1 /* $Id: mdoc_macro.c,v 1.157 2014/12/13 13:14:39 schwarze Exp $ */
3 * Copyright (c) 2008-2012 Kristaps Dzonsons <kristaps@bsd.lv>
4 * Copyright (c) 2010, 2012, 2013, 2014 Ingo Schwarze <schwarze@openbsd.org>
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.
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.
20 #include <sys/types.h>
32 #include "libmandoc.h"
34 enum rew { /* see rew_dohalt() */
43 static void blk_full(MACRO_PROT_ARGS);
44 static void blk_exp_close(MACRO_PROT_ARGS);
45 static void blk_part_exp(MACRO_PROT_ARGS);
46 static void blk_part_imp(MACRO_PROT_ARGS);
47 static void ctx_synopsis(MACRO_PROT_ARGS);
48 static void in_line_eoln(MACRO_PROT_ARGS);
49 static void in_line_argn(MACRO_PROT_ARGS);
50 static void in_line(MACRO_PROT_ARGS);
51 static void phrase_ta(MACRO_PROT_ARGS);
53 static void dword(struct mdoc *, int, int, const char *,
55 static void append_delims(struct mdoc *, int, int *, char *);
56 static enum mdoct lookup(struct mdoc *, enum mdoct,
57 int, int, const char *);
58 static int macro_or_word(MACRO_PROT_ARGS, int);
59 static int make_pending(struct mdoc_node *, enum mdoct,
60 struct mdoc *, int, int);
61 static int parse_rest(struct mdoc *, enum mdoct,
63 static enum mdoct rew_alt(enum mdoct);
64 static enum rew rew_dohalt(enum mdoct, enum mdoc_type,
65 const struct mdoc_node *);
66 static void rew_elem(struct mdoc *, enum mdoct);
67 static void rew_last(struct mdoc *, const struct mdoc_node *);
68 static void rew_sub(enum mdoc_type, struct mdoc *,
69 enum mdoct, int, int);
71 const struct mdoc_macro __mdoc_macros[MDOC_MAX] = {
72 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* Ap */
73 { in_line_eoln, MDOC_PROLOGUE }, /* Dd */
74 { in_line_eoln, MDOC_PROLOGUE }, /* Dt */
75 { in_line_eoln, MDOC_PROLOGUE }, /* Os */
76 { blk_full, MDOC_PARSED | MDOC_JOIN }, /* Sh */
77 { blk_full, MDOC_PARSED | MDOC_JOIN }, /* Ss */
78 { in_line_eoln, 0 }, /* Pp */
79 { blk_part_imp, MDOC_PARSED | MDOC_JOIN }, /* D1 */
80 { blk_part_imp, MDOC_PARSED | MDOC_JOIN }, /* Dl */
81 { blk_full, MDOC_EXPLICIT }, /* Bd */
82 { blk_exp_close, MDOC_EXPLICIT | MDOC_JOIN }, /* Ed */
83 { blk_full, MDOC_EXPLICIT }, /* Bl */
84 { blk_exp_close, MDOC_EXPLICIT | MDOC_JOIN }, /* El */
85 { blk_full, MDOC_PARSED | MDOC_JOIN }, /* It */
86 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Ad */
87 { in_line, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* An */
88 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Ar */
89 { in_line, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* Cd */
90 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Cm */
91 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Dv */
92 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Er */
93 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Ev */
94 { in_line_eoln, 0 }, /* Ex */
95 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Fa */
96 { in_line_eoln, 0 }, /* Fd */
97 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Fl */
98 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Fn */
99 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Ft */
100 { in_line, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* Ic */
101 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* In */
102 { in_line, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* Li */
103 { blk_full, MDOC_JOIN }, /* Nd */
104 { ctx_synopsis, MDOC_CALLABLE | MDOC_PARSED }, /* Nm */
105 { blk_part_imp, MDOC_CALLABLE | MDOC_PARSED }, /* Op */
106 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Ot */
107 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Pa */
108 { in_line_eoln, 0 }, /* Rv */
109 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* St */
110 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Va */
111 { ctx_synopsis, MDOC_CALLABLE | MDOC_PARSED }, /* Vt */
112 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* Xr */
113 { in_line_eoln, MDOC_JOIN }, /* %A */
114 { in_line_eoln, MDOC_JOIN }, /* %B */
115 { in_line_eoln, MDOC_JOIN }, /* %D */
116 { in_line_eoln, MDOC_JOIN }, /* %I */
117 { in_line_eoln, MDOC_JOIN }, /* %J */
118 { in_line_eoln, 0 }, /* %N */
119 { in_line_eoln, MDOC_JOIN }, /* %O */
120 { in_line_eoln, 0 }, /* %P */
121 { in_line_eoln, MDOC_JOIN }, /* %R */
122 { in_line_eoln, MDOC_JOIN }, /* %T */
123 { in_line_eoln, 0 }, /* %V */
124 { blk_exp_close, MDOC_CALLABLE | MDOC_PARSED |
125 MDOC_EXPLICIT | MDOC_JOIN }, /* Ac */
126 { blk_part_exp, MDOC_CALLABLE | MDOC_PARSED |
127 MDOC_EXPLICIT | MDOC_JOIN }, /* Ao */
128 { blk_part_imp, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* Aq */
129 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* At */
130 { blk_exp_close, MDOC_CALLABLE | MDOC_PARSED |
131 MDOC_EXPLICIT | MDOC_JOIN }, /* Bc */
132 { blk_full, MDOC_EXPLICIT }, /* Bf */
133 { blk_part_exp, MDOC_CALLABLE | MDOC_PARSED |
134 MDOC_EXPLICIT | MDOC_JOIN }, /* Bo */
135 { blk_part_imp, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* Bq */
