version 1.27, 2009/06/11 12:07:49 |
version 1.87, 2010/06/12 10:09:19 |
|
|
* ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF |
* ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF |
* OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE. |
* OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE. |
*/ |
*/ |
|
#ifdef HAVE_CONFIG_H |
|
#include "config.h" |
|
#endif |
|
|
|
#include <sys/mman.h> |
#include <sys/stat.h> |
#include <sys/stat.h> |
|
|
#include <assert.h> |
#include <assert.h> |
#include <err.h> |
|
#include <fcntl.h> |
#include <fcntl.h> |
#include <stdio.h> |
#include <stdio.h> |
|
#include <stdint.h> |
#include <stdlib.h> |
#include <stdlib.h> |
#include <string.h> |
#include <string.h> |
#include <unistd.h> |
#include <unistd.h> |
|
|
|
#include "mandoc.h" |
#include "mdoc.h" |
#include "mdoc.h" |
#include "man.h" |
#include "man.h" |
|
#include "roff.h" |
|
#include "main.h" |
|
|
/* Account for FreeBSD and Linux in our declarations. */ |
#define UNCONST(a) ((void *)(uintptr_t)(const void *)(a)) |
|
|
#ifdef __linux__ |
/* FIXME: Intel's compiler? LLVM? pcc? */ |
extern int getsubopt(char **, char * const *, char **); |
|
# ifndef __dead |
#if !defined(__GNUC__) || (__GNUC__ < 2) |
# define __dead __attribute__((__noreturn__)) |
# if !defined(lint) |
|
# define __attribute__(x) |
# endif |
# endif |
#elif defined(__dead2) |
#endif /* !defined(__GNUC__) || (__GNUC__ < 2) */ |
# ifndef __dead |
|
# define __dead __dead2 |
|
# endif |
|
#endif |
|
|
|
typedef int (*out_mdoc)(void *, const struct mdoc *); |
typedef void (*out_mdoc)(void *, const struct mdoc *); |
typedef int (*out_man)(void *, const struct man *); |
typedef void (*out_man)(void *, const struct man *); |
typedef void (*out_free)(void *); |
typedef void (*out_free)(void *); |
|
|
struct buf { |
struct buf { |
|
|
enum outt { |
enum outt { |
OUTT_ASCII = 0, |
OUTT_ASCII = 0, |
OUTT_TREE, |
OUTT_TREE, |
OUTT_LINT |
OUTT_HTML, |
|
OUTT_XHTML, |
|
OUTT_LINT, |
|
OUTT_PS |
}; |
}; |
|
|
struct curparse { |
struct curparse { |
const char *file; /* Current parse. */ |
const char *file; /* Current parse. */ |
int fd; /* Current parse. */ |
int fd; /* Current parse. */ |
int wflags; |
int wflags; |
#define WARN_WALL 0x03 /* All-warnings mask. */ |
/* FIXME: set by max error */ |
#define WARN_WCOMPAT (1 << 0) /* Compatibility warnings. */ |
#define WARN_WALL (1 << 0) /* All-warnings mask. */ |
#define WARN_WSYNTAX (1 << 1) /* Syntax warnings. */ |
|
#define WARN_WERR (1 << 2) /* Warnings->errors. */ |
#define WARN_WERR (1 << 2) /* Warnings->errors. */ |
int fflags; |
int fflags; |
#define IGN_SCOPE (1 << 0) /* Ignore scope errors. */ |
#define FL_IGN_SCOPE (1 << 0) /* Ignore scope errors. */ |
#define NO_IGN_ESCAPE (1 << 1) /* Don't ignore bad escapes. */ |
#define FL_NIGN_ESCAPE (1 << 1) /* Don't ignore bad escapes. */ |
#define NO_IGN_MACRO (1 << 2) /* Don't ignore bad macros. */ |
#define FL_NIGN_MACRO (1 << 2) /* Don't ignore bad macros. */ |
#define NO_IGN_CHARS (1 << 3) /* Don't ignore bad chars. */ |
#define FL_IGN_ERRORS (1 << 4) /* Ignore failed parse. */ |
enum intt inttype; /* Input parsers. */ |
#define FL_STRICT FL_NIGN_ESCAPE | \ |
struct man *man; |
FL_NIGN_MACRO /* ignore nothing */ |
struct man *lastman; |
enum intt inttype; /* which parser to use */ |
struct mdoc *mdoc; |
struct man *man; /* man parser */ |
struct mdoc *lastmdoc; |
struct mdoc *mdoc; /* mdoc parser */ |
enum outt outtype; /* Output devices. */ |
struct roff *roff; /* roff parser (!NULL) */ |
out_mdoc outmdoc; |
enum outt outtype; /* which output to use */ |
out_man outman; |
out_mdoc outmdoc; /* mdoc output ptr */ |
out_free outfree; |
out_man outman; /* man output ptr */ |
void *outdata; |
out_free outfree; /* free output ptr */ |
|
void *outdata; /* data for output */ |
|
char outopts[BUFSIZ]; /* buf of output opts */ |
}; |
}; |
|
|
extern void *ascii_alloc(void); |
static const char * const mandocerrs[MANDOCERR_MAX] = { |
extern int tree_mdoc(void *, const struct mdoc *); |
"ok", |
extern int tree_man(void *, const struct man *); |
"text should be uppercase", |
extern int terminal_mdoc(void *, const struct mdoc *); |
