version 1.95, 2014/11/01 04:08:43 |
version 1.110, 2015/01/15 02:29:26 |
|
|
/* $Id$ */ |
/* $Id$ */ |
/* |
/* |
* Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv> |
* Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv> |
* Copyright (c) 2010-2014 Ingo Schwarze <schwarze@openbsd.org> |
* Copyright (c) 2010-2015 Ingo Schwarze <schwarze@openbsd.org> |
* Copyright (c) 2010, 2012 Joerg Sonnenberger <joerg@netbsd.org> |
* Copyright (c) 2010, 2012 Joerg Sonnenberger <joerg@netbsd.org> |
* |
* |
* Permission to use, copy, modify, and distribute this software for any |
* Permission to use, copy, modify, and distribute this software for any |
|
|
#include "libmandoc.h" |
#include "libmandoc.h" |
#include "mdoc.h" |
#include "mdoc.h" |
#include "man.h" |
#include "man.h" |
#include "main.h" |
|
|
|
#define REPARSE_LIMIT 1000 |
#define REPARSE_LIMIT 1000 |
|
|
|
|
int filenc; /* encoding of the current file */ |
int filenc; /* encoding of the current file */ |
int reparse_count; /* finite interp. stack */ |
int reparse_count; /* finite interp. stack */ |
int line; /* line number in the file */ |
int line; /* line number in the file */ |
|
pid_t child; /* the gunzip(1) process */ |
}; |
}; |
|
|
static void choose_parser(struct mparse *); |
static void choose_parser(struct mparse *); |
Line 96 static const char * const mandocerrs[MANDOCERR_MAX] = |
|
Line 96 static const char * const mandocerrs[MANDOCERR_MAX] = |
|
"lower case character in document title", |
"lower case character in document title", |
"missing manual section, using \"\"", |
"missing manual section, using \"\"", |
"unknown manual section", |
"unknown manual section", |
"unknown manual volume or arch", |
|
"missing date, using today's date", |
"missing date, using today's date", |
"cannot parse date, using it verbatim", |
"cannot parse date, using it verbatim", |
"missing Os macro, using \"\"", |
"missing Os macro, using \"\"", |
Line 120 static const char * const mandocerrs[MANDOCERR_MAX] = |
|
Line 119 static const char * const mandocerrs[MANDOCERR_MAX] = |
|
|
|
/* related to macros and nesting */ |
/* related to macros and nesting */ |
"obsolete macro", |
"obsolete macro", |
|
"macro neither callable nor escaped", |
"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", |
Line 145 static const char * const mandocerrs[MANDOCERR_MAX] = |
|
Line 145 static const char * const mandocerrs[MANDOCERR_MAX] = |
|
"empty list item", |
"empty list item", |
"missing font type, using \\fR", |
"missing font type, using \\fR", |
"unknown font type, using \\fR", |
"unknown font type, using \\fR", |
|
"nothing follows prefix", |
"missing -std argument, adding it", |
"missing -std argument, adding it", |
"missing eqn box, using \"\"", |
"missing eqn box, using \"\"", |
|
|
Line 187 static const char * const mandocerrs[MANDOCERR_MAX] = |
|
Line 188 static const char * const mandocerrs[MANDOCERR_MAX] = |
|
"ignore data in cell", |
"ignore data in cell", |
"data block still open", |
"data block still open", |
"ignoring extra data cells", |
"ignoring extra data cells", |
|
"ignoring macro in table", |
|
|
/* related to document structure and macros */ |
/* related to document structure and macros */ |
|
NULL, |
"input stack limit exceeded, infinite loop?", |
"input stack limit exceeded, infinite loop?", |
"skipping bad character", |
"skipping bad character", |
"skipping unknown macro", |
"skipping unknown macro", |
