version 1.204, 2018/12/14 01:24:49 |
version 1.209, 2018/12/30 00:49:55 |
|
|
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 65 static void resize_buf(struct buf *, size_t); |
|
Line 64 static void resize_buf(struct buf *, size_t); |
|
static int mparse_buf_r(struct mparse *, struct buf, size_t, int); |
static int mparse_buf_r(struct mparse *, struct buf, size_t, int); |
static int read_whole_file(struct mparse *, int, struct buf *, int *); |
static int read_whole_file(struct mparse *, int, struct buf *, int *); |
static void mparse_end(struct mparse *); |
static void mparse_end(struct mparse *); |
static void mparse_parse_buffer(struct mparse *, struct buf, |
|
const char *); |
|
|
|
|
|
static void |
static void |
Line 125 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 210 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 223 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++] = '?'; |
|
|
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 431 read_whole_file(struct mparse *curp, int fd, struct bu |
|
Line 429 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 445 read_whole_file(struct mparse *curp, int fd, struct bu |
|
Line 443 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 464 read_whole_file(struct mparse *curp, int fd, struct bu |
|
Line 462 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 490 read_whole_file(struct mparse *curp, int fd, struct bu |
|
Line 488 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 507 read_whole_file(struct mparse *curp, int fd, struct bu |
|
Line 504 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 516 read_whole_file(struct mparse *curp, int fd, struct bu |
|
Line 513 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 529 read_whole_file(struct mparse *curp, int fd, struct bu |
|
Line 526 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); |
roff_endparse(curp->roff); |
roff_endparse(curp->roff); |
} |
} |
|
|
static void |
/* |
mparse_parse_buffer(struct mparse *curp, struct buf blk, const char *file) |
* Read the whole file into memory and call the parsers. |
|
* Called recursively when an .so request is encountered. |
|
*/ |
|
void |
|
mparse_readfd(struct mparse *curp, int fd, const char *filename) |
{ |
{ |
struct buf *svprimary; |
|
const char *svfile; |
|
size_t offset; |
|
static int recursion_depth; |
static int recursion_depth; |
|
|
if (64 < recursion_depth) { |
struct buf blk; |
mandoc_msg(MANDOCERR_ROFFLOOP, curp, curp->line, 0, NULL); |
struct buf *save_primary; |
|
const char *save_filename; |
|
size_t offset; |
|
int save_filenc, save_lineno; |
|
int with_mmap; |
|
|
|
if (recursion_depth > 64) { |
|
mandoc_msg(MANDOCERR_ROFFLOOP, curp->line, 0, NULL); |
return; |
return; |
} |
} |
|
if (read_whole_file(curp, fd, &blk, &with_mmap) == 0) |
|
return; |
|
|
/* Line number is per-file. */ |
/* |
svfile = mandoc_msg_getinfilename(); |
* Save some properties of the parent file. |
mandoc_msg_setinfilename(file); |
*/ |
svprimary = curp->primary; |
|
|
save_primary = curp->primary; |
|
save_filenc = curp->filenc; |
|
save_lineno = curp->line; |
|
save_filename = mandoc_msg_getinfilename(); |
|
|
curp->primary = &blk; |
curp->primary = &blk; |
|
curp->filenc = curp->options & (MPARSE_UTF8 | MPARSE_LATIN1); |
curp->line = 1; |
curp->line = 1; |
recursion_depth++; |
mandoc_msg_setinfilename(filename); |
|
|
/* Skip an UTF-8 byte order mark. */ |
/* Skip an UTF-8 byte order mark. */ |
if (curp->filenc & MPARSE_UTF8 && blk.sz > 2 && |
if (curp->filenc & MPARSE_UTF8 && blk.sz > 2 && |
Line 569 mparse_parse_buffer(struct mparse *curp, struct buf bl |
|
Line 582 mparse_parse_buffer(struct mparse *curp, struct buf bl |
|
} else |
} else |
offset = 0; |
offset = 0; |
|
|
|
recursion_depth++; |
mparse_buf_r(curp, blk, offset, 1); |
mparse_buf_r(curp, blk, offset, 1); |
|
|
if (--recursion_depth == 0) |
if (--recursion_depth == 0) |
mparse_end(curp); |
mparse_end(curp); |
|
|
curp->primary = svprimary; |
/* |
if (svfile != NULL) |
* Clean up and restore saved parent properties. |
mandoc_msg_setinfilename(svfile); |
*/ |
} |
|
|
|
/* |
if (with_mmap) |
* Read the whole file into memory and call the parsers. |
munmap(blk.buf, blk.sz); |
* Called recursively when an .so request is encountered. |
else |
*/ |
free(blk.buf); |
void |
|
mparse_readfd(struct mparse *curp, int fd, const char *file) |
|
{ |
|
struct buf blk; |
|
int with_mmap; |
|
int save_filenc; |
|
|
|
if (read_whole_file(curp, fd, &blk, &with_mmap)) { |
curp->primary = save_primary; |
save_filenc = curp->filenc; |
curp->filenc = save_filenc; |
curp->filenc = curp->options & |
curp->line = save_lineno; |
(MPARSE_UTF8 | MPARSE_LATIN1); |
if (save_filename != NULL) |
mparse_parse_buffer(curp, blk, file); |
mandoc_msg_setinfilename(save_filename); |
curp->filenc = save_filenc; |
|
if (with_mmap) |
|
munmap(blk.buf, blk.sz); |
|
else |
|
free(blk.buf); |
|
} |
|
} |
} |
|
|
int |
int |
Line 634 mparse_open(struct mparse *curp, const char *file) |
|
Line 634 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 648 mparse_alloc(int options, enum mandoc_os os_e, const c |
|
Line 647 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 672 mparse_reset(struct mparse *curp) |
|
Line 671 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 685 mparse_free(struct mparse *curp) |
|
Line 682 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) |
|
{ |
{ |
|
if (curp->options & MPARSE_VALIDATE) { |
if (sodest && NULL != (*sodest = curp->sodest)) { |
if (curp->man->meta.macroset == MACROSET_MDOC) |
*man = NULL; |
mdoc_validate(curp->man); |
return; |
else |
|
man_validate(curp->man); |
} |
} |
if (man) |
return &curp->man->meta; |
*man = curp->man; |
|
} |
} |
|
|
void |
void |