"sections out of conventional order", |
extern int terminal_man(void *, const struct man *); |
"section name repeats", |
extern void terminal_free(void *); |
"out of order prologue", |
|
"repeated prologue entry", |
|
"list type must come first", |
|
"bad standard", |
|
"bad library", |
|
"bad escape sequence", |
|
"unterminated quoted string", |
|
"argument requires the width argument", |
|
"superfluous width argument", |
|
"bad date argument", |
|
"bad width argument", |
|
"unknown manual section", |
|
"section not in conventional manual section", |
|
"end of line whitespace", |
|
"scope open on exit", |
|
"NAME section must come first", |
|
"bad Boolean value", |
|
"child violates parent syntax", |
|
"bad AT&T symbol", |
|
"list type repeated", |
|
"display type repeated", |
|
"argument repeated", |
|
"manual name not yet set", |
|
"obsolete macro ignored", |
|
"empty macro ignored", |
|
"macro not allowed in body", |
|
"macro not allowed in prologue", |
|
"bad character", |
|
"bad NAME section contents", |
|
"no blank lines", |
|
"no text in this context", |
|
"bad comment style", |
|
"unknown macro will be lost", |
|
"line scope broken", |
|
"scope broken", |
|
"argument count wrong", |
|
"request scope close w/none open", |
|
"scope already open", |
|
"macro requires line argument(s)", |
|
"macro requires body argument(s)", |
|
"macro requires argument(s)", |
|
"no title in document", |
|
"missing list type", |
|
"missing display type", |
|
"line argument(s) will be lost", |
|
"body argument(s) will be lost", |
|
"column syntax is inconsistent", |
|
"missing font type", |
|
"displays may not be nested", |
|
"unsupported display type", |
|
"no scope to rewind: syntax violated", |
|
"scope broken, syntax violated", |
|
"line scope broken, syntax violated", |
|
"argument count wrong, violates syntax", |
|
"child violates parent syntax", |
|
"argument count wrong, violates syntax", |
|
"no document body", |
|
"no document prologue", |
|
"utsname system call failed", |
|
"memory exhausted", |
|
}; |
|
|
|
static void fdesc(struct curparse *); |
|
static void ffile(const char *, struct curparse *); |
static int foptions(int *, char *); |
static int foptions(int *, char *); |
static int toptions(enum outt *, char *); |
static struct man *man_init(struct curparse *); |
|
static struct mdoc *mdoc_init(struct curparse *); |
|
static struct roff *roff_init(struct curparse *); |
static int moptions(enum intt *, char *); |
static int moptions(enum intt *, char *); |
static int woptions(int *, char *); |
static int mmsg(enum mandocerr, void *, |
static int merr(void *, int, int, const char *); |
int, int, const char *); |
static int manwarn(void *, int, int, const char *); |
|
static int mdocwarn(void *, int, int, |
|
enum mdoc_warn, const char *); |
|
static int fstdin(struct buf *, struct buf *, |
|
struct curparse *); |
|
static int ffile(struct buf *, struct buf *, |
|
const char *, struct curparse *); |
|
static int fdesc(struct buf *, struct buf *, |
|
struct curparse *); |
|
static int pset(const char *, int, struct curparse *, |
static int pset(const char *, int, struct curparse *, |
struct man **, struct mdoc **); |
struct man **, struct mdoc **); |
static struct man *man_init(struct curparse *); |
static int toptions(struct curparse *, char *); |
static struct mdoc *mdoc_init(struct curparse *); |
static void usage(void) __attribute__((noreturn)); |
__dead static void version(void); |
static void version(void) __attribute__((noreturn)); |
__dead static void usage(void); |
static int woptions(int *, char *); |
|
|
extern char *__progname; |
static const char *progname; |
|
static int with_error; |
|
static int with_warning; |
|
|
|
|
int |
int |
main(int argc, char *argv[]) |
main(int argc, char *argv[]) |
{ |
{ |
int c, rc; |
int c; |
struct buf ln, blk; |
|
struct curparse curp; |
struct curparse curp; |
|
|
bzero(&curp, sizeof(struct curparse)); |
progname = strrchr(argv[0], '/'); |
|
if (progname == NULL) |
|
progname = argv[0]; |
|
else |
|
++progname; |
|
|
|
memset(&curp, 0, sizeof(struct curparse)); |
|
|
curp.inttype = INTT_AUTO; |
