version 1.140, 2015/07/19 06:05:16 |
version 1.150.2.4, 2017/01/09 02:24:01 |
|
|
/* $Id$ */ |
/* $Id$ */ |
/* |
/* |
* Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv> |
* Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv> |
* Copyright (c) 2010-2015 Ingo Schwarze <schwarze@openbsd.org> |
* Copyright (c) 2010-2017 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 <assert.h> |
#include <assert.h> |
#include <ctype.h> |
#include <ctype.h> |
|
#if HAVE_ERR |
|
#include <err.h> |
|
#endif |
#include <errno.h> |
#include <errno.h> |
#include <fcntl.h> |
#include <fcntl.h> |
#include <stdarg.h> |
#include <stdarg.h> |
|
|
struct mparse { |
struct mparse { |
struct roff_man *man; /* man parser */ |
struct roff_man *man; /* man parser */ |
struct roff *roff; /* roff parser (!NULL) */ |
struct roff *roff; /* roff parser (!NULL) */ |
const struct mchars *mchars; /* character table */ |
|
char *sodest; /* filename pointed to by .so */ |
char *sodest; /* filename pointed to by .so */ |
const char *file; /* filename of current input file */ |
const char *file; /* filename of current input file */ |
struct buf *primary; /* buffer currently being parsed */ |
struct buf *primary; /* buffer currently being parsed */ |
Line 108 static const char * const mandocerrs[MANDOCERR_MAX] = |
|
Line 110 static const char * const mandocerrs[MANDOCERR_MAX] = |
|
"no document body", |
"no document body", |
"content before first section header", |
"content before first section header", |
"first section is not \"NAME\"", |
"first section is not \"NAME\"", |
"NAME section without name", |
"NAME section without Nm before Nd", |
"NAME section without description", |
"NAME section without description", |
"description not at the end of NAME", |
"description not at the end of NAME", |
"bad NAME section content", |
"bad NAME section content", |
|
"missing comma before name", |
"missing description line, using \"\"", |
"missing description line, using \"\"", |
"sections out of conventional order", |
"sections out of conventional order", |
"duplicate section title", |
"duplicate section title", |
Line 129 static const char * const mandocerrs[MANDOCERR_MAX] = |
|
Line 132 static const char * const mandocerrs[MANDOCERR_MAX] = |
|
"blocks badly nested", |
"blocks badly nested", |
"nested displays are not portable", |
"nested displays are not portable", |
"moving content out of list", |
"moving content out of list", |
".Vt block has child macro", |
|
"fill mode already enabled, skipping", |
"fill mode already enabled, skipping", |
"fill mode already disabled, skipping", |
"fill mode already disabled, skipping", |
"line scope broken", |
"line scope broken", |
Line 142 static const char * const mandocerrs[MANDOCERR_MAX] = |
|
Line 144 static const char * const mandocerrs[MANDOCERR_MAX] = |
|
"empty argument, using 0n", |
"empty argument, using 0n", |
"missing display type, using -ragged", |
"missing display type, using -ragged", |
"list type is not the first argument", |
"list type is not the first argument", |
"missing -width in -tag list, using 8n", |
"missing -width in -tag list, using 6n", |
"missing utility name, using \"\"", |
"missing utility name, using \"\"", |
"missing function name, using \"\"", |
"missing function name, using \"\"", |
"empty head in list item", |
"empty head in list item", |
Line 151 static const char * const mandocerrs[MANDOCERR_MAX] = |
|
Line 153 static const char * const mandocerrs[MANDOCERR_MAX] = |
|
