xref: /openbsd-src/usr.bin/mandoc/read.c (revision 07aa05098856a503e2d4cff5708794aef9318348)
1 /* $OpenBSD: read.c,v 1.192 2022/05/19 14:47:47 schwarze Exp $ */
2 /*
3  * Copyright (c) 2010-2020 Ingo Schwarze <schwarze@openbsd.org>
4  * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
5  * Copyright (c) 2010, 2012 Joerg Sonnenberger <joerg@netbsd.org>
6  *
7  * Permission to use, copy, modify, and distribute this software for any
8  * purpose with or without fee is hereby granted, provided that the above
9  * copyright notice and this permission notice appear in all copies.
10  *
11  * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHORS DISCLAIM ALL WARRANTIES
12  * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
13  * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHORS BE LIABLE FOR
14  * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
15  * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
16  * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
17  * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
18  *
19  * Top-level functions of the mandoc(3) parser:
20  * Parser and input encoding selection, decompression,
21  * handling of input bytes, characters, lines, and files,
22  * handling of roff(7) loops and file inclusion,
23  * and steering of the various parsers.
24  */
25 #include <sys/types.h>
26 #include <sys/mman.h>
27 #include <sys/stat.h>
28 
29 #include <assert.h>
30 #include <ctype.h>
31 #include <errno.h>
32 #include <fcntl.h>
33 #include <stdarg.h>
34 #include <stdio.h>
35 #include <stdlib.h>
36 #include <string.h>
37 #include <unistd.h>
38 #include <zlib.h>
39 
40 #include "mandoc_aux.h"
41 #include "mandoc.h"
42 #include "roff.h"
43 #include "mdoc.h"
44 #include "man.h"
45 #include "mandoc_parse.h"
46 #include "libmandoc.h"
47 #include "roff_int.h"
48 #include "tag.h"
49 
50 #define	REPARSE_LIMIT	1000
51 
52 struct	mparse {
53 	struct roff	 *roff; /* roff parser (!NULL) */
54 	struct roff_man	 *man; /* man parser */
55 	struct buf	 *primary; /* buffer currently being parsed */
56 	struct buf	 *secondary; /* copy of top level input */
57 	struct buf	 *loop; /* open .while request line */
58 	const char	 *os_s; /* default operating system */
59 	int		  options; /* parser options */
60 	int		  gzip; /* current input file is gzipped */
61 	int		  filenc; /* encoding of the current file */
62 	int		  reparse_count; /* finite interp. stack */
63 	int		  line; /* line number in the file */
64 };
65 
66 static	void	  choose_parser(struct mparse *);
67 static	void	  free_buf_list(struct buf *);
68 static	void	  resize_buf(struct buf *, size_t);
69 static	int	  mparse_buf_r(struct mparse *, struct buf, size_t, int);
70 static	int	  read_whole_file(struct mparse *, int, struct buf *, int *);
71 static	void	  mparse_end(struct mparse *);
72 
73 
74 static void
resize_buf(struct buf * buf,size_t initial)75 resize_buf(struct buf *buf, size_t initial)
76 {
77 
78 	buf->sz = buf->sz > initial/2 ? 2 * buf->sz : initial;
79 	buf->buf = mandoc_realloc(buf->buf, buf->sz);
80 }
81 
82 static void
free_buf_list(struct buf * buf)83 free_buf_list(struct buf *buf)
84 {
85 	struct buf *tmp;
86 
87 	while (buf != NULL) {
88 		tmp = buf;
89 		buf = tmp->next;
90 		free(tmp->buf);
91 		free(tmp);
92 	}
93 }
94 
95 static void
choose_parser(struct mparse * curp)96 choose_parser(struct mparse *curp)
97 {
98 	char		*cp, *ep;
99 	int		 format;
100 
101 	/*
102 	 * If neither command line arguments -mdoc or -man select
103 	 * a parser nor the roff parser found a .Dd or .TH macro
104 	 * yet, look ahead in the main input buffer.
