version 1.61, 2010/04/12 19:27:22 |
version 1.67, 2010/05/15 05:50:19 |
|
|
#include "config.h" |
#include "config.h" |
#endif |
#endif |
|
|
|
#include <sys/mman.h> |
#include <sys/stat.h> |
#include <sys/stat.h> |
|
|
#include <assert.h> |
#include <assert.h> |
Line 75 struct curparse { |
|
Line 76 struct curparse { |
|
#define FL_IGN_SCOPE (1 << 0) /* Ignore scope errors. */ |
#define FL_IGN_SCOPE (1 << 0) /* Ignore scope errors. */ |
#define FL_NIGN_ESCAPE (1 << 1) /* Don't ignore bad escapes. */ |
#define FL_NIGN_ESCAPE (1 << 1) /* Don't ignore bad escapes. */ |
#define FL_NIGN_MACRO (1 << 2) /* Don't ignore bad macros. */ |
#define FL_NIGN_MACRO (1 << 2) /* Don't ignore bad macros. */ |
#define FL_NIGN_CHARS (1 << 3) /* Don't ignore bad chars. */ |
|
#define FL_IGN_ERRORS (1 << 4) /* Ignore failed parse. */ |
#define FL_IGN_ERRORS (1 << 4) /* Ignore failed parse. */ |
|
#define FL_STRICT FL_NIGN_ESCAPE | \ |
|
FL_NIGN_MACRO |
enum intt inttype; /* Input parsers... */ |
enum intt inttype; /* Input parsers... */ |
struct man *man; |
struct man *man; |
struct man *lastman; |
|
struct mdoc *mdoc; |
struct mdoc *mdoc; |
struct mdoc *lastmdoc; |
|
enum outt outtype; /* Output devices... */ |
enum outt outtype; /* Output devices... */ |
out_mdoc outmdoc; |
out_mdoc outmdoc; |
out_man outman; |
out_man outman; |
Line 90 struct curparse { |
|
Line 90 struct curparse { |
|
char outopts[BUFSIZ]; |
char outopts[BUFSIZ]; |
}; |
}; |
|
|
#define FL_STRICT FL_NIGN_ESCAPE | \ |
static void fdesc(struct curparse *); |
FL_NIGN_MACRO | \ |
static void ffile(const char *, struct curparse *); |
FL_NIGN_CHARS |
|
|
|
static int foptions(int *, char *); |
static int foptions(int *, char *); |
static int toptions(struct curparse *, char *); |
static struct man *man_init(struct curparse *); |
static int moptions(enum intt *, char *); |
static struct mdoc *mdoc_init(struct curparse *); |
static int woptions(int *, char *); |
|
static int merr(void *, int, int, const char *); |
static int merr(void *, int, int, const char *); |
|
static int moptions(enum intt *, char *); |
static int mwarn(void *, int, int, const char *); |
static int mwarn(void *, int, int, const char *); |
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 version(void) __attribute__((noreturn)); |
|
static void usage(void) __attribute__((noreturn)); |
static void usage(void) __attribute__((noreturn)); |
|
static void version(void) __attribute__((noreturn)); |
|
static int woptions(int *, char *); |
|
|
static const 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; |
|
|
progname = strrchr(argv[0], '/'); |
progname = strrchr(argv[0], '/'); |
Line 166 main(int argc, char *argv[]) |
|
Line 160 main(int argc, char *argv[]) |
|
argc -= optind; |
argc -= optind; |
argv += optind; |
argv += optind; |
|
|
memset(&ln, 0, sizeof(struct buf)); |
|
