version 1.205, 2018/12/14 02:16:21 |
version 1.213, 2019/06/03 19:58:02 |
|
|
/* $Id$ */ |
/* $Id$ */ |
/* |
/* |
* Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv> |
* Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv> |
* Copyright (c) 2010-2018 Ingo Schwarze <schwarze@openbsd.org> |
* Copyright (c) 2010-2019 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 |
|
|
struct mparse { |
struct mparse { |
struct roff *roff; /* roff parser (!NULL) */ |
struct roff *roff; /* roff parser (!NULL) */ |
struct roff_man *man; /* man parser */ |
struct roff_man *man; /* man parser */ |
char *sodest; /* filename pointed to by .so */ |
|
struct buf *primary; /* buffer currently being parsed */ |
struct buf *primary; /* buffer currently being parsed */ |
struct buf *secondary; /* copy of top level input */ |
struct buf *secondary; /* copy of top level input */ |
struct buf *loop; /* open .while request line */ |
struct buf *loop; /* open .while request line */ |
Line 123 choose_parser(struct mparse *curp) |
|
Line 122 choose_parser(struct mparse *curp) |
|
} |
} |
|
|
if (format == MPARSE_MDOC) { |
if (format == MPARSE_MDOC) { |
curp->man->macroset = MACROSET_MDOC; |
curp->man->meta.macroset = MACROSET_MDOC; |
if (curp->man->mdocmac == NULL) |
if (curp->man->mdocmac == NULL) |
curp->man->mdocmac = roffhash_alloc(MDOC_Dd, MDOC_MAX); |
curp->man->mdocmac = roffhash_alloc(MDOC_Dd, MDOC_MAX); |
} else { |
} else { |
curp->man->macroset = MACROSET_MAN; |
curp->man->meta.macroset = MACROSET_MAN; |
if (curp->man->manmac == NULL) |
if (curp->man->manmac == NULL) |
curp->man->manmac = roffhash_alloc(MAN_TH, MAN_MAX); |
curp->man->manmac = roffhash_alloc(MAN_TH, MAN_MAX); |
} |
} |
curp->man->first->tok = TOKEN_NONE; |
curp->man->meta.first->tok = TOKEN_NONE; |
} |
} |
|
|
/* |
/* |
Line 158 mparse_buf_r(struct mparse *curp, struct buf blk, size |
|
Line 157 mparse_buf_r(struct mparse *curp, struct buf blk, size |
|
ln.sz = 256; |
ln.sz = 256; |
ln.buf = mandoc_malloc(ln.sz); |
ln.buf = mandoc_malloc(ln.sz); |
ln.next = NULL; |
ln.next = NULL; |
firstln = loop = NULL; |
firstln = lastln = loop = NULL; |
lnn = curp->line; |
lnn = curp->line; |
pos = 0; |
pos = 0; |
inloop = 0; |
inloop = 0; |
Line 208 mparse_buf_r(struct mparse *curp, struct buf blk, size |
|
Line 207 mparse_buf_r(struct mparse *curp, struct buf blk, size |
|
if (c & 0x80) { |
if (c & 0x80) { |
if ( ! (curp->filenc && preconv_encode( |
if ( ! (curp->filenc && preconv_encode( |
&blk, &i, &ln, &pos, &curp->filenc))) { |
&blk, &i, &ln, &pos, &curp->filenc))) { |
mandoc_vmsg(MANDOCERR_CHAR_BAD, curp, |
mandoc_msg(MANDOCERR_CHAR_BAD, |
curp->line, pos, "0x%x", c); |
curp->line, pos, "0x%x", c); |
ln.buf[pos++] = '?'; |
ln.buf[pos++] = '?'; |
i++; |
i++; |
Line 221 mparse_buf_r(struct mparse *curp, struct buf blk, size |
|
Line 220 mparse_buf_r(struct mparse *curp, struct buf blk, size |
