]> git.cameronkatri.com Git - mandoc.git/blob - read.c
a60f42f970776e1abe2593ed9553ce1ab7f02341
[mandoc.git] / read.c
1 /* $Id: read.c,v 1.57 2014/07/02 13:10:45 schwarze Exp $ */
2 /*
3 * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
4 * Copyright (c) 2010-2014 Ingo Schwarze <schwarze@openbsd.org>
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 AUTHOR DISCLAIMS ALL WARRANTIES
12 * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
13 * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR 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 #ifdef HAVE_CONFIG_H
20 #include "config.h"
21 #endif
22
23 #ifdef HAVE_MMAP
24 # include <sys/stat.h>
25 # include <sys/mman.h>
26 #endif
27
28 #include <assert.h>
29 #include <ctype.h>
30 #include <errno.h>
31 #include <fcntl.h>
32 #include <stdarg.h>
33 #include <stdint.h>
34 #include <stdio.h>
35 #include <stdlib.h>
36 #include <string.h>
37 #include <unistd.h>
38
39 #include "mandoc.h"
40 #include "mandoc_aux.h"
41 #include "libmandoc.h"
42 #include "mdoc.h"
43 #include "man.h"
44 #include "main.h"
45
46 #define REPARSE_LIMIT 1000
47
48 struct buf {
49 char *buf; /* binary input buffer */
50 size_t sz; /* size of binary buffer */
51 };
52
53 struct mparse {
54 enum mandoclevel file_status; /* status of current parse */
55 enum mandoclevel wlevel; /* ignore messages below this */
56 int line; /* line number in the file */
57 int options; /* parser options */
58 struct man *pman; /* persistent man parser */
59 struct mdoc *pmdoc; /* persistent mdoc parser */
60 struct man *man; /* man parser */
61 struct mdoc *mdoc; /* mdoc parser */
62 struct roff *roff; /* roff parser (!NULL) */
63 char *sodest; /* filename pointed to by .so */
64 int reparse_count; /* finite interp. stack */
65 mandocmsg mmsg; /* warning/error message handler */
66 const char *file;
67 struct buf *secondary;
68 char *defos; /* default operating system */
69 };
70
71 static void resize_buf(struct buf *, size_t);
72 static void mparse_buf_r(struct mparse *, struct buf, int);
73 static void pset(const char *, int, struct mparse *);
74 static int read_whole_file(struct mparse *, const char *, int,
75 struct buf *, int *);
76 static void mparse_end(struct mparse *);
77 static void mparse_parse_buffer(struct mparse *, struct buf,
78 const char *);
79
80 static const enum mandocerr mandoclimits[MANDOCLEVEL_MAX] = {
81 MANDOCERR_OK,
82 MANDOCERR_WARNING,
83 MANDOCERR_WARNING,
84 MANDOCERR_ERROR,
85 MANDOCERR_FATAL,
86 MANDOCERR_MAX,
87 MANDOCERR_MAX
88 };
89
90 static const char * const mandocerrs[MANDOCERR_MAX] = {
91 "ok",
92
93 "generic warning",
94
95 /* related to the prologue */
96 "missing .TH macro, using \"unknown 1\"",
97 "lower case character in document title",
98 "unknown manual section",
99 "unknown manual volume or arch",
100 "missing date, using today's date",
101 "cannot parse date, using it verbatim",
102 "prologue macros out of order",
103 "duplicate prologue macro",
104 "incomplete prologue, terminated by",
105 "skipping prologue macro in body",
106
107 /* related to document structure */
108 ".so is fragile, better use ln(1)",
109 "no document body",
110 "content before first section header",
111 "first section is not \"NAME\"",
112 "bad NAME section contents",
113 "sections out of conventional order",
114 "duplicate section title",
115 "unexpected section",
116
117 /* related to macros and nesting */
118 "obsolete macro",
119 "skipping paragraph macro",
120 "moving paragraph macro out of list",
121 "skipping no-space macro",
122 "blocks badly nested",
123 "nested displays are not portable",
124 "moving content out of list",
