) &&
876 !cnt && !scope && nc && mayopen) {
877 mdoc_elem_alloc(mdoc, line, ppos, tok, arg);
878 scope = 1;
879 cnt++;
880 if (tok == MDOC_Nm)
881 mayopen = 0;
882 }
883 /*
884 * Close out our scope, if one is open, before
885 * any punctuation.
886 */
887 if (scope && tok != MDOC_Lk) {
888 rew_elem(mdoc, tok);
889 scope = 0;
890 if (tok == MDOC_Fn)
891 mayopen = 0;
892 }
893 } else if (mayopen && !scope) {
894 mdoc_elem_alloc(mdoc, line, ppos, tok, arg);
895 scope = 1;
896 cnt++;
897 }
898
899 dword(mdoc, line, la, p, d,
900 mdoc_macro(tok)->flags & MDOC_JOIN);
901
902 if (ac == ARGS_ALLOC)
903 free(p);
904
905 /*
906 * If the first argument is a closing delimiter,
907 * do not suppress spacing before it.
908 */
909
910 if (firstarg && d == DELIM_CLOSE && !nc)
911 mdoc->last->flags &= ~NODE_DELIMC;
912 firstarg = 0;
913
914 /*
915 * `Fl' macros have their scope re-opened with each new
916 * word so that the `-' can be added to each one without
917 * having to parse out spaces.
918 */
919 if (scope && tok == MDOC_Fl) {
920 rew_elem(mdoc, tok);
921 scope = 0;
922 }
923 }
924
925 if (scope && tok != MDOC_Lk) {
926 rew_elem(mdoc, tok);
927 scope = 0;
928 }
929
930 /*
931 * If no elements have been collected and we're allowed to have
932 * empties (nc), open a scope and close it out. Otherwise,
933 * raise a warning.
934 */
935
936 if ( ! cnt) {
937 if (nc) {
938 mdoc_elem_alloc(mdoc, line, ppos, tok, arg);
939 rew_last(mdoc, mdoc->last);
940 } else {
941 mdoc_argv_free(arg);
942 mandoc_msg(MANDOCERR_MACRO_EMPTY,
943 line, ppos, "%s", roff_name[tok]);
944 }
945 }
946 if (nl)
947 append_delims(mdoc, line, pos, buf);
948 if (scope)
949 rew_elem(mdoc, tok);
950 }
951
952 static void
blk_full(MACRO_PROT_ARGS)953 blk_full(MACRO_PROT_ARGS)
954 {
955 struct mdoc_arg *arg;
956 struct roff_node *blk; /* Our own or a broken block. */
957 struct roff_node *head; /* Our own head. */
958 struct roff_node *body; /* Our own body. */
959 struct roff_node *n;
960 char *p;
961 size_t iarg;
962 int done, la, nl, parsed;
963 enum margserr ac, lac;
964
965 nl = MDOC_NEWLINE & mdoc->flags;
966
967 if (buf[*pos] == '\0' && (tok == MDOC_Sh || tok == MDOC_Ss)) {
968 mandoc_msg(MANDOCERR_MACRO_EMPTY,
969 line, ppos, "%s", roff_name[tok]);
970 return;
971 }
972
973 if ((mdoc_macro(tok)->flags & MDOC_EXPLICIT) == 0) {
974
975 /* Here, tok is one of Sh Ss Nm Nd It. */
976
977 blk = NULL;
978 for (n = mdoc->last; n != NULL; n = n->parent) {
979 if (n->flags & NODE_ENDED) {
980 if ( ! (n->flags & NODE_VALID))
981 n->flags |= NODE_BROKEN;
982 continue;
983 }
984 if (n->type != ROFFT_BLOCK)
985 continue;
986
987 if (tok == MDOC_It && n->tok == MDOC_Bl) {
988 if (blk != NULL) {
989 mandoc_msg(MANDOCERR_BLK_BROKEN,
990 line, ppos, "It breaks %s",
991 roff_name[blk->tok]);
992 rew_pending(mdoc, blk);
993 }
994 break;
995 }
996
997 if (mdoc_macro(n->tok)->flags & MDOC_EXPLICIT) {
998 switch (tok) {
999 case MDOC_Sh:
1000 case MDOC_Ss:
1001 mandoc_msg(MANDOCERR_BLK_BROKEN,
1002 line, ppos,