|
*/ |
*/ |
|
|
if (c == 0x7f || (c < 0x20 && c != 0x09)) { |
if (c == 0x7f || (c < 0x20 && c != 0x09)) { |
mandoc_vmsg(c == 0x00 || c == 0x04 || |
mandoc_msg(c == 0x00 || c == 0x04 || |
c > 0x0a ? MANDOCERR_CHAR_BAD : |
c > 0x0a ? MANDOCERR_CHAR_BAD : |
MANDOCERR_CHAR_UNSUPP, |
MANDOCERR_CHAR_UNSUPP, |
curp, curp->line, pos, "0x%x", c); |
curp->line, pos, "0x%x", c); |
i++; |
i++; |
if (c != '\r') |
if (c != '\r') |
ln.buf[pos++] = '?'; |
ln.buf[pos++] = '?'; |
Line 256 mparse_buf_r(struct mparse *curp, struct buf blk, size |
|
Line 255 mparse_buf_r(struct mparse *curp, struct buf blk, size |
|
/* XXX Ugly hack to mark the end of the input. */ |
/* XXX Ugly hack to mark the end of the input. */ |
|
|
if (i == blk.sz || blk.buf[i] == '\0') { |
if (i == blk.sz || blk.buf[i] == '\0') { |
|
if (pos + 2 > ln.sz) |
|
resize_buf(&ln, 256); |
ln.buf[pos++] = '\n'; |
ln.buf[pos++] = '\n'; |
ln.buf[pos] = '\0'; |
ln.buf[pos] = '\0'; |
} |
} |
|
|
if (loop == curp->loop) |
if (loop == curp->loop) |
break; |
break; |
mandoc_msg(MANDOCERR_WHILE_NEST, |
mandoc_msg(MANDOCERR_WHILE_NEST, |
curp, curp->line, pos, NULL); |
curp->line, pos, NULL); |
} |
} |
curp->loop = thisln; |
curp->loop = thisln; |
loop = NULL; |
loop = NULL; |
|
|
case ROFF_LOOPEXIT: |
case ROFF_LOOPEXIT: |
if (curp->loop == NULL) { |
if (curp->loop == NULL) { |
mandoc_msg(MANDOCERR_WHILE_FAIL, |
mandoc_msg(MANDOCERR_WHILE_FAIL, |
curp, curp->line, pos, NULL); |
curp->line, pos, NULL); |
break; |
break; |
} |
} |
if (inloop == 0) { |
if (inloop == 0) { |
mandoc_msg(MANDOCERR_WHILE_INTO, |
mandoc_msg(MANDOCERR_WHILE_INTO, |
curp, curp->line, pos, NULL); |
curp->line, pos, NULL); |
curp->loop = loop = NULL; |
curp->loop = loop = NULL; |
break; |
break; |
} |
} |
|
|
case ROFF_IGN: |
case ROFF_IGN: |
break; |
break; |
case ROFF_CONT: |
case ROFF_CONT: |
if (curp->man->macroset == MACROSET_NONE) |
if (curp->man->meta.macroset == MACROSET_NONE) |
choose_parser(curp); |
choose_parser(curp); |
if ((curp->man->macroset == MACROSET_MDOC ? |
if ((curp->man->meta.macroset == MACROSET_MDOC ? |
mdoc_parseln(curp->man, curp->line, ln.buf, of) : |
mdoc_parseln(curp->man, curp->line, ln.buf, of) : |
man_parseln(curp->man, curp->line, ln.buf, of) |
man_parseln(curp->man, curp->line, ln.buf, of) |
) == 2) |
) == 2) |
|
|
if (++curp->reparse_count > REPARSE_LIMIT) { |
if (++curp->reparse_count > REPARSE_LIMIT) { |
/* Abort and return to the top level. */ |
/* Abort and return to the top level. */ |
result = ROFF_IGN; |
result = ROFF_IGN; |
mandoc_msg(MANDOCERR_ROFFLOOP, curp, |
mandoc_msg(MANDOCERR_ROFFLOOP, |
curp->line, pos, NULL); |
curp->line, pos, NULL); |
goto out; |
goto out; |
} |
} |
|
|
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')) { |
curp->sodest = mandoc_strdup(ln.buf + of); |
curp->man->meta.sodest = |
|
mandoc_strdup(ln.buf + of); |
goto out; |
goto out; |
} |
} |