136 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* Bsx */
137 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* Bx */
138 { in_line_eoln, 0 }, /* Db */
139 { blk_exp_close, MDOC_CALLABLE | MDOC_PARSED |
140 MDOC_EXPLICIT | MDOC_JOIN }, /* Dc */
141 { blk_part_exp, MDOC_CALLABLE | MDOC_PARSED |
142 MDOC_EXPLICIT | MDOC_JOIN }, /* Do */
143 { blk_part_imp, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* Dq */
144 { blk_exp_close, MDOC_CALLABLE | MDOC_PARSED | MDOC_EXPLICIT }, /* Ec */
145 { blk_exp_close, MDOC_EXPLICIT | MDOC_JOIN }, /* Ef */
146 { in_line, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* Em */
147 { blk_part_exp, MDOC_CALLABLE | MDOC_PARSED | MDOC_EXPLICIT }, /* Eo */
148 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* Fx */
149 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Ms */
150 { in_line, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* No */
151 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED |
152 MDOC_IGNDELIM | MDOC_JOIN }, /* Ns */
153 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* Nx */
154 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* Ox */
155 { blk_exp_close, MDOC_CALLABLE | MDOC_PARSED |
156 MDOC_EXPLICIT | MDOC_JOIN }, /* Pc */
157 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED | MDOC_IGNDELIM }, /* Pf */
158 { blk_part_exp, MDOC_CALLABLE | MDOC_PARSED |
159 MDOC_EXPLICIT | MDOC_JOIN }, /* Po */
160 { blk_part_imp, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* Pq */
161 { blk_exp_close, MDOC_CALLABLE | MDOC_PARSED |
162 MDOC_EXPLICIT | MDOC_JOIN }, /* Qc */
163 { blk_part_imp, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* Ql */
164 { blk_part_exp, MDOC_CALLABLE | MDOC_PARSED |
165 MDOC_EXPLICIT | MDOC_JOIN }, /* Qo */
166 { blk_part_imp, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* Qq */
167 { blk_exp_close, MDOC_EXPLICIT | MDOC_JOIN }, /* Re */
168 { blk_full, MDOC_EXPLICIT }, /* Rs */
169 { blk_exp_close, MDOC_CALLABLE | MDOC_PARSED |
170 MDOC_EXPLICIT | MDOC_JOIN }, /* Sc */
171 { blk_part_exp, MDOC_CALLABLE | MDOC_PARSED |
172 MDOC_EXPLICIT | MDOC_JOIN }, /* So */
173 { blk_part_imp, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* Sq */
174 { in_line_argn, 0 }, /* Sm */
175 { in_line, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* Sx */
176 { in_line, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* Sy */
177 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Tn */
178 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* Ux */
179 { blk_exp_close, MDOC_EXPLICIT | MDOC_CALLABLE | MDOC_PARSED }, /* Xc */
180 { blk_part_exp, MDOC_CALLABLE | MDOC_PARSED | MDOC_EXPLICIT }, /* Xo */
181 { blk_full, MDOC_EXPLICIT | MDOC_CALLABLE }, /* Fo */
182 { blk_exp_close, MDOC_CALLABLE | MDOC_PARSED |
183 MDOC_EXPLICIT | MDOC_JOIN }, /* Fc */
184 { blk_part_exp, MDOC_CALLABLE | MDOC_PARSED |
185 MDOC_EXPLICIT | MDOC_JOIN }, /* Oo */
186 { blk_exp_close, MDOC_CALLABLE | MDOC_PARSED |
187 MDOC_EXPLICIT | MDOC_JOIN }, /* Oc */
188 { blk_full, MDOC_EXPLICIT }, /* Bk */
189 { blk_exp_close, MDOC_EXPLICIT | MDOC_JOIN }, /* Ek */
190 { in_line_eoln, 0 }, /* Bt */
191 { in_line_eoln, 0 }, /* Hf */
192 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Fr */
193 { in_line_eoln, 0 }, /* Ud */
194 { in_line, 0 }, /* Lb */
195 { in_line_eoln, 0 }, /* Lp */
196 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Lk */
197 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Mt */
198 { blk_part_imp, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* Brq */
199 { blk_part_exp, MDOC_CALLABLE | MDOC_PARSED |
200 MDOC_EXPLICIT | MDOC_JOIN }, /* Bro */
201 { blk_exp_close, MDOC_CALLABLE | MDOC_PARSED |
202 MDOC_EXPLICIT | MDOC_JOIN }, /* Brc */
203 { in_line_eoln, MDOC_JOIN }, /* %C */
204 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* Es */
205 { blk_part_imp, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* En */
206 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* Dx */
207 { in_line_eoln, MDOC_JOIN }, /* %Q */
208 { in_line_eoln, 0 }, /* br */
209 { in_line_eoln, 0 }, /* sp */
210 { in_line_eoln, 0 }, /* %U */
211 { phrase_ta, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* Ta */
212 { in_line_eoln, MDOC_PROLOGUE }, /* ll */
215 const struct mdoc_macro * const mdoc_macros = __mdoc_macros;
219 * This is called at the end of parsing. It must traverse up the tree,
220 * closing out open [implicit] scopes. Obviously, open explicit scopes
224 mdoc_macroend(struct mdoc *mdoc)
228 /* Scan for open explicit scopes. */
230 n = mdoc->last->flags & MDOC_VALID ?
231 mdoc->last->parent : mdoc->last;
233 for ( ; n; n = n->parent)
234 if (n->type == MDOC_BLOCK &&
235 mdoc_macros[n->tok].flags & MDOC_EXPLICIT)
236 mandoc_msg(MANDOCERR_BLK_NOEND, mdoc->parse,
237 n->line, n->pos, mdoc_macronames[n->tok]);
239 /* Rewind to the first. */
241 rew_last(mdoc, mdoc->first);
245 * Look up the macro at *p called by "from",
246 * or as a line macro if from == MDOC_MAX.
