1 /* $Id: mdoc_macro.c,v 1.154 2014/11/29 04:31:35 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(enum mdoct, const char *);
57 static int macro_or_word(MACRO_PROT_ARGS, int);
58 static int make_pending(struct mdoc_node *, enum mdoct,
59 struct mdoc *, int, int);
60 static int parse_rest(struct mdoc *, enum mdoct,
62 static enum mdoct rew_alt(enum mdoct);
63 static enum rew rew_dohalt(enum mdoct, enum mdoc_type,
64 const struct mdoc_node *);
65 static void rew_elem(struct mdoc *, enum mdoct);
66 static void rew_last(struct mdoc *, const struct mdoc_node *);
67 static void rew_sub(enum mdoc_type, struct mdoc *,
68 enum mdoct, int, int);
70 const struct mdoc_macro __mdoc_macros[MDOC_MAX] = {
71 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* Ap */
72 { in_line_eoln, MDOC_PROLOGUE }, /* Dd */
73 { in_line_eoln, MDOC_PROLOGUE }, /* Dt */
74 { in_line_eoln, MDOC_PROLOGUE }, /* Os */
75 { blk_full, MDOC_PARSED | MDOC_JOIN }, /* Sh */
76 { blk_full, MDOC_PARSED | MDOC_JOIN }, /* Ss */
77 { in_line_eoln, 0 }, /* Pp */
78 { blk_part_imp, MDOC_PARSED | MDOC_JOIN }, /* D1 */
79 { blk_part_imp, MDOC_PARSED | MDOC_JOIN }, /* Dl */
80 { blk_full, MDOC_EXPLICIT }, /* Bd */
81 { blk_exp_close, MDOC_EXPLICIT | MDOC_JOIN }, /* Ed */
82 { blk_full, MDOC_EXPLICIT }, /* Bl */
83 { blk_exp_close, MDOC_EXPLICIT | MDOC_JOIN }, /* El */
84 { blk_full, MDOC_PARSED | MDOC_JOIN }, /* It */
85 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Ad */
86 { in_line, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* An */
87 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Ar */
88 { in_line, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* Cd */
89 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Cm */
90 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Dv */
91 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Er */
92 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Ev */
93 { in_line_eoln, 0 }, /* Ex */
94 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Fa */
95 { in_line_eoln, 0 }, /* Fd */
96 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Fl */
97 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Fn */
98 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Ft */
99 { in_line, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* Ic */
100 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* In */
101 { in_line, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* Li */
102 { blk_full, MDOC_JOIN }, /* Nd */
103 { ctx_synopsis, MDOC_CALLABLE | MDOC_PARSED }, /* Nm */
104 { blk_part_imp, MDOC_CALLABLE | MDOC_PARSED }, /* Op */
105 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Ot */
106 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Pa */
107 { in_line_eoln, 0 }, /* Rv */
108 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* St */
109 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Va */
110 { ctx_synopsis, MDOC_CALLABLE | MDOC_PARSED }, /* Vt */
111 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* Xr */
112 { in_line_eoln, MDOC_JOIN }, /* %A */
113 { in_line_eoln, MDOC_JOIN }, /* %B */
114 { in_line_eoln, MDOC_JOIN }, /* %D */
115 { in_line_eoln, MDOC_JOIN }, /* %I */
116 { in_line_eoln, MDOC_JOIN }, /* %J */
117 { in_line_eoln, 0 }, /* %N */
118 { in_line_eoln, MDOC_JOIN }, /* %O */
119 { in_line_eoln, 0 }, /* %P */
120 { in_line_eoln, MDOC_JOIN }, /* %R */
121 { in_line_eoln, MDOC_JOIN }, /* %T */
122 { in_line_eoln, 0 }, /* %V */
123 { blk_exp_close, MDOC_CALLABLE | MDOC_PARSED |
124 MDOC_EXPLICIT | MDOC_JOIN }, /* Ac */
125 { blk_part_exp, MDOC_CALLABLE | MDOC_PARSED |
126 MDOC_EXPLICIT | MDOC_JOIN }, /* Ao */
127 { blk_part_imp, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* Aq */
128 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* At */
129 { blk_exp_close, MDOC_CALLABLE | MDOC_PARSED |
130 MDOC_EXPLICIT | MDOC_JOIN }, /* Bc */
131 { blk_full, MDOC_EXPLICIT }, /* Bf */
132 { blk_part_exp, MDOC_CALLABLE | MDOC_PARSED |
133 MDOC_EXPLICIT | MDOC_JOIN }, /* Bo */
134 { blk_part_imp, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* Bq */
135 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* Bsx */
136 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* Bx */
137 { in_line_eoln, 0 }, /* Db */
138 { blk_exp_close, MDOC_CALLABLE | MDOC_PARSED |
139 MDOC_EXPLICIT | MDOC_JOIN }, /* Dc */
140 { blk_part_exp, MDOC_CALLABLE | MDOC_PARSED |
141 MDOC_EXPLICIT | MDOC_JOIN }, /* Do */
142 { blk_part_imp, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* Dq */