if ((fd = mparse_open(curp, ln.buf + of)) != -1) { |
if ((fd = mparse_open(curp, ln.buf + of)) != -1) { |
mparse_readfd(curp, fd, ln.buf + of); |
mparse_readfd(curp, fd, ln.buf + of); |
close(fd); |
close(fd); |
} else { |
} else { |
mandoc_vmsg(MANDOCERR_SO_FAIL, |
mandoc_msg(MANDOCERR_SO_FAIL, |
curp, curp->line, pos, |
curp->line, of, ".so %s: %s", |
".so %s", ln.buf + of); |
ln.buf + of, strerror(errno)); |
ln.sz = mandoc_asprintf(&cp, |
ln.sz = mandoc_asprintf(&cp, |
".sp\nSee the file %s.\n.sp", |
".sp\nSee the file %s.\n.sp", |
ln.buf + of); |
ln.buf + of); |
|
|
out: |
out: |
if (inloop) { |
if (inloop) { |
if (result != ROFF_USERRET) |
if (result != ROFF_USERRET) |
mandoc_msg(MANDOCERR_WHILE_OUTOF, curp, |
mandoc_msg(MANDOCERR_WHILE_OUTOF, |
curp->line, pos, NULL); |
curp->line, pos, NULL); |
curp->loop = NULL; |
curp->loop = NULL; |
} |
} |
Line 429 read_whole_file(struct mparse *curp, int fd, struct bu |
|
Line 431 read_whole_file(struct mparse *curp, int fd, struct bu |
|
int gzerrnum, retval; |
int gzerrnum, retval; |
|
|
if (fstat(fd, &st) == -1) { |
if (fstat(fd, &st) == -1) { |
mandoc_vmsg(MANDOCERR_FILE, curp, 0, 0, |
mandoc_msg(MANDOCERR_FILE, 0, 0, |
"fstat: %s", strerror(errno)); |
"fstat: %s", strerror(errno)); |
return 0; |
return 0; |
} |
} |
Line 443 read_whole_file(struct mparse *curp, int fd, struct bu |
|
Line 445 read_whole_file(struct mparse *curp, int fd, struct bu |
|
|
|
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, 0, 0, NULL); |
return 0; |
return 0; |
} |
} |
*with_mmap = 1; |
*with_mmap = 1; |
Line 462 read_whole_file(struct mparse *curp, int fd, struct bu |
|
Line 464 read_whole_file(struct mparse *curp, int fd, struct bu |
|
* which this function must not do. |
* which this function must not do. |
*/ |
*/ |
if ((fd = dup(fd)) == -1) { |
if ((fd = dup(fd)) == -1) { |
mandoc_vmsg(MANDOCERR_FILE, curp, 0, 0, |
mandoc_msg(MANDOCERR_FILE, 0, 0, |
"dup: %s", strerror(errno)); |
"dup: %s", strerror(errno)); |
return 0; |
return 0; |
} |
} |
if ((gz = gzdopen(fd, "rb")) == NULL) { |
if ((gz = gzdopen(fd, "rb")) == NULL) { |
mandoc_vmsg(MANDOCERR_FILE, curp, 0, 0, |
mandoc_msg(MANDOCERR_FILE, 0, 0, |
"gzdopen: %s", strerror(errno)); |
"gzdopen: %s", strerror(errno)); |
close(fd); |
close(fd); |
return 0; |
return 0; |
Line 488 read_whole_file(struct mparse *curp, int fd, struct bu |
|
Line 490 read_whole_file(struct mparse *curp, int fd, struct bu |
|
for (;;) { |
for (;;) { |
if (off == fb->sz) { |
if (off == fb->sz) { |
if (fb->sz == (1U << 31)) { |
if (fb->sz == (1U << 31)) { |
mandoc_msg(MANDOCERR_TOOLARGE, curp, |
mandoc_msg(MANDOCERR_TOOLARGE, 0, 0, NULL); |
0, 0, NULL); |
|
break; |
break; |
} |
} |
resize_buf(fb, 65536); |
resize_buf(fb, 65536); |
Line 505 read_whole_file(struct mparse *curp, int fd, struct bu |
|
Line 506 read_whole_file(struct mparse *curp, int fd, struct bu |
|
if (ssz == -1) { |