105 	 */
106 
107 	if ((format = roff_getformat(curp->roff)) == 0) {
108 		cp = curp->primary->buf;
109 		ep = cp + curp->primary->sz;
110 		while (cp < ep) {
111 			if (*cp == '.' || *cp == '\'') {
112 				cp++;
113 				if (cp[0] == 'D' && cp[1] == 'd') {
114 					format = MPARSE_MDOC;
115 					break;
116 				}
117 				if (cp[0] == 'T' && cp[1] == 'H') {
118 					format = MPARSE_MAN;
119 					break;
120 				}
121 			}
122 			cp = memchr(cp, '\n', ep - cp);
123 			if (cp == NULL)
124 				break;
125 			cp++;
126 		}
127 	}
128 
129 	if (format == MPARSE_MDOC) {
130 		curp->man->meta.macroset = MACROSET_MDOC;
131 		if (curp->man->mdocmac == NULL)
132 			curp->man->mdocmac = roffhash_alloc(MDOC_Dd, MDOC_MAX);
133 	} else {
134 		curp->man->meta.macroset = MACROSET_MAN;
135 		if (curp->man->manmac == NULL)
136 			curp->man->manmac = roffhash_alloc(MAN_TH, MAN_MAX);
137 	}
138 	curp->man->meta.first->tok = TOKEN_NONE;
139 }
140 
141 /*
142  * Main parse routine for a buffer.
143  * It assumes encoding and line numbering are already set up.
144  * It can recurse directly (for invocations of user-defined
145  * macros, inline equations, and input line traps)
146  * and indirectly (for .so file inclusion).
147  */
148 static int
mparse_buf_r(struct mparse * curp,struct buf blk,size_t i,int start)149 mparse_buf_r(struct mparse *curp, struct buf blk, size_t i, int start)
150 {
151 	struct buf	 ln;
152 	struct buf	*firstln, *lastln, *thisln, *loop;
153 	char		*cp;
154 	size_t		 pos; /* byte number in the ln buffer */
155 	size_t		 spos; /* at the start of the current line parse */
156 	int		 line_result, result;
157 	int		 of;
158 	int		 lnn; /* line number in the real file */
159 	int		 fd;
160 	int		 inloop; /* Saw .while on this level. */
161 	unsigned char	 c;
162 
163 	ln.sz = 256;
164 	ln.buf = mandoc_malloc(ln.sz);
165 	ln.next = NULL;
166 	firstln = lastln = loop = NULL;
167 	lnn = curp->line;
168 	pos = 0;
169 	inloop = 0;
170 	result = ROFF_CONT;
171 
172 	while (i < blk.sz && (blk.buf[i] != '\0' || pos != 0)) {
173 		if (start) {
174 			curp->line = lnn;
175 			curp->reparse_count = 0;
176 
177 			if (lnn < 3 &&
178 			    curp->filenc & MPARSE_UTF8 &&
179 			    curp->filenc & MPARSE_LATIN1)
180 				curp->filenc = preconv_cue(&blk, i);
181 		}
182 		spos = pos;
183 
184 		while (i < blk.sz && (start || blk.buf[i] != '\0')) {
185 
186 			/*
187 			 * When finding an unescaped newline character,
188 			 * leave the character loop to process the line.
189 			 * Skip a preceding carriage return, if any.
190 			 */
191 
192 			if ('\r' == blk.buf[i] && i + 1 < blk.sz &&
193 			    '\n' == blk.buf[i + 1])
194 				++i;
195 			if ('\n' == blk.buf[i]) {
196 				++i;
197 				++lnn;
198 				break;
199 			}
200 
201 			/*
202 			 * Make sure we have space for the worst
203 			 * case of 12 bytes: "\\[u10ffff]\n\0"
204 			 */
205 
206 			if (pos + 12 > ln.sz)
207 				resize_buf(&ln, 256);
208 
209 			/*
210 			 * Encode 8-bit input.
