1/*	$Id: mdoc_man.c,v 1.9 2011/10/24 21:47:59 schwarze Exp $ */
2/*
3 * Copyright (c) 2011 Ingo Schwarze <schwarze@openbsd.org>
4 *
5 * Permission to use, copy, modify, and distribute this software for any
6 * purpose with or without fee is hereby granted, provided that the above
7 * copyright notice and this permission notice appear in all copies.
8 *
9 * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
10 * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
11 * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
12 * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
13 * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
14 * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
15 * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
16 */
17#ifdef HAVE_CONFIG_H
18#include "config.h"
19#endif
20
21#include <stdio.h>
22#include <string.h>
23
24#include "mandoc.h"
25#include "man.h"
26#include "mdoc.h"
27#include "main.h"
28
29#define	DECL_ARGS const struct mdoc_meta *m, \
30		  const struct mdoc_node *n, \
31		  struct mman *mm
32
33struct	mman {
34	int		  need_space; /* next word needs prior ws */
35	int		  need_nl; /* next word needs prior nl */
36};
37
38struct	manact {
39	int		(*cond)(DECL_ARGS); /* DON'T run actions */
40	int		(*pre)(DECL_ARGS); /* pre-node action */
41	void		(*post)(DECL_ARGS); /* post-node action */
42	const char	 *prefix; /* pre-node string constant */
43	const char	 *suffix; /* post-node string constant */
44};
45
46static	int	  cond_body(DECL_ARGS);
47static	int	  cond_head(DECL_ARGS);
48static	void	  post_bd(DECL_ARGS);
49static	void	  post_dl(DECL_ARGS);
50static	void	  post_enc(DECL_ARGS);
51static	void	  post_nm(DECL_ARGS);
52static	void	  post_percent(DECL_ARGS);
53static	void	  post_pf(DECL_ARGS);
54static	void	  post_sect(DECL_ARGS);
55static	void	  post_sp(DECL_ARGS);
56static	int	  pre_ap(DECL_ARGS);
57static	int	  pre_bd(DECL_ARGS);
58static	int	  pre_br(DECL_ARGS);
59static	int	  pre_bx(DECL_ARGS);
60static	int	  pre_dl(DECL_ARGS);
61static	int	  pre_enc(DECL_ARGS);
62static	int	  pre_it(DECL_ARGS);
63static	int	  pre_nm(DECL_ARGS);
64static	int	  pre_ns(DECL_ARGS);
65static	int	  pre_pp(DECL_ARGS);
66static	int	  pre_sp(DECL_ARGS);
67static	int	  pre_sect(DECL_ARGS);
68static	int	  pre_ux(DECL_ARGS);
69static	int	  pre_xr(DECL_ARGS);
70static	void	  print_word(struct mman *, const char *);
71static	void	  print_node(DECL_ARGS);
72
73static	const struct manact manacts[MDOC_MAX + 1] = {
74	{ NULL, pre_ap, NULL, NULL, NULL }, /* Ap */
75	{ NULL, NULL, NULL, NULL, NULL }, /* Dd */
76	{ NULL, NULL, NULL, NULL, NULL }, /* Dt */
77	{ NULL, NULL, NULL, NULL, NULL }, /* Os */
78	{ NULL, pre_sect, post_sect, ".SH", NULL }, /* Sh */
79	{ NULL, pre_sect, post_sect, ".SS", NULL }, /* Ss */
80	{ NULL, pre_pp, NULL, NULL, NULL }, /* Pp */
81	{ cond_body, pre_dl, post_dl, NULL, NULL }, /* D1 */
82	{ cond_body, pre_dl, post_dl, NULL, NULL }, /* Dl */
83	{ cond_body, pre_bd, post_bd, NULL, NULL }, /* Bd */
84	{ NULL, NULL, NULL, NULL, NULL }, /* Ed */
85	{ NULL, NULL, NULL, NULL, NULL }, /* Bl */
86	{ NULL, NULL, NULL, NULL, NULL }, /* El */
87	{ NULL, pre_it, NULL, NULL, NULL }, /* _It */
88	{ NULL, pre_enc, post_enc, "\\fI", "\\fP" }, /* Ad */
89	{ NULL, NULL, NULL, NULL, NULL }, /* _An */
90	{ NULL, pre_enc, post_enc, "\\fI", "\\fP" }, /* Ar */
91	{ NULL, pre_enc, post_enc, "\\fB", "\\fP" }, /* Cd */
92	{ NULL, pre_enc, post_enc, "\\fB", "\\fP" }, /* Cm */
93	{ NULL, pre_enc, post_enc, "\\fR", "\\fP" }, /* Dv */
94	{ NULL, pre_enc, post_enc, "\\fR", "\\fP" }, /* Er */
95	{ NULL, pre_enc, post_enc, "\\fR", "\\fP" }, /* Ev */
96	{ NULL, pre_enc, post_enc, "The \\fB",
97	    "\\fP\nutility exits 0 on success, and >0 if an error occurs."