249 lookup(struct mdoc *mdoc, enum mdoct from, int line, int ppos, const char *p)
253 if (from == MDOC_MAX || mdoc_macros[from].flags & MDOC_PARSED) {
254 res = mdoc_hash_find(p);
255 if (res != MDOC_MAX) {
256 if (mdoc_macros[res].flags & MDOC_CALLABLE)
258 if (res != MDOC_br && res != MDOC_sp && res != MDOC_ll)
259 mandoc_msg(MANDOCERR_MACRO_CALL,
260 mdoc->parse, line, ppos, p);
267 rew_last(struct mdoc *mdoc, const struct mdoc_node *to)
269 struct mdoc_node *n, *np;
272 mdoc->next = MDOC_NEXT_SIBLING;
273 while (mdoc->last != to) {
275 * Save the parent here, because we may delete the
276 * mdoc->last node in the post-validation phase and reset
277 * it to mdoc->last->parent, causing a step in the closing
280 np = mdoc->last->parent;
281 mdoc_valid_post(mdoc);
285 mdoc->last->last = n;
287 mdoc_valid_post(mdoc);
291 * For a block closing macro, return the corresponding opening one.
292 * Otherwise, return the macro itself.
295 rew_alt(enum mdoct tok)
337 * Rewinding to tok, how do we have to handle *p?
338 * REWIND_NONE: *p would delimit tok, but no tok scope is open
339 * inside *p, so there is no need to rewind anything at all.
340 * REWIND_THIS: *p matches tok, so rewind *p and nothing else.
341 * REWIND_MORE: *p is implicit, rewind it and keep searching for tok.
342 * REWIND_FORCE: *p is explicit, but tok is full, force rewinding *p.
343 * REWIND_LATER: *p is explicit and still open, postpone rewinding.
344 * REWIND_ERROR: No tok block is open at all.
347 rew_dohalt(enum mdoct tok, enum mdoc_type type,
348 const struct mdoc_node *p)
352 * No matching token, no delimiting block, no broken block.
353 * This can happen when full implicit macros are called for
354 * the first time but try to rewind their previous
357 if (MDOC_ROOT == p->type)
358 return(MDOC_BLOCK == type &&
359 MDOC_EXPLICIT & mdoc_macros[tok].flags ?
360 REWIND_ERROR : REWIND_NONE);
363 * When starting to rewind, skip plain text
364 * and nodes that have already been rewound.
366 if (MDOC_TEXT == p->type || MDOC_VALID & p->flags)
370 * The easiest case: Found a matching token.
371 * This applies to both blocks and elements.
375 return(p->end ? REWIND_NONE :
376 type == p->type ? REWIND_THIS : REWIND_MORE);
379 * While elements do require rewinding for themselves,
380 * they never affect rewinding of other nodes.
382 if (MDOC_ELEM == p->type)
386 * Blocks delimited by our target token get REWIND_MORE.
387 * Blocks delimiting our target token get REWIND_NONE.
391 if (MDOC_It == p->tok)
395 if (MDOC_BODY == p->type && MDOC_Bl == p->tok)
399 * XXX Badly nested block handling still fails badly
400 * when one block is breaking two blocks of the same type.
401 * This is an incomplete and extremely ugly workaround,
402 * required to let the OpenBSD tree build.
405 if (MDOC_Op == p->tok)
413 if (MDOC_BODY == p->type && MDOC_Sh == p->tok)
417 if (MDOC_ROOT == p->parent->type)
419 if (MDOC_Nd == p->tok || MDOC_Ss == p->tok ||
428 * Default block rewinding rules.
429 * In particular, always skip block end markers,
430 * and let all blocks rewind Nm children.
431 * Do not warn again when closing a block,
432 * since closing the body already warned.
434 if (ENDBODY_NOT != p->end || MDOC_Nm == p->tok ||
435 MDOC_BLOCK == type || (MDOC_BLOCK == p->type &&
436 ! (MDOC_EXPLICIT & mdoc_macros[tok].flags)))
440 * By default, closing out full blocks
441 * forces closing of broken explicit blocks,
442 * while closing out partial blocks
443 * allows delayed rewinding by default.
445 return (&blk_full == mdoc_macros[tok].fp ?
446 REWIND_FORCE : REWIND_LATER);
450 rew_elem(struct mdoc *mdoc, enum mdoct tok)
455 if (MDOC_ELEM != n->type)
457 assert(MDOC_ELEM == n->type);
458 assert(tok == n->tok);
463 * We are trying to close a block identified by tok,
464 * but the child block *broken is still open.
465 * Thus, postpone closing the tok block
466 * until the rew_sub call closing *broken.
469 make_pending(struct mdoc_node *broken, enum mdoct tok,
470 struct mdoc *mdoc, int line, int ppos)
472 struct mdoc_node *breaker;
475 * Iterate backwards, searching for the block matching tok,
476 * that is, the block breaking the *broken block.
478 for (breaker = broken->parent; breaker; breaker = breaker->parent) {
481 * If the *broken block had already been broken before
482 * and we encounter its breaker, make the tok block
483 * pending on the inner breaker.
484 * Graphically, "[A breaker=[B broken=[C->B B] tok=A] C]"
485 * becomes "[A broken=[B [C->B B] tok=A] C]"
486 * and finally "[A [B->A [C->B B] A] C]".