curp.inttype = INTT_AUTO; |
curp.outtype = OUTT_ASCII; |
curp.outtype = OUTT_ASCII; |
|
|
/* LINTED */ |
/* LINTED */ |
while (-1 != (c = getopt(argc, argv, "f:m:VW:T:"))) |
while (-1 != (c = getopt(argc, argv, "f:m:O:T:VW:"))) |
switch (c) { |
switch (c) { |
case ('f'): |
case ('f'): |
if ( ! foptions(&curp.fflags, optarg)) |
if ( ! foptions(&curp.fflags, optarg)) |
return(0); |
return(EXIT_FAILURE); |
break; |
break; |
case ('m'): |
case ('m'): |
if ( ! moptions(&curp.inttype, optarg)) |
if ( ! moptions(&curp.inttype, optarg)) |
return(0); |
return(EXIT_FAILURE); |
break; |
break; |
|
case ('O'): |
|
(void)strlcat(curp.outopts, optarg, BUFSIZ); |
|
(void)strlcat(curp.outopts, ",", BUFSIZ); |
|
break; |
case ('T'): |
case ('T'): |
if ( ! toptions(&curp.outtype, optarg)) |
if ( ! toptions(&curp, optarg)) |
return(0); |
return(EXIT_FAILURE); |
break; |
break; |
case ('W'): |
case ('W'): |
if ( ! woptions(&curp.wflags, optarg)) |
if ( ! woptions(&curp.wflags, optarg)) |
return(0); |
return(EXIT_FAILURE); |
break; |
break; |
case ('V'): |
case ('V'): |
version(); |
version(); |
Line 159 main(int argc, char *argv[]) |
|
Line 233 main(int argc, char *argv[]) |
|
argc -= optind; |
argc -= optind; |
argv += optind; |
argv += optind; |
|
|
/* Configure buffers. */ |
if (NULL == *argv) { |
|
curp.file = "<stdin>"; |
|
curp.fd = STDIN_FILENO; |
|
|
bzero(&ln, sizeof(struct buf)); |
fdesc(&curp); |
bzero(&blk, sizeof(struct buf)); |
} |
|
|
rc = 1; |
while (*argv) { |
|
ffile(*argv, &curp); |
|
|
if (NULL == *argv) |
if (with_error && !(curp.fflags & FL_IGN_ERRORS)) |
if ( ! fstdin(&blk, &ln, &curp)) |
break; |
rc = 0; |
++argv; |
|
|
while (rc && *argv) { |
|
if ( ! ffile(&blk, &ln, *argv, &curp)) |
|
rc = 0; |
|
argv++; |
|
if (*argv && rc) { |
|
if (curp.lastman) |
|
if ( ! man_reset(curp.lastman)) |
|
rc = 0; |
|
if (curp.lastmdoc) |
|
if ( ! mdoc_reset(curp.lastmdoc)) |
|
rc = 0; |
|
curp.lastman = NULL; |
|
curp.lastmdoc = NULL; |
|
} |
|
} |
} |
|
|
if (blk.buf) |
|
free(blk.buf); |
|
if (ln.buf) |
|
free(ln.buf); |
|
|
|
/* TODO: have a curp_free routine. */ |
|
if (curp.outfree) |
if (curp.outfree) |
(*curp.outfree)(curp.outdata); |
(*curp.outfree)(curp.outdata); |
if (curp.mdoc) |
if (curp.mdoc) |
mdoc_free(curp.mdoc); |
mdoc_free(curp.mdoc); |
if (curp.man) |
if (curp.man) |
man_free(curp.man); |
man_free(curp.man); |
|
if (curp.roff) |
|
roff_free(curp.roff); |
|
|
return(rc ? EXIT_SUCCESS : EXIT_FAILURE); |
return((with_warning || with_error) ? |
|
EXIT_FAILURE : EXIT_SUCCESS); |
} |
} |
|
|
|
|
__dead static void |
static void |
version(void) |
version(void) |
{ |
{ |
|
|
(void)printf("%s %s\n", __progname, VERSION); |
(void)printf("%s %s\n", progname, VERSION); |
exit(EXIT_SUCCESS); |
exit(EXIT_SUCCESS); |
} |
} |
|
|
|
|
__dead static void |
static void |
usage(void) |
usage(void) |
{ |
{ |
|
|
(void)fprintf(stderr, "usage: %s [-V] [-foption...] " |
(void)fprintf(stderr, "usage: %s [-V] [-foption] " |
"[-mformat] [-Toutput] [-Werr...]\n", |
"[-mformat] [-Ooption] [-Toutput] " |
__progname); |
"[-Werr] [file...]\n", progname); |
exit(EXIT_FAILURE); |
exit(EXIT_FAILURE); |
} |
} |
|
|
Line 227 static struct man * |
|
Line 286 static struct man * |
|
man_init(struct curparse *curp) |
man_init(struct curparse *curp) |
{ |
{ |
int pflags; |
int pflags; |
struct man *man; |
|
struct man_cb mancb; |
|
|
|
mancb.man_err = merr; |
/* Defaults from mandoc.1. */ |
mancb.man_warn = manwarn; |
|
|
|
/* |
pflags = MAN_IGN_MACRO | MAN_IGN_ESCAPE; |
* Default behaviour is to ignore unknown macros. This is |
|
* specified in mandoc.1. |
|
*/ |
|
|
|
pflags = MAN_IGN_MACRO; |
if (curp->fflags & FL_NIGN_MACRO) |
|
pflags &= ~MAN_IGN_MACRO; |
|
if (curp->fflags & FL_NIGN_ESCAPE) |
|
pflags &= ~MAN_IGN_ESCAPE; |
|
|
/* Override default behaviour... */ |
return(man_alloc(curp, pflags, mmsg)); |
|
} |
|
|
if (curp->fflags & NO_IGN_MACRO) |
|
pflags &= ~MAN_IGN_MACRO; |
|
|
|
if (NULL == (man = man_alloc(curp, pflags, &mancb))) |
static struct roff * |
warnx("memory exhausted"); |
roff_init(struct curparse *curp) |
|
{ |
|
|
return(man); |
return(roff_alloc(mmsg, curp)); |
} |
} |
|
|
|
|