143 { blk_exp_close, MDOC_CALLABLE | MDOC_PARSED | MDOC_EXPLICIT }, /* Ec */
144 { blk_exp_close, MDOC_EXPLICIT | MDOC_JOIN }, /* Ef */
145 { in_line, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* Em */
146 { blk_part_exp, MDOC_CALLABLE | MDOC_PARSED | MDOC_EXPLICIT }, /* Eo */
147 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* Fx */
148 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Ms */
149 { in_line, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* No */
150 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED |
151 MDOC_IGNDELIM | MDOC_JOIN }, /* Ns */
152 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* Nx */
153 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* Ox */
154 { blk_exp_close, MDOC_CALLABLE | MDOC_PARSED |
155 MDOC_EXPLICIT | MDOC_JOIN }, /* Pc */
156 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED | MDOC_IGNDELIM }, /* Pf */
157 { blk_part_exp, MDOC_CALLABLE | MDOC_PARSED |
158 MDOC_EXPLICIT | MDOC_JOIN }, /* Po */
159 { blk_part_imp, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* Pq */
160 { blk_exp_close, MDOC_CALLABLE | MDOC_PARSED |
161 MDOC_EXPLICIT | MDOC_JOIN }, /* Qc */
162 { blk_part_imp, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* Ql */
163 { blk_part_exp, MDOC_CALLABLE | MDOC_PARSED |
164 MDOC_EXPLICIT | MDOC_JOIN }, /* Qo */
165 { blk_part_imp, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* Qq */
166 { blk_exp_close, MDOC_EXPLICIT | MDOC_JOIN }, /* Re */
167 { blk_full, MDOC_EXPLICIT }, /* Rs */
168 { blk_exp_close, MDOC_CALLABLE | MDOC_PARSED |
169 MDOC_EXPLICIT | MDOC_JOIN }, /* Sc */
170 { blk_part_exp, MDOC_CALLABLE | MDOC_PARSED |
171 MDOC_EXPLICIT | MDOC_JOIN }, /* So */
172 { blk_part_imp, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* Sq */
173 { in_line_argn, 0 }, /* Sm */
174 { in_line, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* Sx */
175 { in_line, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* Sy */
176 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Tn */
177 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* Ux */
178 { blk_exp_close, MDOC_EXPLICIT | MDOC_CALLABLE | MDOC_PARSED }, /* Xc */
179 { blk_part_exp, MDOC_CALLABLE | MDOC_PARSED | MDOC_EXPLICIT }, /* Xo */
180 { blk_full, MDOC_EXPLICIT | MDOC_CALLABLE }, /* Fo */
181 { blk_exp_close, MDOC_CALLABLE | MDOC_PARSED |
182 MDOC_EXPLICIT | MDOC_JOIN }, /* Fc */
183 { blk_part_exp, MDOC_CALLABLE | MDOC_PARSED |
184 MDOC_EXPLICIT | MDOC_JOIN }, /* Oo */
185 { blk_exp_close, MDOC_CALLABLE | MDOC_PARSED |
186 MDOC_EXPLICIT | MDOC_JOIN }, /* Oc */
187 { blk_full, MDOC_EXPLICIT }, /* Bk */
188 { blk_exp_close, MDOC_EXPLICIT | MDOC_JOIN }, /* Ek */
189 { in_line_eoln, 0 }, /* Bt */
190 { in_line_eoln, 0 }, /* Hf */
191 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Fr */
192 { in_line_eoln, 0 }, /* Ud */
193 { in_line, 0 }, /* Lb */
194 { in_line_eoln, 0 }, /* Lp */
195 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Lk */
196 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Mt */
197 { blk_part_imp, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* Brq */
198 { blk_part_exp, MDOC_CALLABLE | MDOC_PARSED |
199 MDOC_EXPLICIT | MDOC_JOIN }, /* Bro */
200 { blk_exp_close, MDOC_CALLABLE | MDOC_PARSED |
201 MDOC_EXPLICIT | MDOC_JOIN }, /* Brc */
202 { in_line_eoln, MDOC_JOIN }, /* %C */
203 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* Es */
204 { blk_part_imp, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* En */
205 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* Dx */
206 { in_line_eoln, MDOC_JOIN }, /* %Q */
207 { in_line_eoln, 0 }, /* br */
208 { in_line_eoln, 0 }, /* sp */
209 { in_line_eoln, 0 }, /* %U */
210 { phrase_ta, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* Ta */
211 { in_line_eoln, MDOC_PROLOGUE }, /* ll */
214 const struct mdoc_macro * const mdoc_macros = __mdoc_macros;
218 * This is called at the end of parsing. It must traverse up the tree,
219 * closing out open [implicit] scopes. Obviously, open explicit scopes
223 mdoc_macroend(struct mdoc *mdoc)
227 /* Scan for open explicit scopes. */
229 n = mdoc->last->flags & MDOC_VALID ?
230 mdoc->last->parent : mdoc->last;
232 for ( ; n; n = n->parent)
233 if (n->type == MDOC_BLOCK &&
234 mdoc_macros[n->tok].flags & MDOC_EXPLICIT)
235 mandoc_msg(MANDOCERR_BLK_NOEND, mdoc->parse,
236 n->line, n->pos, mdoc_macronames[n->tok]);
238 /* Rewind to the first. */
240 rew_last(mdoc, mdoc->first);
244 * Look up the macro at *p called by "from",
245 * or as a line macro if from == MDOC_MAX.
