]>
git.cameronkatri.com Git - mandoc.git/blob - main.c
1 /* $Id: main.c,v 1.101 2010/07/29 22:00:39 joerg Exp $ */
3 * Copyright (c) 2008, 2009, 2010 Kristaps Dzonsons <kristaps@bsd.lv>
4 * Copyright (c) 2010 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.
44 #define UNCONST(a) ((void *)(uintptr_t)(const void *)(a))
46 /* FIXME: Intel's compiler? LLVM? pcc? */
48 #if !defined(__GNUC__) || (__GNUC__ < 2)
50 # define __attribute__(x)
52 #endif /* !defined(__GNUC__) || (__GNUC__ < 2) */
54 typedef void (*out_mdoc
)(void *, const struct mdoc
*);
55 typedef void (*out_man
)(void *, const struct man
*);
56 typedef void (*out_free
)(void *);
80 const char *file
; /* Current parse. */
81 int fd
; /* Current parse. */
83 /* FIXME: set by max error */
84 #define WARN_WALL (1 << 0) /* All-warnings mask. */
85 #define WARN_WERR (1 << 2) /* Warnings->errors. */
87 #define FL_IGN_SCOPE (1 << 0) /* Ignore scope errors. */
88 #define FL_NIGN_ESCAPE (1 << 1) /* Don't ignore bad escapes. */
89 #define FL_NIGN_MACRO (1 << 2) /* Don't ignore bad macros. */
90 #define FL_IGN_ERRORS (1 << 4) /* Ignore failed parse. */
91 #define FL_STRICT FL_NIGN_ESCAPE | \
92 FL_NIGN_MACRO /* ignore nothing */
93 enum intt inttype
; /* which parser to use */
94 struct man
*man
; /* man parser */
95 struct mdoc
*mdoc
; /* mdoc parser */
96 struct roff
*roff
; /* roff parser (!NULL) */
97 struct regset regs
; /* roff registers */
98 enum outt outtype
; /* which output to use */
99 out_mdoc outmdoc
; /* mdoc output ptr */
100 out_man outman
; /* man output ptr */
101 out_free outfree
; /* free output ptr */
102 void *outdata
; /* data for output */
103 char outopts
[BUFSIZ
]; /* buf of output opts */
106 static const char * const mandocerrs
[MANDOCERR_MAX
] = {
111 "text should be uppercase",
112 "sections out of conventional order",
113 "section name repeats",
114 "out of order prologue",
115 "repeated prologue entry",
116 "list type must come first",
119 "tab in non-literal context",
120 "bad escape sequence",
121 "unterminated quoted string",
122 "argument requires the width argument",
123 "superfluous width argument",
126 "bad width argument",
127 "unknown manual section",
128 "section not in conventional manual section",
129 "end of line whitespace",
130 "blocks badly nested",
131 "scope open on exit",
135 "NAME section must come first",
137 "child violates parent syntax",
139 "list type repeated",
140 "display type repeated",
142 "manual name not yet set",
143 "obsolete macro ignored",
144 "empty macro ignored",
145 "macro not allowed in body",
146 "macro not allowed in prologue",
148 "bad NAME section contents",
150 "no text in this context",
152 "unknown macro will be lost",
154 "argument count wrong",
155 "request scope close w/none open",
156 "scope already open",
157 "macro requires line argument(s)",
158 "macro requires body argument(s)",
159 "macro requires argument(s)",
160 "no title in document",
162 "missing display type",
164 "line argument(s) will be lost",
165 "body argument(s) will be lost",
167 "generic fatal error",
169 "column syntax is inconsistent",
170 "displays may not be nested",
171 "unsupported display type",
172 "blocks badly nested",
173 "no such block is open",
174 "scope broken, syntax violated",
175 "line scope broken, syntax violated",
176 "argument count wrong, violates syntax",
177 "child violates parent syntax",
178 "argument count wrong, violates syntax",