125 ".Vt block has child macro",
126 "fill mode already enabled, skipping .fi",
127 "fill mode already disabled, skipping .nf",
128 "line scope broken",
129
130 /* related to missing macro arguments */
131 "skipping empty macro",
132 "argument count wrong",
133 "missing display type",
134 "list type must come first",
135 "tag lists require a width argument",
136 "missing font type",
137
138 /* related to bad macro arguments */
139 "skipping argument",
140 "duplicate argument",
141 "duplicate display type",
142 "duplicate list type",
143 "unknown AT&T UNIX version",
144 "bad Boolean value",
145 "unknown font",
146 "unknown standard specifier",
147 "bad width argument",
148
149 /* related to plain text */
150 "blank line in non-literal context",
151 "tab in non-literal context",
152 "end of line whitespace",
153 "bad comment style",
154 "bad escape sequence",
155 "unterminated quoted string",
156
157 "generic error",
158
159 /* related to equations */
160 "unexpected equation scope closure",
161 "equation scope open on exit",
162 "overlapping equation scopes",
163 "unexpected end of equation",
164 "equation syntax error",
165
166 /* related to tables */
167 "bad table syntax",
168 "bad table option",
169 "bad table layout",
170 "no table layout cells specified",
171 "no table data cells specified",
172 "ignore data in cell",
173 "data block still open",
174 "ignoring extra data cells",
175
176 "input stack limit exceeded, infinite loop?",
177 "skipping bad character",
178 "escaped character not allowed in a name",
179 "manual name not yet set",
180 "skipping text before first section header",
181 "skipping unknown macro",
182 "NOT IMPLEMENTED, please use groff: skipping request",
183 "argument count wrong",
184 "skipping invalid content in .Rs block",
185 "skipping column outside column list",
186 "skipping end of block that is not open",
187 "missing end of block",
188 "scope open on exit",
189 "uname(3) system call failed",
190 "macro requires line argument(s)",
191 "macro requires body argument(s)",
192 "macro requires argument(s)",
193 "request requires a numeric argument",
194 "missing list type",
195 "line argument(s) will be lost",
196
197 "generic fatal error",
198
199 "input too large",
200 "not a manual",
201 "column syntax is inconsistent",
202 "NOT IMPLEMENTED: .Bd -file",
203 "argument count wrong, violates syntax",
204 "child violates parent syntax",
205 "argument count wrong, violates syntax",
206 "NOT IMPLEMENTED: .so with absolute path or \"..\"",
207 ".so request failed",
208 "no document prologue",
209 "static buffer exhausted",
210
211 /* system errors */
212 NULL,
213 "cannot stat file",
214 "cannot read file",
215 };
216
217 static const char * const mandoclevels[MANDOCLEVEL_MAX] = {
218 "SUCCESS",
219 "RESERVED",
220 "WARNING",
221 "ERROR",
222 "FATAL",
223 "BADARG",
224 "SYSERR"
225 };
226
227
228 static void
229 resize_buf(struct buf *buf, size_t initial)
230 {
231
232 buf->sz = buf->sz > initial/2 ? 2 * buf->sz : initial;
233 buf->buf = mandoc_realloc(buf->buf, buf->sz);
234 }
235
236 static void
237 pset(const char *buf, int pos, struct mparse *curp)
238 {
239 int i;
240
241 /*
242 * Try to intuit which kind of manual parser should be used. If
243 * passed in by command-line (-man, -mdoc), then use that
244 * explicitly. If passed as -mandoc, then try to guess from the
245 * line: either skip dot-lines, use -mdoc when finding `.Dt', or
246 * default to -man, which is more lenient.
247 *
248 * Separate out pmdoc/pman from mdoc/man: the first persists
249 * through all parsers, while the latter is used per-parse.