488 if (breaker == broken->pending) {
493 if (REWIND_THIS != rew_dohalt(tok, MDOC_BLOCK, breaker))
495 if (MDOC_BODY == broken->type)
496 broken = broken->parent;
500 * If another, outer breaker is already pending on
501 * the *broken block, we must not clobber the link
502 * to the outer breaker, but make it pending on the
503 * new, now inner breaker.
504 * Graphically, "[A breaker=[B broken=[C->A A] tok=B] C]"
505 * becomes "[A breaker=[B->A broken=[C A] tok=B] C]"
506 * and finally "[A [B->A [C->B A] B] C]".
508 if (broken->pending) {
509 struct mdoc_node *taker;
512 * If the breaker had also been broken before,
513 * it cannot take on the outer breaker itself,
514 * but must hand it on to its own breakers.
515 * Graphically, this is the following situation:
516 * "[A [B breaker=[C->B B] broken=[D->A A] tok=C] D]"
517 * "[A taker=[B->A breaker=[C->B B] [D->C A] C] D]"
520 while (taker->pending)
521 taker = taker->pending;
522 taker->pending = broken->pending;
524 broken->pending = breaker;
525 mandoc_vmsg(MANDOCERR_BLK_NEST, mdoc->parse, line, ppos,
526 "%s breaks %s", mdoc_macronames[tok],
527 mdoc_macronames[broken->tok]);
532 * Found no matching block for tok.
533 * Are you trying to close a block that is not open?
539 rew_sub(enum mdoc_type t, struct mdoc *mdoc,
540 enum mdoct tok, int line, int ppos)
546 switch (rew_dohalt(tok, t, n)) {
551 (mdoc->flags & MDOC_NEWLINE &&
552 ! (mdoc_macros[tok].flags & MDOC_EXPLICIT));
555 mandoc_vmsg(MANDOCERR_BLK_BROKEN, mdoc->parse,
556 line, ppos, "%s breaks %s",
557 mdoc_macronames[tok],
558 mdoc_macronames[n->tok]);
562 (mdoc->flags & MDOC_NEWLINE ? 1 : 0);
566 if (make_pending(n, tok, mdoc, line, ppos) ||
571 mandoc_msg(MANDOCERR_BLK_NOTOPEN,
572 mdoc->parse, line, ppos,
573 mdoc_macronames[tok]);
582 * The current block extends an enclosing block.
583 * Now that the current block ends, close the enclosing block, too.
585 while ((n = n->pending) != NULL) {
587 if (n->type == MDOC_HEAD)
588 mdoc_body_alloc(mdoc, n->line, n->pos, n->tok);
593 * Allocate a word and check whether it's punctuation or not.
594 * Punctuation consists of those tokens found in mdoc_isdelim().
597 dword(struct mdoc *mdoc, int line, int col, const char *p,
598 enum mdelim d, int may_append)
605 ! (mdoc->flags & (MDOC_SYNOPSIS | MDOC_KEEP | MDOC_SMOFF)) &&
606 d == DELIM_NONE && mdoc->last->type == MDOC_TEXT &&
607 mdoc_isdelim(mdoc->last->string) == DELIM_NONE) {
608 mdoc_word_append(mdoc, p);
612 mdoc_word_alloc(mdoc, line, col, p);
615 * If the word consists of a bare delimiter,
616 * flag the new node accordingly,
617 * unless doing so was vetoed by the invoking macro.
618 * Always clear the veto, it is only valid for one word.
622 mdoc->last->flags |= MDOC_DELIMO;
623 else if (d == DELIM_CLOSE &&
624 ! (mdoc->flags & MDOC_NODELIMC) &&
625 mdoc->last->parent->tok != MDOC_Fd)
626 mdoc->last->flags |= MDOC_DELIMC;
627 mdoc->flags &= ~MDOC_NODELIMC;
631 append_delims(struct mdoc *mdoc, int line, int *pos, char *buf)
636 if (buf[*pos] == '\0')
641 if (mdoc_args(mdoc, line, pos, buf, MDOC_MAX, &p) == ARGS_EOLN)
643 dword(mdoc, line, la, p, DELIM_MAX, 1);
646 * If we encounter end-of-sentence symbols, then trigger
649 * XXX: it's easy to allow this to propagate outward to
650 * the last symbol, such that `. )' will cause the
651 * correct double-spacing. However, (1) groff isn't
652 * smart enough to do this and (2) it would require
653 * knowing which symbols break this behaviour, for
654 * example, `. ;' shouldn't propagate the double-space.
657 if (mandoc_eos(p, strlen(p)))
658 mdoc->last->flags |= MDOC_EOS;
664 * If it is a macro, call it and return 1.
665 * Otherwise, allocate it and return 0.
668 macro_or_word(MACRO_PROT_ARGS, int parsed)
677 else if (parsed && ! (mdoc->flags & MDOC_PHRASELIT))
678 ntok = lookup(mdoc, tok, line, ppos, p);
680 if (ntok == MDOC_MAX) {
681 dword(mdoc, line, ppos, p, DELIM_MAX, tok == MDOC_MAX ||
682 mdoc_macros[tok].flags & MDOC_JOIN);
685 if (mdoc_macros[tok].fp == in_line_eoln)
687 mdoc_macro(mdoc, ntok, line, ppos, pos, buf);
689 append_delims(mdoc, line, pos, buf);
695 * Close out block partial/full explicit.
698 blk_exp_close(MACRO_PROT_ARGS)
700 struct mdoc_node *body; /* Our own body. */
701 struct mdoc_node *endbody; /* Our own end marker. */
702 struct mdoc_node *later; /* A sub-block starting later. */
703 struct mdoc_node *n; /* For searching backwards. */
705 int j, lastarg, maxargs, flushed, nl;
707 enum mdoct atok, ntok;
710 nl = MDOC_NEWLINE & mdoc->flags;
717 mdoc->flags &= ~MDOC_KEEP;
725 * Search backwards for beginnings of blocks,
726 * both of our own and of pending sub-blocks.