Line 256 static struct mdoc * |
|
Line 312 static struct mdoc * |
|
mdoc_init(struct curparse *curp) |
mdoc_init(struct curparse *curp) |
{ |
{ |
int pflags; |
int pflags; |
struct mdoc *mdoc; |
|
struct mdoc_cb mdoccb; |
|
|
|
mdoccb.mdoc_msg = NULL; |
/* Defaults from mandoc.1. */ |
mdoccb.mdoc_err = merr; |
|
mdoccb.mdoc_warn = mdocwarn; |
|
|
|
/* |
pflags = MDOC_IGN_MACRO | MDOC_IGN_ESCAPE; |
* Default behaviour is to ignore unknown macros, escape |
|
* sequences and characters (very liberal). This is specified |
|
* in mandoc.1. |
|
*/ |
|
|
|
pflags = MDOC_IGN_MACRO | MDOC_IGN_ESCAPE | MDOC_IGN_CHARS; |
if (curp->fflags & FL_IGN_SCOPE) |
|
|
/* Override default behaviour... */ |
|
|
|
if (curp->fflags & IGN_SCOPE) |
|
pflags |= MDOC_IGN_SCOPE; |
pflags |= MDOC_IGN_SCOPE; |
if (curp->fflags & NO_IGN_ESCAPE) |
if (curp->fflags & FL_NIGN_ESCAPE) |
pflags &= ~MDOC_IGN_ESCAPE; |
pflags &= ~MDOC_IGN_ESCAPE; |
if (curp->fflags & NO_IGN_MACRO) |
if (curp->fflags & FL_NIGN_MACRO) |
pflags &= ~MDOC_IGN_MACRO; |
pflags &= ~MDOC_IGN_MACRO; |
if (curp->fflags & NO_IGN_CHARS) |
|
pflags &= ~MDOC_IGN_CHARS; |
|
|
|
if (NULL == (mdoc = mdoc_alloc(curp, pflags, &mdoccb))) |
return(mdoc_alloc(curp, pflags, mmsg)); |
warnx("memory exhausted"); |
} |
|
|
return(mdoc); |
|
|
static void |
|
ffile(const char *file, struct curparse *curp) |
|
{ |
|
|
|
curp->file = file; |
|
if (-1 == (curp->fd = open(curp->file, O_RDONLY, 0))) { |
|
perror(curp->file); |
|
with_error = 1; |
|
return; |
|
} |
|
|
|
fdesc(curp); |
|
|
|
if (-1 == close(curp->fd)) |
|
perror(curp->file); |
} |
} |
|
|
|
|
static int |
static int |
fstdin(struct buf *blk, struct buf *ln, struct curparse *curp) |
resize_buf(struct buf *buf, size_t initial) |
{ |
{ |
|
void *tmp; |
|
size_t sz; |
|
|
curp->file = "<stdin>"; |
if (buf->sz == 0) |
curp->fd = STDIN_FILENO; |
sz = initial; |
return(fdesc(blk, ln, curp)); |
else |
|
sz = 2 * buf->sz; |
|
tmp = realloc(buf->buf, sz); |
|
if (NULL == tmp) { |
|
perror(NULL); |
|
return(0); |
|
} |
|
buf->buf = tmp; |
|
buf->sz = sz; |
|
return(1); |
} |
} |
|
|
|
|
static int |
static int |
ffile(struct buf *blk, struct buf *ln, |
read_whole_file(struct curparse *curp, struct buf *fb, int *with_mmap) |
const char *file, struct curparse *curp) |
|
{ |
{ |
int c; |
struct stat st; |
|
size_t off; |
|
ssize_t ssz; |
|
|
curp->file = file; |
if (-1 == fstat(curp->fd, &st)) { |
if (-1 == (curp->fd = open(curp->file, O_RDONLY, 0))) { |
perror(curp->file); |
warn("%s", curp->file); |
with_error = 1; |
return(0); |
return(0); |
} |
} |
|
|
c = fdesc(blk, ln, curp); |
/* |
|
* If we're a regular file, try just reading in the whole entry |
|
* via mmap(). This is faster than reading it into blocks, and |
|
* since each file is only a few bytes to begin with, I'm not |
|
* concerned that this is going to tank any machines. |
|
*/ |
|
|
if (-1 == close(curp->fd)) |
if (S_ISREG(st.st_mode)) { |
warn("%s", curp->file); |
if (st.st_size >= (1U << 31)) { |
|
fprintf(stderr, "%s: input too large\n", |
|
curp->file); |
|
with_error = 1; |
|
return(0); |
|
} |
|
*with_mmap = 1; |
|
fb->sz = (size_t)st.st_size; |
|
fb->buf = mmap(NULL, fb->sz, PROT_READ, |
|
MAP_FILE|MAP_SHARED, curp->fd, 0); |
|
if (fb->buf != MAP_FAILED) |
|
return(1); |
|
} |
|
|
return(c); |
/* |
|
* If this isn't a regular file (like, say, stdin), then we must |
|
* go the old way and just read things in bit by bit. |
|
*/ |
|
|
|
*with_mmap = 0; |
|
off = 0; |
|
fb->sz = 0; |
|
fb->buf = NULL; |
|
for (;;) { |
|
if (off == fb->sz) { |
|
if (fb->sz == (1U << 31)) { |
|
fprintf(stderr, "%s: input too large\n", |
|
curp->file); |
|
break; |
|
} |
|
if (! resize_buf(fb, 65536)) |
|
break; |
|
} |
|
ssz = read(curp->fd, fb->buf + (int)off, fb->sz - off); |
|
if (ssz == 0) { |
|
fb->sz = off; |
|
return(1); |
|
} |
|
if (ssz == -1) { |
|
perror(curp->file); |
|
break; |
|
} |
|
off += (size_t)ssz; |
|
} |
|
|
|
free(fb->buf); |
|
fb->buf = NULL; |
|
with_error = 1; |
|
return(0); |
} |
} |
|
|
|
|
static int |
static void |
fdesc(struct buf *blk, struct buf *ln, struct curparse *curp) |
fdesc(struct curparse *curp) |
{ |
{ |
size_t sz; |