Line 201 static const char * const mandocerrs[MANDOCERR_MAX] = |
|
Line 204 static const char * const mandocerrs[MANDOCERR_MAX] = |
|
/* related to request and macro arguments */ |
/* related to request and macro arguments */ |
"escaped character not allowed in a name", |
"escaped character not allowed in a name", |
"argument count wrong", |
"argument count wrong", |
|
"NOT IMPLEMENTED: Bd -file", |
"missing list type, using -item", |
"missing list type, using -item", |
"missing manual name, using \"\"", |
"missing manual name, using \"\"", |
"uname(3) system call failed, using UNKNOWN", |
"uname(3) system call failed, using UNKNOWN", |
"unknown standard specifier", |
"unknown standard specifier", |
"skipping request without numeric argument", |
"skipping request without numeric argument", |
|
"NOT IMPLEMENTED: .so with absolute path or \"..\"", |
|
".so request failed", |
"skipping all arguments", |
"skipping all arguments", |
"skipping excess arguments", |
"skipping excess arguments", |
"divide by zero", |
"divide by zero", |
Line 213 static const char * const mandocerrs[MANDOCERR_MAX] = |
|
Line 219 static const char * const mandocerrs[MANDOCERR_MAX] = |
|
"generic fatal error", |
"generic fatal error", |
|
|
"input too large", |
"input too large", |
"NOT IMPLEMENTED: Bd -file", |
|
"NOT IMPLEMENTED: .so with absolute path or \"..\"", |
|
".so request failed", |
|
|
|
/* system errors */ |
|
"cannot dup file descriptor", |
|
"cannot exec", |
|
"gunzip failed with code", |
|
"cannot fork", |
|
NULL, |
|
"cannot open pipe", |
|
"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 296 choose_parser(struct mparse *curp) |
|
Line 287 choose_parser(struct mparse *curp) |
|
/* Fall back to man(7) as a last resort. */ |
/* Fall back to man(7) as a last resort. */ |
|
|
if (NULL == curp->pman) |
if (NULL == curp->pman) |
curp->pman = man_alloc(curp->roff, curp, |
curp->pman = man_alloc( |
|
curp->roff, curp, curp->defos, |
MPARSE_QUICK & curp->options ? 1 : 0); |
MPARSE_QUICK & curp->options ? 1 : 0); |
assert(curp->pman); |
assert(curp->pman); |
curp->man = curp->pman; |
curp->man = curp->pman; |
Line 314 mparse_buf_r(struct mparse *curp, struct buf blk, size |
|
Line 306 mparse_buf_r(struct mparse *curp, struct buf blk, size |
|
{ |
{ |
const struct tbl_span *span; |
const struct tbl_span *span; |
struct buf ln; |
struct buf ln; |
|
char *cp; |
size_t pos; /* byte number in the ln buffer */ |
size_t pos; /* byte number in the ln buffer */ |
enum rofferr rr; |
enum rofferr rr; |
int of, rc; |
int of; |
int lnn; /* line number in the real file */ |
int lnn; /* line number in the real file */ |
|
int fd; |
|
pid_t save_child; |
unsigned char c; |
unsigned char c; |
|
|
memset(&ln, 0, sizeof(ln)); |
memset(&ln, 0, sizeof(ln)); |
Line 493 mparse_buf_r(struct mparse *curp, struct buf blk, size |
|
Line 488 mparse_buf_r(struct mparse *curp, struct buf blk, size |
|
[curp->secondary->sz] = '\0'; |
[curp->secondary->sz] = '\0'; |
} |
} |
rerun: |
rerun: |
rr = roff_parseln(curp->roff, curp->line, |
rr = roff_parseln(curp->roff, curp->line, &ln, &of); |
&ln.buf, &ln.sz, of, &of); |
|
|
|
switch (rr) { |
switch (rr) { |
case ROFF_REPARSE: |
case ROFF_REPARSE: |
|
|
case ROFF_IGN: |
case ROFF_IGN: |
pos = 0; |
pos = 0; |
continue; |
continue; |
case ROFF_ERR: |
|
assert(MANDOCLEVEL_FATAL <= curp->file_status); |
|
break; |
|
case ROFF_SO: |
case ROFF_SO: |
if ( ! (curp->options & MPARSE_SO) && |
if ( ! (curp->options & MPARSE_SO) && |