248 lookup(enum mdoct from, const char *p)
252 if (from == MDOC_MAX || mdoc_macros[from].flags & MDOC_PARSED) {
253 res = mdoc_hash_find(p);
254 if (res != MDOC_MAX && mdoc_macros[res].flags & MDOC_CALLABLE)
261 rew_last(struct mdoc *mdoc, const struct mdoc_node *to)
263 struct mdoc_node *n, *np;
266 mdoc->next = MDOC_NEXT_SIBLING;
267 while (mdoc->last != to) {
269 * Save the parent here, because we may delete the
270 * mdoc->last node in the post-validation phase and reset
271 * it to mdoc->last->parent, causing a step in the closing
274 np = mdoc->last->parent;
275 mdoc_valid_post(mdoc);
279 mdoc->last->last = n;
281 mdoc_valid_post(mdoc);
285 * For a block closing macro, return the corresponding opening one.
286 * Otherwise, return the macro itself.
289 rew_alt(enum mdoct tok)
331 * Rewinding to tok, how do we have to handle *p?
332 * REWIND_NONE: *p would delimit tok, but no tok scope is open
333 * inside *p, so there is no need to rewind anything at all.
334 * REWIND_THIS: *p matches tok, so rewind *p and nothing else.
335 * REWIND_MORE: *p is implicit, rewind it and keep searching for tok.
336 * REWIND_FORCE: *p is explicit, but tok is full, force rewinding *p.
337 * REWIND_LATER: *p is explicit and still open, postpone rewinding.
338 * REWIND_ERROR: No tok block is open at all.
341 rew_dohalt(enum mdoct tok, enum mdoc_type type,
342 const struct mdoc_node *p)
346 * No matching token, no delimiting block, no broken block.
347 * This can happen when full implicit macros are called for
348 * the first time but try to rewind their previous
351 if (MDOC_ROOT == p->type)
352 return(MDOC_BLOCK == type &&
353 MDOC_EXPLICIT & mdoc_macros[tok].flags ?
354 REWIND_ERROR : REWIND_NONE);
357 * When starting to rewind, skip plain text
358 * and nodes that have already been rewound.
360 if (MDOC_TEXT == p->type || MDOC_VALID & p->flags)
364 * The easiest case: Found a matching token.
365 * This applies to both blocks and elements.
369 return(p->end ? REWIND_NONE :
370 type == p->type ? REWIND_THIS : REWIND_MORE);
373 * While elements do require rewinding for themselves,
374 * they never affect rewinding of other nodes.
376 if (MDOC_ELEM == p->type)
380 * Blocks delimited by our target token get REWIND_MORE.
381 * Blocks delimiting our target token get REWIND_NONE.
385 if (MDOC_It == p->tok)
389 if (MDOC_BODY == p->type && MDOC_Bl == p->tok)
393 * XXX Badly nested block handling still fails badly
394 * when one block is breaking two blocks of the same type.
395 * This is an incomplete and extremely ugly workaround,
396 * required to let the OpenBSD tree build.
399 if (MDOC_Op == p->tok)
407 if (MDOC_BODY == p->type && MDOC_Sh == p->tok)
411 if (MDOC_ROOT == p->parent->type)
413 if (MDOC_Nd == p->tok || MDOC_Ss == p->tok ||
422 * Default block rewinding rules.
423 * In particular, always skip block end markers,
424 * and let all blocks rewind Nm children.
425 * Do not warn again when closing a block,
426 * since closing the body already warned.
428 if (ENDBODY_NOT != p->end || MDOC_Nm == p->tok ||
429 MDOC_BLOCK == type || (MDOC_BLOCK == p->type &&
430 ! (MDOC_EXPLICIT & mdoc_macros[tok].flags)))
434 * By default, closing out full blocks
435 * forces closing of broken explicit blocks,
436 * while closing out partial blocks
437 * allows delayed rewinding by default.
439 return (&blk_full == mdoc_macros[tok].fp ?
440 REWIND_FORCE : REWIND_LATER);
444 rew_elem(struct mdoc *mdoc, enum mdoct tok)
449 if (MDOC_ELEM != n->type)
451 assert(MDOC_ELEM == n->type);
452 assert(tok == n->tok);
457 * We are trying to close a block identified by tok,
458 * but the child block *broken is still open.
459 * Thus, postpone closing the tok block
460 * until the rew_sub call closing *broken.
463 make_pending(struct mdoc_node *broken, enum mdoct tok,
464 struct mdoc *mdoc, int line, int ppos)
466 struct mdoc_node *breaker;
469 * Iterate backwards, searching for the block matching tok,
470 * that is, the block breaking the *broken block.
472 for (breaker = broken->parent; breaker; breaker = breaker->parent) {
475 * If the *broken block had already been broken before
476 * and we encounter its breaker, make the tok block
477 * pending on the inner breaker.
478 * Graphically, "[A breaker=[B broken=[C->B B] tok=A] C]"
479 * becomes "[A broken=[B [C->B B] tok=A] C]"
480 * and finally "[A [B->A [C->B B] A] C]".
482 if (breaker == broken->pending) {
487 if (REWIND_THIS != rew_dohalt(tok, MDOC_BLOCK, breaker))
489 if (MDOC_BODY == broken->type)
490 broken = broken->parent;
494 * If another, outer breaker is already pending on
495 * the *broken block, we must not clobber the link
496 * to the outer breaker, but make it pending on the
497 * new, now inner breaker.
498 * Graphically, "[A breaker=[B broken=[C->A A] tok=B] C]"
499 * becomes "[A breaker=[B->A broken=[C A] tok=B] C]"
500 * and finally "[A [B->A [C->B A] B] C]".
502 if (broken->pending) {
503 struct mdoc_node *taker;
506 * If the breaker had also been broken before,
507 * it cannot take on the outer breaker itself,
508 * but must hand it on to its own breakers.