250 */
251
252 if ('.' == buf[0] || '\'' == buf[0]) {
253 for (i = 1; buf[i]; i++)
254 if (' ' != buf[i] && '\t' != buf[i])
255 break;
256 if ('\0' == buf[i])
257 return;
258 }
259
260 if (MPARSE_MDOC & curp->options) {
261 if (NULL == curp->pmdoc)
262 curp->pmdoc = mdoc_alloc(
263 curp->roff, curp, curp->defos,
264 MPARSE_QUICK & curp->options ? 1 : 0);
265 assert(curp->pmdoc);
266 curp->mdoc = curp->pmdoc;
267 return;
268 } else if (MPARSE_MAN & curp->options) {
269 if (NULL == curp->pman)
270 curp->pman = man_alloc(curp->roff, curp,
271 MPARSE_QUICK & curp->options ? 1 : 0);
272 assert(curp->pman);
273 curp->man = curp->pman;
274 return;
275 }
276
277 if (pos >= 3 && 0 == memcmp(buf, ".Dd", 3)) {
278 if (NULL == curp->pmdoc)
279 curp->pmdoc = mdoc_alloc(
280 curp->roff, curp, curp->defos,
281 MPARSE_QUICK & curp->options ? 1 : 0);
282 assert(curp->pmdoc);
283 curp->mdoc = curp->pmdoc;
284 return;
285 }
286
287 if (NULL == curp->pman)
288 curp->pman = man_alloc(curp->roff, curp,
289 MPARSE_QUICK & curp->options ? 1 : 0);
290 assert(curp->pman);
291 curp->man = curp->pman;
292 }
293
294 /*
295 * Main parse routine for an opened file. This is called for each
296 * opened file and simply loops around the full input file, possibly
297 * nesting (i.e., with `so').
298 */
299 static void
300 mparse_buf_r(struct mparse *curp, struct buf blk, int start)
301 {
302 const struct tbl_span *span;
303 struct buf ln;
304 enum rofferr rr;
305 int i, of, rc;
306 int pos; /* byte number in the ln buffer */
307 int lnn; /* line number in the real file */
308 unsigned char c;
309
310 memset(&ln, 0, sizeof(struct buf));
311
312 lnn = curp->line;
313 pos = 0;
314
315 for (i = 0; i < (int)blk.sz; ) {
316 if (0 == pos && '\0' == blk.buf[i])
317 break;
318
319 if (start) {
320 curp->line = lnn;
321 curp->reparse_count = 0;
322 }
323
324 while (i < (int)blk.sz && (start || '\0' != blk.buf[i])) {
325
326 /*
327 * When finding an unescaped newline character,
328 * leave the character loop to process the line.
329 * Skip a preceding carriage return, if any.
330 */
331
332 if ('\r' == blk.buf[i] && i + 1 < (int)blk.sz &&
333 '\n' == blk.buf[i + 1])
334 ++i;
335 if ('\n' == blk.buf[i]) {
336 ++i;
337 ++lnn;
338 break;
339 }
340
341 /*
342 * Make sure we have space for at least
343 * one backslash and one other character
344 * and the trailing NUL byte.
345 */
346
347 if (pos + 2 >= (int)ln.sz)
348 resize_buf(&ln, 256);
349
350 /*
351 * Warn about bogus characters. If you're using
352 * non-ASCII encoding, you're screwing your
353 * readers. Since I'd rather this not happen,
354 * I'll be helpful and replace these characters
355 * with "?", so we don't display gibberish.
356 * Note to manual writers: use special characters.
357 */
358
359 c = (unsigned char) blk.buf[i];
360
361 if ( ! (isascii(c) &&
362 (isgraph(c) || isblank(c)))) {
363 mandoc_msg(MANDOCERR_BADCHAR, curp,
364 curp->line, pos, NULL);
365 i++;
366 ln.buf[pos++] = '?';
367 continue;
368 }
369
370 /* Trailing backslash = a plain char. */
371
372 if ('\\' != blk.buf[i] || i + 1 == (int)blk.sz) {
373 ln.buf[pos++] = blk.buf[i++];
374 continue;
375 }
376
377 /*
378 * Found escape and at least one other character.
379 * When it's a newline character, skip it.
380 * When there is a carriage return in between,
381 * skip that one as well.