730 body = endbody = later = NULL;
731 for (n = mdoc->last; n; n = n->parent) {
732 if (n->flags & MDOC_VALID)
735 /* Remember the start of our own body. */
737 if (n->type == MDOC_BODY && atok == n->tok) {
738 if (n->end == ENDBODY_NOT)
743 if (n->type != MDOC_BLOCK || n->tok == MDOC_Nm)
745 if (atok == n->tok) {
749 * Found the start of our own block.
750 * When there is no pending sub block,
751 * just proceed to closing out.
758 * When there is a pending sub block,
759 * postpone closing out the current block
760 * until the rew_sub() closing out the sub-block.
763 make_pending(later, tok, mdoc, line, ppos);
766 * Mark the place where the formatting - but not
767 * the scope - of the current block ends.
770 mdoc_endbody_alloc(mdoc, line, ppos,
771 atok, body, ENDBODY_SPACE);
774 * If a block closing macro taking arguments
775 * breaks another block, put the arguments
776 * into the end marker and remeber the
777 * end marker in order to close it out.
781 endbody = mdoc->last;
782 mdoc->next = MDOC_NEXT_CHILD;
788 * When finding an open sub block, remember the last
789 * open explicit block, or, in case there are only
790 * implicit ones, the first open implicit block.
794 mdoc_macros[later->tok].flags & MDOC_EXPLICIT)
796 if (n->tok != MDOC_It)
799 rew_sub(MDOC_BODY, mdoc, tok, line, ppos);
801 if ( ! (mdoc_macros[tok].flags & MDOC_PARSED)) {
802 if (buf[*pos] != '\0')
803 mandoc_vmsg(MANDOCERR_ARG_SKIP,
804 mdoc->parse, line, ppos,
805 "%s %s", mdoc_macronames[tok],
807 rew_sub(MDOC_BLOCK, mdoc, tok, line, ppos);
811 if (maxargs && endbody == NULL) {
814 * Stray .Ec without previous .Eo:
815 * Break the output line, ignore any arguments.
817 mdoc_elem_alloc(mdoc, line, ppos, MDOC_br, NULL);
818 rew_elem(mdoc, MDOC_br);
820 mdoc_tail_alloc(mdoc, line, ppos, atok);
827 if (j == maxargs && ! flushed) {
829 rew_sub(MDOC_BLOCK, mdoc, tok, line, ppos);
831 rew_last(mdoc, endbody);
835 ac = mdoc_args(mdoc, line, pos, buf, tok, &p);
836 if (ac == ARGS_PUNCT || ac == ARGS_EOLN)
839 ntok = ac == ARGS_QWORD ? MDOC_MAX :
840 lookup(mdoc, tok, line, lastarg, p);
842 if (ntok == MDOC_MAX) {
843 dword(mdoc, line, lastarg, p, DELIM_MAX,
844 MDOC_JOIN & mdoc_macros[tok].flags);
850 rew_sub(MDOC_BLOCK, mdoc, tok, line, ppos);
852 rew_last(mdoc, endbody);
855 mdoc->flags &= ~MDOC_NEWLINE;
856 mdoc_macro(mdoc, ntok, line, lastarg, pos, buf);
862 rew_sub(MDOC_BLOCK, mdoc, tok, line, ppos);
864 rew_last(mdoc, endbody);
867 append_delims(mdoc, line, pos, buf);
871 in_line(MACRO_PROT_ARGS)
873 int la, scope, cnt, firstarg, mayopen, nc, nl;
877 struct mdoc_arg *arg;
880 nl = MDOC_NEWLINE & mdoc->flags;
883 * Whether we allow ignored elements (those without content,
884 * usually because of reserved words) to squeak by.
906 mdoc_argv(mdoc, line, tok, &arg, pos, buf);
911 for (cnt = scope = 0;; ) {
913 ac = mdoc_args(mdoc, line, pos, buf, tok, &p);
916 * At the end of a macro line,
917 * opening delimiters do not suppress spacing.
920 if (ac == ARGS_EOLN) {
922 mdoc->last->flags &= ~MDOC_DELIMO;
927 * The rest of the macro line is only punctuation,
928 * to be handled by append_delims().
929 * If there were no other arguments,
930 * do not allow the first one to suppress spacing,
931 * even if it turns out to be a closing one.
934 if (ac == ARGS_PUNCT) {
935 if (cnt == 0 && nc == 0)
936 mdoc->flags |= MDOC_NODELIMC;
940 ntok = (ac == ARGS_QWORD || (tok == MDOC_Fn && !cnt)) ?
941 MDOC_MAX : lookup(mdoc, tok, line, la, p);
944 * In this case, we've located a submacro and must
945 * execute it. Close out scope, if open. If no
946 * elements have been generated, either create one (nc)
947 * or raise a warning.