98	    }, /* Ex */
99	{ NULL, NULL, NULL, NULL, NULL }, /* _Fa */
100	{ NULL, NULL, NULL, NULL, NULL }, /* _Fd */
101	{ NULL, pre_enc, post_enc, "\\fB-", "\\fP" }, /* Fl */
102	{ NULL, NULL, NULL, NULL, NULL }, /* _Fn */
103	{ NULL, NULL, NULL, NULL, NULL }, /* _Ft */
104	{ NULL, pre_enc, post_enc, "\\fB", "\\fP" }, /* Ic */
105	{ NULL, NULL, NULL, NULL, NULL }, /* _In */
106	{ NULL, pre_enc, post_enc, "\\fR", "\\fP" }, /* Li */
107	{ cond_head, pre_enc, NULL, "\\- ", NULL }, /* Nd */
108	{ NULL, pre_nm, post_nm, NULL, NULL }, /* Nm */
109	{ cond_body, pre_enc, post_enc, "[", "]" }, /* Op */
110	{ NULL, NULL, NULL, NULL, NULL }, /* Ot */
111	{ NULL, pre_enc, post_enc, "\\fI", "\\fP" }, /* Pa */
112	{ NULL, pre_enc, post_enc, "The \\fB",
113		"\\fP\nfunction returns the value 0 if successful;\n"
114		"otherwise the value -1 is returned and the global\n"
115		"variable \\fIerrno\\fP is set to indicate the error."
116		}, /* Rv */
117	{ NULL, NULL, NULL, NULL, NULL }, /* St */
118	{ NULL, NULL, NULL, NULL, NULL }, /* _Va */
119	{ NULL, NULL, NULL, NULL, NULL }, /* _Vt */
120	{ NULL, pre_xr, NULL, NULL, NULL }, /* Xr */
121	{ NULL, NULL, post_percent, NULL, NULL }, /* _%A */
122	{ NULL, NULL, NULL, NULL, NULL }, /* _%B */
123	{ NULL, NULL, post_percent, NULL, NULL }, /* _%D */
124	{ NULL, NULL, NULL, NULL, NULL }, /* _%I */
125	{ NULL, pre_enc, post_percent, "\\fI", "\\fP" }, /* %J */
126	{ NULL, NULL, NULL, NULL, NULL }, /* _%N */
127	{ NULL, NULL, NULL, NULL, NULL }, /* _%O */
128	{ NULL, NULL, NULL, NULL, NULL }, /* _%P */
129	{ NULL, NULL, NULL, NULL, NULL }, /* _%R */
130	{ NULL, pre_enc, post_percent, "\"", "\"" }, /* %T */
131	{ NULL, NULL, NULL, NULL, NULL }, /* _%V */
132	{ NULL, NULL, NULL, NULL, NULL }, /* Ac */
133	{ cond_body, pre_enc, post_enc, "<", ">" }, /* Ao */
134	{ cond_body, pre_enc, post_enc, "<", ">" }, /* Aq */
135	{ NULL, NULL, NULL, NULL, NULL }, /* At */
136	{ NULL, NULL, NULL, NULL, NULL }, /* Bc */
137	{ NULL, NULL, NULL, NULL, NULL }, /* _Bf */
138	{ cond_body, pre_enc, post_enc, "[", "]" }, /* Bo */
139	{ cond_body, pre_enc, post_enc, "[", "]" }, /* Bq */
140	{ NULL, pre_ux, NULL, "BSD/OS", NULL }, /* Bsx */
141	{ NULL, pre_bx, NULL, NULL, NULL }, /* Bx */
142	{ NULL, NULL, NULL, NULL, NULL }, /* Db */
143	{ NULL, NULL, NULL, NULL, NULL }, /* Dc */
144	{ cond_body, pre_enc, post_enc, "``", "''" }, /* Do */
145	{ cond_body, pre_enc, post_enc, "``", "''" }, /* Dq */
146	{ NULL, NULL, NULL, NULL, NULL }, /* _Ec */