(i >= blk.sz || blk.buf[i] == '\0')) { |
(i >= blk.sz || blk.buf[i] == '\0')) { |
|
|
*/ |
*/ |
if (curp->secondary) |
if (curp->secondary) |
curp->secondary->sz -= pos + 1; |
curp->secondary->sz -= pos + 1; |
mparse_readfd(curp, -1, ln.buf + of); |
save_child = curp->child; |
if (MANDOCLEVEL_FATAL <= curp->file_status) { |
if (mparse_open(curp, &fd, ln.buf + of) == |
|
MANDOCLEVEL_OK) |
|
mparse_readfd(curp, fd, ln.buf + of); |
|
else { |
mandoc_vmsg(MANDOCERR_SO_FAIL, |
mandoc_vmsg(MANDOCERR_SO_FAIL, |
curp, curp->line, pos, |
curp, curp->line, pos, |
".so %s", ln.buf + of); |
".so %s", ln.buf + of); |
break; |
ln.sz = mandoc_asprintf(&cp, |
|
".sp\nSee the file %s.\n.sp", |
|
ln.buf + of); |
|
free(ln.buf); |
|
ln.buf = cp; |
|
of = 0; |
|
mparse_buf_r(curp, ln, of, 0); |
} |
} |
|
curp->child = save_child; |
pos = 0; |
pos = 0; |
continue; |
continue; |
default: |
default: |
|
|
* Do the same for ROFF_EQN. |
* Do the same for ROFF_EQN. |
*/ |
*/ |
|
|
rc = -1; |
if (rr == ROFF_TBL) { |
|
while ((span = roff_span(curp->roff)) != NULL) |
|
if (curp->man == NULL) |
|
mdoc_addspan(curp->mdoc, span); |
|
else |
|
man_addspan(curp->man, span); |
|
} else if (rr == ROFF_EQN) { |
|
if (curp->man == NULL) |
|
mdoc_addeqn(curp->mdoc, roff_eqn(curp->roff)); |
|
else |
|
man_addeqn(curp->man, roff_eqn(curp->roff)); |
|
} else if ((curp->man == NULL ? |
|
mdoc_parseln(curp->mdoc, curp->line, ln.buf, of) : |
|
man_parseln(curp->man, curp->line, ln.buf, of)) == 2) |
|
break; |
|
|
if (ROFF_TBL == rr) |
|
while (NULL != (span = roff_span(curp->roff))) { |
|
rc = curp->man ? |
|
man_addspan(curp->man, span) : |
|
mdoc_addspan(curp->mdoc, span); |
|
if (0 == rc) |
|
break; |
|
} |
|
else if (ROFF_EQN == rr) |
|
rc = curp->mdoc ? |
|
mdoc_addeqn(curp->mdoc, |
|
roff_eqn(curp->roff)) : |
|
man_addeqn(curp->man, |
|
roff_eqn(curp->roff)); |
|
else if (curp->man || curp->mdoc) |
|
rc = curp->man ? |
|
man_parseln(curp->man, |
|
curp->line, ln.buf, of) : |
|
mdoc_parseln(curp->mdoc, |
|
curp->line, ln.buf, of); |
|
|
|
if (0 == rc) { |
|
assert(MANDOCLEVEL_FATAL <= curp->file_status); |
|
break; |
|
} else if (2 == rc) |
|
break; |
|
|
|
/* Temporary buffers typically are not full. */ |
/* Temporary buffers typically are not full. */ |
|
|
if (0 == start && '\0' == blk.buf[i]) |
if (0 == start && '\0' == blk.buf[i]) |
Line 622 read_whole_file(struct mparse *curp, const char *file, |
|
Line 610 read_whole_file(struct mparse *curp, const char *file, |
|
#if 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; |
perror(file); |
if (curp->mmsg) |
exit((int)MANDOCLEVEL_SYSERR); |
(*curp->mmsg)(MANDOCERR_SYSSTAT, curp->file_status, |
|
file, 0, 0, strerror(errno)); |
|
return(0); |
|
} |
} |
|
|
/* |
/* |
Line 679 read_whole_file(struct mparse *curp, const char *file, |
|
Line 664 read_whole_file(struct mparse *curp, const char *file, |
|
return(1); |
return(1); |
} |
} |
if (ssz == -1) { |
if (ssz == -1) { |
curp->file_status = MANDOCLEVEL_SYSERR; |
perror(file); |
if (curp->mmsg) |
exit((int)MANDOCLEVEL_SYSERR); |
(*curp->mmsg)(MANDOCERR_SYSREAD, |
|
curp->file_status, file, 0, 0, |
|
strerror(errno)); |
|
break; |
|
} |
} |
off += (size_t)ssz; |
off += (size_t)ssz; |
} |
} |
Line 708 mparse_end(struct mparse *curp) |
|
Line 689 mparse_end(struct mparse *curp) |
|
curp->mdoc = curp->pmdoc; |
curp->mdoc = curp->pmdoc; |
else { |
else { |
if (curp->pman == NULL) |
if (curp->pman == NULL) |