382 */
383
384 if ('\r' == blk.buf[i + 1] && i + 2 < (int)blk.sz &&
385 '\n' == blk.buf[i + 2])
386 ++i;
387 if ('\n' == blk.buf[i + 1]) {
388 i += 2;
389 ++lnn;
390 continue;
391 }
392
393 if ('"' == blk.buf[i + 1] || '#' == blk.buf[i + 1]) {
394 i += 2;
395 /* Comment, skip to end of line */
396 for (; i < (int)blk.sz; ++i) {
397 if ('\n' == blk.buf[i]) {
398 ++i;
399 ++lnn;
400 break;
401 }
402 }
403
404 /* Backout trailing whitespaces */
405 for (; pos > 0; --pos) {
406 if (ln.buf[pos - 1] != ' ')
407 break;
408 if (pos > 2 && ln.buf[pos - 2] == '\\')
409 break;
410 }
411 break;
412 }
413
414 /* Catch escaped bogus characters. */
415
416 c = (unsigned char) blk.buf[i+1];
417
418 if ( ! (isascii(c) &&
419 (isgraph(c) || isblank(c)))) {
420 mandoc_msg(MANDOCERR_BADCHAR, curp,
421 curp->line, pos, NULL);
422 i += 2;
423 ln.buf[pos++] = '?';
424 continue;
425 }
426
427 /* Some other escape sequence, copy & cont. */
428
429 ln.buf[pos++] = blk.buf[i++];
430 ln.buf[pos++] = blk.buf[i++];
431 }
432
433 if (pos >= (int)ln.sz)
434 resize_buf(&ln, 256);
435
436 ln.buf[pos] = '\0';
437
438 /*
439 * A significant amount of complexity is contained by
440 * the roff preprocessor. It's line-oriented but can be
441 * expressed on one line, so we need at times to
442 * readjust our starting point and re-run it. The roff
443 * preprocessor can also readjust the buffers with new
444 * data, so we pass them in wholesale.
445 */
446
447 of = 0;
448
449 /*
450 * Maintain a lookaside buffer of all parsed lines. We
451 * only do this if mparse_keep() has been invoked (the
452 * buffer may be accessed with mparse_getkeep()).
453 */
454
455 if (curp->secondary) {
456 curp->secondary->buf = mandoc_realloc(
457 curp->secondary->buf,
458 curp->secondary->sz + pos + 2);
459 memcpy(curp->secondary->buf +
460 curp->secondary->sz,
461 ln.buf, pos);
462 curp->secondary->sz += pos;
463 curp->secondary->buf
464 [curp->secondary->sz] = '\n';
465 curp->secondary->sz++;
466 curp->secondary->buf
467 [curp->secondary->sz] = '\0';
468 }
469 rerun:
470 rr = roff_parseln(curp->roff, curp->line,
471 &ln.buf, &ln.sz, of, &of);
472
473 switch (rr) {
474 case ROFF_REPARSE:
475 if (REPARSE_LIMIT >= ++curp->reparse_count)
476 mparse_buf_r(curp, ln, 0);
477 else
478 mandoc_msg(MANDOCERR_ROFFLOOP, curp,
479 curp->line, pos, NULL);
480 pos = 0;
481 continue;
482 case ROFF_APPEND:
483 pos = (int)strlen(ln.buf);
484 continue;
485 case ROFF_RERUN:
486 goto rerun;
487 case ROFF_IGN:
488 pos = 0;
489 continue;
490 case ROFF_ERR:
491 assert(MANDOCLEVEL_FATAL <= curp->file_status);
492 break;
493 case ROFF_SO:
494 if (0 == (MPARSE_SO & curp->options) &&
495 (i >= (int)blk.sz || '\0' == blk.buf[i])) {
496 curp->sodest = mandoc_strdup(ln.buf + of);
497 free(ln.buf);
498 return;
499 }
500 /*
501 * We remove `so' clauses from our lookaside
502 * buffer because we're going to descend into
503 * the file recursively.