147	{ NULL, NULL, NULL, NULL, NULL }, /* _Ef */
148	{ NULL, pre_enc, post_enc, "\\fI", "\\fP" }, /* Em */
149	{ NULL, NULL, NULL, NULL, NULL }, /* _Eo */
150	{ NULL, pre_ux, NULL, "FreeBSD", NULL }, /* Fx */
151	{ NULL, pre_enc, post_enc, "\\fB", "\\fP" }, /* Ms */
152	{ NULL, NULL, NULL, NULL, NULL }, /* No */
153	{ NULL, pre_ns, NULL, NULL, NULL }, /* Ns */
154	{ NULL, pre_ux, NULL, "NetBSD", NULL }, /* Nx */
155	{ NULL, pre_ux, NULL, "OpenBSD", NULL }, /* Ox */
156	{ NULL, NULL, NULL, NULL, NULL }, /* Pc */
157	{ NULL, NULL, post_pf, NULL, NULL }, /* Pf */
158	{ cond_body, pre_enc, post_enc, "(", ")" }, /* Po */
159	{ cond_body, pre_enc, post_enc, "(", ")" }, /* Pq */
160	{ NULL, NULL, NULL, NULL, NULL }, /* Qc */
161	{ cond_body, pre_enc, post_enc, "`", "'" }, /* Ql */
162	{ cond_body, pre_enc, post_enc, "\"", "\"" }, /* Qo */
163	{ cond_body, pre_enc, post_enc, "\"", "\"" }, /* Qq */
164	{ NULL, NULL, NULL, NULL, NULL }, /* Re */
165	{ cond_body, pre_pp, NULL, NULL, NULL }, /* Rs */
166	{ NULL, NULL, NULL, NULL, NULL }, /* Sc */
167	{ cond_body, pre_enc, post_enc, "`", "'" }, /* So */
168	{ cond_body, pre_enc, post_enc, "`", "'" }, /* Sq */
169	{ NULL, NULL, NULL, NULL, NULL }, /* _Sm */
170	{ NULL, pre_enc, post_enc, "\\fI", "\\fP" }, /* Sx */
171	{ NULL, pre_enc, post_enc, "\\fB", "\\fP" }, /* Sy */
172	{ NULL, pre_enc, post_enc, "\\fR", "\\fP" }, /* Tn */
173	{ NULL, pre_ux, NULL, "UNIX", NULL }, /* Ux */
174	{ NULL, NULL, NULL, NULL, NULL }, /* _Xc */
175	{ NULL, NULL, NULL, NULL, NULL }, /* _Xo */
176	{ NULL, NULL, NULL, NULL, NULL }, /* _Fo */
177	{ NULL, NULL, NULL, NULL, NULL }, /* _Fc */
178	{ cond_body, pre_enc, post_enc, "[", "]" }, /* Oo */
179	{ NULL, NULL, NULL, NULL, NULL }, /* Oc */
180	{ NULL, NULL, NULL, NULL, NULL }, /* _Bk */
181	{ NULL, NULL, NULL, NULL, NULL }, /* _Ek */
182	{ NULL, pre_ux, NULL, "is currently in beta test.", NULL }, /* Bt */
183	{ NULL, NULL, NULL, NULL, NULL }, /* Hf */
184	{ NULL, NULL, NULL, NULL, NULL }, /* Fr */
185	{ NULL, pre_ux, NULL, "currently under development.", NULL }, /* Ud */
186	{ NULL, NULL, NULL, NULL, NULL }, /* _Lb */
187	{ NULL, pre_pp, NULL, NULL, NULL }, /* Lp */
188	{ NULL, NULL, NULL, NULL, NULL }, /* _Lk */
189	{ NULL, NULL, NULL, NULL, NULL }, /* _Mt */
190	{ cond_body, pre_enc, post_enc, "{", "}" }, /* Brq */
191	{ cond_body, pre_enc, post_enc, "{", "}" }, /* Bro */
192	{ NULL, NULL, NULL, NULL, NULL }, /* Brc */
193	{ NULL, NULL, NULL, NULL, NULL }, /* _%C */