1003 "%s breaks %s", roff_name[tok],
1004 roff_name[n->tok]);
1005 rew_pending(mdoc, n);
1006 n = mdoc->last;
1007 continue;
1008 case MDOC_It:
1009 /* Delay in case it's astray. */
1010 blk = n;
1011 continue;
1012 default:
1013 break;
1014 }
1015 break;
1016 }
1017
1018 /* Here, n is one of Sh Ss Nm Nd It. */
1019
1020 if (tok != MDOC_Sh && (n->tok == MDOC_Sh ||
1021 (tok != MDOC_Ss && (n->tok == MDOC_Ss ||
1022 (tok != MDOC_It && n->tok == MDOC_It)))))
1023 break;
1024
1025 /* Item breaking an explicit block. */
1026
1027 if (blk != NULL) {
1028 mandoc_msg(MANDOCERR_BLK_BROKEN, line, ppos,
1029 "It breaks %s", roff_name[blk->tok]);
1030 rew_pending(mdoc, blk);
1031 blk = NULL;
1032 }
1033
1034 /* Close out prior implicit scopes. */
1035
1036 rew_pending(mdoc, n);
1037 }
1038
1039 /* Skip items outside lists. */
1040
1041 if (tok == MDOC_It && (n == NULL || n->tok != MDOC_Bl)) {
1042 mandoc_msg(MANDOCERR_IT_STRAY,
1043 line, ppos, "It %s", buf + *pos);
1044 roff_elem_alloc(mdoc, line, ppos, ROFF_br);
1045 rew_elem(mdoc, ROFF_br);
1046 return;
1047 }
1048 }
1049
1050 /*
1051 * This routine accommodates implicitly- and explicitly-scoped
1052 * macro openings. Implicit ones first close out prior scope
1053 * (seen above). Delay opening the head until necessary to
1054 * allow leading punctuation to print. Special consideration
1055 * for `It -column', which has phrase-part syntax instead of
1056 * regular child nodes.
1057 */
1058
1059 switch (tok) {
1060 case MDOC_Sh:
1061 mdoc->flags &= ~ROFF_NOFILL;
1062 break;
1063 case MDOC_Ss:
1064 mdoc->flags |= ROFF_NONOFILL;
1065 break;
1066 default:
1067 break;
1068 }
1069 mdoc_argv(mdoc, line, tok, &arg, pos, buf);
1070 blk = mdoc_block_alloc(mdoc, line, ppos, tok, arg);
1071 head = body = NULL;
1072
1073 /*
1074 * Exception: Heads of `It' macros in `-diag' lists are not
1075 * parsed, even though `It' macros in general are parsed.
1076 */
1077
1078 parsed = tok != MDOC_It ||
1079 mdoc->last->parent->tok != MDOC_Bl ||
1080 mdoc->last->parent->norm->Bl.type != LIST_diag;
1081
1082 /*
1083 * The `Nd' macro has all arguments in its body: it's a hybrid
1084 * of block partial-explicit and full-implicit. Stupid.
1085 */
1086
1087 if (tok == MDOC_Nd) {
1088 head = roff_head_alloc(mdoc, line, ppos, tok);
1089 rew_last(mdoc, head);
1090 body = roff_body_alloc(mdoc, line, ppos, tok);
1091 }
1092
1093 if (tok == MDOC_Bk)
1094 mdoc->flags |= MDOC_KEEP;
1095
1096 ac = ARGS_EOLN;
1097 for (;;) {
1098
1099 /*
1100 * If we are right after a tab character,
1101 * do not parse the first word for macros.
1102 */
1103
1104 if (mdoc->flags & MDOC_PHRASEQN) {
1105 mdoc->flags &= ~MDOC_PHRASEQN;
1106 mdoc->flags |= MDOC_PHRASEQF;
1107 }
1108
1109 la = *pos;
1110 lac = ac;
1111 ac = mdoc_args(mdoc, line, pos, buf, tok, &p);
1112 if (ac == ARGS_EOLN) {
1113 if (lac != ARGS_PHRASE ||
1114 ! (mdoc->flags & MDOC_PHRASEQF))
1115 break;
1116
1117 /*
1118 * This line ends in a tab; start the next
1119 * column now, with a leading blank.