"unknown font type, using \\fR", |
"unknown font type, using \\fR", |
"nothing follows prefix", |
"nothing follows prefix", |
"empty reference block", |
"empty reference block", |
|
"missing section argument", |
"missing -std argument, adding it", |
"missing -std argument, adding it", |
"missing option string, using \"\"", |
"missing option string, using \"\"", |
"missing resource identifier, using \"\"", |
"missing resource identifier, using \"\"", |
Line 216 static const char * const mandocerrs[MANDOCERR_MAX] = |
|
Line 219 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", |
"NOT IMPLEMENTED: Bd -file", |
"NOT IMPLEMENTED: Bd -file", |
|
"skipping display without arguments", |
"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", |
Line 289 choose_parser(struct mparse *curp) |
|
Line 293 choose_parser(struct mparse *curp) |
|
} |
} |
} |
} |
|
|
if (curp->man == NULL) { |
|
curp->man = roff_man_alloc(curp->roff, curp, curp->defos, |
|
curp->options & MPARSE_QUICK ? 1 : 0); |
|
curp->man->macroset = MACROSET_MAN; |
|
curp->man->first->tok = TOKEN_NONE; |
|
} |
|
|
|
if (format == MPARSE_MDOC) { |
if (format == MPARSE_MDOC) { |
mdoc_hash_init(); |
mdoc_hash_init(); |
curp->man->macroset = MACROSET_MDOC; |
curp->man->macroset = MACROSET_MDOC; |
Line 322 mparse_buf_r(struct mparse *curp, struct buf blk, size |
|
Line 319 mparse_buf_r(struct mparse *curp, struct buf blk, size |
|
const char *save_file; |
const char *save_file; |
char *cp; |
char *cp; |
size_t pos; /* byte number in the ln buffer */ |
size_t pos; /* byte number in the ln buffer */ |
|
size_t j; /* auxiliary byte number in the blk buffer */ |
enum rofferr rr; |
enum rofferr rr; |
int of; |
int of; |
int lnn; /* line number in the real file */ |
int lnn; /* line number in the real file */ |
Line 427 mparse_buf_r(struct mparse *curp, struct buf blk, size |
|
Line 425 mparse_buf_r(struct mparse *curp, struct buf blk, size |
|
} |
} |
|
|
if ('"' == blk.buf[i + 1] || '#' == blk.buf[i + 1]) { |
if ('"' == blk.buf[i + 1] || '#' == blk.buf[i + 1]) { |
|
j = i; |
i += 2; |
i += 2; |
/* Comment, skip to end of line */ |
/* Comment, skip to end of line */ |
for (; i < blk.sz; ++i) { |
for (; i < blk.sz; ++i) { |
if ('\n' == blk.buf[i]) { |
if (blk.buf[i] != '\n') |
++i; |
continue; |
++lnn; |
if (blk.buf[i - 1] == ' ' || |
break; |
blk.buf[i - 1] == '\t') |
} |
mandoc_msg( |
|
MANDOCERR_SPACE_EOL, |
|
curp, curp->line, |
|
pos + i-1 - j, NULL); |
|
++i; |
|
++lnn; |
|
break; |
} |
} |
|
|
/* Backout trailing whitespaces */ |
/* Backout trailing whitespaces */ |
|
|
if (curp->secondary) |
if (curp->secondary) |
curp->secondary->sz -= pos + 1; |
curp->secondary->sz -= pos + 1; |
save_file = curp->file; |
save_file = curp->file; |
if (mparse_open(curp, &fd, ln.buf + of) == |
if ((fd = mparse_open(curp, ln.buf + of)) != -1) { |
MANDOCLEVEL_OK) { |
|
mparse_readfd(curp, fd, ln.buf + of); |
mparse_readfd(curp, fd, ln.buf + of); |
|
close(fd); |
curp->file = save_file; |
curp->file = save_file; |
} else { |
} else { |
curp->file = save_file; |
curp->file = save_file; |
|
|
break; |
break; |
} |
} |
|
|
/* |
if (curp->man->macroset == MACROSET_NONE) |
* If input parsers have not been allocated, do so now. |
|
* We keep these instanced between parsers, but set them |
|
* locally per parse routine since we can use different |
|
* parsers with each one. |
|
*/ |
|
|
|
if (curp->man == NULL || |
|
curp->man->macroset == MACROSET_NONE) |
|
choose_parser(curp); |
choose_parser(curp); |
