version 1.52, 2014/06/25 00:20:19 |
version 1.87, 2014/09/11 23:53:30 |
|
|
* 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" |
#include "config.h" |
#endif |
|
|
|
#ifdef HAVE_MMAP |
#include <sys/types.h> |
# include <sys/stat.h> |
#if HAVE_MMAP |
# include <sys/mman.h> |
#include <sys/mman.h> |
|
#include <sys/stat.h> |
#endif |
#endif |
|
#include <sys/wait.h> |
|
|
#include <assert.h> |
#include <assert.h> |
#include <ctype.h> |
#include <ctype.h> |
|
|
}; |
}; |
|
|
struct mparse { |
struct mparse { |
enum mandoclevel file_status; /* status of current parse */ |
|
enum mandoclevel wlevel; /* ignore messages below this */ |
|
int line; /* line number in the file */ |
|
int options; /* parser options */ |
|
struct man *pman; /* persistent man parser */ |
struct man *pman; /* persistent man parser */ |
struct mdoc *pmdoc; /* persistent mdoc parser */ |
struct mdoc *pmdoc; /* persistent mdoc parser */ |
struct man *man; /* man parser */ |
struct man *man; /* man parser */ |
struct mdoc *mdoc; /* mdoc parser */ |
struct mdoc *mdoc; /* mdoc parser */ |
struct roff *roff; /* roff parser (!NULL) */ |
struct roff *roff; /* roff parser (!NULL) */ |
char *sodest; /* filename pointed to by .so */ |
char *sodest; /* filename pointed to by .so */ |
int reparse_count; /* finite interp. stack */ |
const char *file; /* filename of current input file */ |
|
struct buf *primary; /* buffer currently being parsed */ |
|
struct buf *secondary; /* preprocessed copy of input */ |
|
const char *defos; /* default operating system */ |
mandocmsg mmsg; /* warning/error message handler */ |
mandocmsg mmsg; /* warning/error message handler */ |
const char *file; |
enum mandoclevel file_status; /* status of current parse */ |
struct buf *secondary; |
enum mandoclevel wlevel; /* ignore messages below this */ |
char *defos; /* default operating system */ |
int options; /* parser options */ |
|
int reparse_count; /* finite interp. stack */ |
|
int line; /* line number in the file */ |
}; |
}; |
|
|
|
static void choose_parser(struct mparse *); |
static void resize_buf(struct buf *, size_t); |
static void resize_buf(struct buf *, size_t); |
static void mparse_buf_r(struct mparse *, struct buf, int); |
static void mparse_buf_r(struct mparse *, struct buf, int); |
static void pset(const char *, int, struct mparse *); |
|
static int read_whole_file(struct mparse *, const char *, int, |
static int read_whole_file(struct mparse *, const char *, int, |
struct buf *, int *); |
struct buf *, int *); |
static void mparse_end(struct mparse *); |
static void mparse_end(struct mparse *); |
Line 93 static const char * const mandocerrs[MANDOCERR_MAX] = |
|
Line 94 static const char * const mandocerrs[MANDOCERR_MAX] = |
|
"generic warning", |
"generic warning", |
|
|
/* related to the prologue */ |
/* related to the prologue */ |
"no TH macro in document", |
"missing manual title, using UNTITLED", |
"document title should be all caps", |
"missing manual title, using \"\"", |
|
"lower case character in document title", |
|
"missing manual section, using \"\"", |
"unknown manual section", |
"unknown manual section", |
"unknown manual volume or arch", |
"unknown manual volume or arch", |
"date missing, using today's date", |
"missing date, using today's date", |
"cannot parse date, using it verbatim", |
"cannot parse date, using it verbatim", |