struct buf ln, blk; |
ssize_t ssz; |
int i, pos, lnn, lnn_start, with_mmap, of; |
struct stat st; |
enum rofferr re; |
int j, i, pos, lnn; |
|
struct man *man; |
struct man *man; |
struct mdoc *mdoc; |
struct mdoc *mdoc; |
|
struct roff *roff; |
|
|
sz = BUFSIZ; |
|
man = NULL; |
man = NULL; |
mdoc = NULL; |
mdoc = NULL; |
|
roff = NULL; |
|
memset(&ln, 0, sizeof(struct buf)); |
|
|
/* |
/* |
* Two buffers: ln and buf. buf is the input buffer optimised |
* Two buffers: ln and buf. buf is the input file and may be |
* here for each file's block size. ln is a line buffer. Both |
* memory mapped. ln is a line buffer and grows on-demand. |
* growable, hence passed in by ptr-ptr. |
|
*/ |
*/ |
|
|
if (-1 == fstat(curp->fd, &st)) |
if ( ! read_whole_file(curp, &blk, &with_mmap)) |
warnx("%s", curp->file); |
return; |
else if ((size_t)st.st_blksize > sz) |
|
sz = st.st_blksize; |
|
|
|
if (sz > blk->sz) { |
if (NULL == curp->roff) |
blk->buf = realloc(blk->buf, sz); |
curp->roff = roff_init(curp); |
if (NULL == blk->buf) { |
if (NULL == (roff = curp->roff)) |
warn("realloc"); |
goto bailout; |
return(0); |
|
} |
|
blk->sz = sz; |
|
} |
|
|
|
/* Fill buf with file blocksize. */ |
for (i = 0, lnn = 1; i < (int)blk.sz;) { |
|
pos = 0; |
for (lnn = 0, pos = 0; ; ) { |
lnn_start = lnn; |
if (-1 == (ssz = read(curp->fd, blk->buf, sz))) { |
while (i < (int)blk.sz) { |
warn("%s", curp->file); |
if ('\n' == blk.buf[i]) { |
return(0); |
++i; |
} else if (0 == ssz) |
++lnn; |
break; |
break; |
|
|
/* Parse the read block into partial or full lines. */ |
|
|
|
for (i = 0; i < (int)ssz; i++) { |
|
if (pos >= (int)ln->sz) { |
|
ln->sz += 256; /* Step-size. */ |
|
ln->buf = realloc(ln->buf, ln->sz); |
|
if (NULL == ln->buf) { |
|
warn("realloc"); |
|
return(0); |
|
} |
|
} |
} |
|
/* Trailing backslash is like a plain character. */ |
if ('\n' != blk->buf[i]) { |
if ('\\' != blk.buf[i] || i + 1 == (int)blk.sz) { |
ln->buf[pos++] = blk->buf[i]; |
if (pos >= (int)ln.sz) |
|
if (! resize_buf(&ln, 256)) |
|
goto bailout; |
|
ln.buf[pos++] = blk.buf[i++]; |
continue; |
continue; |
} |
} |
|
/* Found an escape and at least one other character. */ |
/* Check for CPP-escaped newline. */ |
if ('\n' == blk.buf[i + 1]) { |
|
/* Escaped newlines are skipped over */ |
if (pos > 0 && '\\' == ln->buf[pos - 1]) { |
i += 2; |
for (j = pos - 1; j >= 0; j--) |
++lnn; |
if ('\\' != ln->buf[j]) |
continue; |
|
} |
|
if ('"' == blk.buf[i + 1]) { |
|
i += 2; |
|
/* Comment, skip to end of line */ |
|
for (; i < (int)blk.sz; ++i) { |
|
if ('\n' == blk.buf[i]) { |
|
++i; |
|
++lnn; |
break; |
break; |
|
} |
if ( ! ((pos - j) % 2)) { |
|
pos--; |
|
lnn++; |
|
continue; |
|
} |
} |
|
/* Backout trailing whitespaces */ |
|
for (; pos > 0; --pos) { |
|
if (ln.buf[pos - 1] != ' ') |
|
break; |
|
if (pos > 2 && ln.buf[pos - 2] == '\\') |
|
break; |
|
} |
|
break; |
} |
} |
|
/* Some other escape sequence, copy and continue. */ |
|
if (pos + 1 >= (int)ln.sz) |
|
if (! resize_buf(&ln, 256)) |
|
goto bailout; |
|
|
ln->buf[pos] = 0; |
ln.buf[pos++] = blk.buf[i++]; |
lnn++; |
ln.buf[pos++] = blk.buf[i++]; |
|
} |
/* |
|
* If no manual parser has been assigned, then |
|
* try to assign one in pset(), which may do |
|
* nothing at all. After this, parse the manual |
|
* line accordingly. |
|
*/ |
|
|
|
if ( ! (man || mdoc) && ! pset(ln->buf, |
if (pos >= (int)ln.sz) |
pos, curp, &man, &mdoc)) |
if (! resize_buf(&ln, 256)) |
return(0); |
goto bailout; |
|
ln.buf[pos] = '\0'; |
|
|
pos = 0; |
/* |
|
* A significant amount of complexity is contained by |
|
* the roff preprocessor. It's line-oriented but can be |
|
* expressed on one line, so we need at times to |
|
* readjust our starting point and re-run it. The roff |
|
* preprocessor can also readjust the buffers with new |
|
* data, so we pass them in wholesale. |
|
*/ |
|
|
if (man && ! man_parseln(man, lnn, ln->buf)) |
of = 0; |
return(0); |
do { |
if (mdoc && ! mdoc_parseln(mdoc, lnn, ln->buf)) |
re = roff_parseln(roff, lnn_start, |
return(0); |
&ln.buf, &ln.sz, of, &of); |
} |
} while (ROFF_RERUN == re); |
|
|
|
if (ROFF_IGN == re) |
|
continue; |
|
else if (ROFF_ERR == re) |
|