1120 */
1121
1122 if (body != NULL)
1123 rew_last(mdoc, body);
1124 body = roff_body_alloc(mdoc, line, ppos, tok);
1125 roff_word_alloc(mdoc, line, ppos, "\\&");
1126 break;
1127 }
1128
1129 if (tok == MDOC_Bd || tok == MDOC_Bk) {
1130 mandoc_msg(MANDOCERR_ARG_EXCESS, line, la,
1131 "%s ... %s", roff_name[tok], buf + la);
1132 if (ac == ARGS_ALLOC)
1133 free(p);
1134 break;
1135 }
1136 if (tok == MDOC_Rs) {
1137 mandoc_msg(MANDOCERR_ARG_SKIP,
1138 line, la, "Rs %s", buf + la);
1139 if (ac == ARGS_ALLOC)
1140 free(p);
1141 break;
1142 }
1143 if (ac == ARGS_PUNCT)
1144 break;
1145
1146 /*
1147 * Emit leading punctuation (i.e., punctuation before
1148 * the ROFFT_HEAD) for non-phrase types.
1149 */
1150
1151 if (head == NULL &&
1152 ac != ARGS_PHRASE &&
1153 mdoc_isdelim(p) == DELIM_OPEN) {
1154 dword(mdoc, line, la, p, DELIM_OPEN, 0);
1155 if (ac == ARGS_ALLOC)
1156 free(p);
1157 continue;
1158 }
1159
1160 /* Open a head if one hasn't been opened. */
1161
1162 if (head == NULL)
1163 head = roff_head_alloc(mdoc, line, ppos, tok);
1164
1165 if (ac == ARGS_PHRASE) {
1166
1167 /*
1168 * If we haven't opened a body yet, rewind the
1169 * head; if we have, rewind that instead.
1170 */
1171
1172 rew_last(mdoc, body == NULL ? head : body);
1173 body = roff_body_alloc(mdoc, line, ppos, tok);
1174
1175 /* Process to the tab or to the end of the line. */
1176
1177 mdoc->flags |= MDOC_PHRASE;
1178 parse_rest(mdoc, TOKEN_NONE, line, &la, buf);
1179 mdoc->flags &= ~MDOC_PHRASE;
1180
1181 /* There may have been `Ta' macros. */
1182
1183 while (body->next != NULL)
1184 body = body->next;
1185 continue;
1186 }
1187
1188 done = macro_or_word(mdoc, tok, line, la, pos, buf, p, parsed);
1189 if (ac == ARGS_ALLOC)
1190 free(p);
1191 if (done)
1192 break;
1193 }
1194
1195 if (blk->flags & NODE_VALID)
1196 return;
1197 if (head == NULL)
1198 head = roff_head_alloc(mdoc, line, ppos, tok);
1199 if (nl && tok != MDOC_Bd && tok != MDOC_Bl && tok != MDOC_Rs)
1200 append_delims(mdoc, line, pos, buf);
1201 if (body != NULL)
1202 goto out;
1203 if (find_pending(mdoc, tok, line, ppos, head))
1204 return;
1205
1206 /* Close out scopes to remain in a consistent state. */
1207
1208 rew_last(mdoc, head);
1209 body = roff_body_alloc(mdoc, line, ppos, tok);
1210 if (tok == MDOC_Ss)
1211 mdoc->flags &= ~ROFF_NONOFILL;
1212
1213 /*
1214 * Set up fill mode for display blocks.
1215 * This needs to be done here up front rather than during
1216 * validation such that child nodes get the right flags.
1217 */
1218
1219 if (tok == MDOC_Bd && arg != NULL) {
1220 for (iarg = 0; iarg < arg->argc; iarg++) {
1221 switch (arg->argv[iarg].arg) {
1222 case MDOC_Unfilled:
1223 case MDOC_Literal:
1224 mdoc->flags |= ROFF_NOFILL;
1225 break;
1226 case MDOC_Filled:
1227 case MDOC_Ragged:
1228 case MDOC_Centred:
1229 mdoc->flags &= ~ROFF_NOFILL;
1230 break;
1231 default:
1232 continue;
1233 }
1234 break;
1235 }
1236 }
1237 out:
1238 if (mdoc->flags & MDOC_FREECOL) {
1239 rew_last(mdoc, body);
1240 rew_last(mdoc, blk);
1241 mdoc->flags &= ~MDOC_FREECOL;
1242 }
1243 }
1244
1245 static void
blk_part_imp(MACRO_PROT_ARGS)1246 blk_part_imp(MACRO_PROT_ARGS)
1247 {
1248 int done, la, nl;
1249 enum margserr ac;
1250 char *p;
1251 struct roff_node *blk; /* saved block context */
1252 struct roff_node *body; /* saved body context */
1253 struct roff_node *n;
1254
1255 nl = MDOC_NEWLINE & mdoc->flags;
1256
1257 /*
1258 * A macro that spans to the end of the line. This is generally
1259 * (but not necessarily) called as the first macro. The block
1260 * has a head as the immediate child, which is always empty,
1261 * followed by zero or more opening punctuation nodes, then the
1262 * body (which may be empty, depending on the macro), then zero
1263 * or more closing punctuation nodes.