180 "no document prologue",
181 "utsname system call failed",
185 static void fdesc(struct curparse
*);
186 static void ffile(const char *, struct curparse
*);
187 static int foptions(int *, char *);
188 static struct man
*man_init(struct curparse
*);
189 static struct mdoc
*mdoc_init(struct curparse
*);
190 static struct roff
*roff_init(struct curparse
*);
191 static int moptions(enum intt
*, char *);
192 static int mmsg(enum mandocerr
, void *,
193 int, int, const char *);
194 static int pset(const char *, int, struct curparse
*,
195 struct man
**, struct mdoc
**);
196 static int toptions(struct curparse
*, char *);
197 static void usage(void) __attribute__((noreturn
));
198 static void version(void) __attribute__((noreturn
));
199 static int woptions(int *, char *);
201 static const char *progname
;
202 static int with_fatal
;
203 static int with_error
;
206 main(int argc
, char *argv
[])
209 struct curparse curp
;
211 progname
= strrchr(argv
[0], '/');
212 if (progname
== NULL
)
217 memset(&curp
, 0, sizeof(struct curparse
));
219 curp
.inttype
= INTT_AUTO
;
220 curp
.outtype
= OUTT_ASCII
;
223 while (-1 != (c
= getopt(argc
, argv
, "f:m:O:T:VW:")))
226 if ( ! foptions(&curp
.fflags
, optarg
))
227 return(EXIT_FAILURE
);
230 if ( ! moptions(&curp
.inttype
, optarg
))
231 return(EXIT_FAILURE
);
234 (void)strlcat(curp
.outopts
, optarg
, BUFSIZ
);
235 (void)strlcat(curp
.outopts
, ",", BUFSIZ
);
238 if ( ! toptions(&curp
, optarg
))
239 return(EXIT_FAILURE
);
242 if ( ! woptions(&curp
.wflags
, optarg
))
243 return(EXIT_FAILURE
);
257 curp
.file
= "<stdin>";
258 curp
.fd
= STDIN_FILENO
;
266 if (with_fatal
&& !(curp
.fflags
& FL_IGN_ERRORS
))
272 (*curp
.outfree
)(curp
.outdata
);
274 mdoc_free(curp
.mdoc
);
278 roff_free(curp
.roff
);
280 return((with_fatal
|| with_error
) ?
281 EXIT_FAILURE
: EXIT_SUCCESS
);
289 (void)printf("%s %s\n", progname
, VERSION
);
298 (void)fprintf(stderr
, "usage: %s [-V] [-foption] "
299 "[-mformat] [-Ooption] [-Toutput] "
300 "[-Werr] [file...]\n", progname
);
306 man_init(struct curparse
*curp
)
310 /* Defaults from mandoc.1. */
312 pflags
= MAN_IGN_MACRO
| MAN_IGN_ESCAPE
;
314 if (curp
->fflags
& FL_NIGN_MACRO
)
315 pflags
&= ~MAN_IGN_MACRO
;
316 if (curp
->fflags
& FL_NIGN_ESCAPE
)
317 pflags
&= ~MAN_IGN_ESCAPE
;
319 return(man_alloc(&curp
->regs
, curp
, pflags
, mmsg
));
324 roff_init(struct curparse
*curp
)
327 return(roff_alloc(&curp
->regs
, mmsg
, curp
));
332 mdoc_init(struct curparse
*curp
)
336 /* Defaults from mandoc.1. */
338 pflags
= MDOC_IGN_MACRO
| MDOC_IGN_ESCAPE
;
340 if (curp
->fflags
& FL_IGN_SCOPE
)
341 pflags
|= MDOC_IGN_SCOPE
;
342 if (curp
->fflags
& FL_NIGN_ESCAPE
)
343 pflags
&= ~MDOC_IGN_ESCAPE
;
344 if (curp
->fflags
& FL_NIGN_MACRO
)
345 pflags
&= ~MDOC_IGN_MACRO
;
347 return(mdoc_alloc(&curp
->regs
, curp
, pflags
, mmsg
));
352 ffile(const char *file
, struct curparse
*curp
)
356 if (-1 == (curp
->fd
= open(curp
->file
, O_RDONLY
, 0))) {
364 if (-1 == close(curp
->fd
))
370 resize_buf(struct buf
*buf
, size_t initial
)
379 tmp
= realloc(buf
->buf
, sz
);
391 read_whole_file(struct curparse
*curp
, struct buf
*fb
, int *with_mmap
)
397 if (-1 == fstat(curp
->fd
, &st
)) {
404 * If we're a regular file, try just reading in the whole entry
405 * via mmap(). This is faster than reading it into blocks, and
406 * since each file is only a few bytes to begin with, I'm not
407 * concerned that this is going to tank any machines.
