version 1.64, 2010/05/14 17:59:07 |
version 1.76, 2010/05/16 10:59:36 |
|
|
#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> |
|
|
#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" |
#include "main.h" |
|
|
#define UNCONST(a) ((void *)(uintptr_t)(const void *)(a)) |
#define UNCONST(a) ((void *)(uintptr_t)(const void *)(a)) |
Line 69 struct curparse { |
|
Line 72 struct curparse { |
|
const char *file; /* Current parse. */ |
const char *file; /* Current parse. */ |
int fd; /* Current parse. */ |
int fd; /* Current parse. */ |
int wflags; |
int wflags; |
|
/* FIXME: set by max error */ |
#define WARN_WALL (1 << 0) /* All-warnings mask. */ |
#define WARN_WALL (1 << 0) /* All-warnings mask. */ |
#define WARN_WERR (1 << 2) /* Warnings->errors. */ |
#define WARN_WERR (1 << 2) /* Warnings->errors. */ |
int fflags; |
int fflags; |
Line 76 struct curparse { |
|
Line 80 struct curparse { |
|
#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_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; |
struct roff *roff; |
enum outt outtype; /* Output devices... */ |
enum outt outtype; /* Output devices... */ |
out_mdoc outmdoc; |
out_mdoc outmdoc; |
out_man outman; |
out_man outman; |
Line 89 struct curparse { |
|
Line 94 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 *); |
|
|
static int foptions(int *, char *); |
static int foptions(int *, char *); |
static int toptions(struct curparse *, char *); |
static struct man *man_init(struct curparse *); |
|
static struct mdoc *mdoc_init(struct curparse *); |
|
static struct roff *roff_init(struct curparse *); |
|
static int merr(void *, int, int, const char *); /* DEPRECATED */ |
static int moptions(enum intt *, char *); |
static int moptions(enum intt *, char *); |
static int woptions(int *, char *); |
static int mwarn(void *, int, int, const char *); /* DEPRECATED */ |
static int merr(void *, int, int, const char *); |
static int mmsg(enum mandocerr, void *, |
static int mwarn(void *, int, int, const char *); |
int, int, const char *); |
static void ffile(struct buf *, struct buf *, |
|
const char *, struct curparse *); |
|
static void 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, with_warning; |
static int with_error; |
|
static int with_warning; |
|
|
int |
int |
main(int argc, char *argv[]) |
main(int argc, char *argv[]) |
{ |
{ |
int c; |
int c; |
struct buf ln, blk; |
|
struct curparse curp; |
struct curparse curp; |
|
|
progname = strrchr(argv[0], '/'); |
progname = strrchr(argv[0], '/'); |
Line 164 main(int argc, char *argv[]) |
|
Line 167 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)); |
|
|
|
if (NULL == *argv) { |
if (NULL == *argv) { |
curp.file = "<stdin>"; |
curp.file = "<stdin>"; |
curp.fd = STDIN_FILENO; |
curp.fd = STDIN_FILENO; |
|
|
fdesc(&blk, &ln, &curp); |
fdesc(&curp); |
} |
} |
|
|
while (*argv) { |
while (*argv) { |
ffile(&blk, &ln, *argv, &curp); |
ffile(*argv, &curp); |
|
|
if (with_error && !(curp.fflags & FL_IGN_ERRORS)) |
if (with_error && !(curp.fflags & FL_IGN_ERRORS)) |
break; |
break; |
++argv; |
++argv; |
} |
} |
|
|
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); |
|
if (curp.roff) |
|
roff_free(curp.roff); |
|
|
return((with_warning || with_error) ? EXIT_FAILURE : EXIT_SUCCESS ); |