211 			 */
212 
213 			c = blk.buf[i];
214 			if (c & 0x80) {
215 				if ( ! (curp->filenc && preconv_encode(
216 				    &blk, &i, &ln, &pos, &curp->filenc))) {
217 					mandoc_msg(MANDOCERR_CHAR_BAD,
218 					    curp->line, pos, "0x%x", c);
219 					ln.buf[pos++] = '?';
220 					i++;
221 				}
222 				continue;
223 			}
224 
225 			/*
226 			 * Exclude control characters.
227 			 */
228 
229 			if (c == 0x7f || (c < 0x20 && c != 0x09)) {
230 				mandoc_msg(c == 0x00 || c == 0x04 ||
231 				    c > 0x0a ? MANDOCERR_CHAR_BAD :
232 				    MANDOCERR_CHAR_UNSUPP,
233 				    curp->line, pos, "0x%x", c);
234 				i++;
235 				if (c != '\r')
236 					ln.buf[pos++] = '?';
237 				continue;
238 			}
239 
240 			ln.buf[pos++] = blk.buf[i++];
241 		}
242 		ln.buf[pos] = '\0';
243 
244 		/*
245 		 * Maintain a lookaside buffer of all lines.
246 		 * parsed from this input source.
247 		 */
248 
249 		thisln = mandoc_malloc(sizeof(*thisln));
250 		thisln->buf = mandoc_strdup(ln.buf);
251 		thisln->sz = strlen(ln.buf) + 1;
252 		thisln->next = NULL;
253 		if (firstln == NULL) {
254 			firstln = lastln = thisln;
255 			if (curp->secondary == NULL)
256 				curp->secondary = firstln;
257 		} else {
258 			lastln->next = thisln;
259 			lastln = thisln;
260 		}
261 
262 		/*
263 		 * XXX Ugly hack to mark the end of the input,
264 		 * such that the function roff_parse_comment()
265 		 * doesn't attempt to append another line if the
266 		 * last input line ends with an escape character.
267 		 */
268 
269 		if (i == blk.sz || blk.buf[i] == '\0') {
270 			if (pos + 2 > ln.sz)
271 				resize_buf(&ln, 256);
272 			ln.buf[pos++] = '\n';
273 			ln.buf[pos] = '\0';
274 		}
275 
276 		/*
277 		 * A significant amount of complexity is contained by
278 		 * the roff preprocessor.  It's line-oriented but can be
279 		 * expressed on one line, so we need at times to
280 		 * readjust our starting point and re-run it.  The roff
281 		 * preprocessor can also readjust the buffers with new
282 		 * data, so we pass them in wholesale.
283 		 */
284 
285 		of = 0;
286 rerun:
287 		line_result = roff_parseln(curp->roff, curp->line,
288 		    &ln, &of, start && spos == 0 ? pos : 0);
289 
290 		/* Process options. */
291 
292 		if (line_result & ROFF_APPEND)
293 			assert(line_result == (ROFF_IGN | ROFF_APPEND));
294 
295 		if (line_result & ROFF_USERCALL)
296 			assert((line_result & ROFF_MASK) == ROFF_REPARSE);
297 
298 		if (line_result & ROFF_USERRET) {
299 			assert(line_result == (ROFF_IGN | ROFF_USERRET));
300 			if (start == 0) {
301 				/* Return from the current macro. */
302 				result = ROFF_USERRET;
303 				goto out;
304 			}
305 		}
306 
307 		switch (line_result & ROFF_LOOPMASK) {
308 		case ROFF_IGN:
309 			break;
310 		case ROFF_WHILE:
311 			if (curp->loop != NULL) {
312 				if (loop == curp->loop)
313 					break;
314 				mandoc_msg(MANDOCERR_WHILE_NEST,