410 if (S_ISREG(st
.st_mode
)) {
411 if (st
.st_size
>= (1U << 31)) {
412 fprintf(stderr
, "%s: input too large\n",
418 fb
->sz
= (size_t)st
.st_size
;
419 fb
->buf
= mmap(NULL
, fb
->sz
, PROT_READ
,
420 MAP_FILE
|MAP_SHARED
, curp
->fd
, 0);
421 if (fb
->buf
!= MAP_FAILED
)
426 * If this isn't a regular file (like, say, stdin), then we must
427 * go the old way and just read things in bit by bit.
436 if (fb
->sz
== (1U << 31)) {
437 fprintf(stderr
, "%s: input too large\n",
441 if (! resize_buf(fb
, 65536))
444 ssz
= read(curp
->fd
, fb
->buf
+ (int)off
, fb
->sz
- off
);
464 fdesc(struct curparse
*curp
)
467 int i
, pos
, lnn
, lnn_start
, with_mmap
, of
;
477 memset(&ln
, 0, sizeof(struct buf
));
480 * Two buffers: ln and buf. buf is the input file and may be
481 * memory mapped. ln is a line buffer and grows on-demand.
484 if ( ! read_whole_file(curp
, &blk
, &with_mmap
))
487 if (NULL
== curp
->roff
)
488 curp
->roff
= roff_init(curp
);
489 if (NULL
== (roff
= curp
->roff
))
492 for (i
= 0, lnn
= 1; i
< (int)blk
.sz
;) {
495 while (i
< (int)blk
.sz
) {
496 if ('\n' == blk
.buf
[i
]) {
503 * Warn about bogus characters. If you're using
504 * non-ASCII encoding, you're screwing your
505 * readers. Since I'd rather this not happen,
506 * I'll be helpful and drop these characters so
507 * we don't display gibberish. Note to manual
508 * writers: use special characters.
511 if ( ! isgraph((u_char
)blk
.buf
[i
]) &&
512 ! isblank((u_char
)blk
.buf
[i
])) {
513 if ( ! mmsg(MANDOCERR_BADCHAR
, curp
,
521 /* Trailing backslash is like a plain character. */
522 if ('\\' != blk
.buf
[i
] || i
+ 1 == (int)blk
.sz
) {
523 if (pos
>= (int)ln
.sz
)
524 if (! resize_buf(&ln
, 256))
526 ln
.buf
[pos
++] = blk
.buf
[i
++];
529 /* Found an escape and at least one other character. */
530 if ('\n' == blk
.buf
[i
+ 1]) {
531 /* Escaped newlines are skipped over */
536 if ('"' == blk
.buf
[i
+ 1]) {
538 /* Comment, skip to end of line */
539 for (; i
< (int)blk
.sz
; ++i
) {
540 if ('\n' == blk
.buf
[i
]) {
546 /* Backout trailing whitespaces */
547 for (; pos
> 0; --pos
) {
548 if (ln
.buf
[pos
- 1] != ' ')
550 if (pos
> 2 && ln
.buf
[pos
- 2] == '\\')
555 /* Some other escape sequence, copy and continue. */
556 if (pos
+ 1 >= (int)ln
.sz
)
557 if (! resize_buf(&ln
, 256))
560 ln
.buf
[pos
++] = blk
.buf
[i
++];
561 ln
.buf
[pos
++] = blk
.buf
[i
++];
564 if (pos
>= (int)ln
.sz
)
565 if (! resize_buf(&ln
, 256))
570 * A significant amount of complexity is contained by
571 * the roff preprocessor. It's line-oriented but can be
572 * expressed on one line, so we need at times to
573 * readjust our starting point and re-run it. The roff
574 * preprocessor can also readjust the buffers with new
575 * data, so we pass them in wholesale.