memset(&blk, 0, sizeof(struct buf)); |
|
|
|
rc = 1; |
|
|
|
if (NULL == *argv) { |
if (NULL == *argv) { |
curp.file = "<stdin>"; |
curp.file = "<stdin>"; |
curp.fd = STDIN_FILENO; |
curp.fd = STDIN_FILENO; |
|
|
c = fdesc(&blk, &ln, &curp); |
fdesc(&curp); |
if ( ! (FL_IGN_ERRORS & curp.fflags)) |
|
rc = 1 == c ? 1 : 0; |
|
else |
|
rc = -1 == c ? 0 : 1; |
|
} |
} |
|
|
while (rc && *argv) { |
while (*argv) { |
c = ffile(&blk, &ln, *argv, &curp); |
ffile(*argv, &curp); |
if ( ! (FL_IGN_ERRORS & curp.fflags)) |
|
rc = 1 == c ? 1 : 0; |
|
else |
|
rc = -1 == c ? 0 : 1; |
|
|
|
argv++; |
if (with_error && !(curp.fflags & FL_IGN_ERRORS)) |
if (*argv && rc) { |
break; |
if (curp.lastman) |
++argv; |
man_reset(curp.lastman); |
|
if (curp.lastmdoc) |
|
mdoc_reset(curp.lastmdoc); |
|
curp.lastman = NULL; |
|
curp.lastmdoc = NULL; |
|
} |
|
} |
} |
|
|
if (blk.buf) |
|
free(blk.buf); |
|
if (ln.buf) |
|
free(ln.buf); |
|
if (curp.outfree) |
if (curp.outfree) |
(*curp.outfree)(curp.outdata); |
(*curp.outfree)(curp.outdata); |
if (curp.mdoc) |
|
mdoc_free(curp.mdoc); |
|
if (curp.man) |
|
man_free(curp.man); |
|
|
|
return(rc ? EXIT_SUCCESS : EXIT_FAILURE); |
return((with_warning || with_error) ? |
|
EXIT_FAILURE : EXIT_SUCCESS); |
} |
} |
|
|
|
|
Line 246 man_init(struct curparse *curp) |
|
Line 214 man_init(struct curparse *curp) |
|
|
|
/* Defaults from mandoc.1. */ |
/* Defaults from mandoc.1. */ |
|
|
pflags = MAN_IGN_MACRO | MAN_IGN_ESCAPE | MAN_IGN_CHARS; |
pflags = MAN_IGN_MACRO | MAN_IGN_ESCAPE; |
|
|
if (curp->fflags & FL_NIGN_MACRO) |
if (curp->fflags & FL_NIGN_MACRO) |
pflags &= ~MAN_IGN_MACRO; |
pflags &= ~MAN_IGN_MACRO; |
if (curp->fflags & FL_NIGN_CHARS) |
|
pflags &= ~MAN_IGN_CHARS; |
|
if (curp->fflags & FL_NIGN_ESCAPE) |
if (curp->fflags & FL_NIGN_ESCAPE) |
pflags &= ~MAN_IGN_ESCAPE; |
pflags &= ~MAN_IGN_ESCAPE; |
|
|
Line 270 mdoc_init(struct curparse *curp) |
|
Line 236 mdoc_init(struct curparse *curp) |
|
|
|
/* Defaults from mandoc.1. */ |
/* Defaults from mandoc.1. */ |
|
|
pflags = MDOC_IGN_MACRO | MDOC_IGN_ESCAPE | MDOC_IGN_CHARS; |
pflags = MDOC_IGN_MACRO | MDOC_IGN_ESCAPE; |
|
|
if (curp->fflags & FL_IGN_SCOPE) |
if (curp->fflags & FL_IGN_SCOPE) |
pflags |= MDOC_IGN_SCOPE; |
pflags |= MDOC_IGN_SCOPE; |
Line 278 mdoc_init(struct curparse *curp) |
|
Line 244 mdoc_init(struct curparse *curp) |
|
pflags &= ~MDOC_IGN_ESCAPE; |
pflags &= ~MDOC_IGN_ESCAPE; |
if (curp->fflags & FL_NIGN_MACRO) |
if (curp->fflags & FL_NIGN_MACRO) |
pflags &= ~MDOC_IGN_MACRO; |
pflags &= ~MDOC_IGN_MACRO; |
if (curp->fflags & FL_NIGN_CHARS) |
|
pflags &= ~MDOC_IGN_CHARS; |
|
|
|
return(mdoc_alloc(curp, pflags, &mdoccb)); |
return(mdoc_alloc(curp, pflags, &mdoccb)); |
} |
} |
|
|
|
|
static int |
static void |
ffile(struct buf *blk, struct buf *ln, |
ffile(const char *file, struct curparse *curp) |
const char *file, struct curparse *curp) |