315 				    curp->line, pos, NULL);
316 			}
317 			curp->loop = thisln;
318 			loop = NULL;
319 			inloop = 1;
320 			break;
321 		case ROFF_LOOPCONT:
322 		case ROFF_LOOPEXIT:
323 			if (curp->loop == NULL) {
324 				mandoc_msg(MANDOCERR_WHILE_FAIL,
325 				    curp->line, pos, NULL);
326 				break;
327 			}
328 			if (inloop == 0) {
329 				mandoc_msg(MANDOCERR_WHILE_INTO,
330 				    curp->line, pos, NULL);
331 				curp->loop = loop = NULL;
332 				break;
333 			}
334 			if (line_result & ROFF_LOOPCONT)
335 				loop = curp->loop;
336 			else {
337 				curp->loop = loop = NULL;
338 				inloop = 0;
339 			}
340 			break;
341 		default:
342 			abort();
343 		}
344 
345 		/* Process the main instruction from the roff parser. */
346 
347 		switch (line_result & ROFF_MASK) {
348 		case ROFF_IGN:
349 			break;
350 		case ROFF_CONT:
351 			if (curp->man->meta.macroset == MACROSET_NONE)
352 				choose_parser(curp);
353 			if ((curp->man->meta.macroset == MACROSET_MDOC ?
354 			     mdoc_parseln(curp->man, curp->line, ln.buf, of) :
355 			     man_parseln(curp->man, curp->line, ln.buf, of)
356 			    ) == 2)
357 				goto out;
358 			break;
359 		case ROFF_RERUN:
360 			goto rerun;
361 		case ROFF_REPARSE:
362 			if (++curp->reparse_count > REPARSE_LIMIT) {
363 				/* Abort and return to the top level. */
364 				result = ROFF_IGN;
365 				mandoc_msg(MANDOCERR_ROFFLOOP,
366 				    curp->line, pos, NULL);
367 				goto out;
368 			}
369 			result = mparse_buf_r(curp, ln, of, 0);
370 			if (line_result & ROFF_USERCALL) {
371 				roff_userret(curp->roff);
372 				/* Continue normally. */
373 				if (result & ROFF_USERRET)
374 					result = ROFF_CONT;
375 			}
376 			if (start == 0 && result != ROFF_CONT)
377 				goto out;
378 			break;
379 		case ROFF_SO:
380 			if ( ! (curp->options & MPARSE_SO) &&
381 			    (i >= blk.sz || blk.buf[i] == '\0')) {
382 				curp->man->meta.sodest =
383 				    mandoc_strdup(ln.buf + of);
384 				goto out;
385 			}
386 			if ((fd = mparse_open(curp, ln.buf + of)) != -1) {
387 				mparse_readfd(curp, fd, ln.buf + of);
388 				close(fd);
389 			} else {
390 				mandoc_msg(MANDOCERR_SO_FAIL,
391 				    curp->line, of, ".so %s: %s",
392 				    ln.buf + of, strerror(errno));
393 				ln.sz = mandoc_asprintf(&cp,
394 				    ".sp\nSee the file %s.\n.sp",
395 				    ln.buf + of);
396 				free(ln.buf);
397 				ln.buf = cp;
398 				of = 0;
399 				mparse_buf_r(curp, ln, of, 0);
400 			}
401 			break;
402 		default:
403 			abort();
404 		}
405 
406 		/* Start the next input line. */
407 
408 		if (loop != NULL &&
409 		    (line_result & ROFF_LOOPMASK) == ROFF_IGN)
410 			loop = loop->next;
411 
412 		if (loop != NULL) {
413 			if ((line_result & ROFF_APPEND) == 0)
414 				*ln.buf = '\0';
415 			if (ln.sz < loop->sz)
416 				resize_buf(&ln, loop->sz);
417 			(void)strlcat(ln.buf, loop->buf, ln.sz);
418 			of = 0;