580 re
= roff_parseln(roff
, lnn_start
,
581 &ln
.buf
, &ln
.sz
, of
, &of
);
582 } while (ROFF_RERUN
== re
);
586 else if (ROFF_ERR
== re
)
590 * If input parsers have not been allocated, do so now.
591 * We keep these instanced betwen parsers, but set them
592 * locally per parse routine since we can use different
593 * parsers with each one.
596 if ( ! (man
|| mdoc
))
597 if ( ! pset(ln
.buf
+ of
, pos
- of
, curp
, &man
, &mdoc
))
600 /* Lastly, push down into the parsers themselves. */
602 if (man
&& ! man_parseln(man
, lnn_start
, ln
.buf
, of
))
604 if (mdoc
&& ! mdoc_parseln(mdoc
, lnn_start
, ln
.buf
, of
))
608 /* NOTE a parser may not have been assigned, yet. */
610 if ( ! (man
|| mdoc
)) {
611 fprintf(stderr
, "%s: Not a manual\n", curp
->file
);
615 /* Clean up the parse routine ASTs. */
617 if (mdoc
&& ! mdoc_endparse(mdoc
))
619 if (man
&& ! man_endparse(man
))
621 if (roff
&& ! roff_endparse(roff
))
624 /* If unset, allocate output dev now (if applicable). */
626 if ( ! (curp
->outman
&& curp
->outmdoc
)) {
627 switch (curp
->outtype
) {
629 curp
->outdata
= xhtml_alloc(curp
->outopts
);
632 curp
->outdata
= html_alloc(curp
->outopts
);
635 curp
->outdata
= ascii_alloc(curp
->outopts
);
636 curp
->outfree
= ascii_free
;
639 curp
->outdata
= pdf_alloc(curp
->outopts
);
640 curp
->outfree
= pspdf_free
;
643 curp
->outdata
= ps_alloc(curp
->outopts
);
644 curp
->outfree
= pspdf_free
;
650 switch (curp
->outtype
) {
654 curp
->outman
= html_man
;
655 curp
->outmdoc
= html_mdoc
;
656 curp
->outfree
= html_free
;
659 curp
->outman
= tree_man
;
660 curp
->outmdoc
= tree_mdoc
;
667 curp
->outman
= terminal_man
;
668 curp
->outmdoc
= terminal_mdoc
;
675 /* Execute the out device, if it exists. */
677 if (man
&& curp
->outman
)
678 (*curp
->outman
)(curp
->outdata
, man
);
679 if (mdoc
&& curp
->outmdoc
)
680 (*curp
->outmdoc
)(curp
->outdata
, mdoc
);
683 memset(&curp
->regs
, 0, sizeof(struct regset
));
693 munmap(blk
.buf
, blk
.sz
);
706 pset(const char *buf
, int pos
, struct curparse
*curp
,
707 struct man
**man
, struct mdoc
**mdoc
)
712 * Try to intuit which kind of manual parser should be used. If
713 * passed in by command-line (-man, -mdoc), then use that
714 * explicitly. If passed as -mandoc, then try to guess from the
715 * line: either skip dot-lines, use -mdoc when finding `.Dt', or
716 * default to -man, which is more lenient.