194	{ NULL, NULL, NULL, NULL, NULL }, /* _Es */
195	{ NULL, NULL, NULL, NULL, NULL }, /* _En */
196	{ NULL, pre_ux, NULL, "DragonFly", NULL }, /* Dx */
197	{ NULL, NULL, NULL, NULL, NULL }, /* _%Q */
198	{ NULL, pre_br, NULL, NULL, NULL }, /* br */
199	{ NULL, pre_sp, post_sp, NULL, NULL }, /* sp */
200	{ NULL, NULL, NULL, NULL, NULL }, /* _%U */
201	{ NULL, NULL, NULL, NULL, NULL }, /* _Ta */
202	{ NULL, NULL, NULL, NULL, NULL }, /* ROOT */
203};
204
205static void
206print_word(struct mman *mm, const char *s)
207{
208
209	if (mm->need_nl) {
210		/*
211		 * If we need a newline, print it now and start afresh.
212		 */
213		putchar('\n');
214		mm->need_space = 0;
215		mm->need_nl = 0;
216	} else if (mm->need_space && '\0' != s[0])
217		/*
218		 * If we need a space, only print it before
219		 * (1) a nonzero length word;
220		 * (2) a word that is non-punctuation; and
221		 * (3) if punctuation, non-terminating puncutation.
222		 */
223		if (NULL == strchr(".,:;)]?!", s[0]) || '\0' != s[1])
224			putchar(' ');
225
226	/*
227	 * Reassign needing space if we're not following opening
228	 * punctuation.
229	 */
230	mm->need_space =
231		('(' != s[0] && '[' != s[0]) || '\0' != s[1];
232
233	for ( ; *s; s++) {
234		switch (*s) {
235		case (ASCII_NBRSP):
236			printf("\\~");
237			break;
238		case (ASCII_HYPH):
239			putchar('-');
240			break;
241		default:
242			putchar((unsigned char)*s);
243			break;
244		}
245	}
246}
247
248void
249man_man(void *arg, const struct man *man)
250{
251
252	/*
253	 * Dump the keep buffer.
254	 * We're guaranteed by now that this exists (is non-NULL).
255	 * Flush stdout afterward, just in case.
256	 */
257	fputs(mparse_getkeep(man_mparse(man)), stdout);
258	fflush(stdout);
259}
260
261void
262man_mdoc(void *arg, const struct mdoc *mdoc)
263{
264	const struct mdoc_meta *m;
265	const struct mdoc_node *n;
266	struct mman	        mm;
267
268	m = mdoc_meta(mdoc);
269	n = mdoc_node(mdoc);
270
271	printf(".TH \"%s\" \"%s\" \"%s\" \"%s\" \"%s\"",
272			m->title, m->msec, m->date, m->os, m->vol);
273
274	memset(&mm, 0, sizeof(struct mman));
275
276	mm.need_nl = 1;
277	print_node(m, n, &mm);
278	putchar('\n');
279}
280
281static void
282print_node(DECL_ARGS)
283{
284	const struct mdoc_node	*prev, *sub;
285	const struct manact	*act;
286	int			 cond, do_sub;
287
288	/*
289	 * Break the line if we were parsed subsequent the current node.
290	 * This makes the page structure be more consistent.