504 */
505 if (curp->secondary)
506 curp->secondary->sz -= pos + 1;
507 mparse_readfd(curp, -1, ln.buf + of);
508 if (MANDOCLEVEL_FATAL <= curp->file_status) {
509 mandoc_vmsg(MANDOCERR_SO_FAIL,
510 curp, curp->line, pos,
511 ".so %s", ln.buf + of);
512 break;
513 }
514 pos = 0;
515 continue;
516 default:
517 break;
518 }
519
520 /*
521 * If we encounter errors in the recursive parse, make
522 * sure we don't continue parsing.
523 */
524
525 if (MANDOCLEVEL_FATAL <= curp->file_status)
526 break;
527
528 /*
529 * If input parsers have not been allocated, do so now.
530 * We keep these instanced between parsers, but set them
531 * locally per parse routine since we can use different
532 * parsers with each one.
533 */
534
535 if ( ! (curp->man || curp->mdoc))
536 pset(ln.buf + of, pos - of, curp);
537
538 /*
539 * Lastly, push down into the parsers themselves. One
540 * of these will have already been set in the pset()
541 * routine.
542 * If libroff returns ROFF_TBL, then add it to the
543 * currently open parse. Since we only get here if
544 * there does exist data (see tbl_data.c), we're
545 * guaranteed that something's been allocated.
546 * Do the same for ROFF_EQN.
547 */
548
549 rc = -1;
550
551 if (ROFF_TBL == rr)
552 while (NULL != (span = roff_span(curp->roff))) {
553 rc = curp->man ?
554 man_addspan(curp->man, span) :
555 mdoc_addspan(curp->mdoc, span);
556 if (0 == rc)
557 break;
558 }
559 else if (ROFF_EQN == rr)
560 rc = curp->mdoc ?
561 mdoc_addeqn(curp->mdoc,
562 roff_eqn(curp->roff)) :
563 man_addeqn(curp->man,
564 roff_eqn(curp->roff));
565 else if (curp->man || curp->mdoc)
566 rc = curp->man ?
567 man_parseln(curp->man,
568 curp->line, ln.buf, of) :
569 mdoc_parseln(curp->mdoc,
570 curp->line, ln.buf, of);
571
572 if (0 == rc) {
573 assert(MANDOCLEVEL_FATAL <= curp->file_status);
574 break;
575 } else if (2 == rc)
576 break;
577
578 /* Temporary buffers typically are not full. */
579
580 if (0 == start && '\0' == blk.buf[i])
581 break;
582
583 /* Start the next input line. */
584
585 pos = 0;
586 }
587
588 free(ln.buf);
589 }
590
591 static int
592 read_whole_file(struct mparse *curp, const char *file, int fd,
593 struct buf *fb, int *with_mmap)
594 {
595 size_t off;
596 ssize_t ssz;
597
598 #ifdef HAVE_MMAP
599 struct stat st;
600 if (-1 == fstat(fd, &st)) {
601 curp->file_status = MANDOCLEVEL_SYSERR;
602 if (curp->mmsg)
603 (*curp->mmsg)(MANDOCERR_SYSSTAT, curp->file_status,
604 file, 0, 0, strerror(errno));
605 return(0);
606 }
607
608 /*
609 * If we're a regular file, try just reading in the whole entry
610 * via mmap(). This is faster than reading it into blocks, and
611 * since each file is only a few bytes to begin with, I'm not
612 * concerned that this is going to tank any machines.
613 */
614
615 if (S_ISREG(st.st_mode)) {
616 if (st.st_size >= (1U << 31)) {
617 curp->file_status = MANDOCLEVEL_FATAL;
618 if (curp->mmsg)
619 (*curp->mmsg)(MANDOCERR_TOOLARGE,
620 curp->file_status, file, 0, 0, NULL);
621 return(0);
622 }
623 *with_mmap = 1;
624 fb->sz = (size_t)st.st_size;
625 fb->buf = mmap(NULL, fb->sz, PROT_READ, MAP_SHARED, fd, 0);
626 if (fb->buf != MAP_FAILED)
627 return(1);
628 }
629 #endif
630
631 /*
632 * If this isn't a regular file (like, say, stdin), then we must
633 * go the old way and just read things in bit by bit.