"prologue macros out of order", |
"missing Os macro, using \"\"", |
"duplicate prologue macro", |
"duplicate prologue macro", |
"macro not allowed in prologue", |
"late prologue macro", |
"macro not allowed in body", |
"skipping late title macro", |
|
"prologue macros out of order", |
|
|
/* related to document structure */ |
/* related to document structure */ |
".so is fragile, better use ln(1)", |
".so is fragile, better use ln(1)", |
"no document body", |
"no document body", |
"content before the first section header", |
"content before first section header", |
"NAME section must come first", |
"first section is not \"NAME\"", |
"bad NAME section contents", |
"bad NAME section contents", |
"sections out of conventional order", |
"sections out of conventional order", |
"duplicate section name", |
"duplicate section title", |
"section header suited to sections 2, 3, and 9 only", |
"unexpected section", |
|
"unusual Xr order", |
|
"unusual Xr punctuation", |
|
"AUTHORS section without An macro", |
|
|
/* related to macros and nesting */ |
/* related to macros and nesting */ |
"skipping obsolete macro", |
"obsolete macro", |
"skipping paragraph macro", |
"skipping paragraph macro", |
"moving paragraph macro out of list", |
"moving paragraph macro out of list", |
"skipping no-space macro", |
"skipping no-space macro", |
"blocks badly nested", |
"blocks badly nested", |
"child violates parent syntax", |
|
"nested displays are not portable", |
"nested displays are not portable", |
"already in literal mode", |
"moving content out of list", |
|
".Vt block has child macro", |
|
"fill mode already enabled, skipping", |
|
"fill mode already disabled, skipping", |
"line scope broken", |
"line scope broken", |
|
|
/* related to missing macro arguments */ |
/* related to missing macro arguments */ |
|
"skipping empty request", |
|
"conditional request controls empty scope", |
"skipping empty macro", |
"skipping empty macro", |
|
"empty argument, using 0n", |
"argument count wrong", |
"argument count wrong", |
"missing display type", |
"missing display type, using -ragged", |
"list type must come first", |
"list type is not the first argument", |
"tag lists require a width argument", |
"missing -width in -tag list, using 8n", |
"missing font type", |
"missing utility name, using \"\"", |
"skipping end of block that is not open", |
"empty head in list item", |
|
"empty list item", |
|
"missing font type, using \\fR", |
|
"unknown font type, using \\fR", |
|
"missing -std argument, adding it", |
|
|
/* related to bad macro arguments */ |
/* related to bad macro arguments */ |
"skipping argument", |
"unterminated quoted argument", |
"duplicate argument", |
"duplicate argument", |
"duplicate display type", |
"skipping duplicate argument", |
"duplicate list type", |
"skipping duplicate display type", |
|
"skipping duplicate list type", |
|
"skipping -width argument", |
"unknown AT&T UNIX version", |
"unknown AT&T UNIX version", |
"bad Boolean value", |
"invalid content in Rs block", |
"unknown font", |
"invalid Boolean argument", |
"unknown standard specifier", |
"unknown font, skipping request", |
"bad width argument", |
|
|
|
/* related to plain text */ |
/* related to plain text */ |
"blank line in non-literal context", |
"blank line in fill mode, using .sp", |
"tab in non-literal context", |