if (ssz == -1) { |
if (curp->gzip) |
if (curp->gzip) |
(void)gzerror(gz, &gzerrnum); |
(void)gzerror(gz, &gzerrnum); |
mandoc_vmsg(MANDOCERR_FILE, curp, 0, 0, "read: %s", |
mandoc_msg(MANDOCERR_FILE, 0, 0, "read: %s", |
curp->gzip && gzerrnum != Z_ERRNO ? |
curp->gzip && gzerrnum != Z_ERRNO ? |
zError(gzerrnum) : strerror(errno)); |
zError(gzerrnum) : strerror(errno)); |
break; |
break; |
Line 514 read_whole_file(struct mparse *curp, int fd, struct bu |
|
Line 515 read_whole_file(struct mparse *curp, int fd, struct bu |
|
} |
} |
|
|
if (curp->gzip && (gzerrnum = gzclose(gz)) != Z_OK) |
if (curp->gzip && (gzerrnum = gzclose(gz)) != Z_OK) |
mandoc_vmsg(MANDOCERR_FILE, curp, 0, 0, "gzclose: %s", |
mandoc_msg(MANDOCERR_FILE, 0, 0, "gzclose: %s", |
gzerrnum == Z_ERRNO ? strerror(errno) : |
gzerrnum == Z_ERRNO ? strerror(errno) : |
zError(gzerrnum)); |
zError(gzerrnum)); |
if (retval == 0) { |
if (retval == 0) { |
Line 527 read_whole_file(struct mparse *curp, int fd, struct bu |
|
Line 528 read_whole_file(struct mparse *curp, int fd, struct bu |
|
static void |
static void |
mparse_end(struct mparse *curp) |
mparse_end(struct mparse *curp) |
{ |
{ |
if (curp->man->macroset == MACROSET_NONE) |
if (curp->man->meta.macroset == MACROSET_NONE) |
curp->man->macroset = MACROSET_MAN; |
curp->man->meta.macroset = MACROSET_MAN; |
if (curp->man->macroset == MACROSET_MDOC) |
if (curp->man->meta.macroset == MACROSET_MDOC) |
mdoc_endparse(curp->man); |
mdoc_endparse(curp->man); |
else |
else |
man_endparse(curp->man); |
man_endparse(curp->man); |
Line 553 mparse_readfd(struct mparse *curp, int fd, const char |
|
Line 554 mparse_readfd(struct mparse *curp, int fd, const char |
|
int with_mmap; |
int with_mmap; |
|
|
if (recursion_depth > 64) { |
if (recursion_depth > 64) { |
mandoc_msg(MANDOCERR_ROFFLOOP, curp, curp->line, 0, NULL); |
mandoc_msg(MANDOCERR_ROFFLOOP, curp->line, 0, NULL); |
return; |
return; |
} |
} |
if (read_whole_file(curp, fd, &blk, &with_mmap) == 0) |
if (read_whole_file(curp, fd, &blk, &with_mmap) == 0) |
|
|
mparse_open(struct mparse *curp, const char *file) |
mparse_open(struct mparse *curp, const char *file) |
{ |
{ |
char *cp; |
char *cp; |
int fd; |
int fd, save_errno; |
|
|
cp = strrchr(file, '.'); |
cp = strrchr(file, '.'); |
curp->gzip = (cp != NULL && ! strcmp(cp + 1, "gz")); |
curp->gzip = (cp != NULL && ! strcmp(cp + 1, "gz")); |
Line 624 mparse_open(struct mparse *curp, const char *file) |
|
Line 625 mparse_open(struct mparse *curp, const char *file) |
|
*/ |
*/ |
|
|
if ( ! curp->gzip) { |
if ( ! curp->gzip) { |
|
save_errno = errno; |
mandoc_asprintf(&cp, "%s.gz", file); |
mandoc_asprintf(&cp, "%s.gz", file); |
fd = open(cp, O_RDONLY); |
fd = open(cp, O_RDONLY); |
free(cp); |
free(cp); |
|
errno = save_errno; |
if (fd != -1) { |
if (fd != -1) { |
curp->gzip = 1; |
curp->gzip = 1; |
return fd; |
return fd; |