634 */
635
636 *with_mmap = 0;
637 off = 0;
638 fb->sz = 0;
639 fb->buf = NULL;
640 for (;;) {
641 if (off == fb->sz) {
642 if (fb->sz == (1U << 31)) {
643 curp->file_status = MANDOCLEVEL_FATAL;
644 if (curp->mmsg)
645 (*curp->mmsg)(MANDOCERR_TOOLARGE,
646 curp->file_status,
647 file, 0, 0, NULL);
648 break;
649 }
650 resize_buf(fb, 65536);
651 }
652 ssz = read(fd, fb->buf + (int)off, fb->sz - off);
653 if (ssz == 0) {
654 fb->sz = off;
655 return(1);
656 }
657 if (ssz == -1) {
658 curp->file_status = MANDOCLEVEL_SYSERR;
659 if (curp->mmsg)
660 (*curp->mmsg)(MANDOCERR_SYSREAD,
661 curp->file_status, file, 0, 0,
662 strerror(errno));
663 break;
664 }
665 off += (size_t)ssz;
666 }
667
668 free(fb->buf);
669 fb->buf = NULL;
670 return(0);
671 }
672
673 static void
674 mparse_end(struct mparse *curp)
675 {
676
677 if (MANDOCLEVEL_FATAL <= curp->file_status)
678 return;
679
680 if (curp->mdoc && ! mdoc_endparse(curp->mdoc)) {
681 assert(MANDOCLEVEL_FATAL <= curp->file_status);
682 return;
683 }
684
685 if (curp->man && ! man_endparse(curp->man)) {
686 assert(MANDOCLEVEL_FATAL <= curp->file_status);
687 return;
688 }
689
690 if ( ! (curp->mdoc || curp->man || curp->sodest)) {
691 mandoc_msg(MANDOCERR_NOTMANUAL, curp, 0, 0, NULL);
692 curp->file_status = MANDOCLEVEL_FATAL;
693 return;
694 }
695
696 roff_endparse(curp->roff);
697 }
698
699 static void
700 mparse_parse_buffer(struct mparse *curp, struct buf blk, const char *file)
701 {
702 const char *svfile;
703 static int recursion_depth;
704
705 if (64 < recursion_depth) {
706 mandoc_msg(MANDOCERR_ROFFLOOP, curp, curp->line, 0, NULL);
707 return;
708 }
709
710 /* Line number is per-file. */
711 svfile = curp->file;
712 curp->file = file;
713 curp->line = 1;
714 recursion_depth++;
715
716 mparse_buf_r(curp, blk, 1);
717
718 if (0 == --recursion_depth && MANDOCLEVEL_FATAL > curp->file_status)
719 mparse_end(curp);
720
721 curp->file = svfile;
722 }
723
724 enum mandoclevel
725 mparse_readmem(struct mparse *curp, const void *buf, size_t len,
726 const char *file)
727 {
728 struct buf blk;
729
730 blk.buf = UNCONST(buf);
731 blk.sz = len;
732
733 mparse_parse_buffer(curp, blk, file);
734 return(curp->file_status);
735 }
736
737 enum mandoclevel
738 mparse_readfd(struct mparse *curp, int fd, const char *file)
739 {
740 struct buf blk;
741 int with_mmap;
742
743 if (-1 == fd && -1 == (fd = open(file, O_RDONLY, 0))) {
744 curp->file_status = MANDOCLEVEL_SYSERR;
745 if (curp->mmsg)
746 (*curp->mmsg)(MANDOCERR_SYSOPEN,
747 curp->file_status,
748 file, 0, 0, strerror(errno));
749 goto out;
750 }
751
752 /*
753 * Run for each opened file; may be called more than once for
754 * each full parse sequence if the opened file is nested (i.e.,
755 * from `so'). Simply sucks in the whole file and moves into
756 * the parse phase for the file.