return((with_warning || with_error) ? |
|
EXIT_FAILURE : EXIT_SUCCESS); |
} |
} |
|
|
|
|
Line 235 man_init(struct curparse *curp) |
|
Line 238 man_init(struct curparse *curp) |
|
} |
} |
|
|
|
|
|
static struct roff * |
|
roff_init(struct curparse *curp) |
|
{ |
|
|
|
return(roff_alloc(mmsg, curp)); |
|
} |
|
|
|
|
static struct mdoc * |
static struct mdoc * |
mdoc_init(struct curparse *curp) |
mdoc_init(struct curparse *curp) |
{ |
{ |
Line 260 mdoc_init(struct curparse *curp) |
|
Line 271 mdoc_init(struct curparse *curp) |
|
|
|
|
|
static void |
static void |
ffile(struct buf *blk, struct buf *ln, |
ffile(const char *file, struct curparse *curp) |
const char *file, struct curparse *curp) |
|
{ |
{ |
|
|
curp->file = file; |
curp->file = file; |
Line 271 ffile(struct buf *blk, struct buf *ln, |
|
Line 281 ffile(struct buf *blk, struct buf *ln, |
|
return; |
return; |
} |
} |
|
|
fdesc(blk, ln, curp); |
fdesc(curp); |
|
|
if (-1 == close(curp->fd)) |
if (-1 == close(curp->fd)) |
perror(curp->file); |
perror(curp->file); |
} |
} |
|
|
|
|
static void |
static int |
fdesc(struct buf *blk, struct buf *ln, struct curparse *curp) |
resize_buf(struct buf *buf, size_t initial) |
{ |
{ |
size_t sz; |
void *tmp; |
ssize_t ssz; |
size_t sz; |
struct stat st; |
|
int j, i, pos, lnn, comment; |
|
struct man *man; |
|
struct mdoc *mdoc; |
|
|
|
sz = BUFSIZ; |
if (buf->sz == 0) |
man = NULL; |
sz = initial; |
mdoc = NULL; |
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); |
|
} |
|
|
/* |
|
* Two buffers: ln and buf. buf is the input buffer optimised |
|
* here for each file's block size. ln is a line buffer. Both |
|
* growable, hence passed in by ptr-ptr. |
|
*/ |
|
|
|
|
static int |
|
read_whole_file(struct curparse *curp, struct buf *fb, int *with_mmap) |
|
{ |
|
struct stat st; |
|
size_t off; |
|
ssize_t ssz; |
|
|
if (-1 == fstat(curp->fd, &st)) { |
if (-1 == fstat(curp->fd, &st)) { |
perror(curp->file); |
perror(curp->file); |
with_error = 1; |
with_error = 1; |
return; |
return(0); |
} |
} |
if ((size_t)st.st_blksize > sz) |
|
sz = st.st_blksize; |
|
|
|
if (sz > blk->sz) { |
/* |
void *buf = realloc(blk->buf, sz); |
* 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 (NULL == buf) { |
if (S_ISREG(st.st_mode)) { |
perror(NULL); |
if (st.st_size >= (1U << 31)) { |
|
fprintf(stderr, "%s: input too large\n", |
|
curp->file); |
with_error = 1; |
with_error = 1; |
return; |
return(0); |
} |
} |
blk->buf = buf; |
*with_mmap = 1; |
blk->sz = sz; |
fb->sz = (size_t)st.st_size; |
|
fb->buf = mmap(NULL, fb->sz, PROT_READ, |
|
MAP_FILE, curp->fd, 0); |
|
if (fb->buf != MAP_FAILED) |
|
return(1); |
} |
} |
|
|
/* Fill buf with file blocksize. */ |
/* |
|
* 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. |
|
*/ |
|
|
for (lnn = pos = comment = 0; ; ) { |
*with_mmap = 0; |
if (-1 == (ssz = read(curp->fd, blk->buf, sz))) { |
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); |
perror(curp->file); |
goto bailout; |
|
} else if (0 == ssz) |
|
break; |
break; |
|
} |
|
off += (size_t)ssz; |
|
} |
|
|
/* Parse the read block into partial or full lines. */ |
free(fb->buf); |
|
fb->buf = NULL; |
|
with_error = 1; |
|
return(0); |
|
} |
|
|
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); |
|
goto bailout; |