"tab in filled text", |
"end of line whitespace", |
"whitespace at end of input line", |
"bad comment style", |
"bad comment style", |
"bad escape sequence", |
"invalid escape sequence", |
"unterminated quoted string", |
"undefined string, using \"\"", |
|
|
/* related to equations */ |
|
"unexpected literal in equation", |
|
|
|
"generic error", |
"generic error", |
|
|
/* related to equations */ |
/* related to equations */ |
Line 175 static const char * const mandocerrs[MANDOCERR_MAX] = |
|
Line 189 static const char * const mandocerrs[MANDOCERR_MAX] = |
|
"data block still open", |
"data block still open", |
"ignoring extra data cells", |
"ignoring extra data cells", |
|
|
|
/* related to document structure and macros */ |
"input stack limit exceeded, infinite loop?", |
"input stack limit exceeded, infinite loop?", |
"skipping bad character", |
"skipping bad character", |
"escaped character not allowed in a name", |
|
"manual name not yet set", |
|
"skipping text before the first section header", |
|
"skipping unknown macro", |
"skipping unknown macro", |
"NOT IMPLEMENTED, please use groff: skipping request", |
"skipping item outside list", |
"argument count wrong", |
|
"skipping column outside column list", |
"skipping column outside column list", |
"skipping end of block that is not open", |
"skipping end of block that is not open", |
"missing end of block", |
"inserting missing end of block", |
"scope open on exit", |
"appending missing end of block", |
"uname(3) system call failed", |
|
"macro requires line argument(s)", |
|
"macro requires body argument(s)", |
|
"macro requires argument(s)", |
|
"request requires a numeric argument", |
|
"missing list type", |
|
"line argument(s) will be lost", |
|
"body argument(s) will be lost", |
|
|
|
|
/* related to request and macro arguments */ |
|
"escaped character not allowed in a name", |
|
"argument count wrong", |
|
"missing list type, using -item", |
|
"missing manual name, using \"\"", |
|
"uname(3) system call failed, using UNKNOWN", |
|
"unknown standard specifier", |
|
"skipping request without numeric argument", |
|
"skipping all arguments", |
|
"skipping excess arguments", |
|
|
"generic fatal error", |
"generic fatal error", |
|
|
"input too large", |
"input too large", |
"not a manual", |
"NOT IMPLEMENTED: Bd -file", |
"column syntax is inconsistent", |
|
"NOT IMPLEMENTED: .Bd -file", |
|
"argument count wrong, violates syntax", |
|
"child violates parent syntax", |
|
"argument count wrong, violates syntax", |
|
"NOT IMPLEMENTED: .so with absolute path or \"..\"", |
"NOT IMPLEMENTED: .so with absolute path or \"..\"", |
".so request failed", |
".so request failed", |
"no document prologue", |
|
"static buffer exhausted", |
|
|
|
/* system errors */ |
/* system errors */ |
|
"cannot dup file descriptor", |
|
"cannot exec", |
|
"gunzip failed with code", |
|
"cannot fork", |
NULL, |
NULL, |
"cannot stat file", |
"cannot open pipe", |
"cannot read file", |
"cannot read file", |
|
"gunzip died from signal", |
|
"cannot stat file", |
|
"wait failed", |
}; |
}; |
|
|
static const char * const mandoclevels[MANDOCLEVEL_MAX] = { |
static const char * const mandoclevels[MANDOCLEVEL_MAX] = { |
Line 236 resize_buf(struct buf *buf, size_t initial) |
|
Line 250 resize_buf(struct buf *buf, size_t initial) |
|
} |
} |
|
|
static void |