curp->pman = man_alloc(curp->roff, curp, |
curp->pman = man_alloc( |
|
curp->roff, curp, curp->defos, |
curp->options & MPARSE_QUICK ? 1 : 0); |
curp->options & MPARSE_QUICK ? 1 : 0); |
curp->man = curp->pman; |
curp->man = curp->pman; |
} |
} |
Line 768 mparse_parse_buffer(struct mparse *curp, struct buf bl |
|
Line 750 mparse_parse_buffer(struct mparse *curp, struct buf bl |
|
} |
} |
|
|
enum mandoclevel |
enum mandoclevel |
mparse_readmem(struct mparse *curp, const void *buf, size_t len, |
mparse_readmem(struct mparse *curp, void *buf, size_t len, |
const char *file) |
const char *file) |
{ |
{ |
struct buf blk; |
struct buf blk; |
|
|
blk.buf = UNCONST(buf); |
blk.buf = buf; |
blk.sz = len; |
blk.sz = len; |
|
|
mparse_parse_buffer(curp, blk, file); |
mparse_parse_buffer(curp, blk, file); |
return(curp->file_status); |
return(curp->file_status); |
} |
} |
|
|
|
/* |
|
* Read the whole file into memory and call the parsers. |
|
* Called recursively when an .so request is encountered. |
|
*/ |
enum mandoclevel |
enum mandoclevel |
mparse_readfd(struct mparse *curp, int fd, const char *file) |
mparse_readfd(struct mparse *curp, int fd, const char *file) |
{ |
{ |
Line 787 mparse_readfd(struct mparse *curp, int fd, const char |
|
Line 773 mparse_readfd(struct mparse *curp, int fd, const char |
|
int with_mmap; |
int with_mmap; |
int save_filenc; |
int save_filenc; |
|
|
if (-1 == fd && -1 == (fd = open(file, O_RDONLY, 0))) { |
|
curp->file_status = MANDOCLEVEL_SYSERR; |
|
if (curp->mmsg) |
|
(*curp->mmsg)(MANDOCERR_SYSOPEN, |
|
curp->file_status, |
|
file, 0, 0, strerror(errno)); |
|
return(curp->file_status); |
|
} |
|
|
|
/* |
|
* Run for each opened file; may be called more than once for |
|
* each full parse sequence if the opened file is nested (i.e., |
|
* from `so'). Simply sucks in the whole file and moves into |
|
* the parse phase for the file. |
|
*/ |
|
|
|
if (read_whole_file(curp, file, fd, &blk, &with_mmap)) { |
if (read_whole_file(curp, file, fd, &blk, &with_mmap)) { |
save_filenc = curp->filenc; |
save_filenc = curp->filenc; |
curp->filenc = curp->options & |
curp->filenc = curp->options & |
Line 817 mparse_readfd(struct mparse *curp, int fd, const char |
|
Line 787 mparse_readfd(struct mparse *curp, int fd, const char |
|
free(blk.buf); |
free(blk.buf); |
} |
} |
|
|
if (STDIN_FILENO != fd && -1 == close(fd)) |
if (fd != STDIN_FILENO && close(fd) == -1) |
perror(file); |
perror(file); |
|
|
|
mparse_wait(curp); |
return(curp->file_status); |
return(curp->file_status); |
} |
} |
|
|
enum mandoclevel |
enum mandoclevel |
mparse_open(struct mparse *curp, int *fd, const char *file, |
mparse_open(struct mparse *curp, int *fd, const char *file) |
pid_t *child_pid) |
|
{ |
{ |
int pfd[2]; |
int pfd[2]; |
|
int save_errno; |
char *cp; |
char *cp; |
enum mandocerr err; |
|
|
|
pfd[1] = -1; |
|
curp->file = file; |
curp->file = file; |
|
|
|
/* Unless zipped, try to just open the file. */ |
|
|
if ((cp = strrchr(file, '.')) == NULL || |
if ((cp = strrchr(file, '.')) == NULL || |
strcmp(cp + 1, "gz")) { |
strcmp(cp + 1, "gz")) { |
*child_pid = 0; |
curp->child = 0; |
if ((*fd = open(file, O_RDONLY)) == -1) { |
if ((*fd = open(file, O_RDONLY)) != -1) |
err = MANDOCERR_SYSOPEN; |
return(MANDOCLEVEL_OK); |
goto out; |
|
} |
/* Open failed; try to append ".gz". */ |
return(MANDOCLEVEL_OK); |
|
|
mandoc_asprintf(&cp, "%s.gz", file); |
|
file = cp; |
|
} else |
|
cp = NULL; |
|
|
|