950 if (ntok != MDOC_MAX) {
954 mdoc_elem_alloc(mdoc, line, ppos, tok, arg);
955 rew_last(mdoc, mdoc->last);
956 } else if ( ! nc && ! cnt) {
958 mandoc_msg(MANDOCERR_MACRO_EMPTY,
959 mdoc->parse, line, ppos,
960 mdoc_macronames[tok]);
962 mdoc_macro(mdoc, ntok, line, la, pos, buf);
964 append_delims(mdoc, line, pos, buf);
969 * Non-quote-enclosed punctuation. Set up our scope, if
970 * a word; rewind the scope, if a delimiter; then append
974 d = ac == ARGS_QWORD ? DELIM_NONE : mdoc_isdelim(p);
976 if (DELIM_NONE != d) {
978 * If we encounter closing punctuation, no word
979 * has been emitted, no scope is open, and we're
980 * allowed to have an empty element, then start
983 if ((d == DELIM_CLOSE ||
984 (d == DELIM_MIDDLE && tok == MDOC_Fl)) &&
985 !cnt && !scope && nc && mayopen) {
986 mdoc_elem_alloc(mdoc, line, ppos, tok, arg);
993 * Close out our scope, if one is open, before
1001 } else if (mayopen && !scope) {
1002 mdoc_elem_alloc(mdoc, line, ppos, tok, arg);
1007 dword(mdoc, line, la, p, d,
1008 MDOC_JOIN & mdoc_macros[tok].flags);
1011 * If the first argument is a closing delimiter,
1012 * do not suppress spacing before it.
1015 if (firstarg && d == DELIM_CLOSE && !nc)
1016 mdoc->last->flags &= ~MDOC_DELIMC;
1020 * `Fl' macros have their scope re-opened with each new
1021 * word so that the `-' can be added to each one without
1022 * having to parse out spaces.
1024 if (scope && tok == MDOC_Fl) {
1025 rew_elem(mdoc, tok);
1031 rew_elem(mdoc, tok);
1034 * If no elements have been collected and we're allowed to have
1035 * empties (nc), open a scope and close it out. Otherwise,
1041 mdoc_elem_alloc(mdoc, line, ppos, tok, arg);
1042 rew_last(mdoc, mdoc->last);
1044 mdoc_argv_free(arg);
1045 mandoc_msg(MANDOCERR_MACRO_EMPTY, mdoc->parse,
1046 line, ppos, mdoc_macronames[tok]);
1050 append_delims(mdoc, line, pos, buf);
1054 blk_full(MACRO_PROT_ARGS)
1057 struct mdoc_arg *arg;
1058 struct mdoc_node *head; /* save of head macro */
1059 struct mdoc_node *body; /* save of body macro */
1060 struct mdoc_node *n;
1061 enum margserr ac, lac;
1064 nl = MDOC_NEWLINE & mdoc->flags;
1066 /* Skip items outside lists. */
1068 if (tok == MDOC_It) {
1069 for (n = mdoc->last; n; n = n->parent)
1070 if (n->tok == MDOC_Bl &&
1071 ! (n->flags & MDOC_VALID))
1074 mandoc_vmsg(MANDOCERR_IT_STRAY, mdoc->parse,
1075 line, ppos, "It %s", buf + *pos);
1076 mdoc_elem_alloc(mdoc, line, ppos, MDOC_br, NULL);
1077 rew_elem(mdoc, MDOC_br);
1082 /* Close out prior implicit scope. */
1084 if ( ! (mdoc_macros[tok].flags & MDOC_EXPLICIT)) {
1085 rew_sub(MDOC_BODY, mdoc, tok, line, ppos);
1086 rew_sub(MDOC_BLOCK, mdoc, tok, line, ppos);
1090 * This routine accommodates implicitly- and explicitly-scoped
1091 * macro openings. Implicit ones first close out prior scope
1092 * (seen above). Delay opening the head until necessary to
1093 * allow leading punctuation to print. Special consideration
1094 * for `It -column', which has phrase-part syntax instead of
1095 * regular child nodes.
1098 mdoc_argv(mdoc, line, tok, &arg, pos, buf);
1099 mdoc_block_alloc(mdoc, line, ppos, tok, arg);
1103 * Exception: Heads of `It' macros in `-diag' lists are not
1104 * parsed, even though `It' macros in general are parsed.
1107 parsed = tok != MDOC_It ||
1108 mdoc->last->parent->tok != MDOC_Bl ||
1109 mdoc->last->parent->norm->Bl.type != LIST_diag;
1112 * The `Nd' macro has all arguments in its body: it's a hybrid
1113 * of block partial-explicit and full-implicit. Stupid.
1116 if (tok == MDOC_Nd) {
1117 head = mdoc_head_alloc(mdoc, line, ppos, tok);
1118 rew_sub(MDOC_HEAD, mdoc, tok, line, ppos);
1119 body = mdoc_body_alloc(mdoc, line, ppos, tok);
1123 mdoc->flags |= MDOC_KEEP;
1129 ac = mdoc_args(mdoc, line, pos, buf, tok, &p);
1130 if (ac == ARGS_PUNCT)
1132 if (ac == ARGS_EOLN) {
1133 if (lac != ARGS_PPHRASE && lac != ARGS_PHRASE)
1136 * This is necessary: if the last token on a
1137 * line is a `Ta' or tab, then we'll get
1138 * ARGS_EOLN, so we must be smart enough to
1139 * reopen our scope if the last parse was a
1140 * phrase or partial phrase.
1142 rew_sub(MDOC_BODY, mdoc, tok, line, ppos);
1143 body = mdoc_body_alloc(mdoc, line, ppos, tok);
1148 * Emit leading punctuation (i.e., punctuation before
1149 * the MDOC_HEAD) for non-phrase types.
1154 ac != ARGS_PHRASE &&
1155 ac != ARGS_PPHRASE &&
1157 mdoc_isdelim(p) == DELIM_OPEN) {
1158 dword(mdoc, line, la, p, DELIM_OPEN, 0);
1162 /* Open a head if one hasn't been opened. */
1165 head = mdoc_head_alloc(mdoc, line, ppos, tok);
1167 if (ac == ARGS_PHRASE ||
1169 ac == ARGS_PPHRASE) {
1172 * If we haven't opened a body yet, rewind the
1173 * head; if we have, rewind that instead.