419 			goto rerun;
420 		}
421 
422 		pos = (line_result & ROFF_APPEND) ? strlen(ln.buf) : 0;
423 	}
424 out:
425 	if (inloop) {
426 		if (result != ROFF_USERRET)
427 			mandoc_msg(MANDOCERR_WHILE_OUTOF,
428 			    curp->line, pos, NULL);
429 		curp->loop = NULL;
430 	}
431 	free(ln.buf);
432 	if (firstln != curp->secondary)
433 		free_buf_list(firstln);
434 	return result;
435 }
436 
437 static int
read_whole_file(struct mparse * curp,int fd,struct buf * fb,int * with_mmap)438 read_whole_file(struct mparse *curp, int fd, struct buf *fb, int *with_mmap)
439 {
440 	struct stat	 st;
441 	gzFile		 gz;
442 	size_t		 off;
443 	ssize_t		 ssz;
444 	int		 gzerrnum, retval;
445 
446 	if (fstat(fd, &st) == -1) {
447 		mandoc_msg(MANDOCERR_FSTAT, 0, 0, "%s", strerror(errno));
448 		return -1;
449 	}
450 
451 	/*
452 	 * If we're a regular file, try just reading in the whole entry
453 	 * via mmap().  This is faster than reading it into blocks, and
454 	 * since each file is only a few bytes to begin with, I'm not
455 	 * concerned that this is going to tank any machines.
456 	 */
457 
458 	if (curp->gzip == 0 && S_ISREG(st.st_mode)) {
459 		if (st.st_size > 0x7fffffff) {
460 			mandoc_msg(MANDOCERR_TOOLARGE, 0, 0, NULL);
461 			return -1;
462 		}
463 		*with_mmap = 1;
464 		fb->sz = (size_t)st.st_size;
465 		fb->buf = mmap(NULL, fb->sz, PROT_READ, MAP_SHARED, fd, 0);
466 		if (fb->buf != MAP_FAILED)
467 			return 0;
468 	}
469 
470 	if (curp->gzip) {
471 		/*
472 		 * Duplicating the file descriptor is required
473 		 * because we will have to call gzclose(3)
474 		 * to free memory used internally by zlib,
475 		 * but that will also close the file descriptor,
476 		 * which this function must not do.
477 		 */
478 		if ((fd = dup(fd)) == -1) {
479 			mandoc_msg(MANDOCERR_DUP, 0, 0,
480 			    "%s", strerror(errno));
481 			return -1;
482 		}
483 		if ((gz = gzdopen(fd, "rb")) == NULL) {
484 			mandoc_msg(MANDOCERR_GZDOPEN, 0, 0,
485 			    "%s", strerror(errno));
486 			close(fd);
487 			return -1;
488 		}
489 	} else
490 		gz = NULL;
491 
492 	/*
493 	 * If this isn't a regular file (like, say, stdin), then we must
494 	 * go the old way and just read things in bit by bit.
495 	 */
496 
497 	*with_mmap = 0;
498 	off = 0;
499 	retval = -1;
500 	fb->sz = 0;
501 	fb->buf = NULL;
502 	for (;;) {
503 		if (off == fb->sz) {
504 			if (fb->sz == (1U << 31)) {
505 				mandoc_msg(MANDOCERR_TOOLARGE, 0, 0, NULL);
506 				break;
507 			}
508 			resize_buf(fb, 65536);
509 		}
510 		ssz = curp->gzip ?
511 		    gzread(gz, fb->buf + (int)off, fb->sz - off) :
512 		    read(fd, fb->buf + (int)off, fb->sz - off);
513 		if (ssz == 0) {
514 			fb->sz = off;
515 			retval = 0;
516 			break;
517 		}
518 		if (ssz == -1) {
519 			if (curp->gzip)
520 				(void)gzerror(gz, &gzerrnum);
521 			mandoc_msg(MANDOCERR_READ, 0, 0, "%s",
522 			    curp->gzip && gzerrnum != Z_ERRNO ?