509 * Graphically, this is the following situation:
510 * "[A [B breaker=[C->B B] broken=[D->A A] tok=C] D]"
511 * "[A taker=[B->A breaker=[C->B B] [D->C A] C] D]"
514 while (taker->pending)
515 taker = taker->pending;
516 taker->pending = broken->pending;
518 broken->pending = breaker;
519 mandoc_vmsg(MANDOCERR_BLK_NEST, mdoc->parse, line, ppos,
520 "%s breaks %s", mdoc_macronames[tok],
521 mdoc_macronames[broken->tok]);
526 * Found no matching block for tok.
527 * Are you trying to close a block that is not open?
533 rew_sub(enum mdoc_type t, struct mdoc *mdoc,
534 enum mdoct tok, int line, int ppos)
540 switch (rew_dohalt(tok, t, n)) {
545 (mdoc->flags & MDOC_NEWLINE &&
546 ! (mdoc_macros[tok].flags & MDOC_EXPLICIT));
549 mandoc_vmsg(MANDOCERR_BLK_BROKEN, mdoc->parse,
550 line, ppos, "%s breaks %s",
551 mdoc_macronames[tok],
552 mdoc_macronames[n->tok]);
556 (mdoc->flags & MDOC_NEWLINE ? 1 : 0);
560 if (make_pending(n, tok, mdoc, line, ppos) ||
565 mandoc_msg(MANDOCERR_BLK_NOTOPEN,
566 mdoc->parse, line, ppos,
567 mdoc_macronames[tok]);
576 * The current block extends an enclosing block.
577 * Now that the current block ends, close the enclosing block, too.
579 while ((n = n->pending) != NULL) {
581 if (n->type == MDOC_HEAD)
582 mdoc_body_alloc(mdoc, n->line, n->pos, n->tok);
587 * Allocate a word and check whether it's punctuation or not.
588 * Punctuation consists of those tokens found in mdoc_isdelim().
591 dword(struct mdoc *mdoc, int line, int col, const char *p,
592 enum mdelim d, int may_append)
599 ! (mdoc->flags & (MDOC_SYNOPSIS | MDOC_KEEP | MDOC_SMOFF)) &&
600 d == DELIM_NONE && mdoc->last->type == MDOC_TEXT &&
601 mdoc_isdelim(mdoc->last->string) == DELIM_NONE) {
602 mdoc_word_append(mdoc, p);
606 mdoc_word_alloc(mdoc, line, col, p);
609 * If the word consists of a bare delimiter,
610 * flag the new node accordingly,
611 * unless doing so was vetoed by the invoking macro.
612 * Always clear the veto, it is only valid for one word.
616 mdoc->last->flags |= MDOC_DELIMO;
617 else if (d == DELIM_CLOSE &&
618 ! (mdoc->flags & MDOC_NODELIMC) &&
619 mdoc->last->parent->tok != MDOC_Fd)
620 mdoc->last->flags |= MDOC_DELIMC;
621 mdoc->flags &= ~MDOC_NODELIMC;
625 append_delims(struct mdoc *mdoc, int line, int *pos, char *buf)
630 if (buf[*pos] == '\0')
635 if (mdoc_args(mdoc, line, pos, buf, MDOC_MAX, &p) == ARGS_EOLN)
637 dword(mdoc, line, la, p, DELIM_MAX, 1);
640 * If we encounter end-of-sentence symbols, then trigger
643 * XXX: it's easy to allow this to propagate outward to
644 * the last symbol, such that `. )' will cause the
645 * correct double-spacing. However, (1) groff isn't
646 * smart enough to do this and (2) it would require
647 * knowing which symbols break this behaviour, for
648 * example, `. ;' shouldn't propagate the double-space.
651 if (mandoc_eos(p, strlen(p)))
652 mdoc->last->flags |= MDOC_EOS;
658 * If it is a macro, call it and return 1.
659 * Otherwise, allocate it and return 0.
662 macro_or_word(MACRO_PROT_ARGS, int parsed)
669 if (mdoc->flags & MDOC_PHRASELIT)
674 ntok = lookup(tok, p);
676 if (ntok == MDOC_MAX) {
677 dword(mdoc, line, ppos, p, DELIM_MAX, tok == MDOC_MAX ||
678 mdoc_macros[tok].flags & MDOC_JOIN);
681 if (mdoc_macros[tok].fp == in_line_eoln)
683 mdoc_macro(mdoc, ntok, line, ppos, pos, buf);
685 append_delims(mdoc, line, pos, buf);
691 * Close out block partial/full explicit.
694 blk_exp_close(MACRO_PROT_ARGS)
696 struct mdoc_node *body; /* Our own body. */
697 struct mdoc_node *endbody; /* Our own end marker. */
698 struct mdoc_node *later; /* A sub-block starting later. */
699 struct mdoc_node *n; /* For searching backwards. */
701 int j, lastarg, maxargs, flushed, nl;
703 enum mdoct atok, ntok;
706 nl = MDOC_NEWLINE & mdoc->flags;
713 mdoc->flags &= ~MDOC_KEEP;
721 * Search backwards for beginnings of blocks,
722 * both of our own and of pending sub-blocks.
726 body = endbody = later = NULL;
727 for (n = mdoc->last; n; n = n->parent) {
728 if (n->flags & MDOC_VALID)
731 /* Remember the start of our own body. */
733 if (n->type == MDOC_BODY && atok == n->tok) {
734 if (n->end == ENDBODY_NOT)
739 if (n->type != MDOC_BLOCK || n->tok == MDOC_Nm)
741 if (atok == n->tok) {
745 * Found the start of our own block.