|
|
/* |
/* |
Line 613 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)) { |
|
perror(file); |
|
exit((int)MANDOCLEVEL_SYSERR); |
|
} |
|
|
|
|
if (fstat(fd, &st) == -1) |
|
err((int)MANDOCLEVEL_SYSERR, "%s", file); |
|
|
/* |
/* |
* If we're a regular file, try just reading in the whole entry |
* If we're a regular file, try just reading in the whole entry |
* via mmap(). This is faster than reading it into blocks, and |
* via mmap(). This is faster than reading it into blocks, and |
Line 628 read_whole_file(struct mparse *curp, const char *file, |
|
Line 624 read_whole_file(struct mparse *curp, const char *file, |
|
if (curp->gzip == 0 && S_ISREG(st.st_mode)) { |
if (curp->gzip == 0 && S_ISREG(st.st_mode)) { |
if (st.st_size > 0x7fffffff) { |
if (st.st_size > 0x7fffffff) { |
mandoc_msg(MANDOCERR_TOOLARGE, curp, 0, 0, NULL); |
mandoc_msg(MANDOCERR_TOOLARGE, curp, 0, 0, NULL); |
return(0); |
return 0; |
} |
} |
*with_mmap = 1; |
*with_mmap = 1; |
fb->sz = (size_t)st.st_size; |
fb->sz = (size_t)st.st_size; |
fb->buf = mmap(NULL, fb->sz, PROT_READ, MAP_SHARED, fd, 0); |
fb->buf = mmap(NULL, fb->sz, PROT_READ, MAP_SHARED, fd, 0); |
if (fb->buf != MAP_FAILED) |
if (fb->buf != MAP_FAILED) |
return(1); |
return 1; |
} |
} |
#endif |
#endif |
|
|
if (curp->gzip) { |
if (curp->gzip) { |
if ((gz = gzdopen(fd, "rb")) == NULL) { |
if ((gz = gzdopen(fd, "rb")) == NULL) |
perror(file); |
err((int)MANDOCLEVEL_SYSERR, "%s", file); |
exit((int)MANDOCLEVEL_SYSERR); |
|
} |
|
} else |
} else |
gz = NULL; |
gz = NULL; |
|
|
Line 669 read_whole_file(struct mparse *curp, const char *file, |
|
Line 663 read_whole_file(struct mparse *curp, const char *file, |
|
read(fd, fb->buf + (int)off, fb->sz - off); |
read(fd, fb->buf + (int)off, fb->sz - off); |
if (ssz == 0) { |
if (ssz == 0) { |
fb->sz = off; |
fb->sz = off; |
return(1); |
return 1; |
} |
} |
if (ssz == -1) { |
if (ssz == -1) |
perror(file); |
err((int)MANDOCLEVEL_SYSERR, "%s", file); |
exit((int)MANDOCLEVEL_SYSERR); |
|
} |
|
off += (size_t)ssz; |
off += (size_t)ssz; |
} |
} |
|
|
free(fb->buf); |
free(fb->buf); |
fb->buf = NULL; |
fb->buf = NULL; |
return(0); |
return 0; |
} |
} |
|
|
static void |
static void |
mparse_end(struct mparse *curp) |
mparse_end(struct mparse *curp) |
{ |
{ |
|
|
if (curp->man == NULL && curp->sodest == NULL) |
|
curp->man = roff_man_alloc(curp->roff, curp, curp->defos, |
|
curp->options & MPARSE_QUICK ? 1 : 0); |
|
if (curp->man->macroset == MACROSET_NONE) |
if (curp->man->macroset == MACROSET_NONE) |
curp->man->macroset = MACROSET_MAN; |
curp->man->macroset = MACROSET_MAN; |
if (curp->man->macroset == MACROSET_MDOC) |
if (curp->man->macroset == MACROSET_MDOC) |
Line 749 mparse_readmem(struct mparse *curp, void *buf, size_t |
|
Line 737 mparse_readmem(struct mparse *curp, void *buf, size_t |
|
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; |
} |
} |
|
|
/* |
/* |
Line 776 mparse_readfd(struct mparse *curp, int fd, const char |
|
Line 764 mparse_readfd(struct mparse *curp, int fd, const char |
|
#endif |
#endif |
free(blk.buf); |
free(blk.buf); |
} |
} |
|
return curp->file_status; |
if (fd != STDIN_FILENO && close(fd) == -1) |
|
perror(file); |
|
|
|
return(curp->file_status); |
|
} |
} |
|
|
enum mandoclevel |
int |
mparse_open(struct mparse *curp, int *fd, const char *file) |
mparse_open(struct mparse *curp, const char *file) |
{ |
{ |
char *cp; |
char *cp; |
|
int fd; |
|
|
curp->file = file; |
curp->file = file; |
cp = strrchr(file, '.'); |