1176 rew_sub(body ? MDOC_BODY : MDOC_HEAD,
1177 mdoc, tok, line, ppos);
1178 body = mdoc_body_alloc(mdoc, line, ppos, tok);
1181 * Process phrases: set whether we're in a
1182 * partial-phrase (this effects line handling)
1183 * then call down into the phrase parser.
1186 if (ac == ARGS_PPHRASE)
1187 mdoc->flags |= MDOC_PPHRASE;
1188 if (ac == ARGS_PEND && lac == ARGS_PPHRASE)
1189 mdoc->flags |= MDOC_PPHRASE;
1190 parse_rest(mdoc, MDOC_MAX, line, &la, buf);
1191 mdoc->flags &= ~MDOC_PPHRASE;
1195 if (macro_or_word(mdoc, tok, line, la, pos, buf, parsed))
1200 head = mdoc_head_alloc(mdoc, line, ppos, tok);
1202 append_delims(mdoc, line, pos, buf);
1207 * If there is an open (i.e., unvalidated) sub-block requiring
1208 * explicit close-out, postpone switching the current block from
1209 * head to body until the rew_sub() call closing out that
1212 for (n = mdoc->last; n && n != head; n = n->parent) {
1213 if (n->type == MDOC_BLOCK &&
1214 mdoc_macros[n->tok].flags & MDOC_EXPLICIT &&
1215 ! (n->flags & MDOC_VALID)) {
1221 /* Close out scopes to remain in a consistent state. */
1223 rew_sub(MDOC_HEAD, mdoc, tok, line, ppos);
1224 mdoc_body_alloc(mdoc, line, ppos, tok);
1226 if (mdoc->flags & MDOC_FREECOL) {
1227 rew_sub(MDOC_BODY, mdoc, tok, line, ppos);
1228 rew_sub(MDOC_BLOCK, mdoc, tok, line, ppos);
1229 mdoc->flags &= ~MDOC_FREECOL;
1234 blk_part_imp(MACRO_PROT_ARGS)
1239 struct mdoc_node *blk; /* saved block context */
1240 struct mdoc_node *body; /* saved body context */
1241 struct mdoc_node *n;
1243 nl = MDOC_NEWLINE & mdoc->flags;
1246 * A macro that spans to the end of the line. This is generally
1247 * (but not necessarily) called as the first macro. The block
1248 * has a head as the immediate child, which is always empty,
1249 * followed by zero or more opening punctuation nodes, then the
1250 * body (which may be empty, depending on the macro), then zero
1251 * or more closing punctuation nodes.
1254 blk = mdoc_block_alloc(mdoc, line, ppos, tok, NULL);
1255 mdoc_head_alloc(mdoc, line, ppos, tok);
1256 rew_sub(MDOC_HEAD, mdoc, tok, line, ppos);
1259 * Open the body scope "on-demand", that is, after we've
1260 * processed all our the leading delimiters (open parenthesis,
1264 for (body = NULL; ; ) {
1266 ac = mdoc_args(mdoc, line, pos, buf, tok, &p);
1267 if (ac == ARGS_EOLN || ac == ARGS_PUNCT)
1270 if (body == NULL && ac != ARGS_QWORD &&
1271 mdoc_isdelim(p) == DELIM_OPEN) {
1272 dword(mdoc, line, la, p, DELIM_OPEN, 0);
1277 body = mdoc_body_alloc(mdoc, line, ppos, tok);
1279 if (macro_or_word(mdoc, tok, line, la, pos, buf, 1))
1283 body = mdoc_body_alloc(mdoc, line, ppos, tok);
1286 * If there is an open sub-block requiring explicit close-out,
1287 * postpone closing out the current block
1288 * until the rew_sub() call closing out the sub-block.
1291 for (n = mdoc->last; n && n != body && n != blk->parent;
1293 if (n->type == MDOC_BLOCK &&
1294 mdoc_macros[n->tok].flags & MDOC_EXPLICIT &&
1295 ! (n->flags & MDOC_VALID)) {
1296 make_pending(n, tok, mdoc, line, ppos);
1297 mdoc_endbody_alloc(mdoc, line, ppos,
1298 tok, body, ENDBODY_NOSPACE);
1303 rew_sub(MDOC_BODY, mdoc, tok, line, ppos);
1305 append_delims(mdoc, line, pos, buf);
1306 rew_sub(MDOC_BLOCK, mdoc, tok, line, ppos);
1308 /* Move trailing .Ns out of scope. */
1310 for (n = body->child; n && n->next; n = n->next)
1312 if (n && n->tok == MDOC_Ns)
1313 mdoc_node_relink(mdoc, n);
1317 blk_part_exp(MACRO_PROT_ARGS)
1321 struct mdoc_node *head; /* keep track of head */
1322 struct mdoc_node *body; /* keep track of body */
1325 nl = MDOC_NEWLINE & mdoc->flags;
1328 * The opening of an explicit macro having zero or more leading
1329 * punctuation nodes; a head with optional single element (the
1330 * case of `Eo'); and a body that may be empty.
1333 mdoc_block_alloc(mdoc, line, ppos, tok, NULL);
1334 for (head = body = NULL; ; ) {
1336 ac = mdoc_args(mdoc, line, pos, buf, tok, &p);
1337 if (ac == ARGS_PUNCT || ac == ARGS_EOLN)
1340 /* Flush out leading punctuation. */
1342 if (head == NULL && ac != ARGS_QWORD &&
1343 mdoc_isdelim(p) == DELIM_OPEN) {
1344 assert(NULL == body);
1345 dword(mdoc, line, la, p, DELIM_OPEN, 0);
1350 assert(body == NULL);
1351 head = mdoc_head_alloc(mdoc, line, ppos, tok);
1355 * `Eo' gobbles any data into the head, but most other
1356 * macros just immediately close out and begin the body.