523 			    zError(gzerrnum) : strerror(errno));
524 			break;
525 		}
526 		off += (size_t)ssz;
527 	}
528 
529 	if (curp->gzip && (gzerrnum = gzclose(gz)) != Z_OK)
530 		mandoc_msg(MANDOCERR_GZCLOSE, 0, 0, "%s",
531 		    gzerrnum == Z_ERRNO ? strerror(errno) :
532 		    zError(gzerrnum));
533 	if (retval == -1) {
534 		free(fb->buf);
535 		fb->buf = NULL;
536 	}
537 	return retval;
538 }
539 
540 static void
mparse_end(struct mparse * curp)541 mparse_end(struct mparse *curp)
542 {
543 	if (curp->man->meta.macroset == MACROSET_NONE)
544 		curp->man->meta.macroset = MACROSET_MAN;
545 	if (curp->man->meta.macroset == MACROSET_MDOC)
546 		mdoc_endparse(curp->man);
547 	else
548 		man_endparse(curp->man);
549 	roff_endparse(curp->roff);
550 }
551 
552 /*
553  * Read the whole file into memory and call the parsers.
554  * Called recursively when an .so request is encountered.
555  */
556 void
mparse_readfd(struct mparse * curp,int fd,const char * filename)557 mparse_readfd(struct mparse *curp, int fd, const char *filename)
558 {
559 	static int	 recursion_depth;
560 
561 	struct buf	 blk;
562 	struct buf	*save_primary;
563 	const char	*save_filename, *cp;
564 	size_t		 offset;
565 	int		 save_filenc, save_lineno;
566 	int		 with_mmap;
567 
568 	if (recursion_depth > 64) {
569 		mandoc_msg(MANDOCERR_ROFFLOOP, curp->line, 0, NULL);
570 		return;
571 	} else if (recursion_depth == 0 &&
572 	    (cp = strrchr(filename, '.')) != NULL &&
573             cp[1] >= '1' && cp[1] <= '9')
574                 curp->man->filesec = cp[1];
575         else
576                 curp->man->filesec = '\0';
577 
578 	if (read_whole_file(curp, fd, &blk, &with_mmap) == -1)
579 		return;
580 
581 	/*
582 	 * Save some properties of the parent file.
583 	 */
584 
585 	save_primary = curp->primary;
586 	save_filenc = curp->filenc;
587 	save_lineno = curp->line;
588 	save_filename = mandoc_msg_getinfilename();
589 
590 	curp->primary = &blk;
591 	curp->filenc = curp->options & (MPARSE_UTF8 | MPARSE_LATIN1);
592 	curp->line = 1;
593 	mandoc_msg_setinfilename(filename);
594 
595 	/* Skip an UTF-8 byte order mark. */
596 	if (curp->filenc & MPARSE_UTF8 && blk.sz > 2 &&
597 	    (unsigned char)blk.buf[0] == 0xef &&
598 	    (unsigned char)blk.buf[1] == 0xbb &&
599 	    (unsigned char)blk.buf[2] == 0xbf) {
600 		offset = 3;
601 		curp->filenc &= ~MPARSE_LATIN1;
602 	} else
603 		offset = 0;
604 
605 	recursion_depth++;
606 	mparse_buf_r(curp, blk, offset, 1);
607 	if (--recursion_depth == 0)
608 		mparse_end(curp);
609 
610 	/*
611 	 * Clean up and restore saved parent properties.