719 if ('.' == buf
[0] || '\'' == buf
[0]) {
720 for (i
= 1; buf
[i
]; i
++)
721 if (' ' != buf
[i
] && '\t' != buf
[i
])
727 switch (curp
->inttype
) {
729 if (NULL
== curp
->mdoc
)
730 curp
->mdoc
= mdoc_init(curp
);
731 if (NULL
== (*mdoc
= curp
->mdoc
))
735 if (NULL
== curp
->man
)
736 curp
->man
= man_init(curp
);
737 if (NULL
== (*man
= curp
->man
))
744 if (pos
>= 3 && 0 == memcmp(buf
, ".Dd", 3)) {
745 if (NULL
== curp
->mdoc
)
746 curp
->mdoc
= mdoc_init(curp
);
747 if (NULL
== (*mdoc
= curp
->mdoc
))
752 if (NULL
== curp
->man
)
753 curp
->man
= man_init(curp
);
754 if (NULL
== (*man
= curp
->man
))
761 moptions(enum intt
*tflags
, char *arg
)
764 if (0 == strcmp(arg
, "doc"))
766 else if (0 == strcmp(arg
, "andoc"))
768 else if (0 == strcmp(arg
, "an"))
771 fprintf(stderr
, "%s: Bad argument\n", arg
);
780 toptions(struct curparse
*curp
, char *arg
)
783 if (0 == strcmp(arg
, "ascii"))
784 curp
->outtype
= OUTT_ASCII
;
785 else if (0 == strcmp(arg
, "lint")) {
786 curp
->outtype
= OUTT_LINT
;
787 curp
->wflags
|= WARN_WALL
;
788 curp
->fflags
|= FL_STRICT
;
790 else if (0 == strcmp(arg
, "tree"))
791 curp
->outtype
= OUTT_TREE
;
792 else if (0 == strcmp(arg
, "html"))
793 curp
->outtype
= OUTT_HTML
;
794 else if (0 == strcmp(arg
, "xhtml"))
795 curp
->outtype
= OUTT_XHTML
;
796 else if (0 == strcmp(arg
, "ps"))
797 curp
->outtype
= OUTT_PS
;
798 else if (0 == strcmp(arg
, "pdf"))
799 curp
->outtype
= OUTT_PDF
;
801 fprintf(stderr
, "%s: Bad argument\n", arg
);
810 foptions(int *fflags
, char *arg
)
815 toks
[0] = "ign-scope";
816 toks
[1] = "no-ign-escape";
817 toks
[2] = "no-ign-macro";
818 toks
[3] = "ign-errors";
820 toks
[5] = "ign-escape";
825 switch (getsubopt(&arg
, UNCONST(toks
), &v
)) {
827 *fflags
|= FL_IGN_SCOPE
;
830 *fflags
|= FL_NIGN_ESCAPE
;
833 *fflags
|= FL_NIGN_MACRO
;
836 *fflags
|= FL_IGN_ERRORS
;
839 *fflags
|= FL_STRICT
;
842 *fflags
&= ~FL_NIGN_ESCAPE
;
845 fprintf(stderr
, "%s: Bad argument\n", o
);
855 woptions(int *wflags
, char *arg
)
866 switch (getsubopt(&arg
, UNCONST(toks
), &v
)) {
868 *wflags
|= WARN_WALL
;
871 *wflags
|= WARN_WERR
;
874 fprintf(stderr
, "%s: Bad argument\n", o
);
884 mmsg(enum mandocerr t
, void *arg
, int ln
, int col
, const char *msg
)
890 cp
= (struct curparse
*)arg
;
894 if (t
>= MANDOCERR_FATAL
) {
899 if ( ! (WARN_WALL
& cp
->wflags
))
901 if (t
>= MANDOCERR_ERROR
) {
905 if (WARN_WERR
& cp
->wflags
) {
911 fprintf(stderr
, "%s:%d:%d:", cp
->file
, ln
, col
+ 1);
913 fprintf(stderr
, " %s:", level
);
914 fprintf(stderr
, " %s", mandocerrs
[t
]);
916 fprintf(stderr
, ": %s", msg
);