goto bailout; |
|
|
|
/* |
|
* If input parsers have not been allocated, do so now. |
|
* We keep these instanced betwen parsers, but set them |
|
* locally per parse routine since we can use different |
|
* parsers with each one. |
|
*/ |
|
|
|
if ( ! (man || mdoc)) |
|
if ( ! pset(ln.buf + of, pos - of, curp, &man, &mdoc)) |
|
goto bailout; |
|
|
|
/* Lastly, push down into the parsers themselves. */ |
|
|
|
if (man && ! man_parseln(man, lnn_start, ln.buf, of)) |
|
goto bailout; |
|
if (mdoc && ! mdoc_parseln(mdoc, lnn_start, ln.buf, of)) |
|
goto bailout; |
} |
} |
|
|
/* Note that a parser may not have been assigned, yet. */ |
/* NOTE a parser may not have been assigned, yet. */ |
|
|
if ( ! (man || mdoc)) { |
if ( ! (man || mdoc)) { |
warnx("%s: not a manual", curp->file); |
fprintf(stderr, "%s: Not a manual\n", curp->file); |
return(0); |
goto bailout; |
} |
} |
|
|
|
/* Clean up the parse routine ASTs. */ |
|
|
if (mdoc && ! mdoc_endparse(mdoc)) |
if (mdoc && ! mdoc_endparse(mdoc)) |
return(0); |
goto bailout; |
if (man && ! man_endparse(man)) |
if (man && ! man_endparse(man)) |
return(0); |
goto bailout; |
|
if (roff && ! roff_endparse(roff)) |
|
goto bailout; |
|
|
/* |
/* If unset, allocate output dev now (if applicable). */ |
* If an output device hasn't been allocated, see if we should |
|
* do so now. Note that not all outtypes have functions, so |
|
* this switch statement may be superfluous, but it's |
|
* low-overhead enough not to matter very much. |
|
*/ |
|
|
|
if ( ! (curp->outman && curp->outmdoc)) { |
if ( ! (curp->outman && curp->outmdoc)) { |
switch (curp->outtype) { |
switch (curp->outtype) { |
|
case (OUTT_XHTML): |
|
curp->outdata = xhtml_alloc(curp->outopts); |
|
break; |
|
case (OUTT_HTML): |
|
curp->outdata = html_alloc(curp->outopts); |
|
break; |
|
case (OUTT_ASCII): |
|
curp->outdata = ascii_alloc(curp->outopts); |
|
curp->outfree = ascii_free; |
|
break; |
|
case (OUTT_PS): |
|
curp->outdata = ps_alloc(); |
|
curp->outfree = ps_free; |
|
break; |
|
default: |
|
break; |
|
} |
|
|
|
switch (curp->outtype) { |
|
case (OUTT_HTML): |
|
/* FALLTHROUGH */ |
|
case (OUTT_XHTML): |
|
curp->outman = html_man; |
|
curp->outmdoc = html_mdoc; |
|
curp->outfree = html_free; |
|
break; |
case (OUTT_TREE): |
case (OUTT_TREE): |
curp->outman = tree_man; |
curp->outman = tree_man; |
curp->outmdoc = tree_mdoc; |
curp->outmdoc = tree_mdoc; |
break; |
break; |
case (OUTT_LINT): |
case (OUTT_ASCII): |
break; |
/* FALLTHROUGH */ |
default: |
case (OUTT_PS): |
curp->outdata = ascii_alloc(); |
|
curp->outman = terminal_man; |
curp->outman = terminal_man; |
curp->outmdoc = terminal_mdoc; |
curp->outmdoc = terminal_mdoc; |
curp->outfree = terminal_free; |
|
break; |
break; |
|
default: |
|
break; |
} |
} |
} |
} |
|
|
/* Execute the out device, if it exists. */ |
/* Execute the out device, if it exists. */ |
|
|
if (man && curp->outman) |
if (man && curp->outman) |
if ( ! (*curp->outman)(curp->outdata, man)) |
(*curp->outman)(curp->outdata, man); |
return(0); |
|
if (mdoc && curp->outmdoc) |
if (mdoc && curp->outmdoc) |
if ( ! (*curp->outmdoc)(curp->outdata, mdoc)) |
(*curp->outmdoc)(curp->outdata, mdoc); |
return(0); |
|
|
|
return(1); |
cleanup: |
|
if (mdoc) |
|
mdoc_reset(mdoc); |
|
if (man) |
|
man_reset(man); |
|
if (roff) |
|
roff_reset(roff); |
|
if (ln.buf) |
|
free(ln.buf); |
|
if (with_mmap) |
|
munmap(blk.buf, blk.sz); |
|
else |
|
free(blk.buf); |
|
|
|
return; |
|
|
|
bailout: |
|
with_error = 1; |
|
goto cleanup; |
} |
} |
|
|
|
|
|
|
pset(const char *buf, int pos, struct curparse *curp, |
pset(const char *buf, int pos, struct curparse *curp, |
struct man **man, struct mdoc **mdoc) |
struct man **man, struct mdoc **mdoc) |
{ |
{ |
|
int i; |
|
|
/* |
/* |
* Try to intuit which kind of manual parser should be used. If |
* Try to intuit which kind of manual parser should be used. If |
* passed in by command-line (-man, -mdoc), then use that |
* passed in by command-line (-man, -mdoc), then use that |
* explicitly. If passed as -mandoc, then try to guess from the |
* explicitly. If passed as -mandoc, then try to guess from the |