|
{ |
{ |
int c; |
|
|
|
curp->file = file; |
curp->file = file; |
if (-1 == (curp->fd = open(curp->file, O_RDONLY, 0))) { |
if (-1 == (curp->fd = open(curp->file, O_RDONLY, 0))) { |
perror(curp->file); |
perror(curp->file); |
return(-1); |
with_error = 1; |
|
return; |
} |
} |
|
|
c = fdesc(blk, ln, curp); |
fdesc(curp); |
|
|
if (-1 == close(curp->fd)) |
if (-1 == close(curp->fd)) |
perror(curp->file); |
perror(curp->file); |
|
|
return(c); |
|
} |
} |
|
|
|
|
static int |
static int |
fdesc(struct buf *blk, struct buf *ln, struct curparse *curp) |
read_whole_file(struct curparse *curp, struct buf *fb, int *with_mmap) |
{ |
{ |
size_t sz; |
|
ssize_t ssz; |
|
struct stat st; |
struct stat st; |
int j, i, pos, lnn, comment; |
char *buf; |
|
size_t sz, off; |
|
ssize_t ssz; |
|
|
|
if (-1 == fstat(curp->fd, &st)) { |
|
perror(curp->file); |
|
with_error = 1; |
|
return(0); |
|
} |
|
|
|
/* |
|
* 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 (S_ISREG(st.st_mode)) { |
|
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 = st.st_size; |
|
fb->buf = mmap(NULL, fb->sz, PROT_READ, |
|
MAP_FILE, curp->fd, 0); |
|
if (fb->buf != MAP_FAILED) |
|
return(1); |
|
} |
|
|
|
/* |
|
* 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 (fb->sz == 0) |
|
sz = 65536; |
|
else |
|
sz = 2 * fb->sz; |
|
buf = realloc(fb->buf, sz); |
|
if (NULL == buf) { |
|
perror(NULL); |
|
break; |
|
} |
|
fb->buf = buf; |
|
fb->sz = sz; |
|
} |
|
ssz = read(curp->fd, fb->buf + off, fb->sz - off); |
|
if (ssz == 0) { |
|
fb->sz = off; |
|
return(1); |
|
} |
|
if (ssz == -1) { |
|
perror(curp->file); |
|
break; |
|
} |
|
off += ssz; |
|
} |
|
|
|
free(fb->buf); |
|
fb->buf = NULL; |
|
with_error = 1; |
|
return(0); |
|
} |
|
|
|
|
|
static void |
|
fdesc(struct curparse *curp) |
|
{ |
|
size_t sz; |
|
struct buf ln, blk; |
|
int j, i, pos, lnn, comment, with_mmap; |
struct man *man; |
struct man *man; |
struct mdoc *mdoc; |
struct mdoc *mdoc; |
|
|
sz = BUFSIZ; |
sz = BUFSIZ; |
man = NULL; |
man = NULL; |
mdoc = NULL; |
mdoc = 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 buffer optimised |
Line 326 fdesc(struct buf *blk, struct buf *ln, struct curparse |
|
Line 370 fdesc(struct buf *blk, struct buf *ln, struct curparse |
|
* growable, hence passed in by ptr-ptr. |
* growable, hence passed in by ptr-ptr. |
*/ |
*/ |
|
|
if (-1 == fstat(curp->fd, &st)) |
if (!read_whole_file(curp, &blk, &with_mmap)) |
perror(curp->file); |
return; |
else if ((size_t)st.st_blksize > sz) |
|
sz = st.st_blksize; |
|
|
|
if (sz > blk->sz) { |
|
blk->buf = realloc(blk->buf, sz); |
|
if (NULL == blk->buf) { |
|
perror(NULL); |
|
exit(EXIT_FAILURE); |
|
} |
|
blk->sz = sz; |
|
} |
|
|
|
/* Fill buf with file blocksize. */ |
/* Fill buf with file blocksize. */ |
|
|
for (lnn = pos = comment = 0; ; ) { |
for (i = lnn = pos = comment = 0; i < (int)blk.sz; ++i) { |