1361 /* No check whether it's a macro! */
1363 dword(mdoc, line, la, p, DELIM_MAX, 0);
1364 rew_sub(MDOC_HEAD, mdoc, tok, line, ppos);
1365 body = mdoc_body_alloc(mdoc, line, ppos, tok);
1369 assert(head != NULL && body != NULL);
1371 if (macro_or_word(mdoc, tok, line, la, pos, buf, 1))
1375 /* Clean-up to leave in a consistent state. */
1378 mdoc_head_alloc(mdoc, line, ppos, tok);
1381 rew_sub(MDOC_HEAD, mdoc, tok, line, ppos);
1382 mdoc_body_alloc(mdoc, line, ppos, tok);
1385 append_delims(mdoc, line, pos, buf);
1389 in_line_argn(MACRO_PROT_ARGS)
1391 int la, flushed, j, maxargs, nl;
1393 struct mdoc_arg *arg;
1397 nl = mdoc->flags & MDOC_NEWLINE;
1400 * A line macro that has a fixed number of arguments (maxargs).
1401 * Only open the scope once the first non-leading-punctuation is
1402 * found (unless MDOC_IGNDELIM is noted, like in `Pf'), then
1403 * keep it open until the maximum number of arguments are
1427 mdoc_argv(mdoc, line, tok, &arg, pos, buf);
1433 ac = mdoc_args(mdoc, line, pos, buf, tok, &p);
1434 if (ac == ARGS_PUNCT || ac == ARGS_EOLN) {
1435 if (j < 2 && tok == MDOC_Pf)
1436 mandoc_vmsg(MANDOCERR_PF_SKIP,
1437 mdoc->parse, line, ppos, "Pf %s",
1438 p == NULL ? "at eol" : p);
1442 if ( ! (mdoc_macros[tok].flags & MDOC_IGNDELIM) &&
1443 ac != ARGS_QWORD && j == 0 &&
1444 mdoc_isdelim(p) == DELIM_OPEN) {
1445 dword(mdoc, line, la, p, DELIM_OPEN, 0);
1448 mdoc_elem_alloc(mdoc, line, ppos, tok, arg);
1450 if (j == maxargs && ! flushed) {
1451 rew_elem(mdoc, tok);
1455 ntok = (ac == ARGS_QWORD || (tok == MDOC_Pf && j == 0)) ?
1456 MDOC_MAX : lookup(mdoc, tok, line, la, p);
1458 if (ntok != MDOC_MAX) {
1460 rew_elem(mdoc, tok);
1462 mdoc_macro(mdoc, ntok, line, la, pos, buf);
1467 if ( ! (mdoc_macros[tok].flags & MDOC_IGNDELIM) &&
1468 ac != ARGS_QWORD && ! flushed &&
1469 mdoc_isdelim(p) != DELIM_NONE) {
1470 rew_elem(mdoc, tok);
1474 dword(mdoc, line, la, p, DELIM_MAX,
1475 MDOC_JOIN & mdoc_macros[tok].flags);
1480 mdoc_elem_alloc(mdoc, line, ppos, tok, arg);
1481 if (ac == ARGS_PUNCT && tok == MDOC_Pf)
1482 append_delims(mdoc, line, pos, buf);
1485 rew_elem(mdoc, tok);
1487 append_delims(mdoc, line, pos, buf);
1491 in_line_eoln(MACRO_PROT_ARGS)
1493 struct mdoc_arg *arg;
1496 rew_sub(MDOC_BLOCK, mdoc, MDOC_Nm, line, ppos);
1498 mdoc_argv(mdoc, line, tok, &arg, pos, buf);
1499 mdoc_elem_alloc(mdoc, line, ppos, tok, arg);
1500 if (parse_rest(mdoc, tok, line, pos, buf))
1502 rew_elem(mdoc, tok);
1506 * The simplest argument parser available: Parse the remaining
1507 * words until the end of the phrase or line and return 0
1508 * or until the next macro, call that macro, and return 1.
1511 parse_rest(struct mdoc *mdoc, enum mdoct tok, int line, int *pos, char *buf)
1517 if (mdoc_args(mdoc, line, pos, buf, tok, NULL) == ARGS_EOLN)
1519 if (macro_or_word(mdoc, tok, line, la, pos, buf, 1))
1525 ctx_synopsis(MACRO_PROT_ARGS)
1528 if (~mdoc->flags & (MDOC_SYNOPSIS | MDOC_NEWLINE))
1529 in_line(mdoc, tok, line, ppos, pos, buf);
1530 else if (tok == MDOC_Nm)
1531 blk_full(mdoc, tok, line, ppos, pos, buf);
1533 assert(tok == MDOC_Vt);
1534 blk_part_imp(mdoc, tok, line, ppos, pos, buf);
1539 * Phrases occur within `Bl -column' entries, separated by `Ta' or tabs.
1540 * They're unusual because they're basically free-form text until a
1541 * macro is encountered.
1544 phrase_ta(MACRO_PROT_ARGS)
1546 struct mdoc_node *n;
1548 /* Make sure we are in a column list or ignore this macro. */
1551 while (n != NULL && n->tok != MDOC_Bl)
1553 if (n == NULL || n->norm->Bl.type != LIST_column) {
1554 mandoc_msg(MANDOCERR_TA_STRAY, mdoc->parse,
1559 /* Advance to the next column. */
1561 rew_sub(MDOC_BODY, mdoc, MDOC_It, line, ppos);
1562 mdoc_body_alloc(mdoc, line, ppos, MDOC_It);
1563 parse_rest(mdoc, MDOC_MAX, line, pos, buf);