Line 635 mparse_open(struct mparse *curp, const char *file) |
|
Line 638 mparse_open(struct mparse *curp, const char *file) |
|
|
|
/* Neither worked, give up. */ |
/* Neither worked, give up. */ |
|
|
mandoc_msg(MANDOCERR_FILE, curp, 0, 0, strerror(errno)); |
|
return -1; |
return -1; |
} |
} |
|
|
Line 649 mparse_alloc(int options, enum mandoc_os os_e, const c |
|
Line 651 mparse_alloc(int options, enum mandoc_os os_e, const c |
|
curp->options = options; |
curp->options = options; |
curp->os_s = os_s; |
curp->os_s = os_s; |
|
|
curp->roff = roff_alloc(curp, options); |
curp->roff = roff_alloc(options); |
curp->man = roff_man_alloc(curp->roff, curp, curp->os_s, |
curp->man = roff_man_alloc(curp->roff, curp->os_s, |
curp->options & MPARSE_QUICK ? 1 : 0); |
curp->options & MPARSE_QUICK ? 1 : 0); |
if (curp->options & MPARSE_MDOC) { |
if (curp->options & MPARSE_MDOC) { |
curp->man->macroset = MACROSET_MDOC; |
curp->man->meta.macroset = MACROSET_MDOC; |
if (curp->man->mdocmac == NULL) |
if (curp->man->mdocmac == NULL) |
curp->man->mdocmac = roffhash_alloc(MDOC_Dd, MDOC_MAX); |
curp->man->mdocmac = roffhash_alloc(MDOC_Dd, MDOC_MAX); |
} else if (curp->options & MPARSE_MAN) { |
} else if (curp->options & MPARSE_MAN) { |
curp->man->macroset = MACROSET_MAN; |
curp->man->meta.macroset = MACROSET_MAN; |
if (curp->man->manmac == NULL) |
if (curp->man->manmac == NULL) |
curp->man->manmac = roffhash_alloc(MAN_TH, MAN_MAX); |
curp->man->manmac = roffhash_alloc(MAN_TH, MAN_MAX); |
} |
} |
curp->man->first->tok = TOKEN_NONE; |
curp->man->meta.first->tok = TOKEN_NONE; |
curp->man->meta.os_e = os_e; |
curp->man->meta.os_e = os_e; |
return curp; |
return curp; |
} |
} |
Line 673 mparse_reset(struct mparse *curp) |
|
Line 675 mparse_reset(struct mparse *curp) |
|
roff_man_reset(curp->man); |
roff_man_reset(curp->man); |
free_buf_list(curp->secondary); |
free_buf_list(curp->secondary); |
curp->secondary = NULL; |
curp->secondary = NULL; |
free(curp->sodest); |
|
curp->sodest = NULL; |
|
curp->gzip = 0; |
curp->gzip = 0; |
} |
} |
|
|
Line 686 mparse_free(struct mparse *curp) |
|
Line 686 mparse_free(struct mparse *curp) |
|
roff_man_free(curp->man); |
roff_man_free(curp->man); |
roff_free(curp->roff); |
roff_free(curp->roff); |
free_buf_list(curp->secondary); |
free_buf_list(curp->secondary); |
free(curp->sodest); |
|
free(curp); |
free(curp); |
} |
} |
|
|
void |
struct roff_meta * |
mparse_result(struct mparse *curp, struct roff_man **man, |
mparse_result(struct mparse *curp) |
char **sodest) |
|
{ |
{ |
|
roff_state_reset(curp->man); |
if (sodest && NULL != (*sodest = curp->sodest)) { |
if (curp->options & MPARSE_VALIDATE) { |
*man = NULL; |
if (curp->man->meta.macroset == MACROSET_MDOC) |
return; |
mdoc_validate(curp->man); |
|
else |
|
man_validate(curp->man); |
} |
} |
if (man) |
return &curp->man->meta; |
*man = curp->man; |
|
} |
} |
|
|
void |
void |