746 * When there is no pending sub block,
747 * just proceed to closing out.
754 * When there is a pending sub block,
755 * postpone closing out the current block
756 * until the rew_sub() closing out the sub-block.
759 make_pending(later, tok, mdoc, line, ppos);
762 * Mark the place where the formatting - but not
763 * the scope - of the current block ends.
766 mdoc_endbody_alloc(mdoc, line, ppos,
767 atok, body, ENDBODY_SPACE);
770 * If a block closing macro taking arguments
771 * breaks another block, put the arguments
772 * into the end marker and remeber the
773 * end marker in order to close it out.
777 endbody = mdoc->last;
778 mdoc->next = MDOC_NEXT_CHILD;
784 * When finding an open sub block, remember the last
785 * open explicit block, or, in case there are only
786 * implicit ones, the first open implicit block.
790 mdoc_macros[later->tok].flags & MDOC_EXPLICIT)
792 if (n->tok != MDOC_It)
795 rew_sub(MDOC_BODY, mdoc, tok, line, ppos);
797 if ( ! (mdoc_macros[tok].flags & MDOC_PARSED)) {
798 if (buf[*pos] != '\0')
799 mandoc_vmsg(MANDOCERR_ARG_SKIP,
800 mdoc->parse, line, ppos,
801 "%s %s", mdoc_macronames[tok],
803 rew_sub(MDOC_BLOCK, mdoc, tok, line, ppos);
807 if (maxargs && endbody == NULL) {
810 * Stray .Ec without previous .Eo:
811 * Break the output line, ignore any arguments.
813 mdoc_elem_alloc(mdoc, line, ppos, MDOC_br, NULL);
814 rew_elem(mdoc, MDOC_br);
816 mdoc_tail_alloc(mdoc, line, ppos, atok);
823 if (j == maxargs && ! flushed) {
825 rew_sub(MDOC_BLOCK, mdoc, tok, line, ppos);
827 rew_last(mdoc, endbody);
831 ac = mdoc_args(mdoc, line, pos, buf, tok, &p);
832 if (ac == ARGS_PUNCT || ac == ARGS_EOLN)
835 ntok = ac == ARGS_QWORD ? MDOC_MAX : lookup(tok, p);
837 if (ntok == MDOC_MAX) {
838 dword(mdoc, line, lastarg, p, DELIM_MAX,
839 MDOC_JOIN & mdoc_macros[tok].flags);
845 rew_sub(MDOC_BLOCK, mdoc, tok, line, ppos);
847 rew_last(mdoc, endbody);
850 mdoc->flags &= ~MDOC_NEWLINE;
851 mdoc_macro(mdoc, ntok, line, lastarg, pos, buf);
857 rew_sub(MDOC_BLOCK, mdoc, tok, line, ppos);
859 rew_last(mdoc, endbody);
862 append_delims(mdoc, line, pos, buf);
866 in_line(MACRO_PROT_ARGS)
868 int la, scope, cnt, firstarg, mayopen, nc, nl;
872 struct mdoc_arg *arg;
875 nl = MDOC_NEWLINE & mdoc->flags;
878 * Whether we allow ignored elements (those without content,
879 * usually because of reserved words) to squeak by.
901 mdoc_argv(mdoc, line, tok, &arg, pos, buf);
906 for (cnt = scope = 0;; ) {
908 ac = mdoc_args(mdoc, line, pos, buf, tok, &p);
911 * At the end of a macro line,
912 * opening delimiters do not suppress spacing.
915 if (ac == ARGS_EOLN) {
917 mdoc->last->flags &= ~MDOC_DELIMO;
922 * The rest of the macro line is only punctuation,
923 * to be handled by append_delims().
924 * If there were no other arguments,
925 * do not allow the first one to suppress spacing,
926 * even if it turns out to be a closing one.
929 if (ac == ARGS_PUNCT) {
930 if (cnt == 0 && nc == 0)
931 mdoc->flags |= MDOC_NODELIMC;
935 ntok = (ac == ARGS_QWORD || (tok == MDOC_Fn && !cnt)) ?
936 MDOC_MAX : lookup(tok, p);
939 * In this case, we've located a submacro and must
940 * execute it. Close out scope, if open. If no
941 * elements have been generated, either create one (nc)
942 * or raise a warning.
945 if (ntok != MDOC_MAX) {
949 mdoc_elem_alloc(mdoc, line, ppos, tok, arg);
950 rew_last(mdoc, mdoc->last);
951 } else if ( ! nc && ! cnt) {
953 mandoc_msg(MANDOCERR_MACRO_EMPTY,
954 mdoc->parse, line, ppos,
955 mdoc_macronames[tok]);
957 mdoc_macro(mdoc, ntok, line, la, pos, buf);
959 append_delims(mdoc, line, pos, buf);
964 * Non-quote-enclosed punctuation. Set up our scope, if
965 * a word; rewind the scope, if a delimiter; then append
969 d = ac == ARGS_QWORD ? DELIM_NONE : mdoc_isdelim(p);
971 if (DELIM_NONE != d) {
973 * If we encounter closing punctuation, no word
974 * has been emitted, no scope is open, and we're
975 * allowed to have an empty element, then start
978 if ((d == DELIM_CLOSE ||
979 (d == DELIM_MIDDLE && tok == MDOC_Fl)) &&
980 !cnt && !scope && nc && mayopen) {
981 mdoc_elem_alloc(mdoc, line, ppos, tok, arg);
988 * Close out our scope, if one is open, before
996 } else if (mayopen && !scope) {
997 mdoc_elem_alloc(mdoc, line, ppos, tok, arg);
1002 dword(mdoc, line, la, p, d,
1003 MDOC_JOIN & mdoc_macros[tok].flags);
1006 * If the first argument is a closing delimiter,
1007 * do not suppress spacing before it.