291	 */
292	prev = n->prev ? n->prev : n->parent;
293	if (prev && prev->line < n->line)
294		mm->need_nl = 1;
295
296	act = NULL;
297	cond = 0;
298	do_sub = 1;
299
300	if (MDOC_TEXT == n->type) {
301		/*
302		 * Make sure that we don't happen to start with a
303		 * control character at the start of a line.
304		 */
305		if (mm->need_nl && ('.' == *n->string ||
306					'\'' == *n->string)) {
307			print_word(mm, "\\&");
308			mm->need_space = 0;
309		}
310		print_word(mm, n->string);
311	} else {
312		/*
313		 * Conditionally run the pre-node action handler for a
314		 * node.
315		 */
316		act = manacts + n->tok;
317		cond = NULL == act->cond || (*act->cond)(m, n, mm);
318		if (cond && act->pre)
319			do_sub = (*act->pre)(m, n, mm);
320	}
321
322	/*
323	 * Conditionally run all child nodes.
324	 * Note that this iterates over children instead of using
325	 * recursion.  This prevents unnecessary depth in the stack.
326	 */
327	if (do_sub)
328		for (sub = n->child; sub; sub = sub->next)
329			print_node(m, sub, mm);
330
331	/*
332	 * Lastly, conditionally run the post-node handler.
333	 */
334	if (cond && act->post)
335		(*act->post)(m, n, mm);
336}
337
338static int
339cond_head(DECL_ARGS)
340{
341
342	return(MDOC_HEAD == n->type);
343}
344
345static int
346cond_body(DECL_ARGS)
347{
348
349	return(MDOC_BODY == n->type);
350}
351
352/*
353 * Output a font encoding before a node, e.g., \fR.
354 * This obviously has no trailing space.
355 */
356static int
357pre_enc(DECL_ARGS)
358{
359	const char	*prefix;
360
361	prefix = manacts[n->tok].prefix;
362	if (NULL == prefix)
363		return(1);
364	print_word(mm, prefix);
365	mm->need_space = 0;
366	return(1);
367}
368
369/*
370 * Output a font encoding subsequent a node, e.g., \fP.
371 */
372static void
373post_enc(DECL_ARGS)
374{
375	const char *suffix;
376
377	suffix = manacts[n->tok].suffix;
378	if (NULL == suffix)
379		return;
380	mm->need_space = 0;
381	print_word(mm, suffix);
382}
383
384/*
385 * Used in listings (percent = %A, e.g.).
386 * FIXME: this is incomplete.
387 * It doesn't print a nice ", and" for lists.
388 */
389static void
390post_percent(DECL_ARGS)
391{
392
393	post_enc(m, n, mm);
394	if (n->next)
395		print_word(mm, ",");
396	else {
397		print_word(mm, ".");
398		mm->need_nl = 1;
399	}
400}
401
402/*
403 * Print before a section header.
404 */
405static int
406pre_sect(DECL_ARGS)
407{
408
409	if (MDOC_HEAD != n->type)
410		return(1);
411	mm->need_nl = 1;
412	print_word(mm, manacts[n->tok].prefix);
413	print_word(mm, "\"");
414	mm->need_space = 0;
415	return(1);
416}
417
418/*
419 * Print subsequent a section header.