1264 */
1265
1266 blk = mdoc_block_alloc(mdoc, line, ppos, tok, NULL);
1267 rew_last(mdoc, roff_head_alloc(mdoc, line, ppos, tok));
1268
1269 /*
1270 * Open the body scope "on-demand", that is, after we've
1271 * processed all our the leading delimiters (open parenthesis,
1272 * etc.).
1273 */
1274
1275 for (body = NULL; ; ) {
1276 la = *pos;
1277 ac = mdoc_args(mdoc, line, pos, buf, tok, &p);
1278 if (ac == ARGS_EOLN || ac == ARGS_PUNCT)
1279 break;
1280
1281 if (body == NULL && mdoc_isdelim(p) == DELIM_OPEN) {
1282 dword(mdoc, line, la, p, DELIM_OPEN, 0);
1283 if (ac == ARGS_ALLOC)
1284 free(p);
1285 continue;
1286 }
1287
1288 if (body == NULL)
1289 body = roff_body_alloc(mdoc, line, ppos, tok);
1290
1291 done = macro_or_word(mdoc, tok, line, la, pos, buf, p, 1);
1292 if (ac == ARGS_ALLOC)
1293 free(p);
1294 if (done)
1295 break;
1296 }
1297 if (body == NULL)
1298 body = roff_body_alloc(mdoc, line, ppos, tok);
1299
1300 if (find_pending(mdoc, tok, line, ppos, body))
1301 return;
1302
1303 rew_last(mdoc, body);
1304 if (nl)
1305 append_delims(mdoc, line, pos, buf);
1306 rew_pending(mdoc, blk);
1307
1308 /* Move trailing .Ns out of scope. */
1309
1310 for (n = body->child; n && n->next; n = n->next)
1311 /* Do nothing. */ ;
1312 if (n && n->tok == MDOC_Ns)
1313 roff_node_relink(mdoc, n);
1314 }
1315
1316 static void
blk_part_exp(MACRO_PROT_ARGS)1317 blk_part_exp(MACRO_PROT_ARGS)
1318 {
1319 int done, la, nl;
1320 enum margserr ac;
1321 struct roff_node *head; /* keep track of head */
1322 char *p;
1323
1324 nl = MDOC_NEWLINE & mdoc->flags;
1325
1326 /*
1327 * The opening of an explicit macro having zero or more leading
1328 * punctuation nodes; a head with optional single element (the
1329 * case of `Eo'); and a body that may be empty.
1330 */
1331
1332 roff_block_alloc(mdoc, line, ppos, tok);
1333 head = NULL;
1334 for (;;) {
1335 la = *pos;
1336 ac = mdoc_args(mdoc, line, pos, buf, tok, &p);
1337 if (ac == ARGS_PUNCT || ac == ARGS_EOLN)
1338 break;
1339
1340 /* Flush out leading punctuation. */
1341
1342 if (head == NULL && mdoc_isdelim(p) == DELIM_OPEN) {
1343 dword(mdoc, line, la, p, DELIM_OPEN, 0);
1344 if (ac == ARGS_ALLOC)
1345 free(p);
1346 continue;
1347 }
1348
1349 if (head == NULL) {
1350 head = roff_head_alloc(mdoc, line, ppos, tok);
1351 if (tok == MDOC_Eo) /* Not parsed. */
1352 dword(mdoc, line, la, p, DELIM_MAX, 0);
1353 rew_last(mdoc, head);
1354 roff_body_alloc(mdoc, line, ppos, tok);
1355 if (tok == MDOC_Eo) {
1356 if (ac == ARGS_ALLOC)
1357 free(p);
1358 continue;
1359 }
1360 }
1361
1362 done = macro_or_word(mdoc, tok, line, la, pos, buf, p, 1);
1363 if (ac == ARGS_ALLOC)
1364 free(p);
1365 if (done)
1366 break;
1367 }
1368
1369 /* Clean-up to leave in a consistent state. */
1370
1371 if (head == NULL) {
1372 rew_last(mdoc, roff_head_alloc(mdoc, line, ppos, tok));
1373 roff_body_alloc(mdoc, line, ppos, tok);
1374 }
1375 if (nl)
1376 append_delims(mdoc, line, pos, buf);
1377 }
1378
1379 static void
in_line_argn(MACRO_PROT_ARGS)1380 in_line_argn(MACRO_PROT_ARGS)
1381 {
1382 struct mdoc_arg *arg;
1383 char *p;
1384 enum margserr ac;
1385 enum roff_tok ntok;
1386 int state; /* arg#; -1: not yet open; -2: closed */
1387 int la, maxargs, nl;
1388
1389 nl = mdoc->flags & MDOC_NEWLINE;
1390
1391 /*
1392 * A line macro that has a fixed number of arguments (maxargs).