if (-1 == (ssz = read(curp->fd, blk->buf, sz))) { |
if (pos >= (int)ln.sz) { |
perror(curp->file); |
ln.sz += 256; /* Step-size. */ |
return(-1); |
ln.buf = realloc(ln.buf, ln.sz); |
} else if (0 == ssz) |
if (NULL == ln.buf) { |
break; |
perror(NULL); |
|
goto bailout; |
/* 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) { |
|
perror(NULL); |
|
return(EXIT_FAILURE); |
|
} |
|
} |
} |
|
} |
|
|
if ('\n' != blk->buf[i]) { |
if ('\n' != blk.buf[i]) { |
if (comment) |
if (comment) |
continue; |
continue; |
ln->buf[pos++] = blk->buf[i]; |
ln.buf[pos++] = blk.buf[i]; |
|
|
/* Handle in-line `\"' comments. */ |
/* Handle in-line `\"' comments. */ |
|
|
if (1 == pos || '\"' != ln->buf[pos - 1]) |
if (1 == pos || '\"' != ln.buf[pos - 1]) |
continue; |
continue; |
|
|
for (j = pos - 2; j >= 0; j--) |
for (j = pos - 2; j >= 0; j--) |
if ('\\' != ln->buf[j]) |
if ('\\' != ln.buf[j]) |
break; |
break; |
|
|
if ( ! ((pos - 2 - j) % 2)) |
if ( ! ((pos - 2 - j) % 2)) |
continue; |
|
|
|
comment = 1; |
|
pos -= 2; |
|
for (; pos > 0; --pos) { |
|
if (ln->buf[pos] != ' ') |
|
break; |
|
if (ln->buf[pos - 1] == '\\') |
|
break; |
|
} |
|
continue; |
continue; |
} |
|
|
|
/* Handle escaped `\\n' newlines. */ |
comment = 1; |
|
pos -= 2; |
|
for (; pos > 0; --pos) { |
|
if (ln.buf[pos - 1] != ' ') |
|
break; |
|
if (pos > 2 && ln.buf[pos - 2] == '\\') |
|
break; |
|
} |
|
continue; |
|
} |
|
|
if (pos > 0 && 0 == comment && |
/* Handle escaped `\\n' newlines. */ |
'\\' == ln->buf[pos - 1]) { |
|
for (j = pos - 1; j >= 0; j--) |
if (pos > 0 && 0 == comment && '\\' == ln.buf[pos - 1]) { |
if ('\\' != ln->buf[j]) |
for (j = pos - 1; j >= 0; j--) |
break; |
if ('\\' != ln.buf[j]) |
if ( ! ((pos - j) % 2)) { |
break; |
pos--; |
if ( ! ((pos - j) % 2)) { |
lnn++; |
pos--; |
continue; |
lnn++; |
} |
continue; |
} |
} |
|
} |
|
|
ln->buf[pos] = 0; |
ln.buf[pos] = 0; |
lnn++; |
lnn++; |
|
|
/* If unset, assign parser in pset(). */ |
/* If unset, assign parser in pset(). */ |
|
|
if ( ! (man || mdoc) && ! pset(ln->buf, |
if ( ! (man || mdoc) && ! pset(ln.buf, pos, curp, &man, &mdoc)) |
pos, curp, &man, &mdoc)) |
goto bailout; |
return(-1); |
|
|
|
pos = comment = 0; |
pos = comment = 0; |
|
|
/* Pass down into parsers. */ |
/* Pass down into parsers. */ |
|
|
if (man && ! man_parseln(man, lnn, ln->buf)) |
if (man && ! man_parseln(man, lnn, ln.buf)) |
return(0); |
goto bailout; |
if (mdoc && ! mdoc_parseln(mdoc, lnn, ln->buf)) |
if (mdoc && ! mdoc_parseln(mdoc, lnn, ln.buf)) |
return(0); |
goto bailout; |
} |
|
} |
} |
|
|
/* NOTE a parser may not have been assigned, yet. */ |
/* NOTE a parser may not have been assigned, yet. */ |
|
|
if ( ! (man || mdoc)) { |
if ( ! (man || mdoc)) { |
fprintf(stderr, "%s: Not a manual\n", curp->file); |
fprintf(stderr, "%s: Not a manual\n", curp->file); |
return(0); |