420 */
421static void
422post_sect(DECL_ARGS)
423{
424
425	if (MDOC_HEAD != n->type)
426		return;
427	mm->need_space = 0;
428	print_word(mm, "\"");
429	mm->need_nl = 1;
430}
431
432static int
433pre_ap(DECL_ARGS)
434{
435
436	mm->need_space = 0;
437	print_word(mm, "'");
438	mm->need_space = 0;
439	return(0);
440}
441
442static int
443pre_bd(DECL_ARGS)
444{
445
446	if (DISP_unfilled == n->norm->Bd.type ||
447	    DISP_literal  == n->norm->Bd.type) {
448		mm->need_nl = 1;
449		print_word(mm, ".nf");
450	}
451	mm->need_nl = 1;
452	return(1);
453}
454
455static void
456post_bd(DECL_ARGS)
457{
458
459	if (DISP_unfilled == n->norm->Bd.type ||
460	    DISP_literal  == n->norm->Bd.type) {
461		mm->need_nl = 1;
462		print_word(mm, ".fi");
463	}
464	mm->need_nl = 1;
465}
466
467static int
468pre_br(DECL_ARGS)
469{
470
471	mm->need_nl = 1;
472	print_word(mm, ".br");
473	mm->need_nl = 1;
474	return(0);
475}
476
477static int
478pre_bx(DECL_ARGS)
479{
480
481	n = n->child;
482	if (n) {
483		print_word(mm, n->string);
484		mm->need_space = 0;
485		n = n->next;
486	}
487	print_word(mm, "BSD");
488	if (NULL == n)
489		return(0);
490	mm->need_space = 0;
491	print_word(mm, "-");
492	mm->need_space = 0;
493	print_word(mm, n->string);
494	return(0);
495}
496
497static int
498pre_dl(DECL_ARGS)
499{
500
501	mm->need_nl = 1;
502	print_word(mm, ".RS 6n");
503	mm->need_nl = 1;
504	return(1);
505}
506
507static void
508post_dl(DECL_ARGS)
509{
510
511	mm->need_nl = 1;
512	print_word(mm, ".RE");
513	mm->need_nl = 1;
514}
515
516static int
517pre_it(DECL_ARGS)
518{
519	const struct mdoc_node *bln;
520
521	if (MDOC_HEAD == n->type) {
522		mm->need_nl = 1;
523		print_word(mm, ".TP");
524		bln = n->parent->parent->prev;
525		switch (bln->norm->Bl.type) {
526		case (LIST_bullet):
527			print_word(mm, "4n");
528			mm->need_nl = 1;
529			print_word(mm, "\\fBo\\fP");
530			break;
531		default:
532			if (bln->norm->Bl.width)
533				print_word(mm, bln->norm->Bl.width);
534			break;
535		}
536		mm->need_nl = 1;
537	}
538	return(1);
539}
540
541static int
542pre_nm(DECL_ARGS)
543{
544
545	if (MDOC_ELEM != n->type && MDOC_HEAD != n->type)
546		return(1);
547	print_word(mm, "\\fB");
548	mm->need_space = 0;
549	if (NULL == n->child)
550		print_word(mm, m->name);
551	return(1);
552}
553
554static void
555post_nm(DECL_ARGS)
556{
557
558	if (MDOC_ELEM != n->type && MDOC_HEAD != n->type)
559		return;
560	mm->need_space = 0;
561	print_word(mm, "\\fP");
562}
563
564static int
565pre_ns(DECL_ARGS)
566{
567
568	mm->need_space = 0;
569	return(0);
570}
571
572static void
573post_pf(DECL_ARGS)
574{
575
576	mm->need_space = 0;
577}
578
579static int
580pre_pp(DECL_ARGS)
581{
582
583	mm->need_nl = 1;
584	if (MDOC_It == n->parent->tok)
585		print_word(mm, ".sp");
586	else
587		print_word(mm, ".PP");
588	mm->need_nl = 1;
589	return(1);
590}
591
592static int
593pre_sp(DECL_ARGS)
594{
595
596	mm->need_nl = 1;
597	print_word(mm, ".sp");
598	return(1);
599}
600
601static void
602post_sp(DECL_ARGS)
603{
604
605	mm->need_nl = 1;
606}
607
608static int
609pre_xr(DECL_ARGS)
610{
611
612	n = n->child;
613	if (NULL == n)
614		return(0);
615	print_node(m, n, mm);
616	n = n->next;
617	if (NULL == n)
618		return(0);
619	mm->need_space = 0;
620	print_word(mm, "(");
621	print_node(m, n, mm);
622	print_word(mm, ")");
623	return(0);
624}
625
626static int
627pre_ux(DECL_ARGS)
628{
629
630	print_word(mm, manacts[n->tok].prefix);
631	if (NULL == n->child)
632		return(0);
633	mm->need_space = 0;
634	print_word(mm, "\\~");
635	mm->need_space = 0;
636	return(1);
637}
638