static void |
pset(const char *buf, int pos, struct mparse *curp) |
choose_parser(struct mparse *curp) |
{ |
{ |
int i; |
char *cp, *ep; |
|
int format; |
|
|
/* |
/* |
* Try to intuit which kind of manual parser should be used. If |
* If neither command line arguments -mdoc or -man select |
* passed in by command-line (-man, -mdoc), then use that |
* a parser nor the roff parser found a .Dd or .TH macro |
* explicitly. If passed as -mandoc, then try to guess from the |
* yet, look ahead in the main input buffer. |
* line: either skip dot-lines, use -mdoc when finding `.Dt', or |
|
* default to -man, which is more lenient. |
|
* |
|
* Separate out pmdoc/pman from mdoc/man: the first persists |
|
* through all parsers, while the latter is used per-parse. |
|
*/ |
*/ |
|
|
if ('.' == buf[0] || '\'' == buf[0]) { |
if ((format = roff_getformat(curp->roff)) == 0) { |
for (i = 1; buf[i]; i++) |
cp = curp->primary->buf; |
if (' ' != buf[i] && '\t' != buf[i]) |
ep = cp + curp->primary->sz; |
|
while (cp < ep) { |
|
if (*cp == '.' || *cp == '\'') { |
|
cp++; |
|
if (cp[0] == 'D' && cp[1] == 'd') { |
|
format = MPARSE_MDOC; |
|
break; |
|
} |
|
if (cp[0] == 'T' && cp[1] == 'H') { |
|
format = MPARSE_MAN; |
|
break; |
|
} |
|
} |
|
cp = memchr(cp, '\n', ep - cp); |
|
if (cp == NULL) |
break; |
break; |
if ('\0' == buf[i]) |
cp++; |
return; |
} |
} |
} |
|
|
if (MPARSE_MDOC & curp->options) { |
if (format == MPARSE_MDOC) { |
if (NULL == curp->pmdoc) |
if (NULL == curp->pmdoc) |
curp->pmdoc = mdoc_alloc( |
curp->pmdoc = mdoc_alloc( |
curp->roff, curp, curp->defos, |
curp->roff, curp, curp->defos, |
Line 267 pset(const char *buf, int pos, struct mparse *curp) |
|
Line 291 pset(const char *buf, int pos, struct mparse *curp) |
|
assert(curp->pmdoc); |
assert(curp->pmdoc); |
curp->mdoc = curp->pmdoc; |
curp->mdoc = curp->pmdoc; |
return; |
return; |
} else if (MPARSE_MAN & curp->options) { |
|
if (NULL == curp->pman) |
|
curp->pman = man_alloc(curp->roff, curp, |
|
MPARSE_QUICK & curp->options ? 1 : 0); |
|
assert(curp->pman); |
|
curp->man = curp->pman; |
|
return; |
|
} |
} |
|
|
if (pos >= 3 && 0 == memcmp(buf, ".Dd", 3)) { |
/* Fall back to man(7) as a last resort. */ |
if (NULL == curp->pmdoc) |
|
curp->pmdoc = mdoc_alloc( |
|
curp->roff, curp, curp->defos, |
|
MPARSE_QUICK & curp->options ? 1 : 0); |
|
assert(curp->pmdoc); |
|
curp->mdoc = curp->pmdoc; |
|
return; |
|
} |
|
|
|
if (NULL == curp->pman) |
if (NULL == curp->pman) |
curp->pman = man_alloc(curp->roff, curp, |
curp->pman = man_alloc(curp->roff, curp, |
Line 362 mparse_buf_r(struct mparse *curp, struct buf blk, int |
|
Line 371 mparse_buf_r(struct mparse *curp, struct buf blk, int |
|
|
|
if ( ! (isascii(c) && |
if ( ! (isascii(c) && |
(isgraph(c) || isblank(c)))) { |
(isgraph(c) || isblank(c)))) { |
mandoc_msg(MANDOCERR_BADCHAR, curp, |
mandoc_vmsg(MANDOCERR_BADCHAR, curp, |
curp->line, pos, NULL); |
curp->line, pos, "0x%x", c); |
i++; |
i++; |
ln.buf[pos++] = '?'; |
ln.buf[pos++] = '?'; |
continue; |
continue; |
Line 419 mparse_buf_r(struct mparse *curp, struct buf blk, int |
|
Line 428 mparse_buf_r(struct mparse *curp, struct buf blk, int |
|
|
|
if ( ! (isascii(c) && |
if ( ! (isascii(c) && |
(isgraph(c) || isblank(c)))) { |
(isgraph(c) || isblank(c)))) { |
mandoc_msg(MANDOCERR_BADCHAR, curp, |
mandoc_vmsg(MANDOCERR_BADCHAR, curp, |