|
} |
|
} |
|
|
|
if ('\n' != blk->buf[i]) { |
static void |
if (comment) |
fdesc(struct curparse *curp) |
continue; |
{ |
ln->buf[pos++] = blk->buf[i]; |
struct buf ln, blk; |
|
int i, pos, lnn, lnn_start, with_mmap, of; |
|
enum rofferr re; |
|
struct man *man; |
|
struct mdoc *mdoc; |
|
struct roff *roff; |
|
|
/* Handle in-line `\"' comments. */ |
man = NULL; |
|
mdoc = NULL; |
|
roff = NULL; |
|
memset(&ln, 0, sizeof(struct buf)); |
|
|
if (1 == pos || '\"' != ln->buf[pos - 1]) |
/* |
continue; |
* Two buffers: ln and buf. buf is the input file and may be |
|
* memory mapped. ln is a line buffer and grows on-demand. |
|
*/ |
|
|
for (j = pos - 2; j >= 0; j--) |
if ( ! read_whole_file(curp, &blk, &with_mmap)) |
if ('\\' != ln->buf[j]) |
return; |
break; |
|
|
|
if ( ! ((pos - 2 - j) % 2)) |
if (NULL == curp->roff) |
continue; |
curp->roff = roff_init(curp); |
|
if (NULL == (roff = curp->roff)) |
|
goto bailout; |
|
|
comment = 1; |
for (i = 0, lnn = 1; i < (int)blk.sz;) { |
pos -= 2; |
pos = 0; |
|
lnn_start = lnn; |
|
while (i < (int)blk.sz) { |
|
if ('\n' == blk.buf[i]) { |
|
++i; |
|
++lnn; |
|
break; |
|
} |
|
/* Trailing backslash is like a plain character. */ |
|
if ('\\' != blk.buf[i] || i + 1 == (int)blk.sz) { |
|
if (pos >= (int)ln.sz) |
|
if (! resize_buf(&ln, 256)) |
|
goto bailout; |
|
ln.buf[pos++] = blk.buf[i++]; |
|
continue; |
|
} |
|
/* Found an escape and at least one other character. */ |
|
if ('\n' == blk.buf[i + 1]) { |
|
/* Escaped newlines are skipped over */ |
|
i += 2; |
|
++lnn; |
|
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; |
|
} |
|
} |
|
/* Backout trailing whitespaces */ |
for (; pos > 0; --pos) { |
for (; pos > 0; --pos) { |
if (ln->buf[pos - 1] != ' ') |
if (ln.buf[pos - 1] != ' ') |
break; |
break; |
if (pos > 2 && ln->buf[pos - 2] == '\\') |
if (pos > 2 && ln.buf[pos - 2] == '\\') |
break; |
break; |
} |
} |
continue; |
break; |
} |
|
|
|
/* Handle escaped `\\n' newlines. */ |
|
|
|
if (pos > 0 && 0 == comment && |
|
'\\' == ln->buf[pos - 1]) { |
|
for (j = pos - 1; j >= 0; j--) |
|
if ('\\' != ln->buf[j]) |
|
break; |
|
if ( ! ((pos - j) % 2)) { |
|
pos--; |
|
lnn++; |
|
continue; |
|
} |
|
} |
} |
|
/* 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 unset, assign parser in pset(). */ |
if (pos >= (int)ln.sz) |
|
if (! resize_buf(&ln, 256)) |
if ( ! (man || mdoc) && ! pset(ln->buf, |
|
pos, curp, &man, &mdoc)) |
|
goto bailout; |
goto bailout; |
|
ln.buf[pos] = '\0'; |
|
|
pos = comment = 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. |
|
*/ |
|
|
/* Pass down into parsers. */ |
of = 0; |
|
do { |
|
re = roff_parseln(roff, lnn_start, |
|
&ln.buf, &ln.sz, of, &of); |
|
} while (ROFF_RERUN == re); |
|
|
if (man && ! man_parseln(man, lnn, ln->buf)) |
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; |
goto bailout; |
if (mdoc && ! mdoc_parseln(mdoc, lnn, ln->buf)) |
|
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 a parser may not have been assigned, yet. */ |
/* NOTE a parser may not have been assigned, yet. */ |
Line 408 fdesc(struct buf *blk, struct buf *ln, struct curparse |
|
Line 513 fdesc(struct buf *blk, struct buf *ln, struct curparse |
|
goto bailout; |
goto bailout; |
} |
} |
|
|
|
/* Clean up the parse routine ASTs. */ |