/* Before forking, make sure the file can be read. */ |
|
|
|
save_errno = errno; |
|
if (access(file, R_OK) == -1) { |
|
if (cp != NULL) |
|
errno = save_errno; |
|
free(cp); |
|
*fd = -1; |
|
curp->child = 0; |
|
mandoc_msg(MANDOCERR_FILE, curp, 0, 0, strerror(errno)); |
|
return(MANDOCLEVEL_ERROR); |
} |
} |
|
|
|
/* Run gunzip(1). */ |
|
|
if (pipe(pfd) == -1) { |
if (pipe(pfd) == -1) { |
err = MANDOCERR_SYSPIPE; |
perror("pipe"); |
goto out; |
exit((int)MANDOCLEVEL_SYSERR); |
} |
} |
|
|
switch (*child_pid = fork()) { |
switch (curp->child = fork()) { |
case -1: |
case -1: |
err = MANDOCERR_SYSFORK; |
perror("fork"); |
close(pfd[0]); |
exit((int)MANDOCLEVEL_SYSERR); |
close(pfd[1]); |
|
pfd[1] = -1; |
|
break; |
|
case 0: |
case 0: |
close(pfd[0]); |
close(pfd[0]); |
if (dup2(pfd[1], STDOUT_FILENO) == -1) { |
if (dup2(pfd[1], STDOUT_FILENO) == -1) { |
err = MANDOCERR_SYSDUP; |
perror("dup"); |
break; |
exit((int)MANDOCLEVEL_SYSERR); |
} |
} |
execlp("gunzip", "gunzip", "-c", file, NULL); |
execlp("gunzip", "gunzip", "-c", file, NULL); |
err = MANDOCERR_SYSEXEC; |
perror("exec"); |
break; |
exit((int)MANDOCLEVEL_SYSERR); |
default: |
default: |
close(pfd[1]); |
close(pfd[1]); |
*fd = pfd[0]; |
*fd = pfd[0]; |
return(MANDOCLEVEL_OK); |
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 |
enum mandoclevel |
mparse_wait(struct mparse *curp, pid_t child_pid) |
mparse_wait(struct mparse *curp) |
{ |
{ |
int status; |
int status; |
|
|
if (waitpid(child_pid, &status, 0) == -1) { |
if (curp->child == 0) |
mandoc_msg(MANDOCERR_SYSWAIT, curp, 0, 0, |
return(MANDOCLEVEL_OK); |
strerror(errno)); |
|
curp->file_status = MANDOCLEVEL_SYSERR; |
if (waitpid(curp->child, &status, 0) == -1) { |
return(curp->file_status); |
perror("wait"); |
|
exit((int)MANDOCLEVEL_SYSERR); |
} |
} |
if (WIFSIGNALED(status)) { |
if (WIFSIGNALED(status)) { |
mandoc_vmsg(MANDOCERR_SYSSIG, curp, 0, 0, |
mandoc_vmsg(MANDOCERR_FILE, curp, 0, 0, |
"%d", WTERMSIG(status)); |
"gunzip died from signal %d", WTERMSIG(status)); |
curp->file_status = MANDOCLEVEL_SYSERR; |
return(MANDOCLEVEL_ERROR); |
return(curp->file_status); |
|
} |
} |
if (WEXITSTATUS(status)) { |
if (WEXITSTATUS(status)) { |
mandoc_vmsg(MANDOCERR_SYSEXIT, curp, 0, 0, |
mandoc_vmsg(MANDOCERR_FILE, curp, 0, 0, |
"%d", WEXITSTATUS(status)); |
"gunzip failed with code %d", WEXITSTATUS(status)); |
curp->file_status = MANDOCLEVEL_SYSERR; |
return(MANDOCLEVEL_ERROR); |
return(curp->file_status); |
|
} |
} |
return(MANDOCLEVEL_OK); |
return(MANDOCLEVEL_OK); |
} |
} |
Line 930 mparse_alloc(int options, enum mandoclevel wlevel, man |
|
Line 905 mparse_alloc(int options, enum mandoclevel wlevel, man |
|
curp->roff, curp, curp->defos, |
curp->roff, curp, curp->defos, |
curp->options & MPARSE_QUICK ? 1 : 0); |
curp->options & MPARSE_QUICK ? 1 : 0); |
if (curp->options & MPARSE_MAN) |
if (curp->options & MPARSE_MAN) |
curp->pman = man_alloc(curp->roff, curp, |
curp->pman = man_alloc( |
|
curp->roff, curp, curp->defos, |
curp->options & MPARSE_QUICK ? 1 : 0); |
curp->options & MPARSE_QUICK ? 1 : 0); |
|
|
return(curp); |
return(curp); |
Line 1015 mandoc_msg(enum mandocerr er, struct mparse *m, |
|
Line 991 mandoc_msg(enum mandocerr er, struct mparse *m, |
|
while (er < mandoclimits[level]) |
while (er < mandoclimits[level]) |
level--; |
level--; |
|
|
if (level < m->wlevel) |
if (level < m->wlevel && er != MANDOCERR_FILE) |
return; |
return; |
|
|
if (m->mmsg) |
if (m->mmsg) |