goto bailout; |
} |
} |
|
|
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 unset, allocate output dev now (if applicable). */ |
/* If unset, allocate output dev now (if applicable). */ |
|
|
Line 473 fdesc(struct buf *blk, struct buf *ln, struct curparse |
|
Line 494 fdesc(struct buf *blk, struct buf *ln, struct curparse |
|
if (mdoc && curp->outmdoc) |
if (mdoc && curp->outmdoc) |
(*curp->outmdoc)(curp->outdata, mdoc); |
(*curp->outmdoc)(curp->outdata, mdoc); |
|
|
return(1); |
cleanup: |
|
if (curp->mdoc) { |
|
mdoc_free(curp->mdoc); |
|
curp->mdoc = NULL; |
|
} |
|
if (curp->man) { |
|
man_free(curp->man); |
|
curp->man = NULL; |
|
} |
|
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; |
} |
} |
|
|
|
|
Line 505 pset(const char *buf, int pos, struct curparse *curp, |
|
Line 545 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 523 pset(const char *buf, int pos, struct curparse *curp, |
|
Line 561 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 531 pset(const char *buf, int pos, struct curparse *curp, |
|
Line 568 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 590 foptions(int *fflags, char *arg) |
|
Line 626 foptions(int *fflags, char *arg) |
|
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] = "ign-errors"; |
toks[4] = "strict"; |
toks[5] = "strict"; |
toks[5] = "ign-escape"; |
toks[6] = "ign-escape"; |
toks[6] = NULL; |
toks[7] = NULL; |
|
|
|
while (*arg) { |
while (*arg) { |
o = arg; |
o = arg; |
Line 609 foptions(int *fflags, char *arg) |
|
Line 644 foptions(int *fflags, char *arg) |
|
*fflags |= FL_NIGN_MACRO; |
*fflags |= FL_NIGN_MACRO; |
break; |
break; |
case (3): |
case (3): |
*fflags |= FL_NIGN_CHARS; |
|
break; |
|
case (4): |
|
*fflags |= FL_IGN_ERRORS; |
*fflags |= FL_IGN_ERRORS; |
break; |
break; |
case (5): |
case (4): |
*fflags |= FL_STRICT; |
*fflags |= FL_STRICT; |
break; |
break; |
case (6): |
case (5): |
*fflags &= ~FL_NIGN_ESCAPE; |
*fflags &= ~FL_NIGN_ESCAPE; |
break; |
break; |
default: |
default: |
Line 670 merr(void *arg, int line, int col, const char *msg) |
|
Line 702 merr(void *arg, int line, int col, const char *msg) |
|
(void)fprintf(stderr, "%s:%d:%d: error: %s\n", |
(void)fprintf(stderr, "%s:%d:%d: error: %s\n", |
curp->file, line, col + 1, msg); |
curp->file, line, col + 1, msg); |
|
|
|
with_error = 1; |
|
|
return(0); |
return(0); |
} |
} |
|
|
Line 687 mwarn(void *arg, int line, int col, const char *msg) |
|
Line 721 mwarn(void *arg, int line, int col, const char *msg) |
|
(void)fprintf(stderr, "%s:%d:%d: warning: %s\n", |
(void)fprintf(stderr, "%s:%d:%d: warning: %s\n", |
curp->file, line, col + 1, msg); |
curp->file, line, col + 1, msg); |
|
|
if ( ! (curp->wflags & WARN_WERR)) |
with_warning = 1; |
return(1); |
if (curp->wflags & WARN_WERR) { |
|
with_error = 1; |
return(0); |
return(0); |
|
} |
|
|
|
return(1); |
} |
} |
|
|