757 */
758
759 if ( ! read_whole_file(curp, file, fd, &blk, &with_mmap))
760 goto out;
761
762 mparse_parse_buffer(curp, blk, file);
763
764 #ifdef HAVE_MMAP
765 if (with_mmap)
766 munmap(blk.buf, blk.sz);
767 else
768 #endif
769 free(blk.buf);
770
771 if (STDIN_FILENO != fd && -1 == close(fd))
772 perror(file);
773 out:
774 return(curp->file_status);
775 }
776
777 struct mparse *
778 mparse_alloc(int options, enum mandoclevel wlevel,
779 mandocmsg mmsg, char *defos)
780 {
781 struct mparse *curp;
782
783 assert(wlevel <= MANDOCLEVEL_FATAL);
784
785 curp = mandoc_calloc(1, sizeof(struct mparse));
786
787 curp->options = options;
788 curp->wlevel = wlevel;
789 curp->mmsg = mmsg;
790 curp->defos = defos;
791
792 curp->roff = roff_alloc(curp, options);
793 return(curp);
794 }
795
796 void
797 mparse_reset(struct mparse *curp)
798 {
799
800 roff_reset(curp->roff);
801
802 if (curp->mdoc)
803 mdoc_reset(curp->mdoc);
804 if (curp->man)
805 man_reset(curp->man);
806 if (curp->secondary)
807 curp->secondary->sz = 0;
808
809 curp->file_status = MANDOCLEVEL_OK;
810 curp->mdoc = NULL;
811 curp->man = NULL;
812
813 free(curp->sodest);
814 curp->sodest = NULL;
815 }
816
817 void
818 mparse_free(struct mparse *curp)
819 {
820
821 if (curp->pmdoc)
822 mdoc_free(curp->pmdoc);
823 if (curp->pman)
824 man_free(curp->pman);
825 if (curp->roff)
826 roff_free(curp->roff);
827 if (curp->secondary)
828 free(curp->secondary->buf);
829
830 free(curp->secondary);
831 free(curp->sodest);
832 free(curp);
833 }
834
835 void
836 mparse_result(struct mparse *curp,
837 struct mdoc **mdoc, struct man **man, char **sodest)
838 {
839
840 if (sodest && NULL != (*sodest = curp->sodest)) {
841 *mdoc = NULL;
842 *man = NULL;
843 return;
844 }
845 if (mdoc)
846 *mdoc = curp->mdoc;
847 if (man)
848 *man = curp->man;
849 }
850
851 void
852 mandoc_vmsg(enum mandocerr t, struct mparse *m,
853 int ln, int pos, const char *fmt, ...)
854 {
855 char buf[256];
856 va_list ap;
857
858 va_start(ap, fmt);
859 (void)vsnprintf(buf, sizeof(buf), fmt, ap);
860 va_end(ap);
861
862 mandoc_msg(t, m, ln, pos, buf);
863 }
864
865 void
866 mandoc_msg(enum mandocerr er, struct mparse *m,
867 int ln, int col, const char *msg)
868 {
869 enum mandoclevel level;
870
871 level = MANDOCLEVEL_FATAL;
872 while (er < mandoclimits[level])
873 level--;
874
875 if (level < m->wlevel)
876 return;
877
878 if (m->mmsg)
879 (*m->mmsg)(er, level, m->file, ln, col, msg);
880
881 if (m->file_status < level)
882 m->file_status = level;
883 }
884
885 const char *
886 mparse_strerror(enum mandocerr er)
887 {
888
889 return(mandocerrs[er]);
890 }
891
892 const char *
893 mparse_strlevel(enum mandoclevel lvl)
894 {
895 return(mandoclevels[lvl]);
896 }
897
898 void
899 mparse_keep(struct mparse *p)
900 {
901
902 assert(NULL == p->secondary);
903 p->secondary = mandoc_calloc(1, sizeof(struct buf));
904 }
905
906 const char *
907 mparse_getkeep(const struct mparse *p)
908 {
909
910 assert(p->secondary);
911 return(p->secondary->sz ? p->secondary->buf : NULL);
912 }