1010 if (firstarg && d == DELIM_CLOSE && !nc)
1011 mdoc->last->flags &= ~MDOC_DELIMC;
1015 * `Fl' macros have their scope re-opened with each new
1016 * word so that the `-' can be added to each one without
1017 * having to parse out spaces.
1019 if (scope && tok == MDOC_Fl) {
1020 rew_elem(mdoc, tok);
1026 rew_elem(mdoc, tok);
1029 * If no elements have been collected and we're allowed to have
1030 * empties (nc), open a scope and close it out. Otherwise,
1036 mdoc_elem_alloc(mdoc, line, ppos, tok, arg);
1037 rew_last(mdoc, mdoc->last);
1039 mdoc_argv_free(arg);
1040 mandoc_msg(MANDOCERR_MACRO_EMPTY, mdoc->parse,
1041 line, ppos, mdoc_macronames[tok]);
1045 append_delims(mdoc, line, pos, buf);
1049 blk_full(MACRO_PROT_ARGS)
1052 struct mdoc_arg *arg;
1053 struct mdoc_node *head; /* save of head macro */
1054 struct mdoc_node *body; /* save of body macro */
1055 struct mdoc_node *n;
1056 enum margserr ac, lac;
1059 nl = MDOC_NEWLINE & mdoc->flags;
1061 /* Skip items outside lists. */
1063 if (tok == MDOC_It) {
1064 for (n = mdoc->last; n; n = n->parent)
1065 if (n->tok == MDOC_Bl &&
1066 ! (n->flags & MDOC_VALID))
1069 mandoc_vmsg(MANDOCERR_IT_STRAY, mdoc->parse,
1070 line, ppos, "It %s", buf + *pos);
1071 mdoc_elem_alloc(mdoc, line, ppos, MDOC_br, NULL);
1072 rew_elem(mdoc, MDOC_br);
1077 /* Close out prior implicit scope. */
1079 if ( ! (mdoc_macros[tok].flags & MDOC_EXPLICIT)) {
1080 rew_sub(MDOC_BODY, mdoc, tok, line, ppos);
1081 rew_sub(MDOC_BLOCK, mdoc, tok, line, ppos);
1085 * This routine accommodates implicitly- and explicitly-scoped
1086 * macro openings. Implicit ones first close out prior scope
1087 * (seen above). Delay opening the head until necessary to
1088 * allow leading punctuation to print. Special consideration
1089 * for `It -column', which has phrase-part syntax instead of
1090 * regular child nodes.
1093 mdoc_argv(mdoc, line, tok, &arg, pos, buf);
1094 mdoc_block_alloc(mdoc, line, ppos, tok, arg);
1098 * Exception: Heads of `It' macros in `-diag' lists are not
1099 * parsed, even though `It' macros in general are parsed.
1102 parsed = tok != MDOC_It ||
1103 mdoc->last->parent->tok != MDOC_Bl ||
1104 mdoc->last->parent->norm->Bl.type != LIST_diag;
1107 * The `Nd' macro has all arguments in its body: it's a hybrid
1108 * of block partial-explicit and full-implicit. Stupid.
1111 if (tok == MDOC_Nd) {
1112 head = mdoc_head_alloc(mdoc, line, ppos, tok);
1113 rew_sub(MDOC_HEAD, mdoc, tok, line, ppos);
1114 body = mdoc_body_alloc(mdoc, line, ppos, tok);
1118 mdoc->flags |= MDOC_KEEP;
1124 ac = mdoc_args(mdoc, line, pos, buf, tok, &p);
1125 if (ac == ARGS_PUNCT)
1127 if (ac == ARGS_EOLN) {
1128 if (lac != ARGS_PPHRASE && lac != ARGS_PHRASE)
1131 * This is necessary: if the last token on a
1132 * line is a `Ta' or tab, then we'll get
1133 * ARGS_EOLN, so we must be smart enough to
1134 * reopen our scope if the last parse was a
1135 * phrase or partial phrase.
1137 rew_sub(MDOC_BODY, mdoc, tok, line, ppos);
1138 body = mdoc_body_alloc(mdoc, line, ppos, tok);
1143 * Emit leading punctuation (i.e., punctuation before
1144 * the MDOC_HEAD) for non-phrase types.
1149 ac != ARGS_PHRASE &&
1150 ac != ARGS_PPHRASE &&
1152 mdoc_isdelim(p) == DELIM_OPEN) {
1153 dword(mdoc, line, la, p, DELIM_OPEN, 0);
1157 /* Open a head if one hasn't been opened. */
1160 head = mdoc_head_alloc(mdoc, line, ppos, tok);
1162 if (ac == ARGS_PHRASE ||
1164 ac == ARGS_PPHRASE) {
1167 * If we haven't opened a body yet, rewind the
1168 * head; if we have, rewind that instead.