1393 * Only open the scope once the first non-leading-punctuation is
1394 * found (unless MDOC_IGNDELIM is noted, like in `Pf'), then
1395 * keep it open until the maximum number of arguments are
1396 * exhausted.
1397 */
1398
1399 switch (tok) {
1400 case MDOC_Ap:
1401 case MDOC_Ns:
1402 case MDOC_Ux:
1403 maxargs = 0;
1404 break;
1405 case MDOC_Bx:
1406 case MDOC_Es:
1407 case MDOC_Xr:
1408 maxargs = 2;
1409 break;
1410 default:
1411 maxargs = 1;
1412 break;
1413 }
1414
1415 mdoc_argv(mdoc, line, tok, &arg, pos, buf);
1416
1417 state = -1;
1418 p = NULL;
1419 for (;;) {
1420 la = *pos;
1421 ac = mdoc_args(mdoc, line, pos, buf, tok, &p);
1422
1423 if ((ac == ARGS_WORD || ac == ARGS_ALLOC) && state == -1 &&
1424 (mdoc_macro(tok)->flags & MDOC_IGNDELIM) == 0 &&
1425 mdoc_isdelim(p) == DELIM_OPEN) {
1426 dword(mdoc, line, la, p, DELIM_OPEN, 0);
1427 if (ac == ARGS_ALLOC)
1428 free(p);
1429 continue;
1430 }
1431
1432 if (state == -1 && tok != MDOC_In &&
1433 tok != MDOC_St && tok != MDOC_Xr) {
1434 mdoc_elem_alloc(mdoc, line, ppos, tok, arg);
1435 state = 0;
1436 }
1437
1438 if (ac == ARGS_PUNCT || ac == ARGS_EOLN) {
1439 if (abs(state) < 2 && tok == MDOC_Pf)
1440 mandoc_msg(MANDOCERR_PF_SKIP,
1441 line, ppos, "Pf %s",
1442 p == NULL ? "at eol" : p);
1443 break;
1444 }
1445
1446 if (state == maxargs) {
1447 rew_elem(mdoc, tok);
1448 state = -2;
1449 }
1450
1451 ntok = (tok == MDOC_Pf && state == 0) ?
1452 TOKEN_NONE : lookup(mdoc, tok, line, la, p);
1453
1454 if (ntok != TOKEN_NONE) {
1455 if (state >= 0) {
1456 rew_elem(mdoc, tok);
1457 state = -2;
1458 }
1459 (*mdoc_macro(ntok)->fp)(mdoc, ntok,
1460 line, la, pos, buf);
1461 if (ac == ARGS_ALLOC)
1462 free(p);
1463 break;
1464 }
1465
1466 if (mdoc_macro(tok)->flags & MDOC_IGNDELIM ||
1467 mdoc_isdelim(p) == DELIM_NONE) {
1468 if (state == -1) {
1469 mdoc_elem_alloc(mdoc, line, ppos, tok, arg);
1470 state = 1;
1471 } else if (state >= 0)
1472 state++;
1473 } else if (state >= 0) {
1474 rew_elem(mdoc, tok);
1475 state = -2;
1476 }
1477
1478 dword(mdoc, line, la, p, DELIM_MAX,
1479 mdoc_macro(tok)->flags & MDOC_JOIN);
1480 if (ac == ARGS_ALLOC)
1481 free(p);
1482 p = mdoc->last->string;
1483 }
1484
1485 if (state == -1) {
1486 mandoc_msg(MANDOCERR_MACRO_EMPTY,
1487 line, ppos, "%s", roff_name[tok]);
1488 return;
1489 }
1490
1491 if (state == 0 && tok == MDOC_Pf)
1492 append_delims(mdoc, line, pos, buf);
1493 if (state >= 0)
1494 rew_elem(mdoc, tok);
1495 if (nl)
1496 append_delims(mdoc, line, pos, buf);
1497 }
1498
1499 static void
in_line_eoln(MACRO_PROT_ARGS)1500 in_line_eoln(MACRO_PROT_ARGS)
1501 {
1502 struct roff_node *n;
1503 struct mdoc_arg *arg;
1504
1505 if ((tok == MDOC_Pp || tok == MDOC_Lp) &&