curp->line, pos, NULL); |
curp->line, pos, "0x%x", c); |
i += 2; |
i += 2; |
ln.buf[pos++] = '?'; |
ln.buf[pos++] = '?'; |
continue; |
continue; |
|
|
*/ |
*/ |
|
|
if ( ! (curp->man || curp->mdoc)) |
if ( ! (curp->man || curp->mdoc)) |
pset(ln.buf + of, pos - of, curp); |
choose_parser(curp); |
|
|
/* |
/* |
* Lastly, push down into the parsers themselves. One |
* Lastly, push down into the parsers themselves. |
* of these will have already been set in the pset() |
|
* routine. |
|
* If libroff returns ROFF_TBL, then add it to the |
* If libroff returns ROFF_TBL, then add it to the |
* currently open parse. Since we only get here if |
* currently open parse. Since we only get here if |
* there does exist data (see tbl_data.c), we're |
* there does exist data (see tbl_data.c), we're |
Line 597 read_whole_file(struct mparse *curp, const char *file, |
|
Line 604 read_whole_file(struct mparse *curp, const char *file, |
|
size_t off; |
size_t off; |
ssize_t ssz; |
ssize_t ssz; |
|
|
#ifdef HAVE_MMAP |
#if HAVE_MMAP |
struct stat st; |
struct stat st; |
if (-1 == fstat(fd, &st)) { |
if (-1 == fstat(fd, &st)) { |
curp->file_status = MANDOCLEVEL_SYSERR; |
curp->file_status = MANDOCLEVEL_SYSERR; |
Line 679 mparse_end(struct mparse *curp) |
|
Line 686 mparse_end(struct mparse *curp) |
|
if (MANDOCLEVEL_FATAL <= curp->file_status) |
if (MANDOCLEVEL_FATAL <= curp->file_status) |
return; |
return; |
|
|
|
if (curp->mdoc == NULL && |
|
curp->man == NULL && |
|
curp->sodest == NULL) { |
|
if (curp->options & MPARSE_MDOC) |
|
curp->mdoc = curp->pmdoc; |
|
else { |
|
if (curp->pman == NULL) |
|
curp->pman = man_alloc(curp->roff, curp, |
|
curp->options & MPARSE_QUICK ? 1 : 0); |
|
curp->man = curp->pman; |
|
} |
|
} |
|
|
if (curp->mdoc && ! mdoc_endparse(curp->mdoc)) { |
if (curp->mdoc && ! mdoc_endparse(curp->mdoc)) { |
assert(MANDOCLEVEL_FATAL <= curp->file_status); |
assert(MANDOCLEVEL_FATAL <= curp->file_status); |
return; |
return; |
Line 689 mparse_end(struct mparse *curp) |
|
Line 709 mparse_end(struct mparse *curp) |
|
return; |
return; |
} |
} |
|
|
if ( ! (curp->mdoc || curp->man || curp->sodest)) { |
|
mandoc_msg(MANDOCERR_NOTMANUAL, curp, 0, 0, NULL); |
|
curp->file_status = MANDOCLEVEL_FATAL; |
|
return; |
|
} |
|
|
|
roff_endparse(curp->roff); |
roff_endparse(curp->roff); |
} |
} |
|
|
static void |
static void |
mparse_parse_buffer(struct mparse *curp, struct buf blk, const char *file) |
mparse_parse_buffer(struct mparse *curp, struct buf blk, const char *file) |
{ |
{ |
|
struct buf *svprimary; |
const char *svfile; |
const char *svfile; |
static int recursion_depth; |
static int recursion_depth; |
|
|
Line 712 mparse_parse_buffer(struct mparse *curp, struct buf bl |
|
Line 727 mparse_parse_buffer(struct mparse *curp, struct buf bl |
|
/* Line number is per-file. */ |
/* Line number is per-file. */ |
svfile = curp->file; |
svfile = curp->file; |
curp->file = file; |
curp->file = file; |
|
svprimary = curp->primary; |
|
curp->primary = &blk; |
curp->line = 1; |
curp->line = 1; |
recursion_depth++; |
recursion_depth++; |
|
|
Line 720 mparse_parse_buffer(struct mparse *curp, struct buf bl |
|
Line 737 mparse_parse_buffer(struct mparse *curp, struct buf bl |
|
if (0 == --recursion_depth && MANDOCLEVEL_FATAL > curp->file_status) |
if (0 == --recursion_depth && MANDOCLEVEL_FATAL > curp->file_status) |