1171 rew_sub(body ? MDOC_BODY : MDOC_HEAD,
1172 mdoc, tok, line, ppos);
1173 body = mdoc_body_alloc(mdoc, line, ppos, tok);
1176 * Process phrases: set whether we're in a
1177 * partial-phrase (this effects line handling)
1178 * then call down into the phrase parser.
1181 if (ac == ARGS_PPHRASE)
1182 mdoc->flags |= MDOC_PPHRASE;
1183 if (ac == ARGS_PEND && lac == ARGS_PPHRASE)
1184 mdoc->flags |= MDOC_PPHRASE;
1185 parse_rest(mdoc, MDOC_MAX, line, &la, buf);
1186 mdoc->flags &= ~MDOC_PPHRASE;
1190 if (macro_or_word(mdoc, tok, line, la, pos, buf, parsed))
1195 head = mdoc_head_alloc(mdoc, line, ppos, tok);
1197 append_delims(mdoc, line, pos, buf);
1202 * If there is an open (i.e., unvalidated) sub-block requiring
1203 * explicit close-out, postpone switching the current block from
1204 * head to body until the rew_sub() call closing out that
1207 for (n = mdoc->last; n && n != head; n = n->parent) {
1208 if (n->type == MDOC_BLOCK &&
1209 mdoc_macros[n->tok].flags & MDOC_EXPLICIT &&
1210 ! (n->flags & MDOC_VALID)) {
1216 /* Close out scopes to remain in a consistent state. */
1218 rew_sub(MDOC_HEAD, mdoc, tok, line, ppos);
1219 mdoc_body_alloc(mdoc, line, ppos, tok);
1221 if (mdoc->flags & MDOC_FREECOL) {
1222 rew_sub(MDOC_BODY, mdoc, tok, line, ppos);
1223 rew_sub(MDOC_BLOCK, mdoc, tok, line, ppos);
1224 mdoc->flags &= ~MDOC_FREECOL;
1229 blk_part_imp(MACRO_PROT_ARGS)
1234 struct mdoc_node *blk; /* saved block context */
1235 struct mdoc_node *body; /* saved body context */
1236 struct mdoc_node *n;
1238 nl = MDOC_NEWLINE & mdoc->flags;
1241 * A macro that spans to the end of the line. This is generally
1242 * (but not necessarily) called as the first macro. The block
1243 * has a head as the immediate child, which is always empty,
1244 * followed by zero or more opening punctuation nodes, then the
1245 * body (which may be empty, depending on the macro), then zero
1246 * or more closing punctuation nodes.
1249 blk = mdoc_block_alloc(mdoc, line, ppos, tok, NULL);
1250 mdoc_head_alloc(mdoc, line, ppos, tok);
1251 rew_sub(MDOC_HEAD, mdoc, tok, line, ppos);
1254 * Open the body scope "on-demand", that is, after we've
1255 * processed all our the leading delimiters (open parenthesis,
1259 for (body = NULL; ; ) {
1261 ac = mdoc_args(mdoc, line, pos, buf, tok, &p);
1262 if (ac == ARGS_EOLN || ac == ARGS_PUNCT)
1265 if (body == NULL && ac != ARGS_QWORD &&
1266 mdoc_isdelim(p) == DELIM_OPEN) {
1267 dword(mdoc, line, la, p, DELIM_OPEN, 0);
1272 body = mdoc_body_alloc(mdoc, line, ppos, tok);
1274 if (macro_or_word(mdoc, tok, line, la, pos, buf, 1))
1278 body = mdoc_body_alloc(mdoc, line, ppos, tok);
1281 * If there is an open sub-block requiring explicit close-out,
1282 * postpone closing out the current block
1283 * until the rew_sub() call closing out the sub-block.
1286 for (n = mdoc->last; n && n != body && n != blk->parent;
1288 if (n->type == MDOC_BLOCK &&
1289 mdoc_macros[n->tok].flags & MDOC_EXPLICIT &&
1290 ! (n->flags & MDOC_VALID)) {
1291 make_pending(n, tok, mdoc, line, ppos);
1292 mdoc_endbody_alloc(mdoc, line, ppos,
1293 tok, body, ENDBODY_NOSPACE);
1298 rew_sub(MDOC_BODY, mdoc, tok, line, ppos);
1300 append_delims(mdoc, line, pos, buf);
1301 rew_sub(MDOC_BLOCK, mdoc, tok, line, ppos);
1303 /* Move trailing .Ns out of scope. */
1305 for (n = body->child; n && n->next; n = n->next)
1307 if (n && n->tok == MDOC_Ns)
1308 mdoc_node_relink(mdoc, n);
1312 blk_part_exp(MACRO_PROT_ARGS)
1316 struct mdoc_node *head; /* keep track of head */
1317 struct mdoc_node *body; /* keep track of body */
1320 nl = MDOC_NEWLINE & mdoc->flags;
1323 * The opening of an explicit macro having zero or more leading
1324 * punctuation nodes; a head with optional single element (the
1325 * case of `Eo'); and a body that may be empty.
1328 mdoc_block_alloc(mdoc, line, ppos, tok, NULL);
1329 for (head = body = NULL; ; ) {
1331 ac = mdoc_args(mdoc, line, pos, buf, tok, &p);
1332 if (ac == ARGS_PUNCT || ac == ARGS_EOLN)
1335 /* Flush out leading punctuation. */
1337 if (head == NULL && ac != ARGS_QWORD &&
1338 mdoc_isdelim(p) == DELIM_OPEN) {
1339 assert(NULL == body);
1340 dword(mdoc, line, la, p, DELIM_OPEN, 0);
1345 assert(body == NULL);
1346 head = mdoc_head_alloc(mdoc, line, ppos, tok);
1350 * `Eo' gobbles any data into the head, but most other
1351 * macros just immediately close out and begin the body.