612 	 */
613 
614 	if (with_mmap)
615 		munmap(blk.buf, blk.sz);
616 	else
617 		free(blk.buf);
618 
619 	curp->primary = save_primary;
620 	curp->filenc = save_filenc;
621 	curp->line = save_lineno;
622 	if (save_filename != NULL)
623 		mandoc_msg_setinfilename(save_filename);
624 }
625 
626 int
mparse_open(struct mparse * curp,const char * file)627 mparse_open(struct mparse *curp, const char *file)
628 {
629 	char		 *cp;
630 	int		  fd, save_errno;
631 
632 	cp = strrchr(file, '.');
633 	curp->gzip = (cp != NULL && ! strcmp(cp + 1, "gz"));
634 
635 	/* First try to use the filename as it is. */
636 
637 	if ((fd = open(file, O_RDONLY)) != -1)
638 		return fd;
639 
640 	/*
641 	 * If that doesn't work and the filename doesn't
642 	 * already  end in .gz, try appending .gz.
643 	 */
644 
645 	if ( ! curp->gzip) {
646 		save_errno = errno;
647 		mandoc_asprintf(&cp, "%s.gz", file);
648 		fd = open(cp, O_RDONLY);
649 		free(cp);
650 		errno = save_errno;
651 		if (fd != -1) {
652 			curp->gzip = 1;
653 			return fd;
654 		}
655 	}
656 
657 	/* Neither worked, give up. */
658 
659 	return -1;
660 }
661 
662 struct mparse *
mparse_alloc(int options,enum mandoc_os os_e,const char * os_s)663 mparse_alloc(int options, enum mandoc_os os_e, const char *os_s)
664 {
665 	struct mparse	*curp;
666 
667 	curp = mandoc_calloc(1, sizeof(struct mparse));
668 
669 	curp->options = options;
670 	curp->os_s = os_s;
671 
672 	curp->roff = roff_alloc(options);
673 	curp->man = roff_man_alloc(curp->roff, curp->os_s,
674 		curp->options & MPARSE_QUICK ? 1 : 0);
675 	if (curp->options & MPARSE_MDOC) {
676 		curp->man->meta.macroset = MACROSET_MDOC;
677 		if (curp->man->mdocmac == NULL)
678 			curp->man->mdocmac = roffhash_alloc(MDOC_Dd, MDOC_MAX);
679 	} else if (curp->options & MPARSE_MAN) {
680 		curp->man->meta.macroset = MACROSET_MAN;
681 		if (curp->man->manmac == NULL)
682 			curp->man->manmac = roffhash_alloc(MAN_TH, MAN_MAX);
683 	}
684 	curp->man->meta.first->tok = TOKEN_NONE;
685 	curp->man->meta.os_e = os_e;
686 	tag_alloc();
687 	return curp;
688 }
689 
690 void
mparse_reset(struct mparse * curp)691 mparse_reset(struct mparse *curp)
692 {
693 	tag_free();
694 	roff_reset(curp->roff);
695 	roff_man_reset(curp->man);
696 	free_buf_list(curp->secondary);
697 	curp->secondary = NULL;
698 	curp->gzip = 0;
699 	tag_alloc();
700 }
701 
702 void
mparse_free(struct mparse * curp)703 mparse_free(struct mparse *curp)
704 {
705 	tag_free();
706 	roffhash_free(curp->man->mdocmac);
707 	roffhash_free(curp->man->manmac);
708 	roff_man_free(curp->man);
709 	roff_free(curp->roff);
710 	free_buf_list(curp->secondary);
711 	free(curp);
712 }
713 
714 struct roff_meta *
mparse_result(struct mparse * curp)715 mparse_result(struct mparse *curp)
716 {
717 	roff_state_reset(curp->man);
718 	if (curp->options & MPARSE_VALIDATE) {
719 		if (curp->man->meta.macroset == MACROSET_MDOC)
720 			mdoc_validate(curp->man);
721 		else
722 			man_validate(curp->man);
723 		tag_postprocess(curp->man, curp->man->meta.first);
724 	}
725 	return &curp->man->meta;
726 }
727 
728 void
mparse_copy(const struct mparse * p)729 mparse_copy(const struct mparse *p)
730 {
731 	struct buf	*buf;
732 
733 	for (buf = p->secondary; buf != NULL; buf = buf->next)
734 		puts(buf->buf);
735 }
736