mparse_end(curp); |
mparse_end(curp); |
|
|
|
curp->primary = svprimary; |
curp->file = svfile; |
curp->file = svfile; |
} |
} |
|
|
Line 763 mparse_readfd(struct mparse *curp, int fd, const char |
|
Line 781 mparse_readfd(struct mparse *curp, int fd, const char |
|
|
|
mparse_parse_buffer(curp, blk, file); |
mparse_parse_buffer(curp, blk, file); |
|
|
#ifdef HAVE_MMAP |
#if HAVE_MMAP |
if (with_mmap) |
if (with_mmap) |
munmap(blk.buf, blk.sz); |
munmap(blk.buf, blk.sz); |
else |
else |
|
|
return(curp->file_status); |
return(curp->file_status); |
} |
} |
|
|
|
enum mandoclevel |
|
mparse_open(struct mparse *curp, int *fd, const char *file, |
|
pid_t *child_pid) |
|
{ |
|
int pfd[2]; |
|
char *cp; |
|
enum mandocerr err; |
|
|
|
pfd[1] = -1; |
|
curp->file = file; |
|
if ((cp = strrchr(file, '.')) == NULL || |
|
strcmp(cp + 1, "gz")) { |
|
*child_pid = 0; |
|
if ((*fd = open(file, O_RDONLY)) == -1) { |
|
err = MANDOCERR_SYSOPEN; |
|
goto out; |
|
} |
|
return(MANDOCLEVEL_OK); |
|
} |
|
|
|
if (pipe(pfd) == -1) { |
|
err = MANDOCERR_SYSPIPE; |
|
goto out; |
|
} |
|
|
|
switch (*child_pid = fork()) { |
|
case -1: |
|
err = MANDOCERR_SYSFORK; |
|
close(pfd[0]); |
|
close(pfd[1]); |
|
pfd[1] = -1; |
|
break; |
|
case 0: |
|
close(pfd[0]); |
|
if (dup2(pfd[1], STDOUT_FILENO) == -1) { |
|
err = MANDOCERR_SYSDUP; |
|
break; |
|
} |
|
execlp("gunzip", "gunzip", "-c", file, NULL); |
|
err = MANDOCERR_SYSEXEC; |
|
break; |
|
default: |
|
close(pfd[1]); |
|
*fd = pfd[0]; |
|
return(MANDOCLEVEL_OK); |
|
} |
|
|
|
out: |
|
*fd = -1; |
|
*child_pid = 0; |
|
curp->file_status = MANDOCLEVEL_SYSERR; |
|
if (curp->mmsg) |
|
(*curp->mmsg)(err, curp->file_status, file, |
|
0, 0, strerror(errno)); |
|
if (pfd[1] != -1) |
|
exit(1); |
|
return(curp->file_status); |
|
} |
|
|
|
enum mandoclevel |
|
mparse_wait(struct mparse *curp, pid_t child_pid) |
|
{ |
|
int status; |
|
|
|
if (waitpid(child_pid, &status, 0) == -1) { |
|
mandoc_msg(MANDOCERR_SYSWAIT, curp, 0, 0, |
|
strerror(errno)); |
|
curp->file_status = MANDOCLEVEL_SYSERR; |
|
return(curp->file_status); |
|
} |
|
if (WIFSIGNALED(status)) { |
|
mandoc_vmsg(MANDOCERR_SYSSIG, curp, 0, 0, |
|
"%d", WTERMSIG(status)); |
|
curp->file_status = MANDOCLEVEL_SYSERR; |
|
return(curp->file_status); |
|
} |
|
if (WEXITSTATUS(status)) { |
|
mandoc_vmsg(MANDOCERR_SYSEXIT, curp, 0, 0, |
|
"%d", WEXITSTATUS(status)); |
|
curp->file_status = MANDOCLEVEL_SYSERR; |
|
return(curp->file_status); |
|
} |
|
return(MANDOCLEVEL_OK); |
|
} |
|
|
struct mparse * |
struct mparse * |
mparse_alloc(int options, enum mandoclevel wlevel, |
mparse_alloc(int options, enum mandoclevel wlevel, |
mandocmsg mmsg, char *defos) |
mandocmsg mmsg, const char *defos) |
{ |
{ |
struct mparse *curp; |
struct mparse *curp; |
|
|
Line 792 mparse_alloc(int options, enum mandoclevel wlevel, |
|
Line 895 mparse_alloc(int options, enum mandoclevel wlevel, |
|
curp->defos = defos; |
curp->defos = defos; |
|
|
curp->roff = roff_alloc(curp, options); |
curp->roff = roff_alloc(curp, options); |
|
if (curp->options & MPARSE_MDOC) |
|
curp->pmdoc = mdoc_alloc( |
|
curp->roff, curp, curp->defos, |
|
curp->options & MPARSE_QUICK ? 1 : 0); |
|
if (curp->options & MPARSE_MAN) |
|
curp->pman = man_alloc(curp->roff, curp, |
|
curp->options & MPARSE_QUICK ? 1 : 0); |
|
|
return(curp); |
return(curp); |
} |
} |
|
|