1356 /* No check whether it's a macro! */
1358 dword(mdoc, line, la, p, DELIM_MAX, 0);
1359 rew_sub(MDOC_HEAD, mdoc, tok, line, ppos);
1360 body = mdoc_body_alloc(mdoc, line, ppos, tok);
1364 assert(head != NULL && body != NULL);
1366 if (macro_or_word(mdoc, tok, line, la, pos, buf, 1))
1370 /* Clean-up to leave in a consistent state. */
1373 mdoc_head_alloc(mdoc, line, ppos, tok);
1376 rew_sub(MDOC_HEAD, mdoc, tok, line, ppos);
1377 mdoc_body_alloc(mdoc, line, ppos, tok);
1380 append_delims(mdoc, line, pos, buf);
1384 in_line_argn(MACRO_PROT_ARGS)
1386 int la, flushed, j, maxargs, nl;
1388 struct mdoc_arg *arg;
1392 nl = MDOC_NEWLINE & mdoc->flags;
1395 * A line macro that has a fixed number of arguments (maxargs).
1396 * Only open the scope once the first non-leading-punctuation is
1397 * found (unless MDOC_IGNDELIM is noted, like in `Pf'), then
1398 * keep it open until the maximum number of arguments are
1422 mdoc_argv(mdoc, line, tok, &arg, pos, buf);
1424 for (flushed = j = 0; ; ) {
1426 ac = mdoc_args(mdoc, line, pos, buf, tok, &p);
1427 if (ac == ARGS_PUNCT || ac == ARGS_EOLN)
1430 if ( ! (mdoc_macros[tok].flags & MDOC_IGNDELIM) &&
1431 ac != ARGS_QWORD && j == 0 &&
1432 mdoc_isdelim(p) == DELIM_OPEN) {
1433 dword(mdoc, line, la, p, DELIM_OPEN, 0);
1436 mdoc_elem_alloc(mdoc, line, ppos, tok, arg);
1438 if (j == maxargs && ! flushed) {
1439 rew_elem(mdoc, tok);
1443 ntok = ac == ARGS_QWORD ? MDOC_MAX : lookup(tok, p);
1445 if (ntok != MDOC_MAX) {
1447 rew_elem(mdoc, tok);
1449 mdoc_macro(mdoc, ntok, line, la, pos, buf);
1454 if ( ! (mdoc_macros[tok].flags & MDOC_IGNDELIM) &&
1455 ac != ARGS_QWORD && ! flushed &&
1456 mdoc_isdelim(p) != DELIM_NONE) {
1457 rew_elem(mdoc, tok);
1461 dword(mdoc, line, la, p, DELIM_MAX,
1462 MDOC_JOIN & mdoc_macros[tok].flags);
1467 mdoc_elem_alloc(mdoc, line, ppos, tok, arg);
1469 rew_elem(mdoc, tok);
1471 append_delims(mdoc, line, pos, buf);
1475 in_line_eoln(MACRO_PROT_ARGS)
1477 struct mdoc_arg *arg;
1480 rew_sub(MDOC_BLOCK, mdoc, MDOC_Nm, line, ppos);
1482 mdoc_argv(mdoc, line, tok, &arg, pos, buf);
1483 mdoc_elem_alloc(mdoc, line, ppos, tok, arg);
1484 if (parse_rest(mdoc, tok, line, pos, buf))
1486 rew_elem(mdoc, tok);
1490 * The simplest argument parser available: Parse the remaining
1491 * words until the end of the phrase or line and return 0
1492 * or until the next macro, call that macro, and return 1.
1495 parse_rest(struct mdoc *mdoc, enum mdoct tok, int line, int *pos, char *buf)
1501 if (mdoc_args(mdoc, line, pos, buf, tok, NULL) == ARGS_EOLN)
1503 if (macro_or_word(mdoc, tok, line, la, pos, buf, 1))
1509 ctx_synopsis(MACRO_PROT_ARGS)
1512 if (~mdoc->flags & (MDOC_SYNOPSIS | MDOC_NEWLINE))
1513 in_line(mdoc, tok, line, ppos, pos, buf);
1514 else if (tok == MDOC_Nm)
1515 blk_full(mdoc, tok, line, ppos, pos, buf);
1517 assert(tok == MDOC_Vt);
1518 blk_part_imp(mdoc, tok, line, ppos, pos, buf);
1523 * Phrases occur within `Bl -column' entries, separated by `Ta' or tabs.
1524 * They're unusual because they're basically free-form text until a
1525 * macro is encountered.
1528 phrase_ta(MACRO_PROT_ARGS)
1530 struct mdoc_node *n;
1532 /* Make sure we are in a column list or ignore this macro. */
1535 while (n != NULL && n->tok != MDOC_Bl)
1537 if (n == NULL || n->norm->Bl.type != LIST_column) {
1538 mandoc_msg(MANDOCERR_TA_STRAY, mdoc->parse,
1543 /* Advance to the next column. */
1545 rew_sub(MDOC_BODY, mdoc, MDOC_It, line, ppos);
1546 mdoc_body_alloc(mdoc, line, ppos, MDOC_It);
1547 parse_rest(mdoc, MDOC_MAX, line, pos, buf);