1506 ! (mdoc->flags & MDOC_SYNOPSIS)) {
1507 n = mdoc->last;
1508 if (mdoc->next == ROFF_NEXT_SIBLING)
1509 n = n->parent;
1510 if (n->tok == MDOC_Nm)
1511 rew_last(mdoc, n->parent);
1512 }
1513
1514 if (buf[*pos] == '\0' &&
1515 (tok == MDOC_Fd || *roff_name[tok] == '%')) {
1516 mandoc_msg(MANDOCERR_MACRO_EMPTY,
1517 line, ppos, "%s", roff_name[tok]);
1518 return;
1519 }
1520
1521 mdoc_argv(mdoc, line, tok, &arg, pos, buf);
1522 mdoc_elem_alloc(mdoc, line, ppos, tok, arg);
1523 if (parse_rest(mdoc, tok, line, pos, buf))
1524 return;
1525 rew_elem(mdoc, tok);
1526 }
1527
1528 /*
1529 * The simplest argument parser available: Parse the remaining
1530 * words until the end of the phrase or line and return 0
1531 * or until the next macro, call that macro, and return 1.
1532 */
1533 static int
parse_rest(struct roff_man * mdoc,enum roff_tok tok,int line,int * pos,char * buf)1534 parse_rest(struct roff_man *mdoc, enum roff_tok tok,
1535 int line, int *pos, char *buf)
1536 {
1537 char *p;
1538 int done, la;
1539 enum margserr ac;
1540
1541 for (;;) {
1542 la = *pos;
1543 ac = mdoc_args(mdoc, line, pos, buf, tok, &p);
1544 if (ac == ARGS_EOLN)
1545 return 0;
1546 done = macro_or_word(mdoc, tok, line, la, pos, buf, p, 1);
1547 if (ac == ARGS_ALLOC)
1548 free(p);
1549 if (done)
1550 return 1;
1551 }
1552 }
1553
1554 static void
ctx_synopsis(MACRO_PROT_ARGS)1555 ctx_synopsis(MACRO_PROT_ARGS)
1556 {
1557
1558 if (~mdoc->flags & (MDOC_SYNOPSIS | MDOC_NEWLINE))
1559 in_line(mdoc, tok, line, ppos, pos, buf);
1560 else if (tok == MDOC_Nm)
1561 blk_full(mdoc, tok, line, ppos, pos, buf);
1562 else {
1563 assert(tok == MDOC_Vt);
1564 blk_part_imp(mdoc, tok, line, ppos, pos, buf);
1565 }
1566 }
1567
1568 /*
1569 * Phrases occur within `Bl -column' entries, separated by `Ta' or tabs.
1570 * They're unusual because they're basically free-form text until a
1571 * macro is encountered.
1572 */
1573 static void
phrase_ta(MACRO_PROT_ARGS)1574 phrase_ta(MACRO_PROT_ARGS)
1575 {
1576 struct roff_node *body, *n;
1577
1578 /* Make sure we are in a column list or ignore this macro. */
1579
1580 body = NULL;
1581 for (n = mdoc->last; n != NULL; n = n->parent) {
1582 if (n->flags & NODE_ENDED)
1583 continue;
1584 if (n->tok == MDOC_It && n->type == ROFFT_BODY)
1585 body = n;
1586 if (n->tok == MDOC_Bl && n->end == ENDBODY_NOT)
1587 break;
1588 }
1589
1590 if (n == NULL || n->norm->Bl.type != LIST_column) {
1591 mandoc_msg(MANDOCERR_TA_STRAY, line, ppos, "Ta");
1592 return;
1593 }
1594
1595 /* Advance to the next column. */
1596
1597 rew_last(mdoc, body);
1598 roff_body_alloc(mdoc, line, ppos, MDOC_It);
1599 parse_rest(mdoc, TOKEN_NONE, line, pos, buf);
1600 }
1601