|
|
if (mdoc && ! mdoc_endparse(mdoc)) |
if (mdoc && ! mdoc_endparse(mdoc)) |
goto bailout; |
goto bailout; |
if (man && ! man_endparse(man)) |
if (man && ! man_endparse(man)) |
goto bailout; |
goto bailout; |
|
if (roff && ! roff_endparse(roff)) |
|
goto bailout; |
|
|
/* If unset, allocate output dev now (if applicable). */ |
/* If unset, allocate output dev now (if applicable). */ |
|
|
Line 436 fdesc(struct buf *blk, struct buf *ln, struct curparse |
|
Line 545 fdesc(struct buf *blk, struct buf *ln, struct curparse |
|
case (OUTT_LINT): |
case (OUTT_LINT): |
break; |
break; |
default: |
default: |
curp->outdata = ascii_alloc(); |
curp->outdata = ascii_alloc(80); |
curp->outman = terminal_man; |
curp->outman = terminal_man; |
curp->outmdoc = terminal_mdoc; |
curp->outmdoc = terminal_mdoc; |
curp->outfree = terminal_free; |
curp->outfree = terminal_free; |
Line 452 fdesc(struct buf *blk, struct buf *ln, struct curparse |
|
Line 561 fdesc(struct buf *blk, struct buf *ln, struct curparse |
|
(*curp->outmdoc)(curp->outdata, mdoc); |
(*curp->outmdoc)(curp->outdata, mdoc); |
|
|
cleanup: |
cleanup: |
if (curp->mdoc) { |
if (mdoc) |
mdoc_free(curp->mdoc); |
mdoc_reset(mdoc); |
curp->mdoc = NULL; |
if (man) |
} |
man_reset(man); |
if (curp->man) { |
if (roff) |
man_free(curp->man); |
roff_reset(roff); |
curp->man = NULL; |
if (ln.buf) |
} |
free(ln.buf); |
|
if (with_mmap) |
|
munmap(blk.buf, blk.sz); |
|
else |
|
free(blk.buf); |
|
|
return; |
return; |
|
|
bailout: |
bailout: |
Line 482 pset(const char *buf, int pos, struct curparse *curp, |
|
Line 596 pset(const char *buf, int pos, struct curparse *curp, |
|
* default to -man, which is more lenient. |
* default to -man, which is more lenient. |
*/ |
*/ |
|
|
if (buf[0] == '.') { |
if ('.' == buf[0] || '\'' == buf[0]) { |
for (i = 1; buf[i]; i++) |
for (i = 1; buf[i]; i++) |
if (' ' != buf[i] && '\t' != buf[i]) |
if (' ' != buf[i] && '\t' != buf[i]) |
break; |
break; |
Line 496 pset(const char *buf, int pos, struct curparse *curp, |
|
Line 610 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 514 pset(const char *buf, int pos, struct curparse *curp, |
|
Line 626 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 522 pset(const char *buf, int pos, struct curparse *curp, |
|
Line 633 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 685 mwarn(void *arg, int line, int col, const char *msg) |
|
Line 795 mwarn(void *arg, int line, int col, const char *msg) |
|
return(1); |
return(1); |
} |
} |
|
|
|
static const char * const mandocerrs[MANDOCERR_MAX] = { |
|
"ok", |
|
"multi-line scope open on exit", |
|
"request for scope closure when no matching scope is open", |
|
"line arguments will be lost", |
|
"memory exhausted" |
|
}; |
|
|
|
/* |
|
* XXX: this is experimental code that will eventually become the |
|
* generic means of covering all warnings and errors! |
|
*/ |
|
/* ARGSUSED */ |
|
static int |
|
mmsg(enum mandocerr t, void *arg, int ln, int col, const char *msg) |
|
{ |
|
#if 0 |
|
struct curparse *cp; |
|
|
|
cp = (struct curparse *)arg; |
|
|
|
fprintf(stderr, "%s:%d:%d: %s", cp->file, |
|
ln, col + 1, mandocerrs[t]); |
|
|
|
if (msg) |
|
fprintf(stderr, ": %s", msg); |
|
|
|
fputc('\n', stderr); |
|
#endif |
|
return(1); |
|
} |