* line: either skip comments, use -mdoc when finding `.Dt', or |
* line: either skip dot-lines, use -mdoc when finding `.Dt', or |
* default to -man, which is more lenient. |
* default to -man, which is more lenient. |
*/ |
*/ |
|
|
if (pos >= 3 && 0 == memcmp(buf, ".\\\"", 3)) |
if ('.' == buf[0] || '\'' == buf[0]) { |
return(1); |
for (i = 1; buf[i]; i++) |
|
if (' ' != buf[i] && '\t' != buf[i]) |
|
break; |
|
if (0 == buf[i]) |
|
return(1); |
|
} |
|
|
switch (curp->inttype) { |
switch (curp->inttype) { |
case (INTT_MDOC): |
case (INTT_MDOC): |
Line 488 pset(const char *buf, int pos, struct curparse *curp, |
|
Line 682 pset(const char *buf, int pos, struct curparse *curp, |
|
curp->mdoc = mdoc_init(curp); |
curp->mdoc = mdoc_init(curp); |
if (NULL == (*mdoc = curp->mdoc)) |
if (NULL == (*mdoc = curp->mdoc)) |
return(0); |
return(0); |
curp->lastmdoc = *mdoc; |
|
return(1); |
return(1); |
case (INTT_MAN): |
case (INTT_MAN): |
if (NULL == curp->man) |
if (NULL == curp->man) |
curp->man = man_init(curp); |
curp->man = man_init(curp); |
if (NULL == (*man = curp->man)) |
if (NULL == (*man = curp->man)) |
return(0); |
return(0); |
curp->lastman = *man; |
|
return(1); |
return(1); |
default: |
default: |
break; |
break; |
Line 506 pset(const char *buf, int pos, struct curparse *curp, |
|
Line 698 pset(const char *buf, int pos, struct curparse *curp, |
|
curp->mdoc = mdoc_init(curp); |
curp->mdoc = mdoc_init(curp); |
if (NULL == (*mdoc = curp->mdoc)) |
if (NULL == (*mdoc = curp->mdoc)) |
return(0); |
return(0); |
curp->lastmdoc = *mdoc; |
|
return(1); |
return(1); |
} |
} |
|
|
Line 514 pset(const char *buf, int pos, struct curparse *curp, |
|
Line 705 pset(const char *buf, int pos, struct curparse *curp, |
|
curp->man = man_init(curp); |
curp->man = man_init(curp); |
if (NULL == (*man = curp->man)) |
if (NULL == (*man = curp->man)) |
return(0); |
return(0); |
curp->lastman = *man; |
|
return(1); |
return(1); |
} |
} |
|
|
Line 530 moptions(enum intt *tflags, char *arg) |
|
Line 720 moptions(enum intt *tflags, char *arg) |
|
else if (0 == strcmp(arg, "an")) |
else if (0 == strcmp(arg, "an")) |
*tflags = INTT_MAN; |
*tflags = INTT_MAN; |
else { |
else { |
warnx("bad argument: -m%s", arg); |
fprintf(stderr, "%s: Bad argument\n", arg); |
return(0); |
return(0); |
} |
} |
|
|
Line 539 moptions(enum intt *tflags, char *arg) |
|
Line 729 moptions(enum intt *tflags, char *arg) |
|
|
|
|
|
static int |
static int |
toptions(enum outt *tflags, char *arg) |
toptions(struct curparse *curp, char *arg) |
{ |
{ |
|
|
if (0 == strcmp(arg, "ascii")) |
if (0 == strcmp(arg, "ascii")) |
*tflags = OUTT_ASCII; |
curp->outtype = OUTT_ASCII; |
else if (0 == strcmp(arg, "lint")) |
else if (0 == strcmp(arg, "lint")) { |
*tflags = OUTT_LINT; |
curp->outtype = OUTT_LINT; |
|
curp->wflags |= WARN_WALL; |
|
curp->fflags |= FL_STRICT; |
|
} |
else if (0 == strcmp(arg, "tree")) |
else if (0 == strcmp(arg, "tree")) |
*tflags = OUTT_TREE; |
curp->outtype = OUTT_TREE; |
|
else if (0 == strcmp(arg, "html")) |
|
curp->outtype = OUTT_HTML; |
|
else if (0 == strcmp(arg, "xhtml")) |
|
curp->outtype = OUTT_XHTML; |
|
else if (0 == strcmp(arg, "ps")) |
|
curp->outtype = OUTT_PS; |
else { |
else { |
warnx("bad argument: -T%s", arg); |
fprintf(stderr, "%s: Bad argument\n", arg); |
return(0); |
return(0); |
} |
} |
|
|
Line 557 toptions(enum outt *tflags, char *arg) |
|
Line 756 toptions(enum outt *tflags, char *arg) |
|
} |
} |
|
|
|
|
/* |
|
* Parse out the options for [-fopt...] setting compiler options. These |
|
* can be comma-delimited or called again. |
|
*/ |
|
static int |
static int |
foptions(int *fflags, char *arg) |
foptions(int *fflags, char *arg) |
{ |
{ |
char *v; |
char *v, *o; |
char *toks[6]; |
const char *toks[8]; |
|
|
toks[0] = "ign-scope"; |
toks[0] = "ign-scope"; |
toks[1] = "no-ign-escape"; |
toks[1] = "no-ign-escape"; |
toks[2] = "no-ign-macro"; |
toks[2] = "no-ign-macro"; |
toks[3] = "no-ign-chars"; |
toks[3] = "ign-errors"; |
toks[4] = "strict"; |
toks[4] = "strict"; |
toks[5] = NULL; |
toks[5] = "ign-escape"; |
|
toks[6] = NULL; |
|
|
while (*arg) |
while (*arg) { |
switch (getsubopt(&arg, toks, &v)) { |
o = arg; |