cp = strrchr(file, '.'); |
Line 794 mparse_open(struct mparse *curp, int *fd, const char * |
|
Line 779 mparse_open(struct mparse *curp, int *fd, const char * |
|
|
|
/* First try to use the filename as it is. */ |
/* First try to use the filename as it is. */ |
|
|
if ((*fd = open(file, O_RDONLY)) != -1) |
if ((fd = open(file, O_RDONLY)) != -1) |
return(MANDOCLEVEL_OK); |
return fd; |
|
|
/* |
/* |
* If that doesn't work and the filename doesn't |
* If that doesn't work and the filename doesn't |
Line 804 mparse_open(struct mparse *curp, int *fd, const char * |
|
Line 789 mparse_open(struct mparse *curp, int *fd, const char * |
|
|
|
if ( ! curp->gzip) { |
if ( ! curp->gzip) { |
mandoc_asprintf(&cp, "%s.gz", file); |
mandoc_asprintf(&cp, "%s.gz", file); |
*fd = open(file, O_RDONLY); |
fd = open(cp, O_RDONLY); |
free(cp); |
free(cp); |
if (*fd != -1) { |
if (fd != -1) { |
curp->gzip = 1; |
curp->gzip = 1; |
return(MANDOCLEVEL_OK); |
return fd; |
} |
} |
} |
} |
|
|
/* Neither worked, give up. */ |
/* Neither worked, give up. */ |
|
|
mandoc_msg(MANDOCERR_FILE, curp, 0, 0, strerror(errno)); |
mandoc_msg(MANDOCERR_FILE, curp, 0, 0, strerror(errno)); |
return(MANDOCLEVEL_ERROR); |
return -1; |
} |
} |
|
|
struct mparse * |
struct mparse * |
mparse_alloc(int options, enum mandoclevel wlevel, mandocmsg mmsg, |
mparse_alloc(int options, enum mandoclevel wlevel, mandocmsg mmsg, |
const struct mchars *mchars, const char *defos) |
const char *defos) |
{ |
{ |
struct mparse *curp; |
struct mparse *curp; |
|
|
Line 831 mparse_alloc(int options, enum mandoclevel wlevel, man |
|
Line 816 mparse_alloc(int options, enum mandoclevel wlevel, man |
|
curp->mmsg = mmsg; |
curp->mmsg = mmsg; |
curp->defos = defos; |
curp->defos = defos; |
|
|
curp->mchars = mchars; |
curp->roff = roff_alloc(curp, options); |
curp->roff = roff_alloc(curp, curp->mchars, options); |
|
curp->man = roff_man_alloc( curp->roff, curp, curp->defos, |
curp->man = roff_man_alloc( curp->roff, curp, curp->defos, |
curp->options & MPARSE_QUICK ? 1 : 0); |
curp->options & MPARSE_QUICK ? 1 : 0); |
if (curp->options & MPARSE_MDOC) { |
if (curp->options & MPARSE_MDOC) { |
Line 843 mparse_alloc(int options, enum mandoclevel wlevel, man |
|
Line 827 mparse_alloc(int options, enum mandoclevel wlevel, man |
|
curp->man->macroset = MACROSET_MAN; |
curp->man->macroset = MACROSET_MAN; |
} |
} |
curp->man->first->tok = TOKEN_NONE; |
curp->man->first->tok = TOKEN_NONE; |
return(curp); |
return curp; |
} |
} |
|
|
void |
void |
mparse_reset(struct mparse *curp) |
mparse_reset(struct mparse *curp) |
{ |
{ |
|
|
roff_reset(curp->roff); |
roff_reset(curp->roff); |
|
roff_man_reset(curp->man); |
if (curp->man != NULL) |
|
roff_man_reset(curp->man); |
|
if (curp->secondary) |
if (curp->secondary) |
curp->secondary->sz = 0; |
curp->secondary->sz = 0; |
|
|
|
|
mparse_strerror(enum mandocerr er) |
mparse_strerror(enum mandocerr er) |
{ |
{ |
|
|
return(mandocerrs[er]); |
return mandocerrs[er]; |
} |
} |
|
|
const char * |
const char * |
mparse_strlevel(enum mandoclevel lvl) |
mparse_strlevel(enum mandoclevel lvl) |
{ |
{ |
return(mandoclevels[lvl]); |
return mandoclevels[lvl]; |
} |
} |
|
|
void |
void |
Line 951 mparse_getkeep(const struct mparse *p) |
|
Line 932 mparse_getkeep(const struct mparse *p) |
|
{ |
{ |
|
|
assert(p->secondary); |
assert(p->secondary); |
return(p->secondary->sz ? p->secondary->buf : NULL); |
return p->secondary->sz ? p->secondary->buf : NULL; |
} |
} |