|
switch (getsubopt(&arg, UNCONST(toks), &v)) { |
case (0): |
case (0): |
*fflags |= IGN_SCOPE; |
*fflags |= FL_IGN_SCOPE; |
break; |
break; |
case (1): |
case (1): |
*fflags |= NO_IGN_ESCAPE; |
*fflags |= FL_NIGN_ESCAPE; |
break; |
break; |
case (2): |
case (2): |
*fflags |= NO_IGN_MACRO; |
*fflags |= FL_NIGN_MACRO; |
break; |
break; |
case (3): |
case (3): |
*fflags |= NO_IGN_CHARS; |
*fflags |= FL_IGN_ERRORS; |
break; |
break; |
case (4): |
case (4): |
*fflags |= NO_IGN_ESCAPE | |
*fflags |= FL_STRICT; |
NO_IGN_MACRO | NO_IGN_CHARS; |
|
break; |
break; |
|
case (5): |
|
*fflags &= ~FL_NIGN_ESCAPE; |
|
break; |
default: |
default: |
warnx("bad argument: -f%s", arg); |
fprintf(stderr, "%s: Bad argument\n", o); |
return(0); |
return(0); |
} |
} |
|
} |
|
|
return(1); |
return(1); |
} |
} |
|
|
|
|
/* |
|
* Parse out the options for [-Werr...], which sets warning modes. |
|
* These can be comma-delimited or called again. |
|
*/ |
|
static int |
static int |
woptions(int *wflags, char *arg) |
woptions(int *wflags, char *arg) |
{ |
{ |
char *v; |
char *v, *o; |
char *toks[5]; |
const char *toks[3]; |
|
|
toks[0] = "all"; |
toks[0] = "all"; |
toks[1] = "compat"; |
toks[1] = "error"; |
toks[2] = "syntax"; |
toks[2] = NULL; |
toks[3] = "error"; |
|
toks[4] = NULL; |
|
|
|
while (*arg) |
while (*arg) { |
switch (getsubopt(&arg, toks, &v)) { |
o = arg; |
|
switch (getsubopt(&arg, UNCONST(toks), &v)) { |
case (0): |
case (0): |
*wflags |= WARN_WALL; |
*wflags |= WARN_WALL; |
break; |
break; |
case (1): |
case (1): |
*wflags |= WARN_WCOMPAT; |
|
break; |
|
case (2): |
|
*wflags |= WARN_WSYNTAX; |
|
break; |
|
case (3): |
|
*wflags |= WARN_WERR; |
*wflags |= WARN_WERR; |
break; |
break; |
default: |
default: |
warnx("bad argument: -W%s", arg); |
fprintf(stderr, "%s: Bad argument\n", o); |
return(0); |
return(0); |
} |
} |
|
} |
|
|
return(1); |
return(1); |
} |
} |
|
|
|
|
/* ARGSUSED */ |
|
static int |
static int |
merr(void *arg, int line, int col, const char *msg) |
mmsg(enum mandocerr t, void *arg, int ln, int col, const char *msg) |
{ |
{ |
struct curparse *curp; |
struct curparse *cp; |
|
|
curp = (struct curparse *)arg; |
cp = (struct curparse *)arg; |
warnx("%s:%d: error: %s (column %d)", |
|
curp->file, line, msg, col); |
|
|
|
/* Always exit on errors... */ |
if (t <= MANDOCERR_ERROR) { |
return(0); |
if ( ! (cp->wflags & WARN_WALL)) |
} |
return(1); |
|
with_warning = 1; |
|
} else |
|
with_error = 1; |
|
|
|
fprintf(stderr, "%s:%d:%d: %s", cp->file, |
|
ln, col + 1, mandocerrs[t]); |
|
|
static int |
if (msg) |
mdocwarn(void *arg, int line, int col, |
fprintf(stderr, ": %s", msg); |
enum mdoc_warn type, const char *msg) |
|
{ |
|
struct curparse *curp; |
|
char *wtype; |
|
|
|
curp = (struct curparse *)arg; |
fputc('\n', stderr); |
wtype = NULL; |
|
|
|
switch (type) { |
/* This is superfluous, but whatever. */ |
case (WARN_COMPAT): |
if (t > MANDOCERR_ERROR) |
wtype = "compat"; |
return(0); |
if (curp->wflags & WARN_WCOMPAT) |
if (cp->wflags & WARN_WERR) { |
break; |
with_error = 1; |
return(1); |
return(0); |
case (WARN_SYNTAX): |
|
wtype = "syntax"; |
|
if (curp->wflags & WARN_WSYNTAX) |
|
break; |
|
return(1); |
|
} |
} |
|
return(1); |
assert(wtype); |
|
warnx("%s:%d: %s warning: %s (column %d)", |
|
curp->file, line, wtype, msg, col); |
|
|
|
if ( ! (curp->wflags & WARN_WERR)) |
|
return(1); |
|
|
|
/* |
|
* If the -Werror flag is passed in, as in gcc, then all |
|
* warnings are considered as errors. |
|
*/ |
|
|
|
warnx("%s: considering warnings as errors", |
|
__progname); |
|
return(0); |
|
} |
|
|
|
|
|
static int |
|
manwarn(void *arg, int line, int col, const char *msg) |
|
{ |
|
struct curparse *curp; |
|
|
|
curp = (struct curparse *)arg; |
|
|
|
if ( ! (curp->wflags & WARN_WSYNTAX)) |
|
return(1); |
|
|
|
warnx("%s:%d: syntax warning: %s (column %d)", |
|
curp->file, line, msg, col); |
|
|
|
if ( ! (curp->wflags & WARN_WERR)) |
|
return(1); |
|
|
|
/* |
|
* If the -Werror flag is passed in, as in gcc, then all |
|
* warnings are considered as errors. |
|
*/ |
|
|
|
warnx("%s: considering warnings as errors", |
|
__progname); |
|
return(0); |
|
} |
} |